Genotype-by-environment interactions and causal gene fine-mapping for quantitative trait variation in yeast A DISSERTATION SUBMITTED TO THE FACULTY OF THE UNIVERSITY OF MINNESOTA By Randi Rae Avery IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY Advised by Dr. Frank W. Albert December 2024 MOLECULAR, CELLULAR, DEVELOPMENTAL BIOLOGY & GENETICS GRADUATE PROGRAM © Randi R. Avery 2024 Acknowledgements First and foremost I must thank Dr. Frank Albert for his unending support over the last 6+ years. You have trained me to become a rigorous scientist with thick skin. You impressively balanced being tough on me with empathy to sharpen my skills without ever pushing me past my limits. You provided a safe space to work and grow, particularly through the multiple “unprecedented times” that occurred during my PhD. I really appreciate how you also balanced giving me ownership over my research while still being available for conversations and an infinite amount of feedback. But my favorite moments were probably the couple of times I made you laugh out loud :) To Dr. Mahlon Collins, I’m not sure I would be graduating (at least not yet) if it weren’t for you! Thank you so much for your mentorship during my rotation. It was the introduction I needed into my PhD and confirmed that the Albert Lab was where I should be. I especially thank you for providing the basis for my first first-author paper and the endless discussions, information, and feedback on the project. I am proud of the result of this work, and that is in huge part thanks to you. To the other current and former members of the Albert Lab, especially Dr. Sheila Lutz. You all have assisted me every step of the way, and helped me grow as a scientist. You all provided me with such great input at every lab meeting and round table and have greatly contributed to the advancement of my projects. To the current and former members of my thesis committee, Drs. Kate Adamala, Chad Myers, Nik Somia, Juan Carlos Rivera-Mulia, and Aaron Goldstrohm, thank you for your career advice and helping me see the forest through the trees at each committee meeting. I owe a special thanks to Dr. Nik Somia for the many letters of recommendation over the years! To my dear, dear friends I have made through this graduate program. I knew I would need a great support system during my PhD, but I did not realize my classmates would become my greatest confidants. I can honestly say that our friendships are the best thing that has come out of this degree. I will cherish our many memories (and tons of photos!) together forever. I could not have maintained my mental health over the last 6+ without you all. This acknowledgements section would not be complete without mentioning my mom, Dr. CJ Jundt. You started my career as a biological scientist at the ripe old age of two when you took me with you to your undergraduate biology classes when you couldn’t afford childcare as a single mom. Now look where we are. You showed me that it’s never too late to earn a doctorate and instilled the importance of education in me my entire life. Thank you for always being there for me and helping me with whatever I needed during my PhD, particularly with watching Dexter, that crazy dog! You helped decrease the stress of life in many ways so that I could focus on work. Thank you, I love you. And finally to my dog, Dexter. You honestly made this PhD way harder, but you won’t read this, so I can be frank. I do appreciate the occasional cuddles, cute pictures I get of you, and that you made sure I got outside multiple times a day during the pandemic. i Dedication I dedicate this dissertation to all of my future students. I pursued my PhD for you. The entire reason I applied to PhD programs was so that I could become a college professor. I cannot wait to impart every piece of wisdom and knowledge I have gained from this PhD that may help with your own growth, education, and career. Thinking of you has been what kept me going during the late nights and difficult moments. ii Table of Contents Acknowledgements i Dedication ii Declaration iv Chapter I. Introduction 1 Natural genetic variation and complex traits 1 Yeast as a tool 2 Quantitative trait loci 3 Genotype by environment interactions 5 Protein degradation is an essential cellular process 8 The UPS is genetically complex and influenced by GxE 10 Causal loci 11 References 15 Chapter II. Genotype by environment interactions shape ubiquitin-proteasome system activity 20 Abstract 20 Introduction 21 Results 27 Discussion 57 Methods 61 References 80 Chapter III. Substrate-Specific Effects of Natural Genetic Variation on Proteasome Activity 90 Abstract 90 Author Summary 91 Introduction 92 Results 98 Discussion 122 Materials and Methods 130 References 152 Chapter IV. Quantitative Trait Gene discovery by reciprocal hemizygote scanning 162 Introduction 162 Results 166 Discussion 183 Methods 186 References 195 Chapter V. Conclusion 198 References 204 iii Declaration Chapter II of this dissertation was adopted from work posted on bioRxiv: Avery, R. R., Collins, M. A., & Albert, F. W. (2024). Genotype-by-environment interactions shape ubiquitin-proteasome system activity. bioRxiv, 2024.11.21.624644. https://doi.org/10.1101/2024.11.21.624644 Work acknowledgements Chapter II was completed with significant contributions from Dr. Mahlon Collins and Dr. Frank Albert. Chapter III of this dissertation was adopted from published work: Collins, M. A., Avery, R. & Albert, F. W. Substrate-specific effects of natural genetic variation on proteasome activity. PLOS Genet. 19, e1010734 (2023). Work acknowledgements Chapter III was completed with significant contributions from Dr. Mahlon Collins. He, along with Dr. Frank Albert, conceptualized and designed this project and wrote the manuscript for publication. Dr. Collins performed the majority of lab work and data analysis. I was responsible for creating the Rpn4 reporter and validating its utility under Dr. Collins’ guidance. Chapter II stemmed from this work initiated by Dr. Collins. iv https://www.zotero.org/google-docs/?XnfB6Z https://www.zotero.org/google-docs/?XnfB6Z Chapter I. Introduction Randi R. Avery Department of Genetics, Cell Biology, & Development, University of Minnesota, Minneapolis, MN, USA Natural genetic variation and complex traits Natural genetic variation leads to phenotypic differences among individuals. Gregor Mendel laid the groundwork for understanding how alleles lead to phenotypes and astutely chose to study phenotypes that were caused by one gene (Miko, 2008). However, we now know that most traits are influenced by multiple genes (Yeh et al., 2022). When a trait is affected by more than one gene, this is known as a “polygenic trait.” In fact, as the field of genetics has evolved, it is clear that most traits have a high degree of polygenicity (Fisher, 1918; Serpico et al., 2023). Polygenic traits are ubiquitous: from common human diseases, such as heart disease (Plomin et al., 2009), to gene expression levels in yeast (Albert et al., 2018; Brem et al., 2002; Brem & Kruglyak, 2005). Taking this a step further, the omnigenic model (Boyle et al., 2017) has been proposed to explain that variation throughout the genome, including genes that are functionally distant to the measured trait, may all influence, at least to some extent, the phenotype. 1 https://www.zotero.org/google-docs/?tawQ6t https://www.zotero.org/google-docs/?YY8bhK https://www.zotero.org/google-docs/?WcQYMY https://www.zotero.org/google-docs/?WcQYMY https://www.zotero.org/google-docs/?RYarlK https://www.zotero.org/google-docs/?C0aHtp https://www.zotero.org/google-docs/?EFj8ic One of the outstanding objectives in the field of genetics and genomics is to work toward being able to predict phenotype from genotype (Yeh et al., 2022). A major component of determining the genotype to phenotype map is to ascertain the genetic architecture of complex traits, traits affected by multiple genes and / or various environments. We must elucidate how many loci contain variation between individuals that contributes to differences in traits, what that variation is (e.g. a given single nucleotide or copy number variant), where that variation is located in the genome, and to what magnitude that variation affects the trait. For example, are there only a couple of loci with large effect on a given phenotype with many other loci having a relatively small contribution? Or is it that many loci exhibit similar effect sizes that contribute to the trait? It is clear that it is both and that ‘it depends’. Various traits measured, set of individuals compared, and environmental contexts considered have each revealed unique genetic architectures (relevant review: Boye et al., 2024). Yeast as a tool Genetic effects are numerous and tend to have small effect sizes (Albert et al., 2018; Albert & Kruglyak, 2015). Therefore, large sample sizes are needed to achieve statistical power. The yeast Saccharomyces cerevisiae can be used as a powerful tool to study genetic variation. Over a thousand strains of S. cerevisiae have been extensively studied, providing a toolkit of over 1.6 million SNPs that can be used to investigate naturally-occurring genetic variation, which reflects adaptation and evolutionary history (Peter et al., 2018). In particular, the 2 https://www.zotero.org/google-docs/?9suQpJ https://www.zotero.org/google-docs/?bGvQNb https://www.zotero.org/google-docs/?bGvQNb https://www.zotero.org/google-docs/?vbVysu BY strain, closely related to the genome reference strain S288C, is a ‘lab strain’ and has been modified for ease of laboratory work. This strain is relatively easy to genome engineer, survives freeze / thaw cycles, and a complete map of its genome is available to use as a reference when studying multiple strains of yeast. Furthermore, yeast is inexpensive and easy to grow, allowing for large sample sizes, which increases statistical power, and in turn can contribute to more confidence in the outcomes of a study. The BY lab strain is commonly compared to RM, a vineyard isolate, to map the effects of natural genetic variation. BY and RM are the two strains I used throughout my PhD work. There are about 45,000 single nucleotide variants (SNVs) between BY and RM, making the strains differ, on average, one SNV every 200 bp (Ruderfer et al., 2006). This provides abundant genetic variation for the dissection of complex traits. Hundreds of loci have been shown to significantly affect organismal traits for various growth conditions in high-powered studies using this system (Bloom et al., 2013, 2015; Nguyen Ba et al., 2022). Quantitative trait loci One way to elucidate genetic variation that contributes to phenotypic variation is through quantitative trait loci (QTL) mapping. A QTL is simply an area of the genome (‘locus’) that contains genetic variation, where that variation significantly affects the trait being measured (‘quantitative trait’). QTLs can be categorized based on what is being measured. QTLs determined by: mRNA abundance are termed expression QTLs (eQTLs) (Albert & Kruglyak, 2015), 3 https://www.zotero.org/google-docs/?KA2URW https://www.zotero.org/google-docs/?QgUOyG https://www.zotero.org/google-docs/?33zhF5 protein abundance: pQTLs (Albert et al., 2014), growth rate: gQTLs (Bloom et al., 2013, 2015; Nguyen Ba et al., 2022), for example. In yeast, almost all (98.6%) genes have at least one eQTL (Albert et al., 2018). The expression of a gene may be affected by cis or trans effects. Cis effects arise from variation within or near a gene, such as a SNP in the promotor. Trans effects alter expression through variation in a diffusible factor, such as a transcription factor. Trans effects on a single gene are numerous, such that their joint effects are more important than cis effects (Albert et al., 2018). While cis effects only affect the gene at its locus, a single gene can be influenced by trans effects from anywhere in the genome. Although the QTL patterns described for the phenotypes in each of these studies are unique, genomic ‘hotspots’ have been a recurring theme in the field (Albert et al., 2018; Yeh et al., 2022). The term hotspot is used to denote loci that have been shown to affect numerous traits, including multiple growth phenotypes and the expression of thousands of genes. Remarkably, a few hotspots have shown up across studies and labs. In particular, three hotspots that contain the genes HAP1 (Nguyen Ba et al., 2022; Yeh et al., 2022), MKT1 (Steinmetz et al., 2002), and IRA2 (Lutz et al., 2021; Smith & Kruglyak, 2008), respectively, have come up time and time again for containing variation that causes changes in the measured trait, including the present work (Chapter II). QTL mapping approaches are continually advancing and studies are becoming increasingly higher throughput. Advances include genotyping large panels of individual segregants (Bloom et al., 2013), sequencing of pools of 4 https://www.zotero.org/google-docs/?Mmgkaq https://www.zotero.org/google-docs/?SJlHNc https://www.zotero.org/google-docs/?SJlHNc https://www.zotero.org/google-docs/?JE5q9M https://www.zotero.org/google-docs/?Hc6FML https://www.zotero.org/google-docs/?dYnxc6 https://www.zotero.org/google-docs/?tK7tVg https://www.zotero.org/google-docs/?KMXKWb https://www.zotero.org/google-docs/?KMXKWb https://www.zotero.org/google-docs/?T5QoIl https://www.zotero.org/google-docs/?tsjMd7 phenotypically extreme cells, known as X-QTL mapping (Albert et al., 2014; Ehrenreich et al., 2010), and pooled barcoding approaches that have increased throughput about 100-fold, allowing for the phenotyping of 100,000 segregants at once (Nguyen Ba et al., 2022). As the statistical power of genetic studies increases, it has been shown that complex, quantitative traits tend to be regulated by many genes in diverse biological processes (Duveau et al., 2021). The above studies have revealed that eQTL patterns do not directly match pQTL patterns for the same trait. This would be expected even if all other variables are held constant (i.e. same individuals and environment), because the number of protein molecules does not equal the number of mRNA molecules. Therefore, we do not know QTLs that affect the proteome based on eQTLs alone. This was exemplified in work from our lab that used dual fluorescent reporters to measure mRNA and protein abundance for the same gene in the same live cells (Brion et al., 2020). Brion et al. found that less than 20% of QTLs had concordant effects at the mRNA and protein levels. This raises important questions about what QTLs affect other processes that affect gene expression. Protein degradation is an obvious cellular process that could explain discrepancies between eQTL and pQTL patterns. Genotype by environment interactions Many of the studies mentioned above mapped QTLs in multiple environments, which revealed that gene or genotype by environment interactions (GxE) are common and contribute to phenotypic variation (Yadav & Sinha, 2018). 5 https://www.zotero.org/google-docs/?mLYvnN https://www.zotero.org/google-docs/?mLYvnN https://www.zotero.org/google-docs/?fOA19V https://www.zotero.org/google-docs/?9K4Gu9 https://www.zotero.org/google-docs/?b5Md88 https://www.zotero.org/google-docs/?ZCxcck GxE can modulate the contributions of genetic variation when the effect of a variant depends on an individual’s environment. A remarkable display of the dependence of a phenotype on environment in yeast is that 97% of genes were essential for growth in at least one of 178 conditions tested (Hillenmeyer et al., 2008), while fewer than 20% of yeast genes are essential in nutrient-rich conditions (Giaever et al., 2002; Winzeler et al., 1999). Polygenic traits are often more susceptible to environmental perturbations than Mendelian traits, making the study of GxE particularly important for the field of complex traits genetics (Yadav & Sinha, 2018). Multiple approaches have been used to investigate GxE in the model S. cerevisiae. Early work by Smith & Kruglyak (2008) used microarray to elucidate GxE in transcript abundance for BY and RM and their recombinant offspring (“segregants”) in media with glucose versus ethanol as the carbon source. They found GxE at about 40% of loci using linkage analysis of 109 segregants (Smith & Kruglyak, 2008). Further work using RNA-seq to study transcript abundance as a phenotype (Boye et al., 2024; Grishkevich & Yanai, 2013) has shown that GxE predominantly occurred at eQTLs via trans-acting mechanisms. In contrast, cis-acting eQTLs tended to show less GxE. Besides transcript abundance, growth in various conditions has been used to examine GxE in yeast. A survey of natural yeast isolates showed that isolates from genetically different populations grew differently in nearly half of 200 assayed environments (Warringer et al., 2011). Linkage mapping in BY-RM crosses revealed considerable heterogeneity in the genetic architecture of yeast growth across environments, including loci 6 https://www.zotero.org/google-docs/?gqNLcn https://www.zotero.org/google-docs/?gqNLcn https://www.zotero.org/google-docs/?Cxd8Xr https://www.zotero.org/google-docs/?FcfPH7 https://www.zotero.org/google-docs/?sEY5JH https://www.zotero.org/google-docs/?sEY5JH https://www.zotero.org/google-docs/?Xm7DpN https://www.zotero.org/google-docs/?XYUcMC https://www.zotero.org/google-docs/?XYUcMC that only affected growth in specific environments and loci whose direction of effect differed between environments (Bloom et al., 2013; Nguyen Ba et al., 2022). Recently, (Chen et al., 2023) revealed extensive GxE for individual DNA variants by measuring the effect of over 4,000 engineered natural variants on growth in six environments. They found that 93.7% of natural variants that had a significant effect on growth in at least one condition showed evidence of GxE (Chen et al., 2023). Taking the concept of GxE a step further, genetic interactions may change across environments, revealing “GxGxE” interactions. For example, Costanzo et al. devised a high throughput study to look at how genetic networks (GxG) changed across 14 conditions. They first looked at single-mutant fitness interactions (GxE) in the 14 conditions and found that over half of the genes they examined exhibited GxE in at least one condition. When analyzing double mutants across the same 14 conditions, they found that the genetic network (GxG) stayed remarkably stable (Costanzo et al., 2021), showing a relatively low frequency of GxGxE compared to GxG or GxE. These advancements have shaped the field of yeast genetics, particularly for GxE, with the majority of studies measuring either 1) growth or 2) transcript abundance. As stated in the previous section, eQTL and pQTL patterns do not match due to other regulators of gene expression, such as protein degradation. This suggests that mapping QTLs for protein degradation across various environments would reveal novel instances of GxE. 7 https://www.zotero.org/google-docs/?BHfDdD https://www.zotero.org/google-docs/?BHfDdD https://www.zotero.org/google-docs/?WbCZz6 https://www.zotero.org/google-docs/?i9q6f3 https://www.zotero.org/google-docs/?EaCDEi Protein degradation is an essential cellular process One important complex trait is protein degradation, an essential biological process that affects gene expression and maintains homeostasis by removing misfolded and damaged proteins from cells (G. A. Collins & Goldberg, 2017; Hanna & Finley, 2007; Varshavsky, 2011). The cell breaks down proteins through two main processes: autophagy and the ubiquitin-proteasome system (UPS). Autophagy is a general degradation process, while the UPS degrades proteins in a highly substrate-specific manner. In eukaryotes, the UPS, consisting of the ubiquitin system and the proteasome, is responsible for 70-80% of cellular protein degradation (Bachmair et al., 1986; G. A. Collins & Goldberg, 2017; Coux et al., 1996; Hershko & Ciechanover, 1998). The ubiquitin system marks proteins for degradation by covalently bonding the small protein ubiquitin to the target protein, which signals the target protein for degradation via the proteasome (Bett, 2016; Finley et al., 2012; Hershko & Ciechanover, 1998). Various E3 ligases recognize different substrates for degradation. Yeast have about 80 E3 ligases that contribute to degradation specificity by the UPS (Singh et al., 2012). Additionally, the proteasome can degrade certain substrates directly, independently of the ubiquitin system (Finley et al., 2012). A “degron” is a signal within a protein that is recognized by E3 ligases or receptors on the proteasome (Finley et al., 2012; Ha et al., 2012; Varshavsky, 1991) Protein degradation can be measured using multiple approaches, including radioactive labeling and fluorescent reporters. In the former, cells are 8 https://www.zotero.org/google-docs/?2hoBaD https://www.zotero.org/google-docs/?2hoBaD https://www.zotero.org/google-docs/?tivJw1 https://www.zotero.org/google-docs/?tivJw1 https://www.zotero.org/google-docs/?nCLEky https://www.zotero.org/google-docs/?nCLEky https://www.zotero.org/google-docs/?3Bk7Nt https://www.zotero.org/google-docs/?SjgRpE https://www.zotero.org/google-docs/?SjgRpE https://www.zotero.org/google-docs/?aNilbB https://www.zotero.org/google-docs/?aNilbB exposed to a “pulse” of radioactive amino acids, so that nascent proteins are labeled, followed by a “chase” in baseline medium, and then radioactivity is measured over time. As new non-labeled proteins are synthesized, they replace the radioactive labeled proteins, and degradation can be measured via a decrease in radioactivity over time (Burgis & Samson, 2007; Ross et al., 2021). An example of fluorescent reporters for degradation is tandem fluorescent protein timers (TFTs), which I used in this work. TFTs are linear fusions of two fluorescent proteins with distinct maturation kinetics and spectral profiles (Khmelinskii et al., 2012; Khmelinskii & Knop, 2014). Typically, a TFT consists of a faster-maturing green fluorescent protein (GFP) and a more slowly-maturing red fluorescent protein (RFP). As a result of these properties, the RFP / GFP ratio of the TFT increases over time. Because the RFP and GFP are synthesized from the same mRNA transcript, the TFT ratio is independent of the expression level of the TFT (Kats et al., 2018; Khmelinskii et al., 2012; Khmelinskii & Knop, 2014; Kong et al., 2021). Therefore, the TFT can be used to measure UPS activity in genetically diverse cell populations in which genetic variation may influence reporter abundance. The TFT ratio provides a direct, quantitative measure of UPS activity in live, single cells. Fusing a UPS degron to a TFT causes the construct, referred to here as a reporter, to be targeted and degraded by the UPS according to the pathway that recognizes the degron. For example, an N-terminal asparagine as the N-degron in a TFT will be bound by the type 1 binding site of the E3 ligase Ubr1p and degraded via the Arg/N-degron pathway (Khmelinskii et al., 2012; Varshavsky, 2024). These 9 https://www.zotero.org/google-docs/?J4UfSx https://www.zotero.org/google-docs/?0xONGj https://www.zotero.org/google-docs/?0xONGj https://www.zotero.org/google-docs/?Hyu7Wk https://www.zotero.org/google-docs/?Hyu7Wk https://www.zotero.org/google-docs/?i8V98N reporters provide high-throughput measurements of UPS activity that are sensitive to chemical and genetic perturbations that alter UPS activity. The UPS is genetically complex and influenced by GxE Recently, we used these TFT reporters to map QTLs for UPS activity in segregant offspring of the BY / RM cross in a baseline medium (Chapter III and Collins et al. 2022). This work revealed dozens of substrate-specific loci that affect UPS activity. We therefore know that UPS activity is genetically complex. However, up until now, we did not know to what extent GxE affects UPS activity. Because 1) protein degradation is environmentally-dependent (Bajorek et al., 2003; Finley & Prado, 2020; Grimm et al., 2012; Laporte et al., 2008; Sontag et al., 2014; Waite et al., 2016), 2) GxE is prevalent for many traits studied, and 3) USP activity is genetically complex, this motivated me to investigate GxE for UPS activity, which is the work described here in Chapter II. Segregant offspring from the BY / RM cross with six reporters for distinct components of the UPS were exposed to the eight environments. The segregants were then sorted using fluorescence activated cell sorting to collect cells with extremely high and low UPS activity, followed by whole-genome sequencing. We found a total of 416 QTLs for UPS activity across 47 combinations of UPS pathways and environments. Pairwise comparisons of the baseline condition to other environments revealed 254 and 17 loci comparisons that exhibited a presence / absence GxE and sign change GxE, respectively. A majority of the sign change 10 https://www.zotero.org/google-docs/?THsbXB https://www.zotero.org/google-docs/?THsbXB https://www.zotero.org/google-docs/?THsbXB GxE loci mapped to known loci that have been shown to be causal for UPS activity or to the known hotspots MKT1, HAP1, and IRA2. Causal loci The field of complex trait genetics has uncovered QTLs across model organisms, humans, numerous traits, and environments, providing major advancements in elucidating genetic architecture. However, relatively few causal variants have been determined within known QTLs. Without knowing the causal gene or variant(s) within each QTL, using known QTLs to predict phenotype from genotype remains a challenge. When causal variants are known, we can use statistical modeling and computational tools, such as machine learning, to find patterns in genomic sequences that lead to a particular phenotype (Renganaath et al., 2020). We can also uncover the biological mechanisms that cause an allelic difference to lead to a phenotypic difference. For example, an SNV in the promoter of a gene can weaken the affinity between a transcription factor and its binding site, leading to decreased transcription. A key challenge to elucidating a causal locus is that the confidence intervals of most QTLs contain many genes and variants. The peak of a QTL interval does not necessarily indicate the variant that is responsible for the change in phenotype displayed. There could be multiple variants within a small region that individually have an effect on the trait measured, and together, have a greater effect on the trait (Bloom et al., 2015). Even more nuanced is the concept of epistasis, which means that the sum of the effects of individual variants do not 11 https://www.zotero.org/google-docs/?oIvm0c https://www.zotero.org/google-docs/?oIvm0c https://www.zotero.org/google-docs/?IuIWBc equal the effects of the variants together (Lutz et al., 2021), indicating a genetic interaction (Bloom et al., 2015). Additionally, QTLs could even miss causal variation: if two variants are close to each other and they have opposing effects on the trait measured, the effect could be canceled out and no QTL called. Additional work, known as fine-mapping, must be done in order to determine the causal gene(s) of a QTL, that is, the gene that contains variation that affects the trait being measured. Engineering variants or blocks of variants with homologous recombination into the background strains in question, and measuring the effects of those variants, can reveal if they are causal or not (Brion et al., 2020; M. A. Collins et al., 2022; Lutz et al., 2019). Another way to determine causality is the reciprocal hemizygosity (RH) test (Steinmetz et al., 2002; Stern, 2014). The RH test can be used to test individual genes for causality. An RH test is performed in a hybrid between two genetically different strains. Knocking out one allele of a candidate gene creates a “hemizygous” genotype for that gene. This strain is compared to the “reciprocal” strain, where the corresponding allele on the homologous chromosome is knocked out. The trait being studied is then measured, and a significant difference in the quantitative trait values between the hemizygotes indicates that the candidate gene is causal (Stern, 2014). These direct approaches are beneficial when there are few and / or obvious candidate genes. However, if a candidate gene is unclear, or the goal is to determine the causal gene for multiple loci, these approaches are too low-throughput to be feasible. Furthermore, testing genes based on their known 12 https://www.zotero.org/google-docs/?qepnCV https://www.zotero.org/google-docs/?N1csd3 https://www.zotero.org/google-docs/?FZToYJ https://www.zotero.org/google-docs/?FZToYJ https://www.zotero.org/google-docs/?SVj0qR https://www.zotero.org/google-docs/?SVj0qR https://www.zotero.org/google-docs/?EV4AwS or speculated function may miss important genes that a priori appear to not be involved in the phenotype (Boyle et al., 2017). Therefore, genome-wide scans are advantageous for mapping loci at a finer resolution. For example, (Chen et al., 2023) engineered thousands of natural variants to determine causality. An additional approach is to apply the RH test genome-wide in a hybrid between the strains of interest, known as “RH scanning.” One attempt at this was done by crossing strains from the deletion collection to a genetically distant, intact strain (Wilkening et al., 2014). However, the approach produced many false positive results due to the deletion strains being prone to chromosomal abnormalities and suppressor mutations that accumulated during long growth periods (Teng et al., 2013). A successful attempt at RH scanning was performed by (Weiss et al., 2018) in an interspecific hybrid using S. cerevisiae and Saccharomyces paradoxus for the trait of thermotolerance. They used transposon mutagenesis to create a reciprocal hemizygote pool in the diploid hybrid. Sequencing data showed evidence of insertions for both alleles at 4,888 genes. They found eight genes that significantly affected thermotolerance from the scan, and these results were recapitulated by measuring thermotolerance in hemizygotic strains with the top hit genes deleted directly. For my work described in Chapter IV, I aimed to apply this approach in the BY / RM hybrid diploid for growth in YPD at 30°C. I created a large RH pool, grew the pool in triplicate over the course of a week, and sequenced multiple time points to determine changes in hemizygote frequency over time. After 13 https://www.zotero.org/google-docs/?mPcijf https://www.zotero.org/google-docs/?PWNoMg https://www.zotero.org/google-docs/?PWNoMg https://www.zotero.org/google-docs/?RcWL1X https://www.zotero.org/google-docs/?KiOaGp https://www.zotero.org/google-docs/?KiOaGp https://www.zotero.org/google-docs/?QbedNy https://www.zotero.org/google-docs/?QbedNy multiple validation assays, the results of my attempts thus far have shown to consist of false positives and false negatives. In Chapter IV I discuss why this might be the case and what next steps could be taken to improve the approach. 14 References Albert, F. W., Bloom, J. S., Siegel, J., Day, L., & Kruglyak, L. (2018). Genetics of trans-regulatory variation in gene expression. eLife, 7, e35471. https://doi.org/10.7554/eLife.35471 Albert, F. W., & Kruglyak, L. (2015). The role of regulatory variation in complex traits and disease. Nature Reviews Genetics, 16(4), 197–212. https://doi.org/10.1038/nrg3891 Albert, F. W., Treusch, S., Shockley, A. H., Bloom, J. S., & Kruglyak, L. (2014). Genetics of single-cell protein abundance variation in large yeast populations. Nature, 506(7489), 494–497. https://doi.org/10.1038/nature12904 Bachmair, A., Finley, D., & Varshavsky, A. (1986). In vivo half-life of a protein is a function of its amino-terminal residue. Science (New York, N.Y.), 234(4773), 179–186. https://doi.org/10.1126/science.3018930 Bajorek, M., Finley, D., & Glickman, M. H. (2003). Proteasome disassembly and downregulation is correlated with viability during stationary phase. Current Biology: CB, 13(13), 1140–1144. https://doi.org/10.1016/s0960-9822(03)00417-2 Bett, J. S. (2016). Proteostasis regulation by the ubiquitin system. Essays in Biochemistry, 60(2), 143–151. https://doi.org/10.1042/EBC20160001 Bloom, J. S., Ehrenreich, I. M., Loo, W. T., Lite, T.-L. V., & Kruglyak, L. (2013). Finding the sources of missing heritability in a yeast cross. Nature, 494(7436), 234–237. https://doi.org/10.1038/nature11867 Bloom, J. S., Kotenko, I., Sadhu, M. J., Treusch, S., Albert, F. W., & Kruglyak, L. (2015). Genetic interactions contribute less than additive effects to quantitative trait variation in yeast. Nature Communications, 6(1), 8712. https://doi.org/10.1038/ncomms9712 Boye, C., Nirmalan, S., Ranjbaran, A., & Luca, F. (2024). Genotype × environment interactions in gene regulation and complex traits. Nature Genetics, 56(6), 1057–1068. https://doi.org/10.1038/s41588-024-01776-w Boyle, E. A., Li, Y. I., & Pritchard, J. K. (2017). An Expanded View of Complex Traits: From Polygenic to Omnigenic. Cell, 169(7), 1177–1186. https://doi.org/10.1016/j.cell.2017.05.038 Brem, R. B., & Kruglyak, L. (2005). The landscape of genetic complexity across 5,700 gene expression traits in yeast. Proceedings of the National Academy of Sciences, 102(5), 1572–1577. https://doi.org/10.1073/pnas.0408709102 Brem, R. B., Yvert, G., Clinton, R., & Kruglyak, L. (2002). Genetic Dissection of Transcriptional Regulation in Budding Yeast. Science, 296(5568), 752–755. https://doi.org/10.1126/science.1069516 Brion, C., Lutz, S. M., & Albert, F. W. (2020). Simultaneous quantification of mRNA and protein in single cells reveals post-transcriptional effects of genetic variation. eLife, 9, e60645. https://doi.org/10.7554/eLife.60645 Burgis, N. E., & Samson, L. D. (2007). The Protein Degradation Response of Saccharomyces cerevisiae to Classical DNA-Damaging Agents. Chemical 15 https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw Research in Toxicology, 20(12), 1843–1853. https://doi.org/10.1021/tx700126e Chen, S.-A. A., Kern, A. F., Ang, R. M. L., Xie, Y., & Fraser, H. B. (2023). Gene-by-environment interactions are pervasive among natural genetic variants. Cell Genomics, 3(4). https://doi.org/10.1016/j.xgen.2023.100273 Collins, G. A., & Goldberg, A. L. (2017). The Logic of the 26S Proteasome. Cell, 169(5), 792–806. https://doi.org/10.1016/j.cell.2017.04.023 Collins, M. A., Avery, R., & Albert, F. W. (2023). Substrate-specific effects of natural genetic variation on proteasome activity. PLOS Genetics, 19(5), e1010734. https://doi.org/10.1371/journal.pgen.1010734 Collins, M. A., Mekonnen, G., & Albert, F. W. (2022). Variation in ubiquitin system genes creates substrate-specific effects on proteasomal protein degradation. eLife, 11, e79570. https://doi.org/10.7554/eLife.79570 Costanzo, M., Hou, J., Messier, V., Nelson, J., Rahman, M., VanderSluis, B., Wang, W., Pons, C., Ross, C., Ušaj, M., San Luis, B.-J., Shuteriqi, E., Koch, E. N., Aloy, P., Myers, C. L., Boone, C., & Andrews, B. (2021). Environmental robustness of the global yeast genetic interaction network. Science, 372(6542), eabf8424. https://doi.org/10.1126/science.abf8424 Coux, O., Tanaka, K., & Goldberg, A. L. (1996). Structure and functions of the 20S and 26S proteasomes. Annual Review of Biochemistry, 65, 801–847. https://doi.org/10.1146/annurev.bi.65.070196.004101 Duveau, F., Vande Zande, P., Metzger, B. P., Diaz, C. J., Walker, E. A., Tryban, S., Siddiq, M. A., Yang, B., & Wittkopp, P. J. (2021). Mutational sources of trans-regulatory variation affecting gene expression in Saccharomyces cerevisiae. eLife, 10, e67806. https://doi.org/10.7554/eLife.67806 Ehrenreich, I. M., Torabi, N., Jia, Y., Kent, J., Martis, S., Shapiro, J. A., Gresham, D., Caudy, A. A., & Kruglyak, L. (2010). Dissection of genetically complex traits with extremely large pools of yeast segregants. Nature, 464(7291), 1039–1042. https://doi.org/10.1038/nature08923 Finley, D., & Prado, M. A. (2020). The Proteasome and Its Network: Engineering for Adaptability. Cold Spring Harbor Perspectives in Biology, 12(1), a033985. https://doi.org/10.1101/cshperspect.a033985 Finley, D., Ulrich, H. D., Sommer, T., & Kaiser, P. (2012). The Ubiquitin–Proteasome System of Saccharomyces cerevisiae. Genetics, 192(2), 319–360. https://doi.org/10.1534/genetics.112.140467 Fisher, R. A. (1918). The correlation between relatives on the supposition of Mendelian inheritance. Transactions of the Royal Society of Edin-Burgh, 52, 339–433. Giaever, G., Chu, A. M., Ni, L., Connelly, C., Riles, L., Véronneau, S., Dow, S., Lucau-Danila, A., Anderson, K., André, B., Arkin, A. P., Astromoff, A., El Bakkoury, M., Bangham, R., Benito, R., Brachat, S., Campanaro, S., Curtiss, M., Davis, K., … Johnston, M. (2002). Functional profiling of the Saccharomyces cerevisiae genome. Nature, 418(6896), 387–391. https://doi.org/10.1038/nature00935 Grimm, S., Höhn, A., & Grune, T. (2012). Oxidative protein damage and the proteasome. 16 https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw Amino Acids, 42(1), 23–38. https://doi.org/10.1007/s00726-010-0646-8 Grishkevich, V., & Yanai, I. (2013). The genomic determinants of genotype × environment interactions in gene expression. Trends in Genetics, 29(8), 479–487. https://doi.org/10.1016/j.tig.2013.05.006 Ha, S.-W., Ju, D., & Xie, Y. (2012). The N-terminal domain of Rpn4 serves as a portable ubiquitin-independent degron and is recognized by specific 19S RP subunits. Biochemical and Biophysical Research Communications, 419(2), 226–231. https://doi.org/10.1016/j.bbrc.2012.01.152 Hanna, J., & Finley, D. (2007). A proteasome for all occasions. FEBS Letters, 581(15), 2854–2861. https://doi.org/10.1016/j.febslet.2007.03.053 Hershko, A., & Ciechanover, A. (1998). The ubiquitin system. Annual Review of Biochemistry, 67, 425–479. https://doi.org/10.1146/annurev.biochem.67.1.425 Hillenmeyer, M. E., Fung, E., Wildenhain, J., Pierce, S. E., Hoon, S., Lee, W., Proctor, M., St Onge, R. P., Tyers, M., Koller, D., Altman, R. B., Davis, R. W., Nislow, C., & Giaever, G. (2008). The chemical genomic portrait of yeast: Uncovering a phenotype for all genes. Science (New York, N.Y.), 320(5874), 362–365. https://doi.org/10.1126/science.1150021 Kats, I., Khmelinskii, A., Kschonsak, M., Huber, F., Knieß, R. A., Bartosik, A., & Knop, M. (2018). Mapping Degradation Signals and Pathways in a Eukaryotic N-terminome. Molecular Cell, 70(3), 488-501.e5. https://doi.org/10.1016/j.molcel.2018.03.033 Khmelinskii, A., Keller, P. J., Bartosik, A., Meurer, M., Barry, J. D., Mardin, B. R., Kaufmann, A., Trautmann, S., Wachsmuth, M., Pereira, G., Huber, W., Schiebel, E., & Knop, M. (2012). Tandem fluorescent protein timers for in vivo analysis of protein dynamics. Nature Biotechnology, 30(7), 708–714. https://doi.org/10.1038/nbt.2281 Khmelinskii, A., & Knop, M. (2014). Analysis of Protein Dynamics with Tandem Fluorescent Protein Timers. In A. I. Ivanov (Ed.), Exocytosis and Endocytosis (Vol. 1174, pp. 195–210). Springer New York. https://doi.org/10.1007/978-1-4939-0944-5_13 Kong, K.-Y. E., Fischer, B., Meurer, M., Kats, I., Li, Z., Rühle, F., Barry, J. D., Kirrmaier, D., Chevyreva, V., San Luis, B.-J., Costanzo, M., Huber, W., Andrews, B. J., Boone, C., Knop, M., & Khmelinskii, A. (2021). Timer-based proteomic profiling of the ubiquitin-proteasome system reveals a substrate receptor of the GID ubiquitin ligase. Molecular Cell, 81(11), 2460-2476.e11. https://doi.org/10.1016/j.molcel.2021.04.018 Laporte, D., Salin, B., Daignan-Fornier, B., & Sagot, I. (2008). Reversible cytoplasmic localization of the proteasome in quiescent yeast cells. The Journal of Cell Biology, 181(5), 737–745. https://doi.org/10.1083/jcb.200711154 Lutz, S., Brion, C., Kliebhan, M., & Albert, F. W. (2019). DNA variants affecting the expression of numerous genes in trans have diverse mechanisms of action and evolutionary histories. PLOS Genetics, 15(11), e1008375. https://doi.org/10.1371/journal.pgen.1008375 17 https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw Lutz, S., Van Dyke, K., Feraru, M. A., & Albert, F. W. (2021). Multiple epistatic DNA variants in a single gene affect gene expression in trans. Genetics, iyab208. https://doi.org/10.1093/genetics/iyab208 Miko, I. (2008). Gregor Mendel and the Principles of Inheritance. Nature Education, 1(1), 134. Nguyen Ba, A. N., Lawrence, K. R., Rego-Costa, A., Gopalakrishnan, S., Temko, D., Michor, F., & Desai, M. M. (2022). Barcoded bulk QTL mapping reveals highly polygenic and epistatic architecture of complex traits in yeast. eLife, 11, e73983. https://doi.org/10.7554/eLife.73983 Peter, J., De Chiara, M., Friedrich, A., Yue, J.-X., Pflieger, D., Bergström, A., Sigwalt, A., Barre, B., Freel, K., Llored, A., Cruaud, C., Labadie, K., Aury, J.-M., Istace, B., Lebrigand, K., Barbry, P., Engelen, S., Lemainque, A., Wincker, P., … Schacherer, J. (2018). Genome evolution across 1,011 Saccharomyces cerevisiae isolates. Nature, 556(7701), 339–344. https://doi.org/10.1038/s41586-018-0030-5 Plomin, R., Haworth, C. M. A., & Davis, O. S. P. (2009). Common disorders are quantitative traits. Nature Reviews Genetics, 10(12), 872–878. https://doi.org/10.1038/nrg2670 Renganaath, K., Chong, R., Day, L., Kosuri, S., Kruglyak, L., & Albert, F. W. (2020). Systematic identification of cis-regulatory variants that cause gene expression differences in a yeast cross. eLife, 9, e62669. https://doi.org/10.7554/eLife.62669 Ross, A. B., Langer, J. D., & Jovanovic, M. (2021). Proteome Turnover in the Spotlight: Approaches, Applications, and Perspectives. Molecular & Cellular Proteomics, 20, 100016. https://doi.org/10.1074/mcp.R120.002190 Ruderfer, D. M., Pratt, S. C., Seidel, H. S., & Kruglyak, L. (2006). Population genomic analysis of outcrossing and recombination in yeast. Nature Genetics, 38(9), 1077–1081. https://doi.org/10.1038/ng1859 Serpico, D., Lynch, K. E., & Porter, T. M. (2023). New historical and philosophical perspectives on quantitative genetics. Studies in History and Philosophy of Science, 97, 29–33. https://doi.org/10.1016/j.shpsa.2022.11.009 Singh, R. K., Gonzalez, M., Kabbaj, M.-H. M., & Gunjan, A. (2012). Novel E3 Ubiquitin Ligases That Regulate Histone Protein Levels in the Budding Yeast Saccharomyces cerevisiae. PLoS ONE, 7(5), e36295. https://doi.org/10.1371/journal.pone.0036295 Smith, E. N., & Kruglyak, L. (2008). Gene–Environment Interaction in Yeast Gene Expression. PLoS Biology, 6(4), e83. https://doi.org/10.1371/journal.pbio.0060083 Sontag, E. M., Vonk, W. I. M., & Frydman, J. (2014). Sorting out the trash: The spatial nature of eukaryotic protein quality control. Current Opinion in Cell Biology, 26, 139–146. https://doi.org/10.1016/j.ceb.2013.12.006 Steinmetz, L. M., Sinha, H., Richards, D. R., Spiegelman, J. I., Oefner, P. J., McCusker, J. H., & Davis, R. W. (2002). Dissecting the architecture of a quantitative trait locus in yeast. Nature, 416(6878), 326–330. https://doi.org/10.1038/416326a 18 https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw Stern, D. L. (2014). Identification of loci that cause phenotypic variation in diverse species with the reciprocal hemizygosity test. Trends in Genetics, 30(12), 547–554. https://doi.org/10.1016/j.tig.2014.09.006 Teng, X., Dayhoff-Brannigan, M., Cheng, W.-C., Gilbert, C. E., Sing, C. N., Diny, N. L., Wheelan, S. J., Dunham, M. J., Boeke, J. D., Pineda, F. J., & Hardwick, J. M. (2013). Genome-wide Consequences of Deleting Any Single Gene. Molecular Cell, 52(4), 485–494. https://doi.org/10.1016/j.molcel.2013.09.026 Varshavsky, A. (1991). Naming a targeting signal. Cell, 64(1), 13–15. https://doi.org/10.1016/0092-8674(91)90202-a Varshavsky, A. (2011). The N-end rule pathway and regulation by proteolysis. Protein Science: A Publication of the Protein Society, 20(8), 1298–1345. https://doi.org/10.1002/pro.666 Varshavsky, A. (2024). N-degron pathways. Proceedings of the National Academy of Sciences, 121(39). https://doi.org/10.1073/pnas.2408697121 Waite, K. A., Mota-Peynado, A. D.-L., Vontz, G., & Roelofs, J. (2016). Starvation Induces Proteasome Autophagy with Different Pathways for Core and Regulatory Particles. Journal of Biological Chemistry, 291(7), 3239–3253. https://doi.org/10.1074/jbc.M115.699124 Warringer, J., Zörgö, E., Cubillos, F. A., Zia, A., Gjuvsland, A., Simpson, J. T., Forsmark, A., Durbin, R., Omholt, S. W., Louis, E. J., Liti, G., Moses, A., & Blomberg, A. (2011). Trait Variation in Yeast Is Defined by Population History. PLoS Genetics, 7(6), e1002111. https://doi.org/10.1371/journal.pgen.1002111 Weiss, C. V., Roop, J. I., Hackley, R. K., Chuong, J. N., Grigoriev, I. V., Arkin, A. P., Skerker, J. M., & Brem, R. B. (2018). Genetic dissection of interspecific differences in yeast thermotolerance. Nature Genetics, 50(11), 1501–1504. https://doi.org/10.1038/s41588-018-0243-4 Wilkening, S., Lin, G., Fritsch, E. S., Tekkedil, M. M., Anders, S., Kuehn, R., Nguyen, M., Aiyar, R. S., Proctor, M., Sakhanenko, N. A., Galas, D. J., Gagneur, J., Deutschbauer, A., & Steinmetz, L. M. (2014). An Evaluation of High-Throughput Approaches to QTL Mapping in Saccharomyces cerevisiae. Genetics, 196(3), 853–865. https://doi.org/10.1534/genetics.113.160291 Winzeler, E. A., Shoemaker, D. D., Astromoff, A., Liang, H., Anderson, K., Andre, B., Bangham, R., Benito, R., Boeke, J. D., Bussey, H., Chu, A. M., Connelly, C., Davis, K., Dietrich, F., Dow, S. W., El Bakkoury, M., Foury, F., Friend, S. H., Gentalen, E., … Davis, R. W. (1999). Functional Characterization of the S. cerevisiae Genome by Gene Deletion and Parallel Analysis. Science, 285(5429), 901–906. https://doi.org/10.1126/science.285.5429.901 Yadav, A., & Sinha, H. (2018). Gene–gene and gene–environment interactions in complex traits in yeast. Yeast, 35(6), 403–416. https://doi.org/10.1002/yea.3304 Yeh, C.-L. C., Jiang, P., & Dunham, M. J. (2022). High-throughput approaches to functional characterization of genetic variation in yeast. Current Opinion in Genetics & Development, 76, 101979. https://doi.org/10.1016/j.gde.2022.101979 19 https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw https://www.zotero.org/google-docs/?tjQZgw Chapter II. Genotype-by-environment interactions shape ubiquitin-proteasome system activity Randi R. Avery, Mahlon A. Collins*, Frank W. Albert* Department of Genetics, Cell Biology, & Development, University of Minnesota, Minneapolis, MN, USA * To whom correspondence should be addressed: mahlon@umn.edu, falbert@umn.edu MAC and FWA jointly supervised this work Abstract In genotype-by-environment interactions (GxE), the effect of a genetic variant on a trait depends on the environment. GxE influences numerous organismal traits across eukaryotic life. However, we have a limited understanding of how GxE shapes the molecular processes that give rise to organismal traits. Here, we characterized how GxE shapes protein degradation, an essential molecular process that influences numerous aspects of cellular and organismal physiology. Using the yeast Saccharomyces cerevisiae, we characterized GxE in the activity of the ubiquitin-proteasome system (UPS), the primary protein degradation system in eukaryotes. By mapping genetic influences on the degradation of six substrates that engage multiple distinct UPS pathways across eight diverse environments, we discovered extensive GxE in the genetics of UPS activity. Hundreds of locus effects on UPS activity varied depending on the substrate, the environment, or both. Most of these cases corresponded to loci that were present 20 in one environment but not another (“presence / absence” GxE), while a smaller number of loci had opposing effects in different environments (“sign change” GxE). The number of loci exhibiting GxE, their genomic location, and the type of GxE (presence / absence or sign change) varied across UPS substrates. Loci exhibiting GxE were clustered at genomic regions that contain core UPS genes and especially at regions containing variation that affects the expression of thousands of genes, suggesting indirect contributions to UPS activity. Our results reveal highly complex interactions at the level of substrates and environments in the genetics of protein degradation. Introduction Genotype-by-environment interactions (GxE) occur when a genetic variant's effect on a trait is environment-dependent. GxE can have profound effects on organismal physiology. For example, in the disease phenylketonuria individuals with two defective copies of the phenylalanine hydroxylase gene develop severe symptoms, including brain damage and intellectual disabilities, if they consume a diet with standard amounts of phenylalanine. However they can avoid most symptoms by consuming a diet with reduced phenylalanine (Bickel et al., 1953; Guthrie, 1961; Shostak, 2003; Widaman, 2009). Other prominent examples of GxE exist in pharmacogenetics, where genetic differences modulate drug efficacy (Pirmohamed, 2023). Thus, understanding the extent and genetic basis of GxE has been a longstanding goal in biomedical research. 21 Efforts to this end have revealed that GxE is widespread at the level of organismal traits. GxE has been observed for a variety of morphological (e.g., Drosophila bristle number; Gurganus et al., 1998) and developmental (e.g., flowering time in various temperatures; Sasaki et al., 2015) traits in numerous organisms. Recent work in humans has begun to explore the impact of environmental factors on traits related to health and disease. By using self-reported and demographic information to integrate environmental factors into genome-wide association studies, these efforts have revealed that GxE shapes the genetics of a variety of clinical syndromes, including depression (C. Li et al., 2022), cancer (Yang et al., 2020), and health-related traits, such as body mass index (Robinson et al., 2017). However, to what extent GxE occurs at the level of the molecular processes that give rise to organismal phenotypes in humans and other species is poorly understood. A key challenge is that most traits are genetically complex, influenced by variation at loci throughout the genome. Profiling sufficiently large samples to achieve the statistical power needed to detect the effects of multiple loci and their interaction with environmental factors requires assays with high-throughput and quantitative precision. This has limited our ability to understand how environmental factors modulate genetic influences on all but a small number of molecular processes, in particular gene expression and cellular growth. Considerable work has focused on GxE in gene expression (Boye et al., 2024; Grishkevich & Yanai, 2013). For example, studies in flies (Huang et al., 2020), plants (Cubillos et al., 2014), roundworms (Y. Li et al., 2006), and mice 22 (Ballinger et al., 2023) have assayed gene expression in genetically different individuals in different temperatures. In these studies, GxE predominantly occurred at loci that influence gene expression (“expression quantitative trait loci,” eQTLs) via trans-acting mechanisms. In humans, eQTLs identified in immune cells display considerable GxE from variation in genes in pathways that become activated upon exposure to various immunogenic stimuli, demonstrating that GxE can occur via direct effects on genes in a relevant pathway (Fairfax et al., 2014; Kim-Hellmuth et al., 2017; Nédélec et al., 2016; Quach et al., 2016). Profiling stimulated human immune cells has also revealed condition-specific trans-eQTLs, showing that GxE can also occur via indirect mechanisms (Fairfax et al., 2014; Lee et al., 2014). Context-specific eQTLs displaying GxE are enriched for GWAS signals for complex organismal traits (Kim-Hellmuth et al., 2017; Lea et al., 2022), highlighting the value of profiling molecular traits that give rise to organismal phenotypes. The yeast Saccharomyces cerevisiae has served as a powerful model for dissecting GxE (Yadav & Sinha, 2018) because thousands of natural, genetically different isolates (Warringer et al., 2011), their cross progeny (Bloom et al., 2019; Nguyen Ba et al., 2022; Smith & Kruglyak, 2008), or strains harboring engineered natural variants (Chen et al., 2023) can be exposed to tightly controlled environments at high levels of replication. These approaches have revealed that GxE in cellular growth is widespread. A survey of natural yeast isolates showed that isolates from genetically different populations grew differently in nearly half of 200 assayed environments (Warringer et al., 2011). Later work using linkage 23 mapping in crosses revealed considerable heterogeneity in the genetic architecture of yeast growth in dozens of environments, including loci that only affected growth in specific environments and loci whose direction of effect differed between environments (Bloom et al., 2013; Nguyen Ba et al., 2022). Recently, Chen et al. (2023) revealed that 93.7% of natural variants that had a significant effect on growth in at least one condition showed evidence of GxE (Chen et al., 2023). Additionally, in a cross between two yeast strains, Smith & Kruglyak (2008) found GxE at about 40% of loci that shape transcript abundance in media with glucose versus ethanol as the carbon source. Despite these foundational insights from humans and model systems, our knowledge about GxE in molecular traits remains limited. Profiling additional molecular processes with known roles in organismal physiology would expand our understanding of how GxE shapes health, disease, and evolution. Protein degradation is an essential molecular process that influences numerous aspects of cellular and organismal physiology. In eukaryotes, most protein degradation (70-80%) is carried out by the ubiquitin-proteasome system (UPS) (Bachmair et al., 1986; G. A. Collins & Goldberg, 2017; Coux et al., 1996; Hershko & Ciechanover, 1998). By degrading substrate proteins, the UPS regulates protein abundance and removes misfolded and damaged proteins from cells (G. A. Collins & Goldberg, 2017; Hanna & Finley, 2007; Varshavsky, 2011). The central importance of UPS protein degradation is illustrated by the defects in this process that occur in numerous human diseases, including cancers, immune disorders, and neurodegenerative diseases (Dantuma & Bott, 2014; Schwartz & 24 Ciechanover, 1999; Shringarpure & Davies, 2002; Zheng et al., 2014). The UPS comprises the ubiquitin system, a collection of enzymes that mark substrate proteins for degradation, and the proteasome, a multi-protein complex that degrades substrate proteins to small peptides. The ubiquitin system recognizes short signal sequences (termed “degrons”; Varshavsky, 1991) in proteins, then marks the substrate protein for degradation by covalently attaching the small protein ubiquitin. Ubiquitinated substrate proteins are bound by the proteasome’s 19S regulatory particle, unfolded, and degraded to short peptides by the proteasome’s 20S core particle (Bett, 2016; Finley et al., 2012; Hershko & Ciechanover, 1998). The proteasome can also bind and degrade certain substrates directly, independent of the ubiquitin system (Finley et al., 2012). UPS protein degradation is organized into multiple distinct pathways based on the ubiquitin system enzymes and proteasome receptors involved in targeting and binding substrates of a given pathway. This can result in highly pathway- and even substrate-specific UPS activity that is tailored to the physiological needs of the cell. For example, during proteotoxic stress, UPS activity towards misfolded proteins can selectively increase (Gardner et al., 2005; Ibarra et al., 2016; Rosenbaum et al., 2011). We recently showed that UPS activity is a genetically complex trait (M. A. Collins et al., 2022, 2023). By measuring UPS activity towards multiple substrates that engage distinct UPS targeting and degradation pathways, we revealed that many variant effects are substrate-specific in that the magnitude and, in some cases, direction of their effects on UPS activity varied between 25 substrates. Some loci exerting substrate-specific effects contained causal variants in UPS genes, while other loci did not contain any genes with known roles in UPS activity, suggesting indirect effects. However, these experiments were performed in a single environment. The extent of GxE in the genetics of UPS activity is unknown. Protein degradation is highly environment-dependent, raising the possibility that the genetics of UPS activity is subject to GxE. For example, in environmental conditions that cause misfolded or oxidatively damaged proteins to accumulate, UPS activity increases to clear these molecules from the cell (Finley & Prado, 2020; Grimm et al., 2012; Sontag et al., 2014). In contrast, UPS protein degradation, an ATP-dependent process, decreases in nutrient-poor conditions, to conserve cellular resources (Bajorek et al., 2003; Laporte et al., 2008; Waite et al., 2016). This process is well-characterized in yeast cells, which decrease UPS activity in low glucose environments by sequestering proteasomes in inactive aggregates and in low nitrogen environments by autophagically degrading proteasomes (Laporte et al., 2008; J. Li et al., 2019). These environmental influences on UPS activity, combined with the fact that UPS activity is affected by complex natural genetic variation, suggests that genetic influences on UPS activity are influenced by GxE. Here, we quantified and mapped GxE in the genetics of UPS activity. By measuring genetic influences on UPS activity towards six substrates that engage distinct UPS pathways in eight environments, we discovered extensive GxE in 26 the genetics of protein degradation. We found hundreds of instances where a locus altered UPS activity in one environment but not another, as well as a smaller number of instances where a locus effect changed direction between environments. Patterns of loci exhibiting GxE were also highly specific to individual UPS substrates. Our results reveal a high degree of environment- and pathway-specific GxE in the genetic architecture of UPS activity and expand our understanding of how GxE shapes molecular traits. Results Experimental Design Overview To study GxE in the UPS in a pathway-specific manner, we compared the UPS activity of two genetically divergent yeast strains for six UPS substrates that engage multiple distinct UPS pathways in eight environments comprising multiple starvation and chemical stressors (Fig. 1A). We compared the BY laboratory strain, a close relative of the S288C reference strain, to RM, a vineyard isolate. These strains differ on average one nucleotide per 200 base pairs (bp), providing abundant genetic variation that is known to affect molecular and cellular traits (Albert et al., 2018; Bloom et al., 2013; Brem et al., 2002; Brion et al., 2020; Nguyen Ba et al., 2022) that may be subject to GxE, including UPS activity (M. A. Collins et al., 2022, 2023). We selected eight environments predicted to alter UPS activity (Table 1). Throughout this paper we consider synthetic complete medium (“SC”), a 27 nutrient-rich medium, as the baseline environment for normal growth. UPS activity in SC was compared to seven other environments intended to produce diverse impacts on cellular physiology, some of which have known effects on UPS activity. The environments included three conditions with reduced nutrients compared to SC, such that these “starvation” conditions are predicted to decrease UPS activity: low glucose, low nitrogen, and yeast nitrogen base (YNB) without amino acids. Protein degradation via the UPS plays a critical role in the response to multiple forms of chemical stress. We assayed four chemical stress conditions by adding bortezomib (BTZ), L-azetidine-2-carboxylic acid (AZC), 4NQO, or lithium acetate (LiAc) to SC. Bortezomib (BTZ) inhibits proteasomal protein degradation by tightly and selectively binding the 20S proteasome’s catalytically active site, causing proteolytic stress (Nunes & Annunziata, 2017; Work & Brandman, 2021). The proline analog AZC causes misfolding of nascent proteins, leading to cellular “folding stress” (Rodgers & Shiozawa, 2008; Work & Brandman, 2021) wherein misfolded proteins accumulate in protein aggregates, resulting in increased UPS activity (Work & Brandman, 2021). The mutagen 4NQO has been shown to increase global protein degradation (Burgis & Samson, 2007). Finally, we chose LiAc, a chemical used in yeast transformations (Gietz & Schiestl, 2007). High salt concentrations cause the proteasome’s 19S regulatory particle to dissociate from the 20S core particle (Glickman et al., 1998; Saeki et al., 2000), and lithium chloride has been shown to inhibit purified 20S core particles (Holtz et al., 2003), suggesting that LiAc could decrease UPS activity. 28 Table 1: Environments Environment Abbreviation Concentration Description Reference Synthetic complete SC - Baseline Collins et al., 2022, 2023 Low glucose Low G 0.025% glucose (typically 2%) Starvation Laporte et al., 2008; Li et al., 2019 Low nitrogen Low N Low sources of nitrogen from components of yeast nitrogen base w/o amino acids and w/o ammonium sulfate. See Media and Chemical table Starvation Li et al., 2019 4NQO 4NQO 2 µg/mL DNA damaging agent Burgis & Samson, 2007 Bortezomib BTZ 40 µM Proteasome inhibitor Work & Brandman, 2021 L-azetidine-2- carboxylic acid AZC 4 mM Proline analog → protein misfolding Work & Brandman, 2021 Lithium acetate LiAc 20 mM Salt - Yeast nitrogen base YNB - Starvation - To measure UPS activity, we used tandem fluorescent protein timers (TFTs). TFTs are two-color fluorescent protein constructs that provide high-throughput measurements of protein turnover (Khmelinskii et al., 2012; Khmelinskii & Knop, 2014) (Fig. 1B; Table 2). In the most common implementation, which is used here, the TFT consists of a linear fusion of a faster-maturing green fluorescent protein (GFP) and a more slowly-maturing red fluorescent protein (RFP). If the TFT’s degradation rate is faster than the RFP’s 29 maturation rate, then the -log2 (RFP / GFP) ratio is directly proportional to the TFT’s degradation rate (Khmelinskii et al., 2012; Khmelinskii & Knop, 2014). Because the RFP and GFP are synthesized from the same mRNA transcript, the TFT ratio is independent of the expression level of the TFT (Kats et al., 2018; Khmelinskii et al., 2012; Khmelinskii & Knop, 2014; Kong et al., 2021). TFTs can thus be used to measure protein degradation in genetically distinct cell populations where reporter expression may vary (M. A. Collins et al., 2022, 2023; Khmelinskii et al., 2012). An additional key feature of the TFT system is its transferability. By fusing a TFT to a UPS substrate, the substrate's degradation rate can be measured, provided its degradation rate is faster than the RFP's maturation rate (Khmelinskii et al., 2012; Khmelinskii & Knop, 2014). Prior efforts have fused a variety of distinct UPS substrates to TFTs to serve as high-throughput, pathway- and substrate-specific reporters of UPS activity in live cells (M. A. Collins et al., 2022, 2023). Table 2: Reporters Reporter long name Reporter abbreviation Ubiquitin system- Threonine N-degron reporter Thr Dependent Asparagine N-degron reporter Asn Dependent Phenylalanine N-degron reporter Phe Dependent Ubiquitin fusion degradation reporter UFD Dependent Rpn4 reporter Rpn4 Independent Linear tetra-ubiquitin reporter 4xUb Independent 30 We attached six degron-containing substrate sequences that engage multiple distinct UPS pathways to our TFTs to serve as six reporters of UPS activity (Fig. 1B). The UPS reporters studied here included four substrates targeted by the ubiquitin system (“ubiquitin system-dependent” reporters) and two substrates that are directly bound and degraded by the proteasome (“ubiquitin system-independent” reporters). Three of the ubiquitin system-dependent reporters probe the three branches of the N-degron pathway, a UPS pathway in which a protein’s N-terminal amino acid functions as a degron (Varshavsky, 2011, 2019, 2024). These include the Type-1 Arg/N-degron pathway (using asparagine as the N-terminal amino acid; “Asn”), which targets basic N-terminal amino acids; the Type-2 Arg/N-degron pathway (phenylalanine; “Phe”), which targets bulky hydrophobic N-terminal amino acids; and the Ac/N-degron pathway (threonine; “Thr”), which targets acetylated uncharged N-terminal amino acids (Varshavsky, 2011, 2019, 2024). These N-degron pathways influence multiple aspects of cellular physiology by regulating protein abundance. To capture genetic influences on protein quality control-associated UPS activity, we constructed a TFT reporter that measures the activity of the ubiquitin fusion degradation (UFD) pathway (Johnson et al., 1995). In the UFD pathway, a non-cleavable ubiquitin moiety acts as a degron that is recognized by the Ufd4p E3 ligase, which targets misfolded proteins and is involved in the response to proteotoxic stress (Devarajan et al., 2020; Johnson et al., 1995; Theodoraki et al., 2012). 31 To measure genetic effects on proteasome activity separately from the ubiquitin system, we used two reporters containing degrons that are directly bound and degraded by the proteasome. First, the Rpn4 reporter (M. A. Collins et al., 2023) contains the first 80 amino acids of the Rpn4 protein, which are directly bound by the Rpn2p and Rpn5p receptors of the 19S regulatory particle of the proteasome (Ha et al., 2012; Ju & Xie, 2004; Prakash et al., 2004; Xie & Varshavsky, 2001). The second ubiquitin system-independent reporter contains a linear fusion of four ubiquitin molecules (Stack et al., 2000), which functions as a degron that is recognized by the proteasome receptor Rpn13p (Thrower et al., 2000). This “4xUb” reporter serves as a minimal degron that mimics the degradation of the majority of physiological UPS substrates (Inobe et al., 2011; Martinez-Fonts et al., 2020; Thrower et al., 2000; Zhao & Ulrich, 2010). Because the Rpn4 and 4xUb degrons have different sizes, sequence compositions, and structures, we reasoned that they may be influenced by distinct sets of loci, as in our prior studies of ubiquitin-independent substrates (M. A. Collins et al., 2023). Our selection of substrates thus allowed us to capture genetic influences on the activity of multiple UPS pathways involved in physiological protein abundance regulation and protein quality control. The UFD and 4xUb reporters were developed for this study, and we characterized these reporters using flow cytometry in BY, RM, and in a BY strain with reduced UPS activity due to deletion of the RPN4 gene (‘rpn4∆’) (Xie & Varshavsky, 2001). As expected, both reporters showed significantly lower UPS activity in the BY rpn4∆ strain than in the BY and RM strains when grown in SC 32 (Fig. 1C). RM showed higher UPS activity than BY for UFD (T-test: p-value = 1.8e-5), and there was no difference between BY and RM for 4xUb (p = 0.57). Thus, together with our previous work on N-degron pathways and the Rpn4 reporter (M. A. Collins et al., 2022, 2023), all six reporters provide quantitative, substrate-specific, in vivo measurements of UPS activity. Fig. 1: Study design. A. Experimental design overview. FACS: fluorescence activated cell sorting. WGS: whole-genome sequencing. B. Simplified schematics of the six reporters used in this study to measure ubiquitin system-dependent and -independent UPS pathways. UbF: ubiquitin with G76V substitution. Ub*: ubiquitin with G76V and K29/48/63R substitutions. Adapted from (M. A. Collins et al., 2022, 2023). C. UPS activity from UFD and 4xUb reporters in BY, RM, and BY rpn4∆. P-values from two-tailed T-tests are indicated. 33 Widespread GxE in UPS activity between two yeast strains To estimate the extent of GxE in the UPS, we exposed BY and RM strains carrying one of the six reporters to the eight environments and used flow cytometry to assay UPS activity in 20,000 cells in each of eight biological replicates per combination of strain, reporter, and environment. Compared to the SC baseline, all seven environments affected UPS activity in at least one strain and for at least one reporter (Fig. 2). UPS activity significantly decreased in 39 of 80 comparisons and increased in four comparisons (T-test; Bonferroni-corrected p < 0.05). Of the increases, three were seen for BY in AZC (Asn, Rpn4, and 4xUb), in line with the reported modest increases in RPN4 expression caused by this treatment in a BY strain (Work & Brandman, 2021). Notably, the effects of environment on UPS activity were highly strain-dependent. For example, all four cases of environmentally-induced increases in UPS activity were seen in only one of the two strains (Fig. 2A-C). Together, these observations suggest widespread GxE in UPS activity. To search for GxE more formally, we fit linear models that compared UPS activity for a given reporter between BY and RM and between SC and one of the other environments (Methods). GxE was detected in 27 (67.5%) of 40 tests (analysis of variance interaction term p-value < 0.05 after Bonferroni correction), revealing numerous cases in which an environmental effect on UPS activity depended on the strain (Fig. 2A-C). 34 The most significant interaction effect (p = 1e-13) was seen for the phenylalanine N-degron reporter in 4NQO (Fig. 2D). 4NQO reduced the UPS activity measured by this reporter in both BY and RM (environment main effect: p = 8e-19, T-tests: p ≤ 1e-7). However, the reduction was stronger in BY than in RM, such that BY had higher UPS activity than RM in SC, while it had lower activity than RM in 4NQO. The GxE term with the second most significant interaction effect was seen for the asparagine N-degron reporter in AZC (p = 2e-13) (Fig. 2E). In this case, AZC lowered UPS activity in RM but increased it in BY. This also resulted in a rank order change: in SC RM had higher UPS activity than BY (T-test, p = 3e-5), while in AZC RM had lower UPS activity (p = 2e-17). The GxE term with the third-smallest p-value was observed for the threonine N-degron reporter in low glucose (p = 2e-11) (Fig. 2F). Here, RM showed higher UPS activity than BY in SC (T-test, p = 4e-9). However, in low glucose, UPS activity was much lower for both BY and RM, dropping to the two lowest (out of 96) mean UPS activity values in this experiment and removing the strain difference (T-test, p = 0.72) (Supplementary Fig. 1A). Each of the reporters and environments showed at least one significant case of GxE (Fig. 2A-C, Supplementary Fig. 1B & C). Among environments, LiAc and YNB showed GxE for all six reporters when compared to SC, while BTZ had the lowest number of significant interaction effects (1 / 6) (Fig. 2B & C, Supplementary Fig. 1C). The five strongest (ranked by the absolute difference in strain response to the given environment) and most statistically significant cases of GxE were all for N-degron pathway reporters (Fig. 2A & C). N-degron 35 substrates require ubiquitin system targeting and, for two of the three studied substrates (Asn and Thr), pre-processing to produce functional N-degrons. The complex cascade of molecular events required to degrade these substrates may result in stronger GxE in the genetics of UPS activity towards N-degrons relative to the other substrates tested here. Some environments had consistent effects across reporters (Fig. 2B & C). For example, in LiAc, BY showed no significant change in UPS activity for any reporter, while RM showed at least nominally (T-test, p ≤ 0.04) significant reductions for all reporters. Following treatment with AZC, BY had higher UPS activity than RM for all reporters, either because AZC increased activity in BY while activity in RM was unchanged (Rpn4) or even reduced (Asn, 4xUb), or because RM experienced greater reductions in activity than BY (Phe, Thr). Other environmental effects were heterogeneous across UPS pathways. For example, glucose starvation led to a larger decrease in UPS activity in RM than in BY for the Thr N-degron reporter (p ≤ 5e-11, Fig. 2B), but showed the opposite pattern for UFD and 4xUb (p ≤ 4-e6), and even increased degradation of the Phe N-degron reporter in RM (p = 1e-5) with no change in BY (p = 0.25). Our results reveal previously-unappreciated complexities in the influence of strain background, substrate, and environment on UPS activity. Specifically, the effects of multiple environments commonly reported to consistently affect UPS activity were distinct, and in some cases discrepant, between strain backgrounds and UPS substrates (Fig. 2A-C). 36 Fig. 2: GxE in BY and RM across reporters and environments. A & B. Environment effects on UPS activity. Y-axis: the median UPS activity among replicates in SC was subtracted from that in the given environment to visualize environment effects. Negative values indicate that the environment caused a decrease in UPS activity compared to SC, and positive values indicate increased UPS activity. A value of zero means the environment did not affect UPS activity. Significant GxE terms (analysis of variance interaction term p-value < 0.05 after Bonferroni correction) are highlighted by opaque lines. Opaque points indicate a significant difference (T-test; Bonferroni-corrected p < 0.05) in UPS activity between the given environment and SC. A. Data organized by reporter. B. Data as in A, but reorganized by environment. C. Heatmap summarizing strain differences in environment effect. Diamonds indicate significant GxE (Bonferroni-corrected p < 0.05). D-F. Reporter / environment combinations that exhibited the most significant GxE, ranked by p-value of the interaction term in the linear model. Eight replicates were measured for each strain / reporter / environment combination. The center line of each box plot corresponds to the median of the eight replicates, with the lower and upper hinges showing the first and third quartiles, respectively. Whiskers extend to 1.5 times the interquartile range and lines connect the respective BY and RM medians. D. Phe N-degron reporter in 4NQO. E. Asn N-degron reporter in AZC. F. Thr N-degron reporter in low glucose. 37 Supplementary Fig. 1: A. Mean UPS activity across eight replicates for each environment and reporter. B. Proportion of tests with significant (gold) and non-significant (maroon) GxE, by reporter. C. Data as in C, but rearranged by environment. Heterogeneous genetic architectures of UPS activity across pathways and environments To identify genetic loci affecting UPS activity between BY and RM, we used a genetic mapping approach based on bulk segregant analysis (Albert et al., 2014; Brion et al., 2020; Ehrenreich et al., 2010; Michelmore et al., 1991) (Methods). Briefly, UPS activity was measured in large, genetically diverse cell populations 38 of haploid meiotic recombinant progeny (“segregants”) generated by mating RM with BY strains harboring the UPS activity reporters. We exposed two independent segregant populations derived from independent BY / RM matings to each of the eight environments and used fluorescence-activated cell sorting (FACS) to collect pools of segregants from the extreme tails of the UPS activity distribution. Sorted segregant pools were then whole-genome sequenced to determine BY and RM allele frequencies. Genome regions where pools with high and low UPS activity differ in allele frequency indicate quantitative trait loci (QTLs) that influence UPS activity (Fig. 1A). In the baseline SC condition, we identified 46 QTLs across the six UPS reporters. Four of these reporters (Rpn4, Asn, Phe, and Thr) were previously mapped in SC (M. A. Collins et al., 2022, 2023). To assess reproducibility, we compared QTLs identified here and in our previous studies. We observed high concordance with prior results in terms of the QTLs detected, the corresponding allele frequency differences, and the overall shape of the QTL traces (Supplementary Fig. 2). Of the 39 QTLs identified here for these four reporters, 30 were also seen in Collins et al. (2022 & 2023) (Supplementary Fig. 2E). The remaining nine QTLs had significantly lower LODs and effects sizes, as measured by the absolute allele frequency difference (T-test: p = 0.004, and 0.0001, respectively) (Supplementary Fig. 2F), suggesting that they may have been missed due to limited power. Thus, our approach represents a highly reproducible method for characterizing the genetics of UPS activity. 39 40 Supplementary Fig. 2: QTL reproducibility. A-D. QTL traces for the four reporters measured in SC in this study and (M. A. Collins et al., 2022, 2023). E. Table summarizing the number of QTLs that replicated between studies. F. QTLs that did not replicate between studies had significantly lower LOD scores absolute delta allele frequencies. T-test p-values are shown. Across the six UPS reporters and eight environments, we identified a total of 416 QTLs (Fig. 3A-C, Supplementary Fig. 4A). All 47 assayed reporter / environment combinations (Methods) had at least one QTL. The number of QTLs across environments and reporters ranged from one (4xUb in LiAc) to 19 (Asn in low glucose and Thr in low nitrogen) (Fig. 3D). Among reporters, the largest number of QTLs was found for the Thr reporter (n = 111) and the fewest for the 4xUb reporter (n = 30), when summing across all eight environments (Fig. 3B). The ubiquitin system-dependent pathways had 324 QTLs across the eight environments. Among these, the UFD pathway (Fig. 1B), which was not mapped in our previous studies, had five QTLs in the baseline SC condition (Fig. 3B, Supplementary Fig. 3A). The five QTLs for UFD in SC include a QTL on chromosome XII (peak position at 950,450) that was not seen for other reporters here or previously. In this region, the gene RPN13 encodes a subunit of the 19S regulatory particle of the proteasome that acts as a ubiquitin receptor. There are multiple BY-RM promoter and missense variants at RPN13, along with a strong cis-eQTL for this gene (Albert et al., 2018), suggesting RPN13 as a causal gene for this QTL. 41 https://www.zotero.org/google-docs/?20T4v0 The two ubiquitin system-independent reporters had 92 QTLs across environments (Fig. 3). The 4xUb reporter, which we had not assayed in previous work, had two QTLs in the baseline SC condition (n = 2), which is the fewest of all the reporters in SC (Fig. 3, Supplementary Fig. 3B). Both of these 4xUb QTLs were identified for other reporters (M. A. Collins et al., 2023). Specifically, the QTL on chromosome VII contains RPT6, which encodes an ATPase of the 19S regulatory proteasome particle. At a causal variant in the RPT6 promoter, the derived RM allele broadly increases the activity of multiple ubiquitin system-dependent and -independent UPS pathways by increasing RPT6 expression (M. A. Collins et al., 2023), suggesting that this variant also affects 4xUb. The QTL on chromosome XV was previously seen for the Rpn4 reporter (M. A. Collins et al., 2023). The causal gene in this QTL is likely IRA2, a gene that underlies a trans-eQTL hotspot that affects the expression of thousands of genes and numerous growth traits (Lutz et al., 2021; Smith & Kruglyak, 2008). While altered RPT6 expression underlying the QTL on chromosome VII likely affects UPS activity directly, coding variants in IRA2 (Lutz et al., 2021) that alter the activity of the Ira2 RAS signaling regulator likely affect UPS activity indirectly. 42 Fig. 3: UPS activity QTLs across reporters and environments. A. QTL mapping results for the six reporters across eight environments. Colored blocks denote genome bins that contain QTLs detected in each of two independent biological replicates, colored according to the direction and magnitude of the effect size, expressed as the RM allele frequency difference between high and low UPS activity pools. Genes in regions discussed in the text are indicated. No data were collected for 4xUb in low nitrogen (Methods). B. Number of QTLs found per reporter and environment. C. Data as in B, but rearranged by environment. D. Heatmap showing the number of QTLs per reporter / environment combination. 43 Supplementary Fig. 3: QTLs for 4xUb and UFD in SC. The plots show the loess-smoothed allele frequency difference between the high and low UPS activity pools across the genome for each of two independent biological replicates. Asterisks denote QTLs, defined by allele frequency differences that exceed an empirically-derived LOD score significance threshold in the given replicate. Horizontal black lines indicate QTLs that were present in both replicates. The dashed red horizontal lines denote an empirically-derived 99.9% quantile of the allele frequency difference. A. UFD reporter in SC. Five QTLs were present in both replicates. B. 4xUb reporter in SC. Two QTLs were present in both replicates. 44 Supplementary Fig. 4: UPS activity QTLs across environments and reporters. A. QTLs for each reporter / environment combination. Data as in Fig. 3A, but reorganized according to environment. Colored blocks denote genome bins that contain QTLs detected in each of two independent biological replicates, colored according to the direction and magnitude of the effect size, expressed as the RM allele frequency difference between high and low UPS activity pools. Candidate causal genes discussed in the text are indicated. No data was collected for 4xUb in low nitrogen. B. Barplot showing the number of times the BY or RM allele increased degradation in the 416 QTLs, by reporter. C. Data as in B, but reorganized by environment. 45 Collapsing combinations of pathways and environments into physiologically relevant categories revealed that ubiquitin system-dependent pathways had significantly more QTLs (median = 10) than ubiquitin system-independent pathways (median = 6; Wilcoxon test p-value = 0.004; Fig. 3B), indicating greater genetic complexity. The substrates of the ubiquitin system-dependent pathways must undergo binding and processing by various enzymes before being bound by the proteasome. The genes encoding this machinery provide additional targets for genetic variation, likely contributing to the higher number of QTLs observed for these pathways compared to ubiquitin system-independent pathways. Among environments and across all six reporters, most QTLs were found in low glucose (n = 76) and the fewest for AZC (n = 40) (Fig. 3C). The three starvation environments (low glucose, YNB, and low nitrogen) had significantly more QTLs (median = 10) than the non-starvation environments (median = 7) (Wilcoxon test p-value = 0.012). One potential explanation for these results is that starvation environments may have more wide-reaching, systemic effects on cellular physiology than the chemical stressors tested here. As a result, they may create more opportunities for variant effects that alter UPS activity through potentially highly indirect mechanisms. We also note that wild strains, such as RM, commonly undergo periods of nutrient deprivation, akin to the starvation environments tested here (Hong & Gresham, 2014; Wenger et al., 2011). Consequently, they may harbor genetic variation reflecting physiological adaptation to nutrient-poor environments, such as increased protein turnover for 46 amino acid recycling (Vabulas & Hartl, 2005). Consistent with this notion, the RM allele increased UPS activity more often than the BY allele overall (binomial test p = 0.04), and specifically for starvation environments (binomial test p = 0.04) (Supplementary Fig. 4B & C), congruous with Collins et al. (2022 & 2023). Many QTLs mapped to the same genomic locations (Fig. 3A). To quantify the number of unique locations, we counted the number of QTL peaks within 128 genomic bins of 100 thousand base pairs (kb) (M. A. Collins et al., 2022). The top four bins accounted for 28% (116 / 416) of the QTLs, illustrating that variation at a few locations underlies many of the effects on UPS activity. The bin on chromosome VII at 400 – 500 kb, which contains RPT6, harbored the most QTL peaks (n = 32, Fig. 3A). For all of these, the RM allele increased UPS activity, consistent with RPT6 as the causal gene in this bin. This region contains other genes involved in the UPS with sequence variation between BY and RM: SCL1, which encodes the alpha 1 subunit of the 20S proteasome, and RPN14, an assembly-chaperone for the 19S regulatory particle. Thus, this locus likely shapes UPS activity directly via RPT6, and potentially SCL1 and RPN14 as well. Three bins on chromosomes XIV, XII, and XV contained 31, 26, and 26 QTLs, respectively (Fig. 3A). The direction of effect of the QTLs in these bins depended on the pathway and environment (Fig. 3A). These bins contain the genes MKT1, HAP1, and IRA2, respectively, which are all known to harbor variation that results in hotspots that affect the expression of thousands of genes in trans (Albert et al., 2018). None of these genes has obvious connections to 47 UPS function: MKT1 encodes a poorly characterized protein that appears to bind certain mRNAs for genes with mitochondrial functions (Dimitrov et al., 2009; Wickner, 1987); HAP1 encodes a transcription factor that activates genes involved in osmotic stress (Gaisne et al., 1999); and IRA2 encodes a regulator of Ras signaling (Tanaka et al., 1990). Therefore, the wide-reaching effects of these three genes likely alter UPS activity via indirect mechanisms. Most QTLs were not specific to a single environment (Supplementary Fig. 4A), with two notable exceptions. First, is a locus on chromosome IV at 490 – 570 kb for LiAc (Supplementary Fig. 4A). Here, the BY allele was associated with higher UPS activity than the RM allele. This locus contains ENA1, ENA2, and ENA5, which all encode sodium pumps. ENA1 and ENA5 both contain multiple missense and frameshift variants between BY and RM, and the ENA locus harbors structural variation among yeast strains (Treusch et al., 2015; Warringer et al., 2011). Collectively, this variation likely leads to the QTLs seen here in the LiAc environment, where higher activity of BY ENA alleles may reduce LiAc concentrations in the cell compared to RM alleles, alleviating osmotic stress on the UPS. At the RPT6 locus on chromosome VII with QTLs in multiple environments, BTZ stood out in that the QTLs in BTZ had by far the largest effects at this locus (Supplementary Fig. 4A). This locus also contains PDR1, which encodes a transcription factor that regulates genes involved in the yeast pleiotropic drug response (Moye-Rowley, 2003), perhaps affecting cellular BTZ concentrations. Taken together, these QTL mapping results show that in addition 48 to being highly substrate-specific (M. A. Collins et al., 2022, 2023), genetic influences on UPS activity also depend on the environment. GxE arises from loci that affect the UPS through direct and indirect mechanisms The distinct patterns of QTLs across pathways and environments indicated a high degree of GxE in UPS activity. To identify specific loci whose effects depend on the environment, we classified QTLs into three categories based on pairwise comparisons between the baseline SC condition and individual environments. While these comparisons do not account for GxE among non-SC conditions, the patterns described here are broadly representative of other environment comparisons (Fig. 3). In our data, GxE at individual loci can be detected in two forms. First, a QTL may be detected in one environment but not in the other, which we refer to as “presence / absence” GxE. We defined such QTLs as those detected in both replicates of one environment and with no QTLs within 100 kb in either replicate of the other environment (based on QTL peak positions as in prior studies (M. A. Collins et al., 2022, 2023; Methods; Fig. 4A & B). By requiring a locus to be absent or present in both biological replicates, we focus on the strongest cases of presence / absence GxE, reducing the chance that a locus is actually present in both environments but happened to escape detection in one environment due to insufficient statistical power. Second, a QTL may be detected in both environments but with opposing directions of effect, which we call “sign change” GxE. Sign change GxE was defined as QTLs detected in both replicates 49 of one environment and at least one replicate of the other environment, with opposite effect direction (Fig. 4A & B). Fig. 4B displays QTL traces with examples of these two categories. Finally, QTLs within 100 kb of each other with the same direction of effect between environments were considered not to show GxE (Fig. 4B). We exclusively compared QTLs from the same reporter. A total of 507 comparisons between QTLs in different environments were categorized according to this scheme (Fig. 4C & D). All reporters and all environments had loci exhibiting GxE (Fig. 4C & D, Supplementary Fig. 5). Presence / absence GxE was seen for half (254) of the 507 comparisons (Fig. 4C & D). For the Asn N-degron reporter, presence / absence GxE mostly involved QTLs that were absent in SC but present in another environment (Fig. 4C). Conversely, for the Phe N-degron reporter, presence / absence GxE mostly involved QTLs that were present in SC but absent in the other environment (Fig. 4C). Sign change GxE was much more rare (17 / 507 comparisons, Fig. 4C & D), but was nonetheless seen for all reporters (Fig. 4C), and was observed in low nitrogen, AZC, and low glucose environments compared to SC (Fig. 4D). Ubiquitin system-independent pathways had a slightly larger proportion of QTLs with GxE than ubiquitin system-dependent pathways (Wilcoxon p = 0.046, Supplementary Fig. 5). There was no difference in the proportion of loci with GxE between starvation and non-starvation environments (Wilcoxon p = 0.82, Supplementary Fig. 5), as illustrated by the fact that both the fewest (YNB), and the most (low glucose) QTLs exhibiting GxE were seen for starvation conditions (Fig. 4D). These results show that half of the loci that shape 50 UPS activity are subject to GxE, primarily via presence / absence of a given locus in different conditions. Fig. 4: GxE at individual loci. A & B. Examples of pairs of QTLs that show no GxE, presence / absence GxE, and sign change GxE. A. Peaks (short vertical lines) of QTLs (horizontal bars; positions averaged across the two biological replicates) in the two compared conditions must be within 100 kb to be considered present in both environments. Color indicates direction of allelic effect as in Fig. 3A. Blue: BY allele increases UPS activity, red: RM allele increases UPS activity. B. QTL traces for the Thr N-degron reporter in SC (top) and in low nitrogen (bottom). Boxes highlight examples of the three categories of pairwise comparisons of loci. C. Pairwise comparisons of loci between SC and other environments, across reporters. Light gray indicates comparisons where a QTL was present in both replicates of one environment and only one replicate of the other environment, with the same direction of effect. D. Data as in C, but arranged by environment. 51 We examined the loci of QTL comparisons showing GxE using genomic bins as above, and observed a non-uniform distribution (Fig. 5A). Almost all (71/78, 91%) bins that contained any QTLs contained loci that exhibited GxE (Fig. 5A & B). The five bins with the most GxE cases contained 23% (63 / 271) of all GxE cases, showing that a small portion of the genome harbors much of the GxE seen at individual QTLs (Fig. 5A). Only two of these top five bins contained candidate genes clearly related to the UPS: the bin containing RPT6, with 10 GxE cases, and a bin at position 0 – 100 kb on chromosome XIII that contained 14 cases of GxE across all six reporters (Fig. 5A). The latter bin contains the BUL2 gene, which encodes a component of the Rsp5p E3-ubiquitin ligase complex. Two of the top GxE bins correspond to the trans-eQTL hotspots at IRA2 (15 GxE cases) and HAP1 (11 GxE cases) (Fig. 5A). A second bin on chromosome XIII, at 300 – 400 kb, had 13 GxE cases spanning all six reporters 52 Supplementary Fig. 5: A heatmap showing the proportion of locus comparisons that showed GxE out of all comparisons, for combinations of reporters and environments. No data was collected for 4xUb in low nitrogen. and did not contain an obvious candidate gene. Overall, QTLs with cases of GxE tended to be located in bins that also contained a large number of QTLs of any type (Spearman correlation across 78 bins with any QTLs: rho = 0.73, p = 5e-14; Fig. 5B). These data show that loci subject to GxE occur throughout the genome but are clustered in regions with many QTLs, including those caused by variation in core UPS genes as well as indirect, pleiotropic regulators. Fig. 5: Patterns of QTLs with GxE across the genome. A. Distribution of QTL comparisons that exhibited GxE in 100 kb bins. Shown are 71 of 128 bins that contained QTL comparisons with GxE. Bins are sorted based on the number of GxE cases they contain, followed by genomic position to break ties. Candidate causal genes are indicated. B. A comparison of the number of total QTLs and of QTL comparisons exhibiting GxE for each of 78 genomic bins that contained at least one QTL. Spearman correlation: rho = 0.73, p = 5e-14. C. Locus plot showing five QTLs involved in three sign changes (one for UFD, two for 4xUb) in the bin with the most cases of GxE. Genes in this region are indicated, with IRA2 highlighted by the arrow. QTL confidence intervals are shown as horizontal bars and peaks are indicated by the small rectangle within each QTL. QTL colors indicate direction and strength of effect as in Fig. 3A. The QTLs for 4xUb in SC and AZC extend leftwards to position 124,750 and 112,300, respectively. 53 Sign change is a particularly interesting form of GxE, in which the same locus results in opposite effects on UPS activity depending on the environment. The 17 cases of QTL comparisons with sign change GxE involved 29 unique QTLs. Examination of the peaks and confidence intervals of these 29 QTLs revealed that they clustered at seven genomic regions (Fig. 5C and Supplementary Fig. 6A-F). Of these regions, three corresponded to the trans-eQTL hotspots at HAP1 (7 QTLs involved in sign changes, Supplementary Fig. 6A), IRA2 (5 QTLs, Fig. 5C), and MKT1 (3 QTLs, Supplementary Fig. 6B). These genes have no obvious direct connections to the UPS. Two regions had no candidate genes with obvious UPS functions: a region from 280 – 470 kb on chromosome XIII with six sign change QTLs (Supplementary Fig. 6C), and a sign change pair (for Rpn4 in AZC), which was located ~98 kb from UBC6 (Supplementary Fig. 6D). While Collins et al. (2022) identified UBC6 as a causal gene affecting degradation of the Thr N-degron reporter, this gene encodes the E2 ubiquitin-conjugating enzyme of the Ac/N-degron pathway (Varshavsky, 2024). Its function in the ubiquitin system and its fairly large distance from the sign change pair makes it unclear if UBC6 is a causal gene for the ubiquitin system-independent Rpn4 reporter, with no other obvious candidate genes in this region. The remaining two regions with sign change QTLs did contain likely causal genes with direct UPS functions: one sign change pair (for the UFD pathway in AZC) at RPT6 (Supplementary Fig. 6E), and four such QTLs at BUL2 (Supplementary Fig. 6F). Notably, none of the remaining causal genes that we previously determined to shape UPS activity (UBR1, UBC6, NTA1, and DOA10; 54 M. A. Collins et al., 2022) had sign changes between SC and other environments, even though they did show presence / absence GxE. All of these genes encode core UPS components. Thus, loci with genes that may affect UPS activity in a direct fashion (RPT6 and BUL2) accounted for only 21% (6 / 29) of the QTLs involved in sign change GxE. Most of the sign change QTLs (52%, 15 / 29) appear to arise from genes that shape UPS activity indirectly. In sum, these results suggest that GxE in the UPS, especially sign change GxE, is mostly caused by indirect mechanisms, such as widespread changes in gene expression due to trans-eQTL hotspots, rather than by variation in core genes directly involved in the UPS. 55 Supplementary Fig. 6: Locations of QTLs within sign change pairs: Locus plots of the 17 cases of sign change GxE (along with Fig. 5C). The confidence interval (horizontal bars) and peak position (short vertical lines) of the 29 unique QTLs involved in sign changes are shown along the S. cerevisiae genome (sacCer3) using UCSC Genome Browser (Nassar et al., 2023). For the QTLs, colors indicate direction and strength of effect as in Fig. 3A. Candidate causal genes are denoted with black arrows. 56 https://www.zotero.org/google-docs/?upR0pv Discussion To characterize GxE in the genetics of protein degradation, we measured UPS activity towards six distinct substrates in single cells of two strains of S. cerevisiae and their progeny across eight environments. GxE was pervasive between the two strains. The activity of every measured pathway was modified by the environment, and all of the tested environments led to GxE. The BY and RM strains differed greatly in how they responded to a given environment. Remarkably, UPS activity increased in one strain but decreased in another for some combinations of reporter and environment. Previous studies of the UPS using some of the environments studied here were based on lab strains, suggesting that some published treatment effects may not represent the S. cerevisiae species as a whole. Strain-dependency of treatment effects has been widely documented, including between closely related strains of mice (Simon et al., 2013) and yeast (Elserafy & El-Khamisy, 2018; Matheson et al., 2017), and our results reinforce the value of studying physiological effects in multiple genetic backgrounds including those that have evolved in different environments. The interaction of genetics and environment was apparent for loci shaping UPS activity. All reporter / environment combinations had unique QTL patterns, and about half of the QTL comparisons we conducted revealed evidence of GxE. Presence of a given QTL in one but not another environment was by far the predominant form of GxE, making up 94% of the detected GxE cases. In spite of the large number of QTLs showing GxE, almost no QTLs were entirely specific to 57 a particular environment, with the ENA locus in LiAc as the closest exception. Most QTLs were seen in several, but not all environments. Thus, the distinct QTL patterns for specific pathway / environment combinations were formed from subsets of the total set of QTLs identified across the entire study. The QTLs we identified for different environments and pathways, as well as QTLs with GxE, tended to be clustered at certain genomic locations, as seen in our previous work on the UPS (M. A. Collins et al., 2022, 2023) and reflecting work on GxE in yeast gene expression (Smith & Kruglyak, 2008) and in other systems such as flowering time in A. thaliana (Sasaki et al., 2015). GxE tended to be seen in regions that had many QTLs overall (Fig. 5B). Thus, when genetic variation leads to changes in UPS activity, it is likely that this variation will be subject to GxE in some environments. A genome region with numerous QTLs and cases of GxE contained RPT6, which we earlier showed to affect the Rpn4 reporter as well as the Ac/N-degron pathway (M. A. Collins et al., 2022, 2023). Here, the RPT6 locus affected all six assayed reporters in at least one condition. Thus, the causal variant in the RPT6 promoter appears to have wide-reaching effects on UPS activity, with potential downstream effects on protein degradation. RPT6 can be considered a “core” gene under an omnigenic model of complex traits (Boyle et al., 2017), in which core genes encode proteins that are directly related to the given trait, while “peripheral” genes act as indirect regulators that shape complex traits through trans-acting effects on core genes. All core UPS genes we 58 previously showed to cause UPS activity variation (UBR1, UBC6, NTA1, DOA10) had QTLs and presence / absence GxE in this study. Many individual QTLs, QTLs with GxE, and in particular sign change QTLs, occurred at genes known to cause trans-eQTL hotspots, where variation at a single gene affects the expression of hundreds or thousands of genes throughout the genome (Albert et al., 2018; Brem et al., 2002; Smith & Kruglyak, 2008; Zhu et al., 2008). These loci exercise their broad effects through indirect mechanisms in which they alter cellular states that in turn affect many downstream traits, including growth in diverse conditions (Bloom et al., 2013; Renganaath & Albert, 2023). Their many pleiotropic effects include UPS activity (M. A. Collins et al., 2022), for which they can be interpreted as “peripheral” genes under an omnigenic model. Our results here show that these hotspots are also focal points for GxE in UPS activity. Their indirect mode of action likely presents numerous molecular steps before reaching the UPS, offering opportunities for a given environment to change how their effects are propagated. This stands in contrast to variation at core UPS genes, which was less prone to GxE, in particular sign changes. The direct effects of core genes on the UPS may be less responsive to environmental influences compared to the more indirect, pleiotropic hotspot modulators. Our study had several limitations. Because the bulk-segregant mapping design we employed makes it difficult to rigorously detect differences in magnitude of locus effect in the same direction, we did not search for such loci 59 even though magnitude GxE could be prevalent (Cubillos et al., 2014; Smith & Kruglyak, 2008). As such, our GxE QTL results are conservative in that they only search for extreme cases in which a QTL is present in only one condition or switches sign. Due to linkage in the segregant population, a sign change QTL could reflect two presence / absence loci in close proximity. Only experimental determination of causal genes can ultimately rule out this possibility, although we note that GxE in gene expression has previously been shown to arise from variation in the single IRA2 gene (Smith & Kruglyak, 2008). Future work will elucidate the causal genes in the loci identified here and reveal how they cause GxE. In conclusion, our results show that the genetic architecture affecting UPS activity is complex, pathway-specific, and subject to a considerable degree of modulation by the environment. Different UPS pathways affect the degradation of distinct substrates, different environments challenge proteostasis in different ways, and genetic variants differ in how they affect a given pathway in a specific environment. The extent of GxE shown here is similar to the amount of GxE seen previously in yeast for transcript abundance and growth in various environments. Thus, GxE is an important factor in yeast complex traits, with important implications for predicting phenotype from genotype. 60 Methods UPS Activity Reporters To measure ubiquitin-proteasome system activity, we used tandem fluorescent protein timers (TFTs; (Khmelinskii et al., 2012), a two-color fluorescent reporter system that provides high-throughput measurements of protein turnover. TFTs are linear fusions of two fluorescent proteins. In the most common implementation, the TFT consists of a faster-maturing green fluorescent protein (GFP) and a slower-maturing red fluorescent protein (RFP). Because the two fluorophores fold and emit fluorescence over distinct time scales, the ratio of RFP to GFP changes over time. If the TFT’s degradation rate is more rapid than the RFP’s maturation rate, the TFT ratio can be used to measure the construct’s degradation rate (Khmelinskii et al., 2012; Khmelinskii & Knop, 2014). We used superfolder green fluorescent protein (sfGFP) as the GFP in all TFTs and mCherry or mRuby as the RFP as indicated below. As in prior studies, we inserted an unstructured 35 amino acid sequence (GSGSREARHKQKIVAPVKQTLNFDLLKLAGDVESN) between the fluorophores to minimize fluorescence resonance energy transfer (M. A. Collins et al., 2022, 2023; Khmelinskii et al., 2012). To relate the output of our TFT reporters to UPS activity, we utilized degrons that engage distinct UPS pathways to our TFTs. The resulting constructs provide quantitative, high-throughput, pathway-specific readouts of UPS activity in live, single cells that are sensitive to genetic and chemical perturbations that 61 alter UPS activity (M. A. Collins et al., 2022, 2023). We measured the activity of 4 ubiquitin system-dependent UPS pathways. Three of these were N-terminal amino acids that engage distinct branches of UPS N-degron pathways, in which a protein’s N-terminal amino acid functions as a degron (hereafter, an “N-degron”; Varshavsky, 2019). Our N-degron TFT reporters include the Thr N-degron of the Ac/N-degron pathway, the Asn N-degron of the type I Arg/N-degron pathway, and the Phe degron of the type II Arg/N-degron pathway. We also added a non-cleavable ubiquitin moiety (ubiquitin G76V) to the N-terminus of the sfGFP / mCherry TFT. The resulting construct measures the activity of the ubiquitin-fusion domain (UFD) pathway, a UPS pathway involved in protein quality control (Devarajan et al., 2020; Johnson et al., 1995; Theodoraki et al., 2012). We also created two reporters that measure the activity of ubiquitin-system independent UPS pathways. One reporter contains the ubiquitin-independent degron encoded in the N-terminal 80 amino acids of the Rpn4 protein (hereafter, the “Rpn4 degron”). The second reporter contains a linear chain of 4 ubiquitin molecules (hereafter, the “4xUb reporter”). Each of the ubiquitins in 4xUb have the G76V substitution so they cannot be cleaved, and K29/48/63R substitutions to prevent further ubiquitination. Both of the Rpn4 degron and 4xUb reporters are directly bound and degraded by the proteasome without targeting by the ubiquitin system. Their activities thus provide a readout of proteasome activity independent of the ubiquitin system. However, they are each bound by distinct proteasome receptors. Based on the half-lives of the N-degrons (Bachmair et al., 1986; Varshavsky, 2011), we used mRuby, which has 62 a longer half-life than mCherry (168 vs. 40 minutes; Kredel et al., 2009; Shaner et al., 2004) for the Thr N-degron reporter to improve its dynamic range. For all other reporters, mCherry was used as the RFP. The 4xUb and UFD reporters were engineered for this study using procedures described in (M. A. Collins et al., 2022, 2023). We packaged each reporter into plasmid backbone BFA0190 (M. A. Collins et al., 2022, 2023) containing common sequence elements for reporter integration, selection, and expression. Each reporter contains the TDH3 promoter to drive strong, constitutive TFT expression, the ADH1 terminator, codon-optimized sfGFP and mCherry or mRuby, and a KanMX cassette to select for presence of the reporter via resistance to the antibiotic G418. These elements are flanked by sequences homologous to the genomic regions immediately up- and downstream of LYP1. Transformation of the reporter containing these flanking sequences results in integration at the LYP1 locus, which can be selected for using the toxic amino acid analogue thialysine. DNA fragments of the reporters used for transformations were made by PCR amplifying the sequence on the plasmid carrying the reporter sequence. The PCR fragments were purified using Monarch® PCR & DNA Cleanup Kit (5 μg) from New England Biolabs (NEB) Cat#T1030L, or ran on an electrophoresis gel and purified using the Monarch DNA Gel Extraction Kit (NEB) Cat#T1020L, according to the manufacturer’s protocol. 63 Yeast Strains and Handling All experiments used yeast strains derived from two genetically divergent Saccharomyces cerevisiae strains. The haploid BY strain (genotype: MATa his3Δ hoΔ) is closely related to the S288C laboratory strain. The haploid RM strain (genotype: MAT𝛼 can1Δ::STE2pr-SpHIS5 his3Δ::NatMX AMN1-BY hoΔ::HphMX URA3-FY) is derived from a wild strain that was originally isolated from a California vineyard. To characterize UPS reporters, we also used a previously characterized BY strain lacking the RPN4 gene (genotype: MATa his3Δ hoΔ rpn4∆::NatMX) (M. A. Collins et al., 2022). We built strains harboring our UPS activity reporters using the following procedures. Transformations using the reporter sequence fragments were performed using the Zymo Frozen-EZ Yeast Transformation II™ Kit Cat#T2001 according to the manufacturer’s protocol or the lithium acetate / single-stranded carrier DNA / poly-ethylene glycol (PEG) method (Gietz & Schiestl, 2007) as described in (M. A. Collins et al., 2022, 2023). We verified the presence of the reporters in the desired locus with colony PCR. Eight confirmed transformants for each reporter in each strain were collected as independent biological replicates. Yeast mating and segregant populations To create large, genetically diverse cell populations for genetic mapping, we used a previously-described approach (Albert et al., 2014; Brion et al., 2020; Ehrenreich et al., 2010). We created segregant populations containing each UPS activity reporter using a modified synthetic genetic array methodology (Baryshnikova et al., 2010; Kuzmin et al., 2016). Briefly, BY strains (MATa) 64 containing each reporter were mixed with a wild-type RM strain (MAT𝛼, without a reporter: YFA0039) on solid YPD medium and grown overnight at 30°C. This was done independently for two biological replicates based on two different colony PCR-confirmed transformants for each reporter. Diploids from the mating were selected on YPD plates containing G418 and CloNAT to select for the UPS reporter in the BY strain and his3Δ::NatMX in the RM strain, respectively. Five mL of liquid YPD was inoculated with the diploids and grown overnight to saturation at 30°C with rolling. The diploids were then spun down in 15 mL tubes in a tabletop centrifuge at 3000 rpm for two minutes. The cell pellet was resuspended in five mL of sporulation medium and transferred to glass tubes and incubated at room temperature for 10 days on a turning wheel. We evaluated the extent of sporulation in each culture using brightfield microscopy. When the cultures reached approximately 80% sporulation, we harvested the spores. To separate the spores from their asci, we spun the spores for 1.5 minutes at 5000 rpm in a tabletop centrifuge, discarded the supernatant, resuspended in water with 1 mg / mL Zymolyase lytic enzyme (United States Biological, Salem, MA, USA), and incubated for two hours, vortexing every half hour. We again washed the cells and plated them onto solid haploid selection medium with G418 and thialysine. We used this medium to select for recombined haploid cells (“segregants”) that contain the reporter via G418, the MATa mating type locus via the Schizosaccharomyces pombe HIS5 gene under the control of the STE2 promoter (which is only active in MATa cells), and replacement of the LYP1 gene by the reporter via resistance to thialysine. We grew the resulting segregant 65 populations for two days on haploid selection plates at 30°C, harvested the cells from the plates, and stored each population as a separate glycerol stock. We saved two biological replicates for each reporter as separate, individual stocks. Environments To characterize how genetic influences on the UPS are shaped by environmental factors, we measured UPS activity in eight distinct media formulations, which we term “environments.” As a baseline environment, we used synthetic complete (SC), a nutrient-rich medium with glucose, nitrogen, and amino acids. G418 (200 mg/mL) was added to all environments to maintain the reporter sequence in the genome. We compared UPS activity in SC (SC -His -Lys + YNB + 0.1% MSG + 2% glucose) to that in seven different environments: low glucose (SC -His -Lys + YNB + 0.1% MSG + 0.025% glucose), low nitrogen (YNB + 2% glucose), yeast nitrogen base (YNB + 0.1% MSG + 2% glucose), SC + 4NQO (4NQO; 2 µg/mL), SC +L-azetidine-2-carboxylic acid (AZC; 4mM), SC + bortezomib (BTZ; 40µM), and SC + lithium acetate (LiAc; 20 mM). Growth and environmental exposures prior to flow cytometry Eight biological replicates of each of the BY, RM, and rpn4∆ strains containing the reporters were grown overnight to saturation in SC medium. From a common saturated sample for each replicate, 4 µL was used to inoculate 400 µL media in 96 well plates for each environment. G418 (200 mg / mL) was added to all media except for the negative controls. The cultures were incubated at 30°C on a MixMate (Eppendorf, Hamburg, Germany) at 1100 rpm and incubation 66 times in each medium were determined based on a combination of previous literature (Burgis & Samson, 2007; Laporte et al., 2008; J. Li et al., 2019; Marshall et al., 2016; Waite et al., 2016; Work & Brandman, 2021) and preliminary growth rate measurements in a plate reader to ensure that all cultures had similar optical density (O.D.) for flow cytometry and FACS. If cultures showed no growth in a given environment in preliminary experiments, we grew the cells in SC before exposing the cells to that environment (see below for details). For SC, LiAc, and YNB conditions, 400 µL of medium was inoculated with 4 µL of the overnight growth and incubated for 3 hours prior to flow cytometry. For the BTZ samples, 400 µL of SC + BTZ medium was inoculated with 4 µL of the overnight growth culture, incubated for 4 hours until flow cytometry measurements. For samples that were to be exposed to low nitrogen, low glucose, 4NQO, and AZC, 4 µL of the overnight culture was added to 400 µL of SC and grown for 3 hours. After those 3 hours, samples to be exposed to low nitrogen and low glucose were spun down for 5 mins at 3000 rpm in 96 well plates on a tabletop centrifuge. The SC medium was replaced with either the low nitrogen or low glucose media and incubated for 24 hours until measured via flow cytometry. If the low nitrogen or low glucose samples were too dense for flow cytometry, they were diluted in a 1:3 ratio with the same media type. For the 4NQO samples, after the 3 hours of growth in SC, the cells were spun down for 5 mins at 3000 rpm in 96 well plates on a tabletop centrifuge. The SC media was replaced with SC + 4NQO medium and incubated for one hour until measured via flow 67 cytometry. For the AZC samples, after the 3 hours of growth in SC, 3.2 µL of AZC stock was added to each well. After 5 hours of incubation, samples were measured via flow cytometry. For all flow cytometry and cell sorting experiments, we used measurements of the wild-type BY strain (YFA0040) without a TFT reporter grown in SC only to determine background fluorescence levels. When the GFP signal of a TFT did not exceed that of YFA0040 in a given environment, we concluded that the UPS activity measured by the associated reporter could not be accurately measured in that environment. A 400 µL sample of SC -lys was inoculated with 4 µL of the overnight growth of YFA0040 and incubated for 3 hours prior to flow cytometry. Four of the UFD rpn4∆ replicates did not produce GFP fluorescence above the negative control (BY without a reporter). We therefore excluded those four samples without detectable GFP from analysis. Flow cytometry All flow cytometry experiments were performed on the BD FACSymphonyTM A3 flow cytometer (BD, Franklin Lakes, NJ, USA) at the University of Minnesota University Flow Cytometry Resource. The cytometer is equipped with a 20 mW 488 nm laser with a 488 / 10 filter to measure forward scatter (FSC) and side scatter (SSC) and a 525 / 50 filter to measure GFP fluorescence, and a 40 mW 561 nm laser with a 610 / 20 filter to measure RFP fluorescence. We altered voltages for samples containing AZC compared to the other conditions so that GFP did not saturate at the upper end of detection, as 68 GFP fluorescence was higher in those samples, as follows: FSC, 450; GFP, 400; RFP, 600. We used flow cytometry to characterize the two new reporters, 4xUb and UFD, in SC medium. We recorded data for 10,000 cells from each of the eight biological replicates of each strain (BY, RM, rpn4∆) containing these reporters. We used flow cytometry to test for genotype-by-environment interactions (GxE) between the BY and RM strains for all six reporters. For these experiments, we recorded data for 20,000 cells from the eight biological replicates of BY and RM strains containing the six reporters in each of eight environments as described above. Analysis of flow cytometry data Analyses were conducted using code adapted from (M. A. Collins et al., 2022). Briefly, we analyzed flow cytometry data using the R (R Foundation for Statistical Computing, Vienna Austria) package flowCore (Hahne et al., 2009). We first filtered each replicate to include only cells within 10% + / - the FSC median (proxy for cell size). This removed cellular debris, aggregates of multiple cells, and restricted our analyses to cells of the same approximate size. The low nitrogen samples showed two FSC peaks, perhaps due to incomplete budding of daughter cells. In order to only analyze single cells, we selected the smaller of the two peaks for analysis of the low nitrogen samples. The median FSC values of the smaller low nitrogen peaks were similar in size to the FSC medians of all other samples. 69 As in prior studies (Brion et al., 2020; M. A. Collins et al., 2023), we observed that the -log2(RFP / GFP) ratio changed over time within some replicates of the same strain, reporter, and environment. To correct for this, we used the residuals of a loess regression of the -log2(RFP / GFP) ratio, as in (Brion et al., 2020; M. A. Collins et al., 2023). We refer to the time-corrected -log2(RFP / GFP) ratio as “UPS activity” throughout. P-values for differences in UPS activity between the BY, RM, and rpn4∆ strains were calculated using a two-tailed T-test. The GxE effect of each reporter / environment combination was determined using the following linear mixed model: model = UPS activity ~ strain * environment + (1 | replicate) Here, the random effect term “(1 | replicate)” accounts for inter-individual variation among independent biological replicates. We conducted pairwise analyses, in which we compared the effect of each of the seven environments to the baseline SC environment, for one reporter at a time. If one of the eight replicates for a given reporter in a given environment had a median GFP level below that of the negative control (a BY strain with no reporter), all eight replicates for that reporter / environment combination were excluded. This exclusion applied to the 4xUb and UFD reporters in low nitrogen, resulting in 40 tests for GxE. Statistical significance of main effects of strain, environment, and the GxE interaction term between strain and environment was assessed using an ANOVA. We used Bonferroni-corrected p-value thresholds of 0.05 / 40 = 70 0.00125. The magnitude of the environmental effect on UPS activity (Fig. 2A & B) was determined by subtracting the median UPS activity in the given environment from the median UPS activity in SC. To describe GxE patterns revealed by the linear models, we also used T-tests comparing UPS activity for each strain between SC and another condition. For the T-tests, we used Bonferroni-corrected p-value thresholds of 0.05 / 80 = 0.000625. Growth and environmental exposures prior to fluorescence-activated cell sorting (FACS) All incubations were performed at 30°C in glass test tubes with rolling. Two independent biological replicate segregant stocks were thawed and used to inoculate our baseline medium (SC -his -lys + G418) and grown overnight to saturation. This common culture was used to inoculate media for each environment as follows. For SC, LiAc, and YNB, 4.2 mL of media was inoculated with 800 µL of the overnight growth culture and incubated for 3 hours until FACS. For the BTZ samples, 4.5 mL of SC + BTZ medium was inoculated with 500 µL of the overnight growth culture and incubated for 4 hours until FACS. For samples that were exposed to low nitrogen, low glucose, 4NQO, and AZC, 500 µL of the overnight culture was added to 4.5 mL of SC and grown for 3 hours. Samples exposed to low nitrogen and low glucose were then spun down for 2 mins at 3000 rpm in 15 mL tubes on a tabletop centrifuge. The SC medium was replaced with 5 mL of the low nitrogen or low glucose media and incubated for 24 hours until cell sorting. For the 4NQO samples, after the 3 hours of growth 71 in SC, 0.4 µL of 4NQO stock was added to the cultures. The 4NQO cultures were immediately vortexed and incubated for one hour until cell sorting. For the AZC samples, after the 3 hours of growth in SC, 40 µL of AZC stock was added to each sample and vortexed. AZC samples were incubated for 5 hours until FACS. These volumes and incubation times led to the samples being at approximately the same O.D. when sorted. A BY strain without a UPS reporter (YFA0040) was grown overnight to saturation in SC -lys to be used as a negative fluorescence control. A 4.5 mL volume of SC -lys was inoculated with 500 µL of the overnight growth and incubated for 3 hours until FACS. FACS We used FACS to isolate phenotypically extreme cell populations as part of a bulk segregant analysis genetic mapping approach (Albert et al., 2014; Brion et al., 2020). All cell sorting was performed on a FACSAria II cell sorter (BD) by the University of Minnesota Flow Cytometry Resource. To remove doublets from each sample, we used plots of SSC height by width and FSC height by width. We kept cells within the peak of FSC area + / - 7.5%, which maintained our primary haploid cell population and excluded cellular debris and aggregates (M. A. Collins et al., 2022, 2023). We restricted our sorts to populations of cells with GFP fluorescence above that of the negative control BY strain YFA0040, which does not express any fluorescent proteins. We collected populations of cells from the 2% high and low tails of the RFP / GFP ratio distribution. We aimed to collect pools of 20,000 cells for each of these populations. When cultures did not contain 72 a sufficient amount of GFP-positive cells to collect 20,000 cells, we collected fewer cells. We empirically determined reporter / environment combinations for which the cell pools did not grow well after sorting in a preliminary experiment. We therefore collected more cells for those samples when we sorted the cells used in the downstream analyses, up to 100,000 cells.. The segregants with the 4xUb reporter in the low nitrogen environment did not produce fluorescence above the negative control, and therefore no cells were collected for that combination, and all downstream analyses do not include 4xUb in low nitrogen. Cells were collected into sterile 1.5 mL polypropylene tubes with 1 mL of SC -his -lys medium and grown at 30°C with rolling at least 26 hours or until saturation. We added 1 mL of each culture to a 96-well plate with 600 µL of 40% glycerol and stored at -80°C for subsequent genomic DNA extraction. DNA extraction and library preparation We isolated genomic DNA from thawed glycerol stocks of the sorted segregant pools for whole-genome sequencing. We centrifuged 800 µL of each pool at 3700 rpm for 10 minutes to pellet the cells and discarded the supernatant. To digest cell walls we resuspended the cells in 800 µL of 1 M sorbitol, 0.1 M EDTA, 14.3 mM β-mercaptoethanol, and 500 U of Zymolyase lytic enzyme (United States Biological) and incubated for 2 hours at 37°C on a MixMate at 1100 rpm. We re-pelleted the cells, removed the supernatant, and extracted DNA from the cells using the Quick-DNA 96 Plus kit (Zymo Research, Irvine, CA, USA), according to the manufacturer’s instructions, including an overnight protein digestion in 20 mg / mL of proteinase K solution. We eluted the DNA using 35 µL 73 of DNA Elution Buffer (10 mM Tris-HCl [pH 8.5], 0.1 mM EDTA) and determined DNA concentration on a Synergy H1 plate reader (BioTek Instruments, Winooski, VT, USA) in 96-well plates using the Qubit dsDNA BR assay kit (Thermo Fisher Scientific, Waltham, MA, USA). We prepared the genomic DNA for short-read whole-genome sequencing on the Illumina NovaSeq platform using a previously established approach (Albert et al., 2014; Brion et al., 2020; M. A. Collins et al., 2022, 2023). We used the Nextera DNA library kit (Illumina, San Diego, CA, USA) in 96 well plates according to the manufacturer’s instructions, except that we used a 1:20 dilution of the Tagment DNA enzyme in Tagment DNA buffer. After library generation, we quantified the DNA concentration of each sample using the Qubit dsDNA BR assay kit (Thermo Fisher Scientific). For each 96 well plate, 10 µL of each sample was pooled and 1 mL of that pool was run in a large well on a 2% agarose gel. We extracted and purified the DNA in the 400 to 600 bp region using the Monarch Gel Extraction Kit (NEB) according to the manufacturer’s instructions. The DNA from each of the gel extractions was quantified and pooled in equimolar amounts, and submitted for sequencing. The University of Minnesota Genomics Center (UMGC) staff performed quality control assays as described in (M. A. Collins et al., 2022, 2023) on the pooled library before sequencing. Briefly, library concentration was determined using PicoGreen dsDNA quantification reagent (Thermo Fisher Scientific), library size was determined using the Tapestation electrophoresis system (Agilent Technologies, Santa Clara, CA, USA), and library functionality was determined 74 using the KAPA DNA Library Quantification kit (Roche, Penzberg, Germany). The submitted pooled library passed each quality control assay. The pooled library was sequenced on the Illumina NovaSeq with 150 bp paired end reads. Of the samples used for analysis, the median reads produced per sample was 1,692,391, with the minimum being 304,256 reads and the maximum being 5,874,865 reads. UMGC performed sequence data de-multiplexing. Genetic mapping QTLs were determined using an established approach for bulk segregant analysis (Albert et al., 2014; Brion et al., 2020; Ehrenreich et al., 2010; Michelmore et al., 1991). Code from (M. A. Collins et al., 2022, 2023) was used to calculate allele frequencies via the following pipeline. From our whole-genome sequencing reads, we aligned reads to the S. cerevisiae reference genome (version sacCer3) using the BWA "mem" command (H. Li & Durbin, 2009) and retained alignments with a mapping quality score above 30. Using samtools (H. Li et al., 2009), we retained uniquely aligned reads and removed PCR duplicates (command: "samtools markdup -S"). VCF files with allelic read counts at 18,871 high-confidence, reliable SNPs (Bloom et al., 2013; Ehrenreich et al., 2010) were produced using the command: samtools mpileup -vu -t INFO / AD -l. We used adapted code from (M. A. Collins et al., 2022, 2023) to calculate allele counts from the VCF files. Briefly, we excluded variants with allele frequencies lower than 0.1 or higher than 0.9 as in (Albert et al., 2014; Brion et al., 2020). We used MULTIPOOL (Edwards & Gifford, 2012) to estimate logarithm of the odds (LOD) scores comparing a model in which the high and low 75 degradation activity pools come from one population to a model in which these pools come from two different populations with different allele frequencies. As in (M. A. Collins et al., 2022, 2023), we used the following MULTIPOOL settings: bp per centiMorgan = 2,200, bin size = 100 bp, effective pool size = 1,000. We called QTLs as loci with a LOD ≥ 4.5. Previous work has shown that this threshold produces a 0.5% false discovery rate for genetic mapping by bulk segregant analysis using TFT reporters (M. A. Collins et al., 2022). Confidence intervals (CI) for each significant QTL were determined using MULTIPOOL and defined as a 2-LOD drop from the position in the QTL interval with the highest LOD score, which we defined as the QTL peak position. We calculated the RM allele frequency difference (∆AF) between the high and low degradation activity pools using a smoothed allele frequency via a loess regression to account for random counting noise at individual sequence variants. In our scheme, a positive ∆AF indicates that the RM allele of a QTL is associated with higher UPS activity. The loess smoothed values were used for plotting and determining QTL effect sizes. QTLs were called separately for each biological replicate. To determine QTLs that were detected in both biological replicates, we used a previously described approach (M. A. Collins et al., 2022, 2023). QTLs present in both replicates were defined as QTLs on the same chromosome with peaks within 100 kb and with the same effect direction (∆AF sign). From our set of 694 QTLs across replicates, 416 QTLs (60%) were present in both replicates, with the remaining 278 QTLs detected in only one replicate. For QTLs present in both 76 replicates, the left and right CI positions, peak position, LOD, and ∆AF were averaged and used for downstream analyses. For QTLs found in only one replicate, the left and right CI positions, peak position, LOD, and ∆AF were used without alteration. The high and low populations of the second biological replicate of Phe in 4NQO did not have sufficient sequencing coverage to call QTLs. To replace these populations, we used additional populations we had collected during FACS of Phe in 4NQO from the first biological replicate. These additional populations consisted of cells with RFP / GFP ratios in the 3% to 5% area of the distribution. Therefore, both replicates used in data analysis of Phe in 4NQO came from the same original segregant pool. Comparison of QTLs from previous studies QTLs for the Asn, Phe, Rpn4, and Thr reporters had been mapped previously in the same standard SC medium (M. A. Collins et al., 2022, 2023). For these four reporters, we analyzed 39 QTLs that were present in both replicates in SC in the present study and asked whether they were also present in at least one of the two replicates from the previous studies. If a QTL from the present study had a QTL whose peak was within 100 kb from (M. A. Collins et al., 2022, 2023) in at least one replicate, we determined that this QTL was present in both studies. All QTLs present in both studies had the same sign of ∆AF. A two-sample T-test was used to determine if there was a significant difference of LOD scores and absolute ∆AF between QTLs that were present in both studies compared to those found only in the present study. 77 GxE in the QTLs To determine GxE at individual QTLs, we compared loci between SC and each additional, distinct environment for each reporter. GxE at individual QTLs was classified as either 1) presence / absence or 2) sign change. Presence / absence GxE QTLs were defined as loci detected in both replicates of one environment, but where no QTL peak was found within 100 kb in a separate environment. Because a QTL might be absent due to insufficient power, we only considered QTLs that were present in both replicates of one environment (and therefore are likely to be relatively strong in that environment) and absent in both replicates of the other environment. Sign change GxE QTLs were defined as QTLs that were present in SC and a given environment but had ∆AF of a different sign. We considered QTLs to be present in both environments when their peak position occurred within 100 kb. We included QTLs in the sign change pairs that were found in both replicates of one environment and in one or both replicates of the other environment. If a pair of QTLs whose peaks were within 100 kb between SC and a given environment had the same sign of ∆AF, the pair was considered not to exhibit GxE. We included pairs of QTLs where a QTL was present in both replicates of one environment and present in either one or both replicates of the other environment. 78 Acknowledgements We thank members of the Albert laboratory for experimental guidance and feedback on the manuscript. The authors acknowledge the University of Minnesota Genomics Center (UMGC) for providing resources that contributed to the research results reported in this paper. We thank Rashi Arora and the University of Minnesota Flow Cytometry Resource staff for technical assistance with flow cytometry and FACS. Author Contributions Conceptualization: RRA, MAC, FWA Formal Analysis: RRA Funding Acquisition: FWA Investigation: RRA Methodology: RRA, MAC, FWA Resources: FWA Supervision: MAC, FWA Validation: RRA Visualization: RRA Writing - Original Draft: RRA, FWA Writing - Review and Editing: RRA, MAC, FWA Funding This work was supported by NIH grant R35GM124676 to FWA. 79 References Albert, F. W., Bloom, J. S., Siegel, J., Day, L., & Kruglyak, L. (2018). Genetics of trans-regulatory variation in gene expression. eLife, 7, e35471. https://doi.org/10.7554/eLife.35471 Albert, F. W., Treusch, S., Shockley, A. H., Bloom, J. S., & Kruglyak, L. (2014). Genetics of single-cell protein abundance variation in large yeast populations. Nature, 506(7489), 494–497. https://doi.org/10.1038/nature12904 Bachmair, A., Finley, D., & Varshavsky, A. (1986). In vivo half-life of a protein is a function of its amino-terminal residue. Science (New York, N.Y.), 234(4773), 179–186. https://doi.org/10.1126/science.3018930 Bajorek, M., Finley, D., & Glickman, M. H. (2003). Proteasome disassembly and downregulation is correlated with viability during stationary phase. Current Biology: CB, 13(13), 1140–1144. https://doi.org/10.1016/s0960-9822(03)00417-2 Ballinger, M. A., Mack, K. L., Durkin, S. M., Riddell, E. A., & Nachman, M. W. (2023). Environmentally robust cis -regulatory changes underlie rapid climatic adaptation. Proceedings of the National Academy of Sciences, 120(39), e2214614120. https://doi.org/10.1073/pnas.2214614120 Baryshnikova, A., Costanzo, M., Dixon, S., Vizeacoumar, F. J., Myers, C. L., Andrews, B., & Boone, C. (2010). Synthetic genetic array (SGA) analysis in Saccharomyces cerevisiae and Schizosaccharomyces pombe. Methods in Enzymology, 470, 145–179. https://doi.org/10.1016/S0076-6879(10)70007-0 Bett, J. S. (2016). Proteostasis regulation by the ubiquitin system. Essays in Biochemistry, 60(2), 143–151. https://doi.org/10.1042/EBC20160001 Bickel, H., Gerrard, J., & Hickmans, E. M. (1953). Influence of phenylalanine intake on phenylketonuria. Lancet (London, England), 265(6790), 812–813. https://doi.org/10.1016/s0140-6736(53)90473-5 Bloom, J. S., Boocock, J., Treusch, S., Sadhu, M. J., Day, L., Oates-Barker, H., & Kruglyak, L. (2019). Rare variants contribute disproportionately to quantitative trait variation in yeast. eLife, 8, e49212. https://doi.org/10.7554/eLife.49212 Bloom, J. S., Ehrenreich, I. M., Loo, W. T., Lite, T.-L. V., & Kruglyak, L. (2013). Finding the sources of missing heritability in a yeast cross. Nature, 494(7436), 234–237. https://doi.org/10.1038/nature11867 Boye, C., Nirmalan, S., Ranjbaran, A., & Luca, F. (2024). Genotype × environment interactions in gene regulation and complex traits. Nature Genetics, 56(6), 1057–1068. https://doi.org/10.1038/s41588-024-01776-w Boyle, E. A., Li, Y. I., & Pritchard, J. K. (2017). An Expanded View of Complex Traits: From Polygenic to Omnigenic. Cell, 169(7), 1177–1186. https://doi.org/10.1016/j.cell.2017.05.038 Brem, R. B., Yvert, G., Clinton, R., & Kruglyak, L. (2002). Genetic Dissection of Transcriptional Regulation in Budding Yeast. Science, 296(5568), 752–755. https://doi.org/10.1126/science.1069516 Brion, C., Lutz, S. M., & Albert, F. W. (2020). Simultaneous quantification of mRNA and 80 https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO protein in single cells reveals post-transcriptional effects of genetic variation. eLife, 9, e60645. https://doi.org/10.7554/eLife.60645 Burgis, N. E., & Samson, L. D. (2007). The Protein Degradation Response of Saccharomyces cerevisiae to Classical DNA-Damaging Agents. Chemical Research in Toxicology, 20(12), 1843–1853. https://doi.org/10.1021/tx700126e Chen, S.-A. A., Kern, A. F., Ang, R. M. L., Xie, Y., & Fraser, H. B. (2023). Gene-by-environment interactions are pervasive among natural genetic variants. Cell Genomics, 3(4). https://doi.org/10.1016/j.xgen.2023.100273 Collins, G. A., & Goldberg, A. L. (2017). The Logic of the 26S Proteasome. Cell, 169(5), 792–806. https://doi.org/10.1016/j.cell.2017.04.023 Collins, M. A., Avery, R., & Albert, F. W. (2023). Substrate-specific effects of natural genetic variation on proteasome activity. PLOS Genetics, 19(5), e1010734. https://doi.org/10.1371/journal.pgen.1010734 Collins, M. A., Mekonnen, G., & Albert, F. W. (2022). Variation in ubiquitin system genes creates substrate-specific effects on proteasomal protein degradation. eLife, 11, e79570. https://doi.org/10.7554/eLife.79570 Coux, O., Tanaka, K., & Goldberg, A. L. (1996). Structure and functions of the 20S and 26S proteasomes. Annual Review of Biochemistry, 65, 801–847. https://doi.org/10.1146/annurev.bi.65.070196.004101 Cubillos, F. A., Stegle, O., Grondin, C., Canut, M., Tisné, S., Gy, I., & Loudet, O. (2014). Extensive cis -Regulatory Variation Robust to Environmental Perturbation in Arabidopsis. The Plant Cell, 26(11), 4298–4310. https://doi.org/10.1105/tpc.114.130310 Dantuma, N. P., & Bott, L. C. (2014). The ubiquitin-proteasome system in neurodegenerative diseases: Precipitating factor, yet part of the solution. Frontiers in Molecular Neuroscience, 7, 70. https://doi.org/10.3389/fnmol.2014.00070 Devarajan, S., Meurer, M., van Roermund, C. W. T., Chen, X., Hettema, E. H., Kemp, S., Knop, M., & Williams, C. (2020). Proteasome-dependent protein quality control of the peroxisomal membrane protein Pxa1p. Biochimica Et Biophysica Acta. Biomembranes, 1862(9), 183342. https://doi.org/10.1016/j.bbamem.2020.183342 Dimitrov, L. N., Brem, R. B., Kruglyak, L., & Gottschling, D. E. (2009). Polymorphisms in multiple genes contribute to the spontaneous mitochondrial genome instability of Saccharomyces cerevisiae S288C strains. Genetics, 183(1), 365–383. https://doi.org/10.1534/genetics.109.104497 Edwards, M. D., & Gifford, D. K. (2012). High-resolution genetic mapping with pooled sequencing. BMC Bioinformatics, 13 Suppl 6(Suppl 6), S8. https://doi.org/10.1186/1471-2105-13-S6-S8 Ehrenreich, I. M., Torabi, N., Jia, Y., Kent, J., Martis, S., Shapiro, J. A., Gresham, D., Caudy, A. A., & Kruglyak, L. (2010). Dissection of genetically complex traits with extremely large pools of yeast segregants. Nature, 464(7291), 1039–1042. https://doi.org/10.1038/nature08923 Elserafy, M., & El-Khamisy, S. F. (2018). Choose your yeast strain carefully: The RAD5 81 https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO gene matters. Nature Reviews Molecular Cell Biology, 19(6), 343–344. https://doi.org/10.1038/s41580-018-0005-2 Fairfax, B. P., Humburg, P., Makino, S., Naranbhai, V., Wong, D., Lau, E., Jostins, L., Plant, K., Andrews, R., McGee, C., & Knight, J. C. (2014). Innate Immune Activity Conditions the Effect of Regulatory Variants upon Monocyte Gene Expression. Science, 343(6175), 1246949. https://doi.org/10.1126/science.1246949 Finley, D., & Prado, M. A. (2020). The Proteasome and Its Network: Engineering for Adaptability. Cold Spring Harbor Perspectives in Biology, 12(1), a033985. https://doi.org/10.1101/cshperspect.a033985 Finley, D., Ulrich, H. D., Sommer, T., & Kaiser, P. (2012). The Ubiquitin–Proteasome System of Saccharomyces cerevisiae. Genetics, 192(2), 319–360. https://doi.org/10.1534/genetics.112.140467 Gaisne, M., Bécam, A. M., Verdière, J., & Herbert, C. J. (1999). A “natural” mutation in Saccharomyces cerevisiae strains derived from S288c affects the complex regulatory gene HAP1 (CYP1). Current Genetics, 36(4), 195–200. https://doi.org/10.1007/s002940050490 Gardner, R. G., Nelson, Z. W., & Gottschling, D. E. (2005). Degradation-mediated protein quality control in the nucleus. Cell, 120(6), 803–815. https://doi.org/10.1016/j.cell.2005.01.016 Gietz, R. D., & Schiestl, R. H. (2007). High-efficiency yeast transformation using the LiAc/SS carrier DNA/PEG method. Nature Protocols, 2(1), 31–34. https://doi.org/10.1038/nprot.2007.13 Glickman, M. H., Rubin, D. M., Coux, O., Wefes, I., Pfeifer, G., Cjeka, Z., Baumeister, W., Fried, V. A., & Finley, D. (1998). A Subcomplex of the Proteasome Regulatory Particle Required for Ubiquitin-Conjugate Degradation and Related to the COP9-Signalosome and eIF3. Cell, 94(5), 615–623. https://doi.org/10.1016/S0092-8674(00)81603-7 Grimm, S., Höhn, A., & Grune, T. (2012). Oxidative protein damage and the proteasome. Amino Acids, 42(1), 23–38. https://doi.org/10.1007/s00726-010-0646-8 Grishkevich, V., & Yanai, I. (2013). The genomic determinants of genotype × environment interactions in gene expression. Trends in Genetics, 29(8), 479–487. https://doi.org/10.1016/j.tig.2013.05.006 Gurganus, M. C., Fry, J. D., Nuzhdin, S. V., Pasyukova, E. G., Lyman, R. F., & Mackay, T. F. C. (1998). Genotype-Environment Interaction at Quantitative Trait Loci Affecting Sensory Bristle Number in Drosophila melanogaster. Genetics, 149(4), 1883–1898. https://doi.org/10.1093/genetics/149.4.1883 Guthrie, R. (1961). Blood Screening for Phenylketonuria. JAMA, 178(8), 863. https://doi.org/10.1001/jama.1961.03040470079019 Ha, S.-W., Ju, D., & Xie, Y. (2012). The N-terminal domain of Rpn4 serves as a portable ubiquitin-independent degron and is recognized by specific 19S RP subunits. Biochemical and Biophysical Research Communications, 419(2), 226–231. https://doi.org/10.1016/j.bbrc.2012.01.152 Hahne, F., LeMeur, N., Brinkman, R. R., Ellis, B., Haaland, P., Sarkar, D., Spidlen, J., 82 https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO Strain, E., & Gentleman, R. (2009). flowCore: A Bioconductor package for high throughput flow cytometry. BMC Bioinformatics, 10, 106. https://doi.org/10.1186/1471-2105-10-106 Hanna, J., & Finley, D. (2007). A proteasome for all occasions. FEBS Letters, 581(15), 2854–2861. https://doi.org/10.1016/j.febslet.2007.03.053 Hershko, A., & Ciechanover, A. (1998). The ubiquitin system. Annual Review of Biochemistry, 67, 425–479. https://doi.org/10.1146/annurev.biochem.67.1.425 Holtz, K. M., Rice, A. M., & Sartorelli, A. C. (2003). Lithium chloride inactivates the 20S proteasome from WEHI-3B D+ leukemia cells. Biochemical and Biophysical Research Communications, 303(4), 1058–1064. https://doi.org/10.1016/S0006-291X(03)00473-X Hong, J., & Gresham, D. (2014). Molecular specificity, convergence and constraint shape adaptive evolution in nutrient-poor environments. PLoS Genetics, 10(1), e1004041. https://doi.org/10.1371/journal.pgen.1004041 Huang, W., Carbone, M. A., Lyman, R. F., Anholt, R. R. H., & Mackay, T. F. C. (2020). Genotype by environment interaction for gene expression in Drosophila melanogaster. Nature Communications, 11(1), 5451. https://doi.org/10.1038/s41467-020-19131-y Ibarra, R., Sandoval, D., Fredrickson, E. K., Gardner, R. G., & Kleiger, G. (2016). The San1 Ubiquitin Ligase Functions Preferentially with Ubiquitin-conjugating Enzyme Ubc1 during Protein Quality Control. The Journal of Biological Chemistry, 291(36), 18778–18790. https://doi.org/10.1074/jbc.M116.737619 Inobe, T., Fishbain, S., Prakash, S., & Matouschek, A. (2011). Defining the geometry of the two-component proteasome degron. Nature Chemical Biology, 7(3), 161–167. https://doi.org/10.1038/nchembio.521 Johnson, E. S., Ma, P. C. M., Ota, I. M., & Varshavsky, A. (1995). A Proteolytic Pathway That Recognizes Ubiquitin as a Degradation Signal. Journal of Biological Chemistry, 270(29), 17442–17456. https://doi.org/10.1074/jbc.270.29.17442 Ju, D., & Xie, Y. (2004). Proteasomal degradation of RPN4 via two distinct mechanisms, ubiquitin-dependent and -independent. The Journal of Biological Chemistry, 279(23), 23851–23854. https://doi.org/10.1074/jbc.C400111200 Kats, I., Khmelinskii, A., Kschonsak, M., Huber, F., Knieß, R. A., Bartosik, A., & Knop, M. (2018). Mapping Degradation Signals and Pathways in a Eukaryotic N-terminome. Molecular Cell, 70(3), 488-501.e5. https://doi.org/10.1016/j.molcel.2018.03.033 Khmelinskii, A., Keller, P. J., Bartosik, A., Meurer, M., Barry, J. D., Mardin, B. R., Kaufmann, A., Trautmann, S., Wachsmuth, M., Pereira, G., Huber, W., Schiebel, E., & Knop, M. (2012). Tandem fluorescent protein timers for in vivo analysis of protein dynamics. Nature Biotechnology, 30(7), 708–714. https://doi.org/10.1038/nbt.2281 Khmelinskii, A., & Knop, M. (2014). Analysis of Protein Dynamics with Tandem Fluorescent Protein Timers. In A. I. Ivanov (Ed.), Exocytosis and Endocytosis (Vol. 1174, pp. 195–210). Springer New York. 83 https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://doi.org/10.1007/978-1-4939-0944-5_13 Kim-Hellmuth, S., Bechheim, M., Pütz, B., Mohammadi, P., Nédélec, Y., Giangreco, N., Becker, J., Kaiser, V., Fricker, N., Beier, E., Boor, P., Castel, S. E., Nöthen, M. M., Barreiro, L. B., Pickrell, J. K., Müller-Myhsok, B., Lappalainen, T., Schumacher, J., & Hornung, V. (2017). Genetic regulatory effects modified by immune activation contribute to autoimmune disease associations. Nature Communications, 8(1), 266. https://doi.org/10.1038/s41467-017-00366-1 Kong, K.-Y. E., Fischer, B., Meurer, M., Kats, I., Li, Z., Rühle, F., Barry, J. D., Kirrmaier, D., Chevyreva, V., San Luis, B.-J., Costanzo, M., Huber, W., Andrews, B. J., Boone, C., Knop, M., & Khmelinskii, A. (2021). Timer-based proteomic profiling of the ubiquitin-proteasome system reveals a substrate receptor of the GID ubiquitin ligase. Molecular Cell, 81(11), 2460-2476.e11. https://doi.org/10.1016/j.molcel.2021.04.018 Kredel, S., Oswald, F., Nienhaus, K., Deuschle, K., Röcker, C., Wolff, M., Heilker, R., Nienhaus, G. U., & Wiedenmann, J. (2009). mRuby, a bright monomeric red fluorescent protein for labeling of subcellular structures. PloS One, 4(2), e4391. https://doi.org/10.1371/journal.pone.0004391 Kuzmin, E., Costanzo, M., Andrews, B., & Boone, C. (2016). Synthetic Genetic Array Analysis. Cold Spring Harbor Protocols, 2016(4), pdb.prot088807. https://doi.org/10.1101/pdb.prot088807 Laporte, D., Salin, B., Daignan-Fornier, B., & Sagot, I. (2008). Reversible cytoplasmic localization of the proteasome in quiescent yeast cells. The Journal of Cell Biology, 181(5), 737–745. https://doi.org/10.1083/jcb.200711154 Lea, A. J., Peng, J., & Ayroles, J. F. (2022). Diverse environmental perturbations reveal the evolution and context-dependency of genetic effects on gene expression levels. Genome Research, genome;gr.276430.121v1. https://doi.org/10.1101/gr.276430.121 Lee, M. N., Ye, C., Villani, A.-C., Raj, T., Li, W., Eisenhaure, T. M., Imboywa, S. H., Chipendo, P. I., Ran, F. A., Slowikowski, K., Ward, L. D., Raddassi, K., McCabe, C., Lee, M. H., Frohlich, I. Y., Hafler, D. A., Kellis, M., Raychaudhuri, S., Zhang, F., … Hacohen, N. (2014). Common Genetic Variants Modulate Pathogen-Sensing Responses in Human Dendritic Cells. Science, 343(6175), 1246980. https://doi.org/10.1126/science.1246980 Li, C., Liang, X., Cheng, S., Wen, Y., Pan, C., Zhang, H., Chen, Y., Zhang, J., Zhang, Z., Yang, X., Meng, P., & Zhang, F. (2022). A multi-environments-gene interaction study of anxiety, depression and self-harm in the UK Biobank cohort. Journal of Psychiatric Research, 147, 59–66. https://doi.org/10.1016/j.jpsychires.2022.01.009 Li, H., & Durbin, R. (2009). Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics (Oxford, England), 25(14), 1754–1760. https://doi.org/10.1093/bioinformatics/btp324 Li, H., Handsaker, B., Wysoker, A., Fennell, T., Ruan, J., Homer, N., Marth, G., Abecasis, G., Durbin, R., & 1000 Genome Project Data Processing Subgroup. (2009). The 84 https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO Sequence Alignment/Map format and SAMtools. Bioinformatics (Oxford, England), 25(16), 2078–2079. https://doi.org/10.1093/bioinformatics/btp352 Li, J., Breker, M., Graham, M., Schuldiner, M., & Hochstrasser, M. (2019). AMPK regulates ESCRT-dependent microautophagy of proteasomes concomitant with proteasome storage granule assembly during glucose starvation. PLOS Genetics, 15(11), e1008387. https://doi.org/10.1371/journal.pgen.1008387 Li, Y., Álvarez, O. A., Gutteling, E. W., Tijsterman, M., Fu, J., Riksen, J. A. G., Hazendonk, E., Prins, P., Plasterk, R. H. A., Jansen, R. C., Breitling, R., & Kammenga, J. E. (2006). Mapping Determinants of Gene Expression Plasticity by Genetical Genomics in C. elegans. PLoS Genetics, 2(12), e222. https://doi.org/10.1371/journal.pgen.0020222 Lutz, S., Van Dyke, K., Feraru, M. A., & Albert, F. W. (2021). Multiple epistatic DNA variants in a single gene affect gene expression in trans. Genetics, iyab208. https://doi.org/10.1093/genetics/iyab208 Marshall, R. S., McLoughlin, F., & Vierstra, R. D. (2016). Autophagic Turnover of Inactive 26S Proteasomes in Yeast Is Directed by the Ubiquitin Receptor Cue5 and the Hsp42 Chaperone. Cell Reports, 16(6), 1717–1732. https://doi.org/10.1016/j.celrep.2016.07.015 Martinez-Fonts, K., Davis, C., Tomita, T., Elsasser, S., Nager, A. R., Shi, Y., Finley, D., & Matouschek, A. (2020). The proteasome 19S cap and its ubiquitin receptors provide a versatile recognition platform for substrates. Nature Communications, 11(1), 477. https://doi.org/10.1038/s41467-019-13906-8 Matheson, K., Parsons, L., & Gammie, A. (2017). Whole-Genome Sequence and Variant Analysis of W303, a Widely-Used Strain of Saccharomyces cerevisiae. G3 Genes|Genomes|Genetics, 7(7), 2219–2226. https://doi.org/10.1534/g3.117.040022 Michelmore, R. W., Paran, I., & Kesseli, R. V. (1991). Identification of markers linked to disease-resistance genes by bulked segregant analysis: A rapid method to detect markers in specific genomic regions by using segregating populations. Proceedings of the National Academy of Sciences of the United States of America, 88(21), 9828–9832. https://doi.org/10.1073/pnas.88.21.9828 Moye-Rowley, W. S. (2003). Transcriptional control of multidrug resistance in the yeast Saccharomyces. Progress in Nucleic Acid Research and Molecular Biology, 73, 251–279. https://doi.org/10.1016/s0079-6603(03)01008-0 Nassar, L. R., Barber, G. P., Benet-Pagès, A., Casper, J., Clawson, H., Diekhans, M., Fischer, C., Gonzalez, J. N., Hinrichs, A. S., Lee, B. T., Lee, C. M., Muthuraman, P., Nguy, B., Pereira, T., Nejad, P., Perez, G., Raney, B. J., Schmelter, D., Speir, M. L., … Kent, W. J. (2023). The UCSC Genome Browser database: 2023 update. Nucleic Acids Research, 51(D1), D1188–D1195. https://doi.org/10.1093/nar/gkac1072 Nédélec, Y., Sanz, J., Baharian, G., Szpiech, Z. A., Pacis, A., Dumaine, A., Grenier, J.-C., Freiman, A., Sams, A. J., Hebert, S., Pagé Sabourin, A., Luca, F., Blekhman, R., Hernandez, R. D., Pique-Regi, R., Tung, J., Yotova, V., & Barreiro, 85 https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO L. B. (2016). Genetic Ancestry and Natural Selection Drive Population Differences in Immune Responses to Pathogens. Cell, 167(3), 657-669.e21. https://doi.org/10.1016/j.cell.2016.09.025 Nguyen Ba, A. N., Lawrence, K. R., Rego-Costa, A., Gopalakrishnan, S., Temko, D., Michor, F., & Desai, M. M. (2022). Barcoded bulk QTL mapping reveals highly polygenic and epistatic architecture of complex traits in yeast. eLife, 11, e73983. https://doi.org/10.7554/eLife.73983 Nunes, A. T., & Annunziata, C. M. (2017). Proteasome inhibitors: Structure and function. Seminars in Oncology, 44(6), 377–380. https://doi.org/10.1053/j.seminoncol.2018.01.004 Pirmohamed, M. (2023). Pharmacogenomics: Current status and future perspectives. Nature Reviews Genetics, 24(6), 350–362. https://doi.org/10.1038/s41576-022-00572-8 Prakash, S., Tian, L., Ratliff, K. S., Lehotzky, R. E., & Matouschek, A. (2004). An unstructured initiation site is required for efficient proteasome-mediated degradation. Nature Structural & Molecular Biology, 11(9), 830–837. https://doi.org/10.1038/nsmb814 Quach, H., Rotival, M., Pothlichet, J., Loh, Y.-H. E., Dannemann, M., Zidane, N., Laval, G., Patin, E., Harmant, C., Lopez, M., Deschamps, M., Naffakh, N., Duffy, D., Coen, A., Leroux-Roels, G., Clément, F., Boland, A., Deleuze, J.-F., Kelso, J., … Quintana-Murci, L. (2016). Genetic Adaptation and Neandertal Admixture Shaped the Immune System of Human Populations. Cell, 167(3), 643-656.e17. https://doi.org/10.1016/j.cell.2016.09.024 Renganaath, K., & Albert, F. W. (2023). Trans -eQTL hotspots shape complex traits by modulating cellular states. Genetics. https://doi.org/10.1101/2023.11.14.567054 Robinson, M. R., English, G., Moser, G., Lloyd-Jones, L. R., Triplett, M. A., Zhu, Z., Nolte, I. M., Van Vliet-Ostaptchouk, J. V., Snieder, H., Esko, T., Milani, L., Mägi, R., Metspalu, A., Magnusson, P. K. E., Pedersen, N. L., Ingelsson, E., Johannesson, M., Yang, J., Cesarini, D., & Visscher, P. M. (2017). Genotype–covariate interaction effects and the heritability of adult body mass index. Nature Genetics, 49(8), 1174–1181. https://doi.org/10.1038/ng.3912 Rodgers, K. J., & Shiozawa, N. (2008). Misincorporation of amino acid analogues into proteins by biosynthesis. The International Journal of Biochemistry & Cell Biology, 40(8), 1452–1466. https://doi.org/10.1016/j.biocel.2008.01.009 Rosenbaum, J. C., Fredrickson, E. K., Oeser, M. L., Garrett-Engele, C. M., Locke, M. N., Richardson, L. A., Nelson, Z. W., Hetrick, E. D., Milac, T. I., Gottschling, D. E., & Gardner, R. G. (2011). Disorder targets misorder in nuclear quality control degradation: A disordered ubiquitin ligase directly recognizes its misfolded substrates. Molecular Cell, 41(1), 93–106. https://doi.org/10.1016/j.molcel.2010.12.004 Saeki, Y., Toh-e, A., & Yokosawa, H. (2000). Rapid Isolation and Characterization of the Yeast Proteasome Regulatory Complex. Biochemical and Biophysical Research Communications, 273(2), 509–515. https://doi.org/10.1006/bbrc.2000.2980 86 https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO Sasaki, E., Zhang, P., Atwell, S., Meng, D., & Nordborg, M. (2015). “Missing” G x E Variation Controls Flowering Time in Arabidopsis thaliana. PLOS Genetics, 11(10), e1005597. https://doi.org/10.1371/journal.pgen.1005597 Schwartz, A. L., & Ciechanover, A. (1999). The ubiquitin-proteasome pathway and pathogenesis of human diseases. Annual Review of Medicine, 50, 57–74. https://doi.org/10.1146/annurev.med.50.1.57 Shaner, N. C., Campbell, R. E., Steinbach, P. A., Giepmans, B. N. G., Palmer, A. E., & Tsien, R. Y. (2004). Improved monomeric red, orange and yellow fluorescent proteins derived from Discosoma sp. Red fluorescent protein. Nature Biotechnology, 22(12), 1567–1572. https://doi.org/10.1038/nbt1037 Shostak, S. (2003). Locating gene–environment interaction: At the intersections of genetics and public health. Social Science & Medicine, 56(11), 2327–2342. https://doi.org/10.1016/S0277-9536(02)00231-9 Shringarpure, R., & Davies, K. J. A. (2002). Protein turnover by the proteasome in aging and disease. Free Radical Biology & Medicine, 32(11), 1084–1089. https://doi.org/10.1016/s0891-5849(02)00824-9 Simon, M. M., Greenaway, S., White, J. K., Fuchs, H., Gailus-Durner, V., Wells, S., Sorg, T., Wong, K., Bedu, E., Cartwright, E. J., Dacquin, R., Djebali, S., Estabel, J., Graw, J., Ingham, N. J., Jackson, I. J., Lengeling, A., Mandillo, S., Marvel, J., … Brown, S. D. (2013). A comparative phenotypic and genomic analysis of C57BL/6J and C57BL/6N mouse strains. Genome Biology, 14(7), R82. https://doi.org/10.1186/gb-2013-14-7-r82 Smith, E. N., & Kruglyak, L. (2008). Gene–Environment Interaction in Yeast Gene Expression. PLoS Biology, 6(4), e83. https://doi.org/10.1371/journal.pbio.0060083 Sontag, E. M., Vonk, W. I. M., & Frydman, J. (2014). Sorting out the trash: The spatial nature of eukaryotic protein quality control. Current Opinion in Cell Biology, 26, 139–146. https://doi.org/10.1016/j.ceb.2013.12.006 Stack, J. H., Whitney, M., Rodems, S. M., & Pollok, B. A. (2000). A ubiquitin-based tagging system for controlled modulation of protein stability. Nature Biotechnology, 18(12), 1298–1302. https://doi.org/10.1038/82422 Tanaka, K., Nakafuku, M., Tamanoi, F., Kaziro, Y., Matsumoto, K., & Toh-e, A. (1990). IRA2, a second gene of Saccharomyces cerevisiae that encodes a protein with a domain homologous to mammalian ras GTPase-activating protein. Molecular and Cellular Biology, 10(8), 4303–4313. https://doi.org/10.1128/mcb.10.8.4303-4313.1990 Theodoraki, M. A., Nillegoda, N. B., Saini, J., & Caplan, A. J. (2012). A network of ubiquitin ligases is important for the dynamics of misfolded protein aggregates in yeast. The Journal of Biological Chemistry, 287(28), 23911–23922. https://doi.org/10.1074/jbc.M112.341164 Thrower, J. S., Hoffman, L., Rechsteiner, M, & Pickart, C. M. (2000). Recognition of the polyubiquitin proteolytic signal. The EMBO Journal, 19(1), 94–102. https://doi.org/10.1093/emboj/19.1.94 87 https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO Treusch, S., Albert, F. W., Bloom, J. S., Kotenko, I. E., & Kruglyak, L. (2015). Genetic Mapping of MAPK-Mediated Complex Traits Across S. cerevisiae. PLoS Genetics, 11(1), e1004913. https://doi.org/10.1371/journal.pgen.1004913 Vabulas, R. M., & Hartl, F. U. (2005). Protein synthesis upon acute nutrient restriction relies on proteasome function. Science (New York, N.Y.), 310(5756), 1960–1963. https://doi.org/10.1126/science.1121925 Varshavsky, A. (1991). Naming a targeting signal. Cell, 64(1), 13–15. https://doi.org/10.1016/0092-8674(91)90202-a Varshavsky, A. (2011). The N-end rule pathway and regulation by proteolysis. Protein Science: A Publication of the Protein Society, 20(8), 1298–1345. https://doi.org/10.1002/pro.666 Varshavsky, A. (2019). N-degron and C-degron pathways of protein degradation. Proceedings of the National Academy of Sciences, 116(2), 358–366. https://doi.org/10.1073/pnas.1816596116 Varshavsky, A. (2024). N-degron pathways. Proceedings of the National Academy of Sciences, 121(39). https://doi.org/10.1073/pnas.2408697121 Waite, K. A., Mota-Peynado, A. D.-L., Vontz, G., & Roelofs, J. (2016). Starvation Induces Proteasome Autophagy with Different Pathways for Core and Regulatory Particles. Journal of Biological Chemistry, 291(7), 3239–3253. https://doi.org/10.1074/jbc.M115.699124 Warringer, J., Zörgö, E., Cubillos, F. A., Zia, A., Gjuvsland, A., Simpson, J. T., Forsmark, A., Durbin, R., Omholt, S. W., Louis, E. J., Liti, G., Moses, A., & Blomberg, A. (2011). Trait Variation in Yeast Is Defined by Population History. PLoS Genetics, 7(6), e1002111. https://doi.org/10.1371/journal.pgen.1002111 Wenger, J. W., Piotrowski, J., Nagarajan, S., Chiotti, K., Sherlock, G., & Rosenzweig, F. (2011). Hunger artists: Yeast adapted to carbon limitation show trade-offs under carbon sufficiency. PLoS Genetics, 7(8), e1002202. https://doi.org/10.1371/journal.pgen.1002202 Wickner, R. B. (1987). MKT1, a nonessential Saccharomyces cerevisiae gene with a temperature-dependent effect on replication of M2 double-stranded RNA. Journal of Bacteriology, 169(11), 4941–4945. https://doi.org/10.1128/jb.169.11.4941-4945.1987 Widaman, K. F. (2009). Phenylketonuria in Children and Mothers: Genes, Environments, Behavior. Current Directions in Psychological Science, 18(1), 48–52. https://doi.org/10.1111/j.1467-8721.2009.01604.x Work, J. J., & Brandman, O. (2021). Adaptability of the ubiquitin-proteasome system to proteolytic and folding stressors. Journal of Cell Biology, 220(3), e201912041. https://doi.org/10.1083/jcb.201912041 Xie, Y., & Varshavsky, A. (2001). RPN4 is a ligand, substrate, and transcriptional regulator of the 26S proteasome: A negative feedback circuit. Proceedings of the National Academy of Sciences of the United States of America, 98(6), 3056–3061. https://doi.org/10.1073/pnas.071022298 Yadav, A., & Sinha, H. (2018). Gene–gene and gene–environment interactions in 88 https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO complex traits in yeast. Yeast, 35(6), 403–416. https://doi.org/10.1002/yea.3304 Yang, T., Tang, H., Risch, H. A., Olson, S. H., Peterson, G., Bracci, P. M., Gallinger, S., Hung, R. J., Neale, R. E., Scelo, G., Duell, E. J., Kurtz, R. C., Khaw, K.-T., Severi, G., Sund, M., Wareham, N., Amos, C. I., Li, D., & Wei, P. (2020). Incorporating multiple sets of eQTL weights into gene-by-environment interaction analysis identifies novel susceptibility loci for pancreatic cancer. Genetic Epidemiology, 44(8), 880–892. https://doi.org/10.1002/gepi.22348 Zhao, S., & Ulrich, H. D. (2010). Distinct consequences of posttranslational modification by linear versus K63-linked polyubiquitin chains. Proceedings of the National Academy of Sciences, 107(17), 7704–7709. https://doi.org/10.1073/pnas.0908764107 Zheng, C., Geetha, T., & Babu, J. R. (2014). Failure of ubiquitin proteasome system: Risk for neurodegenerative diseases. Neuro-Degenerative Diseases, 14(4), 161–175. https://doi.org/10.1159/000367694 Zhu, J., Zhang, B., Smith, E. N., Drees, B., Brem, R. B., Kruglyak, L., Bumgarner, R. E., & Schadt, E. E. (2008). Integrating large-scale functional genomic data to dissect the complexity of yeast regulatory networks. Nature Genetics, 40(7), 854–861. https://doi.org/10.1038/ng.167 89 https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO https://www.zotero.org/google-docs/?d7TLaO Chapter III. Substrate-Specific Effects of Natural Genetic Variation on Proteasome Activity Mahlon A. Collins, Randi R. Avery, Frank W. Albert* Department of Genetics, Cell Biology, & Development, University of Minnesota, Minneapolis, MN, USA * To whom correspondence should be addressed: falbert@umn.edu Collins, M. A., Avery, R., & Albert, F. W. (2023). Substrate-specific effects of natural genetic variation on proteasome activity. PLOS Genetics, 19(5), e1010734. https://doi.org/10.1371/journal.pgen.1010734 Abstract Protein degradation is an essential biological process that regulates protein abundance and removes misfolded and damaged proteins from cells. In eukaryotes, most protein degradation occurs through the stepwise actions of two functionally distinct entities, the ubiquitin system and the proteasome. Ubiquitin system enzymes attach ubiquitin to cellular proteins, targeting them for degradation. The proteasome then selectively binds and degrades ubiquitinated substrate proteins. Genetic variation in ubiquitin system genes creates heritable differences in the degradation of their substrates. However, the challenges of measuring the degradative activity of the proteasome independently of the ubiquitin system in large samples have limited our understanding of genetic influences on the proteasome. Here, using the yeast Saccharomyces cerevisiae, 90 https://www.zotero.org/google-docs/?nVKCTl https://www.zotero.org/google-docs/?nVKCTl https://www.zotero.org/google-docs/?nVKCTl we built and characterized reporters that provide high-throughput, ubiquitin system-independent measurements of proteasome activity. Using single-cell measurements of proteasome activity from millions of genetically diverse yeast cells, we mapped 15 loci across the genome that influence proteasomal protein degradation. Twelve of these 15 loci exerted specific effects on the degradation of two distinct proteasome substrates, revealing a high degree of substrate-specificity in the genetics of proteasome activity. Using CRISPR-Cas9-based allelic engineering, we resolved a locus to a causal variant in the promoter of RPT6, a gene that encodes a subunit of the proteasome’s 19S regulatory particle. The variant increases RPT6 expression, which we show results in increased proteasome activity. Our results reveal the complex genetic architecture of proteasome activity and suggest that genetic influences on the proteasome may be an important source of variation in the many cellular and organismal traits shaped by protein degradation. Author Summary Protein degradation controls the abundance of cellular proteins and serves an essential role in protein quality control by eliminating misfolded and damaged proteins. In eukaryotes, most protein degradation occurs in two steps. The ubiquitin system first targets proteins for degradation by attaching ubiquitin to them. The proteasome then selectively binds and degrades ubiquitinated proteins. Understanding how individual genetic differences affect the activity of the proteasome could improve our understanding of the many traits influenced by 91 protein degradation. However, most assays that measure proteasomal protein degradation are not suitable for use in large samples or are affected by changes in the activity of the ubiquitin system. Using yeast, we built reporters that provide high-throughput measurements of proteasome activity independently of the ubiquitin system. We used measurements of proteasome activity from millions of live, single cells to identify regions of the genome with DNA variants that affect proteasomal protein degradation. We identified 15 such regions, showing that proteasome activity is a genetically complex trait. Using genome engineering, we found that one locus contained a variant in the promoter of a proteasome subunit gene that affected the activity of the proteasome towards multiple substrates. Our results demonstrate that individual genetic differences shape proteasome activity and suggest that these differences may contribute to variation in the many traits regulated by protein degradation, including gene expression, growth, development, aging, and disease. Introduction Protein degradation helps maintain protein homeostasis by regulating protein abundance and eliminating misfolded and damaged proteins from cells. The primary protein degradation pathway in eukaryotes is the ubiquitin-proteasome system (UPS). The UPS consists of two functionally distinct components, the ubiquitin system and the proteasome1–4. Ubiquitin system enzymes target proteins for degradation by binding degradation-promoting signal sequences (termed “degrons”5) and covalently attaching chains of the small protein ubiquitin 92 (Figure 1A)2,3,6,7. The proteasome then degrades polyubiquitinated proteins using two elements, the 19S regulatory particle and the 20S core particle1,8,9. The 19S regulatory particle selectively binds polyubiquitinated proteins4,10 then deubiquitinates, unfolds, and translocates them to the 20S core particle, which degrades proteins to short peptides11(Figure 1A). The UPS is responsible for 70-80% of intracellular protein degradation4,12 and influences the abundance of much of the proteome13–15. Therefore, UPS activity must be precisely and dynamically regulated at the levels of (1) substrate targeting by the ubiquitin system16–18 and (2) proteasomal protein degradation19,20. Imbalances between UPS activity and the proteolytic needs of the cell adversely impact cellular viability and are associated with a diverse array of human diseases, including cancers, immune disorders, metabolic syndromes, and neurodegenerative diseases3,20–23. Determining the factors that create variation in substrate targeting by the ubiquitin system and proteasomal protein degradation could thus improve our understanding of the many traits influenced by protein degradation. Until recently, it was largely unknown how natural genetic variation affects UPS protein degradation. To begin to address this question, we mapped genetic influences on the N-end Rule, a UPS pathway that recognizes degrons in protein N-termini (termed “N-degrons”5,24). Our results showed that UPS activity is a genetically complex trait, shaped by variation throughout the genome25. Some of the largest genetic effects on N-end rule substrates resulted from variation in ubiquitin system genes. In particular, genes whose products process (NTA1) and recognize N-degrons (UBR1 and DOA10) and ubiquitinate substrates (UBC6) 93 each contained multiple causal variants that altered UPS activity, often in an N-degron-specific manner25. Thus, individual genetic differences in the ubiquitin system are an important source of substrate-specific variation in UPS protein degradation. We do not know whether genetic effects on the proteasome are as prominent as those on the ubiquitin system. Our understanding of genetic influences on proteasome activity is largely limited to the clinical consequences of variation in proteasome genes. Missense mutations in several proteasome genes that alter proteasome activity cause a spectrum of heritable disease phenotypes, including intellectual disability26, lipodystrophy27,28, cataracts29, recurrent fever30, and morphological abnormalities31. Variation in proteasome genes has also been linked to multiple common diseases, including myocardial infarction32, stroke33, type 2 diabetes34,35, and cancer36,37. However, these mutations and polymorphisms were identified through targeted sequencing of a subset of proteasome genes, leaving us with a biased, incomplete view of genetic influences on proteasome activity. Genome-wide association studies have linked variation in the vicinity of proteasome genes to a variety of organismal phenotypes38–41. However, these studies have neither fine-mapped these loci to their individual causal variants nor determined whether they alter proteasome activity. A related question is whether variant effects on proteasome activity result in similar changes in the degradation of distinct proteasome substrates. Variation in 94 protein half-lives spans several orders of magnitude42–44, in part as a result of proteasome-specific factors that are independent of the ubiquitin system, such as how readily proteins are bound, unfolded, and degraded by the proteasome. Substrate protein factors such as unstructured initiation region length45–47, biases in amino acid composition48–50, where in the protein degradation is initiated45, and the stability of a protein’s fold48,51 can all alter how readily a specific protein is degraded by the proteasome. Proteasomes can also be assembled in multiple configurations that impart distinct affinities for different classes of substrates. In yeast, the catalytically active 20S core particle may be uncapped or singly or doubly capped with the 19S regulatory particle or the proteasome activator Blm1052. 20S proteasomes capped with 19S regulatory particles (termed “26S proteasomes”) have high affinity for polyubiquitinated proteins53–55. In contrast, uncapped 20S proteasomes preferentially degrade unfolded or intrinsically disordered substrates that have not been ubiquitinated56–58. Blm10-capped 20S proteasomes preferentially degrade short peptides, rather than proteins59–61. Genetic effects on the composition of the “proteasome pool”62 could, therefore, create substrate-specific on protein degradation. Technical challenges have precluded a more systematic understanding of the genetics of proteasomal protein degradation. The effects of natural DNA polymorphisms are often subtle, necessitating large sample sizes for detection. Statistically powerful genetic mapping of cellular traits such as proteasome 95 activity requires assays that can provide quantitative measurements from thousands of individuals63. At this scale, in vitro biochemical assays of proteasome activity are impractical. Several synthetic reporter systems can measure UPS activity in vivo with high throughput64–66. However, the output of these reporters reflects the activities of both the ubiquitin system and the proteasome, potentially hindering detection of variants that specifically affect the proteasome. In particular, QTLs often span dozens of genes and report the composite signal of multiple linked variants. As a result, the effect of a causal variant may not be reflected in a QTL if its effect direction is opposite one or more additional causal variants with larger effects. Multiple lines of evidence suggest that QTL regions often contain multiple causal variants7,67–69. In our previous work, four QTL regions we fine-mapped to causal ubiquitin system genes each contained multiple causal variants and were, along with QTLs containing HAP1 and MKT1, the most frequently detected and largest effect size QTLs for the N-end Rule69. This suggests that variant effects on the ubiquitin system could mask or obscure specific effects on the proteasome when mapping with UPS activity reporters. Therefore, approaches that can directly and specifically measure proteasome activity are needed to understand the genetics of this trait. The proteasome degrades a handful of endogenous cellular proteins without ubiquitination, providing a means of directly measuring proteasome activity independently of the ubiquitin system (Figure 1B). These proteins contain ubiquitin-independent degrons, short peptides that promote rapid proteasomal degradation without ubiquitination70–74. Ubiquitin-independent degrons 96 simultaneously function as proteasome recognition elements that engage the 19S regulatory particle and unstructured initiation regions for 20S core particle degradation (Figure 1B)71,73–78. The degradation-promoting effect of these peptides is transferable; conjugating a ubiquitin-independent degron to a heterologous protein converts it to a short-lived, ubiquitin-independent proteasome substrate73,74,76,78,79. This property has been leveraged to create genetically encoded, high-throughput reporters of proteasome activity whose readout is independent of ubiquitin system activity71,79,80. Here, we combined ubiquitin-independent degron-based proteasome activity reporters with our recently developed, statistically powerful mapping strategy to study the genetics of proteasome activity in the yeast S. cerevisiae. Our results reveal a polygenic genetic architecture of proteasome activity that is characterized by a high degree of substrate specificity. One locus contained a causal promoter variant that increased the expression of RPT6, a proteasome subunit gene, while other regions contained candidate causal genes with no known links to UPS protein degradation. Our results show that individual genetic differences are an important source of variation in proteasome activity that may contribute to the complex genetic basis of the many cellular and organismal traits influenced by protein degradation. 97 Fig. 1: Ubiquitin-dependent and independent proteasomal protein degradation. A. UPS protein degradation resulting from (1) ubiquitin system targeting followed by (2) proteasomal protein degradation. B. Proteins with ubiquitin-independent degrons are directly bound and degraded by the proteasome without ubiquitin system targeting. Results Single-Cell Measurements Reveal Heritable Variation in Proteasome Activity We sought to develop a reporter system capable of measuring proteasome activity independently of the ubiquitin system in vivo with high throughput and quantitative precision. To do so, we built a series of tandem fluorescent timers (TFTs), fusions of two fluorescent proteins with distinct spectral profiles and 98 maturation kinetics81,82. Our TFTs contained the faster-maturing green fluorescent protein (GFP) superfolder GFP83 (sfGFP) and the slower-maturing red fluorescent protein (RFP) mCherry84 (Figure 2A). The two fluorophores in the TFT mature at different rates and, as a result, the RFP / GFP ratio changes over time. If the TFT’s degradation rate is faster than the RFP’s maturation rate, the TFT’s output, expressed as the RFP / GFP ratio, is directly proportional to its degradation rate (Figure 2B). The sfGFP / mCherry TFT can measure the degradation of substrates with half-lives ranging from several minutes to several hours85, making it an ideal reporter system for studying short-lived proteasomal substrates. The TFT’s output is also independent of the construct’s expression level85, making it possible to use TFTs in genetically diverse cell populations without confounding from genetic influences on reporter expression, which are expected in a genetically diverse cell population14,25,85–88. To relate the TFT’s output to proteasome activity, we fused the ubiquitin-independent degrons from the mouse ornithine decarboxylase (ODC) and yeast Rpn4 proteins to our TFTs (Figure 2C). When expressed in yeast, the mouse ODC degron is recognized, bound, and degraded by the proteasome70,76,79. This property has previously been used to measure proteasome activity in vivo in yeast cells89. We fused amino acids 410 through 461 of mouse ODC to the TFT’s C-terminus, consistent with the sequence properties of the ODC degron71, to create the ODC TFT (Figure 2C). The Rpn4 protein contains a ubiquitin-independent degron in amino acids 1 to 8073,74. We fused this sequence to the TFT’s N-terminus to create the Rpn4 TFT (Figure 2C). 99 We reasoned that the distinct degron positions (C- and N-terminal), sequences, recognition mechanisms, and inferred 19S regulatory particle receptors71,73,90 would allow us to identify potential substrate-specific genetic effects on proteasome activity. We characterized the ODC and Rpn4 TFTs in live, single cells by flow cytometry. We first evaluated the sensitivity of each TFT by comparing each TFT’s output in the BY laboratory strain and a BY strain lacking the RPN4 gene (hereafter “BY rpn4 ”). RPN4 encodes a transcription factor for proteasome genes and deleting RPN4 reduces proteasome activity72,77,91. Deleting RPN4 strongly reduced the output from the ODC and Rpn4 TFTs in BY rpn4 (t-test p = 1.4e-6 and 1.6e-13, respectively; Figure 2D / E), showing that our TFTs provide sensitive in vivo measurements of proteasome activity. Consistent with previous reports75,78,79, in the BY strain the ODC TFT was more rapidly degraded than the Rpn4 TFT (t-test p = 6.9e-10, Figure 2D / E). Taken together, our results show that our TFTs provide quantitative, substrate-specific, in vivo readouts of proteasome activity. To understand how natural genetic variation affects proteasome activity, we measured the output of the ODC and Rpn4 TFTs in two Saccharomyces cerevisiae strains. We compared BY, which is closely related to the S288C reference strain, and the genetically divergent vineyard strain, RM, whose genome differs from BY at an average of one out of every 200 base pairs92. The RM strain showed higher proteasome activity towards the ODC and Rpn4 TFTs than BY (t-test p = 1.9e-4 and 1.2e-8, respectively; Figure 2D / E). We observed 100 a significant interaction between strain background and proteasome substrate such that the magnitude of the BY / RM strain difference was greater for the Rpn4 TFT than the ODC TFT (two-way ANOVA interaction p = 0.013). Together, these results show that individual genetic differences create heritable, substrate-specific variation in proteasome activity. 101 102 Fig 2: Design and characterization of proteasome activity reporters. A. Schematic of the production and maturation of a TFT. B. A bar plot created with simulated data shows how differences in a TFT’s degradation rate influence the reporter’s RFP and GFP levels, as well as the -log RFP / GFP ratio. C. Diagram of mouse ODC and yeast Rpn4 showing the location of each protein’s ubiquitin-independent degron. “AZB” = antizyme binding site, “AS” = active site, “AD” = transcriptional activation domain, “CH” = CH zinc finger DNA binding domain. D. Density plots of proteasome activity from 10,000 cells for each of 8 independent biological replicates per strain per reporter for the indicated strains and TFTs. Thin, opaque lines show individual biological replicates and thicker, transparent lines show the group average for the indicated strains. E. The median from each biological replicate in D. is plotted as a stripchart. t-test p-values are shown for the indicated strain versus BY. Bulk Segregant Analysis Identifies Complex, Polygenic Influences on Proteasome Activity To map genetic influences on proteasome activity, we used our ODC and Rpn4 TFTs to perform bulk segregant analysis, a statistically powerful genetic mapping method that compares large numbers of individuals with extreme values for a trait of interest selected from a genetically diverse population25,87,88,93,94. In our implementation, the method identifies quantitative trait loci (QTLs), regions of the genome with one or more DNA variants that influence a trait, for proteasome activity. We created genetically diverse cell populations by mating BY strains harboring either the ODC or Rpn4 TFT with RM and sporulating the resulting diploids (Figure 3A). Using the resulting populations of haploid, genetically recombined progeny, we collected pools of 20,000 cells from the 2% tails of the proteasome activity distribution using fluorescence-activated cell sorting (FACS) (Figure 3B-E). We then whole-genome sequenced each pool to determine the allele frequency difference between the high and low UPS activity pools at each 103 BY / RM DNA variant. At QTLs affecting proteasome activity, the allele frequencies will be significantly different between pools, while at unlinked loci the allele frequencies will be the same. We called significant QTLs using a logarithm of the odds (LOD) threshold previously determined to produce a 0.5% false discovery rate for TFT-based genetic mapping25 (see “Methods”) and retained only QTLs detected at genome-wide significance in both of two independent biological replicates. We determined the direction of QTL effects by computing the difference in RM allele frequency between the high and low proteasome activity pools at each QTL peak position. When this value is positive, the RM allele of the QTL results in higher proteasome activity, while negative values indicate QTLs where the RM allele decreases proteasome activity. We identified 11 QTLs for the ODC TFT and 7 QTLs for the Rpn4 TFT (Figure 4, Table 1). For the ODC TFT, the RM allele increased proteasome activity for 6 of 11 QTLs, while for the Rpn4 TFT, the RM allele increased proteasome activity for 2 of 7 QTLs. The distribution of proteasome activity QTL effect sizes, as reflected by the allele frequency difference between pools, was continuous and consisted predominantly of QTLs with small effects (Figure 4, Table 1). Together, our mapping results demonstrate that proteasome activity is a polygenic trait, shaped by variation throughout the genome. 104 Fig. 3: Mapping genetic influences on proteasome activity using bulk segregant analysis. A. Schematic of the experimental approach. B. / C. Proteasome activity distributions for the ODC TFT (B.) and Rpn4 TFT (C.). Vertical lines show the gates used to collect cells with extreme high or low proteasome activity. D. / E. Backplot of cells collected using the gates in B. / C. onto a scatter plot of GFP and RFP for the ODC (D.) and Rpn4 (E.) TFTs. 105 Fig. 4: Proteasome activity QTLs detected with the ODC and Rpn4 TFTs. A. The line plot shows the loess-smoothed allele frequency difference between the high and low proteasome activity pools across the S. cerevisiae genome for each of two independent biological replicates per reporter. Asterisks denote QTLs, which are allele frequency differences exceeding an empirically-derived LOD score significance threshold (indicated in B.) in each of two independent biological replicates for a given reporter. The horizontal red lines denote an empirically-derived 99.9% quantile of the allele frequency difference. Magenta horizontal lines above pairs of asterisks denote QTLs detected with both TFTs with the same direction of effect, which are termed “overlapping QTLs”. B. As in A., but for the LOD score for proteasome activity QTLs. The red horizontal line denotes the LOD score significance threshold used to call QTLs at a 0.5% FDR. 106 Table 1: Proteasome activity QTLs detected with the ODC and Rpn4 TFTs. The table lists all detected QTLs, sorted first by reporter, then by chromosome. Lowercase letters following chromosome numbers are used to distinguish QTLs on the same chromosome. “LOD”, logarithm of the odds; “AFD”, RM allele frequency difference (high proteasome activity pool minus low proteasome activity pool) at the QTL peak position. “Peak Position”, “Left Index”, and “Right Index” refer to base pair positions on the indicated chromosome. Each number is the average value calculated from two independent biological replicates for a given QTL. Reporter Chromosome LOD AFD Peak Position Left Index Right Index ODC TFT IIa 9.76 0.10 69800 32850 107100 ODC TFT IIb 7.13 -0.12 418100 358850 462650 ODC TFT IVa 5.64 -0.10 85150 30400 127400 ODC TFT V 12.83 -0.15 291350 247700 325650 ODC TFT VIIa 8.14 -0.15 20000 0 52800 ODC TFT VIIb 28.74 0.23 409000 390050 431700 ODC TFT X 16.36 0.18 666850 649350 691550 ODC TFT XII 8.13 0.11 768150 666200 846700 ODC TFT XIIIa 18.96 0.19 47800 25200 75850 ODC TFT XIIIb 7.96 0.13 410900 377350 450100 ODC TFT XIVa 8.81 -0.11 441750 381400 501600 Rpn4 TFT IVb 12.64 -0.13 240600 213200 309150 Rpn4 TFT V 10.09 -0.13 259650 218250 294900 Rpn4 TFT VIIa 10.21 -0.15 88550 53550 141350 Rpn4 TFT VIIc 6.80 -0.11 882500 840650 926150 Rpn4 TFT XII 40.11 0.23 672850 661800 685750 Rpn4 TFT XIVb 16.58 0.15 544150 497300 574600 Rpn4 TFT XV 30.00 -0.22 167400 142600 186200 107 Genetic Influences on Proteasome Activity are Predominantly Substrate-Specific To study substrate specificity in the genetic architecture of proteasome activity, we evaluated the overlap in the sets of QTLs obtained with the ODC and Rpn4 TFTs. We defined overlapping QTLs as those whose peaks were within 100 kb of each other and that had the same direction of effect. We then calculated the overlap fraction for the two sets of QTLs by dividing the number of overlapping QTLs by the number of overlapping QTLs plus the non-overlapping QTLs for each reporter. Only three proteasome activity QTLs, V, VIIA, and XII, overlapped between the sets of QTLs detected with the ODC and Rpn4 TFTs (overlap fraction = 0.2, Figure 4, Table 1), suggesting a high degree of substrate specificity. To put this result in context, we examined overlap among our previously-described UPS N-end Rule activity QTLs25. The N-end Rule is divided into two primary branches based on how N-degrons are generated and recognized95–98. Based on the molecular mechanisms of Arg/N-degron processing and recognition, we hypothesized that QTLs affecting Arg/N-degrons would have predominantly substrate-specific effects. For example, 2 of 12 Arg/N-degrons require deamidation by Nta198 and 4 of 12 Arg/N-degrons require arginylation by Ate199. QTLs affecting Nta1 or Ate1 would, therefore, be expected to influence, at most, 17% and 33% of Arg/N-degrons. Similarly, the E3 ligase Ubr1 has multiple binding sites and QTLs affecting these sites or their allosteric 108 regulation100 would also be expected to affect only a subset of Arg/N-degrons69. In contrast, Doa10, the E3 ligase for Ac/N-degrons, has a single RING-CH-type finger domain that has not been shown to differentially target distinct Ac/N-degrons. Similarly, the NatA complex acetylates 4 of 8 Ac/N-degrons, suggesting a potentially higher rate of QTL sharing for Ac/N-degrons compared to Arg/N-degrons. Consistent with these predictions, we previously showed that variation at NTA1 and UBR1 differentially affects Arg/N-degrons, while variation at DOA10 and the Ac/N-degron E2 ubiquitin-conjugating enzyme UBC6 affects Ac/N-degrons similarly69. To understand if the distinct patterns of substrate specificity we observed in these specific examples extend to the full sets of QTLs detected for Arg/N-degrons and Ac/N-degrons, we devised an approach to measure the extent of QTL sharing between N-degrons. We computed the QTL overlap fraction among all pairs of Arg/N-degrons or Ac/N-degrons with at least 7 QTLs (to match the minimum number of TFT QTLs detected among proteasome activity reporters) using the criteria above. QTLs for Ac/N-degrons overlapped across multiple reporters (median overlap fraction = 0.54; Figure 5A), while Arg/N-degron QTLs were more substrate specific (median overlap fraction = 0.21; Figure 5A). The distributions of overlap fractions for Arg/N-degrons and Ac/N-degrons were highly distinct (Figure 5A), making them an ideal reference to gauge the substrate-specificity of proteasome activity QTLs. The overlap fraction for the two sets of proteasome activity QTLs (0.2) was close to the median overlap for Arg/N-degrons (0.21, Figure 5A). Thus, genetic influences on proteasome activity are as substrate-specific as those on 109 N-degrons that are engaged by a broad variety of molecular mechanisms in the ubiquitin system95. Overlap among the two sets of proteasome activity QTLs was considerably lower than that for the Ac/N-degrons (Figure 5A). Crucially, the current proteasome and previous N-end Rule QTLs were detected with a similar experimental design with similarly high statistical power. Therefore, these comparisons across datasets provide an estimate of substrate specificity that is immune to potential inflation from QTLs that truly affect multiple substrates but may appear to be substrate-specific because they happened to be detected with only one or a few reporters. The chromosome XIVa and XIVb QTLs, which occur at similar positions but have opposing effects on the degradation of the Rpn4 and ODC TFTs (Figure 4A), provide further evidence that genetic effects on proteasome activity are highly substrate-specific. 110 Fig 5: Overlap of N-end Rule and Proteasome Activity QTLs. A. Analysis of QTL overlap for proteasome activity, Arg/N-degron, and Ac/N-degron QTLs. For all pairs of reporters in the indicated reporter sets, we computed the overlap fraction as overlapping QTLs divided by total QTLs (overlapping QTLs plus reporter-specific QTLs). B. Overlap of proteasome activity and N-end Rule QTLs. The plot shows the number, identity, and N-end Rule branch of the N-degron QTLs that overlap proteasome activity QTLs on the y axis are ordered first by reporter then by chromosomal position and labeled as in Table 1. N-degrons on the x axis are ordered by the distance of their QTL’s peak position from the peak of the corresponding proteasome activity QTL detected with either the ODC or Rpn4 TFT. Effects of Proteasome Activity QTLs on the UPS N-end Rule We previously showed that four QTLs affecting the degradation of N-end Rule substrates contained causal variants in ubiquitin system genes25. As expected, these QTLs did not meet our criteria for overlap with any proteasome activity QTLs (Supplementary Table 1). However, many N-end Rule QTLs did not contain ubiquitin system genes, suggesting that they may result from genetic effects on processes unrelated to ubiquitin system targeting. To understand whether 111 variation in N-end Rule activity could be explained by genetic effects on proteasome activity, we examined the overlap between the proteasome activity QTLs identified here and our previously-identified N-end Rule QTLs25. The set of N-end Rule QTLs comprises 149 QTLs detected with the 20 possible N-degron TFTs. However, many N-end Rule QTLs detected with distinct reporters overlap. To account for this, we applied our criteria for QTL overlap, which reduced the 149 N-end Rule QTLs detected with multiple reporters to 35 distinct, non-overlapping QTLs. Eleven proteasome activity QTLs overlapped one of these 35 N-end Rule QTLs (31%), suggesting that genetic effects on proteasome activity play a prominent role in shaping the activity of the UPS N-end Rule (Figure 5B). Conversely, 4 of 15 proteasome activity QTLs did not overlap any N-end Rule QTLs, demonstrating that genetic variation can specifically alter the turnover of ubiquitin-independent proteasome substrates (Figure 5B). In particular, the chromosome V QTL altered the degradation of both the ODC and Rpn4 TFTs, but no N-end Rule TFTs, suggesting broad effects on ubiquitin-independent proteasomal degradation (Figure 5B). These results and prior genome-wide deletion screens that identified multiple factors, such as Elp2, that specifically affect the ubiquitin-independent degradation of ODC without altering the degradation of ubiquitinated N-end Rule substrates85,89, suggest that multiple mechanisms can independently influence ubiquitin-independent versus ubiquitin-dependent proteasomal protein degradation. 112 Overlapping Proteasome Activity and N-end Rule QTLs Identify Candidate Causal Genes for Proteasome Activity QTLs often span large intervals, complicating efforts to identify the underlying causal genes and variants. We reasoned that we could use overlapping proteasome activity and N-end Rule QTLs to more precisely estimate QTL peak positions and nominate candidate causal genes. To this end, we computed the overlaps between the sets of proteasome activity QTLs and N-end rule QTLs and used this information to identify candidate causal genes (Figure 5B). Two proteasome activity QTLs that were also detected with multiple N-degron TFTs occurred in genomic regions harboring variation that affects a multitude of traits in the BY / RM cross. The chromosome XIVa QTL was detected with the ODC TFT, 6 Arg/N-degron TFTs, and 2 Ac/N-degron TFTs (Figure 5B). The QTL’s average peak position at base pair 462,767 was located approximately 4.5 kb from the MKT1 gene. MKT1 encodes a multifunctional RNA binding protein involved in 3’ UTR-mediated RNA regulation101,102 and variation at MKT1 affects multiple organismal traits in yeast, including sporulation efficiency and growth103,104. The MKT1 locus also occurs in a gene expression QTL “hotspot” that influences the expression of thousands of genes86,87 in the BY / RM cross. The chromosome XV QTL was detected with the Rpn4 TFT, 7 Arg/N-degron TFTs, and 1 Ac/N-degron TFT (Figure 5B). This set of QTL peaks clustered tightly at the average peak position of base pair 164,256. This position is approximately 7 kb away from IRA2, which encodes a negative regulator of RAS signaling105. Variation in IRA2 affects the expression of thousands of genes in this 113 cross of strains106 via multiple causal variants that interact epistatically67. The QTL intervals for the chromosome XIVa and XV QTLs do not contain any genes encoding proteasome subunits or proteasome assembly factors. Therefore, the QTLs at MKT1 and IRA2 illustrate that some genetic effects on proteasome activity likely result from complex, indirect molecular mechanisms involving altered gene expression. The chromosome VIIb QTL detected with the ODC TFT had the highest number of overlapping N-end rule QTLs, with QTLs detected in the same region with 4 Arg/N-degron and 7 Ac/N-degron TFTs (Figure 5B). The high number of overlapping N-end Rule QTLs for both Arg/N-degrons and Ac/N-degrons suggested that this QTL contained variation that broadly affects UPS protein degradation. The average chromosome VIIb QTL peak position at base pair 411,250 is within the RPT6 open reading frame. RPT6 encodes a subunit of the proteasome’s 19S regulatory particle, suggesting that this QTL influences proteasome activity via direct effects on a proteasome subunit. Proteasome Activity is Shaped by a Causal Variant in the RPT6 Promoter We selected the chromosome VIIb QTL for further experimental dissection. There are no missense RPT6 variants between BY and RM. However, a single non-coding variant occurs at base pair 411,461 (Figure 6A) in an intergenic region between RPT6 and the adjacent ALG13, which encodes an enzyme involved in oligosaccharide biosynthesis. We hypothesized that this intergenic 114 variant (hereafter, “RPT6 -175”) was the causal nucleotide for the chromosome VIIb QTL. To test the effect of RPT6 -175 on proteasome activity, we used CRISPR-Cas9 to create BY strains with either the BY or RM alleles at RPT6 -175. We tested the effect of the RPT6 -175 RM allele on the ODC and Rpn4 TFTs, as well as a subset of Ac/N-degron and Arg/N-degron reporters with which the chromosome VIIb QTL was also detected. The RPT6 -175 RM allele significantly increased proteasome activity towards the ODC TFT as compared to the BY RPT6 -175 allele (p = 2.8e-6, Figure 6B), but did not increase proteasome activity towards the Rpn4 TFT (p = 0.42, Figure 6B). The RPT6 -175 RM allele significantly increased the degradation of the proline, serine, and threonine Ac/N-degron TFTs (p = 0.021, 1.1e-3, and 4.9e-5, respectively), while its effect on the degradation of the tryptophan Arg/N-degron was not statistically significant (p = 0.062, Figure 6B). We hypothesized that RPT6 -175 increases proteasome activity by increasing RPT6 expression. Increasing the expression of individual proteasome subunits is a well-established means of increasing proteasome activity107–110. Increasing RPT6 expression in human cells increases proteasome activity110 and the turnover of proteasome substrates111. To understand RPT6 -175’s effect on Rpt6 levels, we created yeast strains with the BY or RM allele at RPT6 -175 and added a tag encoding the green fluorescent protein mNeon to the chromosomal RPT6 locus. Because RPT6 -175 occurs in an intergenic region with putative 115 promoters for RPT6 and the divergently oriented ALG13, we also created strains expressing mNeon-tagged Alg13. We then used flow cytometry to measure Rpt6 and Alg13 expression. RPT6 -175 significantly increased Rpt6 abundance (p = 6.4e-4), but did not affect Alg13 abundance (p = 0.61). Therefore, RPT6 -175 likely increases proteasome activity by increasing RPT6 expression. 116 Fig. 6: Fine-mapping a causal variant for the chromosome VIIb QTL. A. Genomic interval for the chromosome VIIb QTL. The red box depicts the 95% confidence interval of the chromosome VIIb QTL peak position, which was calculated using the chromosome VIIb QTL intervals from the ODC and N-end Rule TFTs with which the QTL was detected. B. CRISPR-Cas9 was used to engineer strains containing either the BY or RM allele at RPT6 -175 and the variant’s effect on proteasome activity was measured by flow cytometry. The variant’s effect was tested on strains harboring the ODC and Rpn4 ubiquitin-independent degron TFTs, as well as the proline (Pro), serine (Ser), and threonine (Thr) Ac/N-end TFTs, and the tryptophan (Trp) Arg/N-degron TFT. C. Rpt6 abundance was measured in strains harboring the BY or RM allele at RPT6 -175. D. Alg13 abundance was measured in strains harboring the BY or RM allele at RPT6 -175. In B-D, each point shows the median of 10,000 cells from independent biological replicates following Z-score normalization to the median of the strain with BY allele at RPT6 -175. 117 To better understand how RPT6 expression levels influence proteasomal protein degradation, we measured how plasmid-based overexpression of RPT6 affected proteasome activity. We created plasmids containing RPT6 expressed from its native promoter or the strong, constitutively active ACT1 promoter. To measure RPT6’s expression level, we created yeast strains in which both the plasmid and chromosomal copy of RPT6 was tagged with mNeon and measured each gene’s abundance using flow cytometry. As expected, plasmid-based overexpression RPT6 led to a significant increase in RPT6 expression compared to strains with an otherwise identical empty vector lacking the extra gene copy (t-test corrected Figure 7A / B). Likewise, overexpressing RPT6 from the ACT1 promoter led to significant increases in the levels of Rpt6 above either strains with empty vector or the native promoter overexpression plasmid (Figure 7A / B). Together with our RPT6 -175 edited strains, these plasmid overexpression strains thus allowed us to determine how Rpt6 influence proteasome activity across a wide range of expression levels. To measure the effect of RPT6 overexpression on proteasome activity, we created plasmids to overexpress RPT6 from its native promoter or the ACT1 promoter, but without an mNeon tag. We built strains harboring one of these plasmids or an empty vector control plasmid as well as either the genomically-integrated ODC or Thr N-degron TFT and measured the degradation of each TFT by flow cytometry. Increasing RPT6 expression either via the native RPT6 promoter or the ACT1pr increased the degradation of both the ODC and Thr N-degron TFTs (Figure 6A), further suggesting that the effect of the causal 118 RPT6 -175 variant results from its effects on RPT6 expression. The causal variant’s effect on RPT6 expression is small (log fold change = 0.12), even in the context of natural variants68. This raises the possibility that we may not have detected true RPT6 -175 effects on the Rpn4 and Trp N-degron TFTs (Figure 6B) due to incomplete statistical power. Nevertheless, our results further establish that increasing the expression of individual proteasome subunits generally and increasing Rpt6 specifically is sufficient to increase proteasome activity. Interestingly, the significant increase in Rpt6 levels resulting from a second copy of RPT6 driven by the ACT1 versus RPT6 promoter did not further enhance the degradation of either the ODC or Thr TFTs (corrected p = 0.14 and 0.53, respectively; Figure 6C). Thus, by using multiple levels of RPT6 overexpression, we have revealed limits to the extent to which increasing RPT6 expression can drive increased proteasome activity. 119 Fig 7: Effect of RPT6 overexpression on proteasome activity. A. A BY strain with the chromosomal RPT6 gene tagged with the GFP mNeon was transformed with plasmids containing the RPT6 gene tagged with mNeon and expressed from the native RPT6 promoter or the strong, constitutively active ACT1 promoter. Rpt6 levels were measured by flow cytometry and compared to the same strain transformed with an otherwise identical empty vector control plasmid lacking the RPT6 gene. B. BY strains harboring either the ODC or Thr N-degron TFTs were transformed with plasmids expressing RPT6 without an mNeon tag from the native RPT6 or ACT1 promoters. Proteasome activity was measured by flow cytometry and compared to strains harboring an otherwise identical empty vector control plasmid lacking the RPT6 gene. In A. and B. each point represents the median of 10,000 cells after Z-score normalization to the median of the corresponding empty vector control strain. To better understand the molecular properties, evolutionary history, and population characteristics of RPT6 -175, we performed several additional analyses. To understand how RPT6 -175 might increase RPT6 expression, we scanned the RPT6 promoter with either the BY or RM allele at RPT6 -175 for putative transcription factor binding motifs. The RPT6 promoter containing the RM, but not BY, allele at RPT6 -175 contains a putative binding site for Yap1 (Figure 8A). Yap1 is a stress-associated transcription factor that indirectly 120 increases proteasome activity during cellular stress, in part, by increasing expression of the proteasome gene transcription factor RPN4112–114. A multi-species alignment of the RPT6 promoter, showed that the RPT6 -175 BY allele is highly conserved among yeast species (Figure 8B). The BY allele is also present in the ancestral Taiwanese S. cerevisiae isolate, further indicating that the RPT6 -175 RM allele is derived. We then examined RPT6 -175 allelic status in a global panel of 1,011 S. cerevisiae isolates115 to better understand its population characteristics and evolutionary origin. Across all strains, the RPT6 -175 RM allele has a 33.7% population frequency (Figure 8C). However, among the “Wine / European” clade that contains RM, the RPT6 -175 RM allele has a population frequency of 91.6% (Figure 8C). No other clades have a comparably high RPT6 -175 RM allele frequency (Figure 8C). Thus, a derived variant whose ancestral allele is highly conserved across yeast species can increase proteasome function, a result in contrast with the often-deleterious consequences of new mutations. 121 Fig. 8: Properties of the causal RPT6 -175 causal variant. A. The RPT6 promoter with either the BY or RM alleles at RPT6 -175 were scanned for transcription factor binding motifs. The motif plot displays the sequence logo of a Yap1 binding motif created by the RM allele at RPT6 -175. B. Multi-species alignment of the RPT6 promoter with the causal -175 variant highlighted in red. “S. pas” = Saccharomyces pastorianus, “S. par” = Saccharomyces paradoxus, “S. mik” = Saccharomyces mikatae, “S. kud” = Saccharomyces kudriavzevii, “S. bay” = Saccharomyces bayanus, “S. pombe” = Saccharomyces pombe. Tree diagram showing the distribution of the RPT6 -175 allele among a panel of 1,011 S. cerevisiae strains. Clades with the RPT6 -175 RM allele are indicated along with its frequency in that clade in parentheses. Discussion Much of the proteome undergoes regulated turnover via proteasomal protein degradation13–15. Proteasome activity is thus a critical determinant of the abundance of individual proteins and, by extension, the functional state of the cell. Physiological variation in proteasome activity enables cells to adapt to changing internal and external environments, such as during cellular 122 stress107,116,117, while pathological variation in proteasome activity is linked to a diverse array of human diseases3,20,23,118. However, a full understanding of the factors that determine proteasome activity has remained elusive. In particular, the challenges of measuring proteasomal protein degradation in large samples has limited our understanding of the genetics of proteasome activity. By combining high-throughput proteasome activity reporters with a statistically powerful genetic mapping method, we have established individual genetic differences as an important source of variation in proteasome activity. Our results add to the emerging picture of the complex effects of genetic variation on protein degradation, which include widespread effects on the activity of the ubiquitin system25 and, as we show here, the proteasome. This work provides several new insights into how individual genetic differences shape the activity of the proteasome. Previous studies identified rare mutations in proteasome genes as the cause of a variety of monogenic disorders27–29,31,118,119. However, it was unclear to what extent these mutations are representative of genetic effects on proteasome activity. Our results suggest that disease-causing mutations and disease-linked polymorphisms with large effects on proteasome activity represent one extreme of a continuous distribution of variant effects on proteasome activity. Aberrant proteasome activity is a hallmark of numerous common human diseases3,20,23. Our results raise the possibility that the risk for these diseases may be subtly influenced by common variants that create heritable variation in proteasome activity. Our unbiased, genome-wide genetic mapping also identified QTLs containing no genes with previously-established 123 connections to the regulation of proteasome activity. In particular, the chromosome XIVa and XV QTLs do not contain any genes encoding proteasome genes or proteasome assembly factors. Instead, the peaks of these QTLs center on MKT1 and IRA2, which encode an RNA-binding protein and a RAS signaling regulator respectively, further highlighting the complexity of genetic effects on proteasome activity and providing support for recent models of the genetics of complex traits, which emphasize the predominant role of weak trans-acting effects120. The proteasome activity QTLs we have identified add new insight into how genetic variation shapes the molecular effectors of cellular protein degradation. We recently mapped the genetics of the UPS N-end rule pathway and discovered multiple DNA variants that alter the activity of four functionally distinct components of the ubiquitin system69. Here, we extend this result by showing that genetic variation also shapes protein degradation through effects on the proteasome. Although many stimuli, such as protein misfolding or heat shock, cause coordinated changes in the activity of the ubiquitin system and the proteasome, recent work shows that these two systems can also be regulated independently and function autonomously of one another19,121. For example, ubiquitination can initiate events besides proteasomal protein degradation, including lysosomal protein degradation, altered protein subcellular localization, and signaling cascade activation121–123. Likewise, a number of cellular proteins are bound and degraded by the proteasome without modification by the ubiquitin system75. Thus, predicting how genetic variation shapes the turnover of individual 124 proteins will require consideration of genetic effects on both the ubiquitin system and the proteasome. Our work further establishes the highly substrate-specific effects of natural genetic variation on protein degradation. This substrate specificity likely results from diverse mechanisms involving both direct and indirect effects on each step in the cascade of molecular events in UPS protein degradation, from substrate targeting by the ubiquitin system to proteasomal protein degradation. Our previous work69 revealed that causal variants in ubiquitin system genes create direct, substrate-specific effects on the protein degradation by altering the sequence or expression of ubiquitin system genes whose products process, recognize, and ubiquitinate distinct sets of UPS substrates. We find a similarly high degree of substrate specificity among the set of proteasome activity QTLs (Figures 4 and 5), raising the question of how direct and indirect genetic mechanisms cause substrate-specific effects on the proteasome. Direct effects on the proteasome could arise through effects on substrate selection by the proteasome’s 19S regulatory particle. Efficient degradation of the proteasome substrates tested here and in our previous study25 require the proteasome’s 19S regulatory particle73,124, which contains multiple substrate receptors. The Rpn4 degron is not bound by the canonical 19S receptors for polyubiquitin chains, Rpn10125 and Rpn13126, but instead is bound by Rpn2 and Rpn573. Genetic variation in 19S subunits that affects their abundance, affinity, or activity could therefore create direct, substrate-specific effects on the proteasome. 125 However, we think it is likely that many substrate-specific variant effects on protein degradation arise through indirect mechanisms. As an example, we map proteasome activity QTLs containing MKT1, HAP1, and IRA2, regions of the genome known to contain variation that affects the expression of thousands of genes, including numerous UPS genes86,88,127–129. Understanding how such highly pleiotropic QTL regions shape protein degradation is a difficult, but important challenge for future studies, particularly for understanding how genetic effects on protein degradation contribute to organismal traits, such as health, aging, and disease, traits that are likely shaped by the collective effects many small, indirect effects. Based on the high degree of substrate specificity in the ubiquitin system and the proteasome13,14, we anticipate that the degradation of individual proteins will also be shaped by genetic effects that are highly substrate-specific. Understanding how natural genetic variation affects the proteome through effects on the degradation of individual proteins will thus require reporters that can sensitively measure the degradation of proteins with half-lives ranging from several minutes to several hours42,43. The mCherry / sfGFP TFT is well-suited to this purpose. Previous studies have shown that this TFT should be capable of measuring the degradation of approximately 80% of yeast proteins based on their half-lives85,85 and assuming the protein tolerates the TFT tag. Recently, a genome-wide TFT tagging approach successfully used the mCherry / SfGFP timer to measure the turnover of approximately 70% (around 4,000 proteins) of the yeast proteome130, suggesting that degradation QTLs for most proteins could be mapped using this 126 reporter. TFTs with red fluorescent proteins that mature over longer time scales, such as mRuby131 or dsRed132, can be used to measure the degradation of longer-lived proteins85. We identified QTLs for proteasome activity using bulk segregant analysis, an approach that has previously been used to characterize the genetic basis of variation in a host of molecular, cellular, and organismal traits88,92–94,133. By assaying large numbers of individuals, bulk segregant analysis provides high statistical power to detect variant effects on a trait92,93. Here, we used high-throughput reporters to measure proteasome activity in millions of recombinant progeny from a cross of the BY and RM strains, which allowed us to reproducibly identify proteasome activity QTLs. Moreover, bulk segregant analysis is efficient in terms of time, labor, and resources as compared to linkage or association mapping. In particular, by generating two “bulks” with extreme phenotypes, we could detect proteasome activity QTLs through pooled whole-genome sequencing, rather than genotyping individual meiotic progeny. However, the choice of bulk segregant analysis also involves limitations that arise from this pooled whole-genome sequencing approach. Because we do not ascertain the genotypes of individual meiotic progeny, we cannot readily estimate the heritability of proteasome activity or the variance explained by the QTLs we detect. For the same reason, we are unable to detect genetic interactions between loci. Recent advances134 could enable efficient, statistically powerful mapping of proteasome activity using individual meiotic progeny in future studies, which would address these limitations. 127 Using CRISPR-Cas9 based allelic engineering, we resolved a QTL on chromosome VII to a noncoding causal nucleotide that increases RPT6 expression. Our results are consistent with previous studies in human cells, where overexpressing RPT6 results in large increases in proteasome activity110 and the turnover of proteasomal substrates111. Here, we observe similar effects by overexpressing RPT6 in yeast, adding to a growing body of evidence that has established increasing proteasome subunit expression as a robust mechanism for increasing proteasome activity. In yeast, overexpression of the SCL1 gene, which encodes the 1 20S core particle, also increases proteasome activity and promotes resistance to cellular stress107. In drosophila, overexpression of the DME1 gene, which encodes the 5 subunit of the 20S core particle, increases proteasome activity and extends lifespan135. Similar effects occur in C. elegans, where overexpression of PBS-5, which encodes the 5 subunit136, and the 19S subunit encoding RPN-6137 each increase proteasome activity and promote resistance to cellular stressors. In human cells, overexpression of 5 of 6 Rpt subunits (Rpt1-4 and Rpt6) of the 19S regulatory particle, PSMA4, which encodes the 4 subunit of the 20S core particle, and PSMD11, which encodes the 19S subunit Rpn2, all increase proteasome activity???,108,110. While increasing proteasome subunit expression is thus an established means of increasing proteasome activity, the mechanism(s) of this effect are not well-understood, but may involve increased proteasome gene expression or enhanced proteasome assembly. In the case of RPT6, one possibility is that increasing expression levels increases the number of 19S regulatory particles 128 and, in turn, the fraction of 26S proteasomes. The “proteasome pool” comprises uncapped 20S core particle and 20S singly or doubly capped with 19S regulatory particles or other proteasome activators such as Blm10. Estimates of the fraction of uncapped 20S proteasomes in the proteasome pool vary across species and cell types, but are generally no less than 30%58,62,138,139, suggesting that a large fraction of the proteasome pool could be converted to 26S proteasomes. Moreover, the 26S fraction is dynamic and responsive to changes in 19S subunit expression. For example, in human cells, decreasing the expression of either of the 19S subunits Rpt6 and Rpn2 reduces the fraction of 26S proteasomes140. Current models of proteasome assembly posit that the 20S core particle can serve as a template for assembling the 19S regulatory particle141–143. Rpt6 plays a critical role in this process - insertion of its C-terminal tail into the 2-3 pocket is the first step in assembling the 19S regulatory particle’s base that sits atop the 20S core particle142–144. After insertion, Rpt6 functions as an anchor to which other RPT heterodimers are added142–144. These findings suggest that increasing RPT6 expression could increase the 26S proteasome fraction by promoting the formation of an assembly intermediate that acts as a scaffold for further 19S assembly onto the 20S core. Importantly, our measures of the change in proteasome activity at multiple levels of RPT6 overexpression suggest limits to the extent to which proteasome activity can be increased by overexpressing a single proteasome subunit (Figure 7). We observed no additional increase in proteasome activity when overexpressing RPT6 from the ACT1 versus RPT6 promoter, despite the former producing a 129 greater than 2-fold increase in RPT6 abundance over the latter (Figure 7). Previous studies have reported increased expression of multiple proteasome genes136 and more efficient proteasome assembly137 in response to overexpression of single proteasome subunits. Potentially, these mechanisms become saturated at high levels of individual subunit overexpression, such that the overexpressed subunit is subject to degradation through quality control pathways that monitor subunit stoichiometry or other subunits become rate-limiting for proteasome assembly145. We have developed a generalizable strategy for mapping genetic effects on proteasomal protein degradation with high statistical power. The elements in our reporters function in many other eukaryotic organisms, including human cells73,76,85. Deploying the reporter systems developed here in genetically diverse cell populations may provide new insights into the genetic basis of a variety of cellular and organismal traits, including the many diseases marked by aberrant proteasome activity. Materials and Methods Tandem Fluorescent Timer (TFT) Reporters of Proteasome Activity and Plasmids We used TFTs, fusions of two fluorescent proteins with distinct spectral profiles and maturation kinetics, to measure proteasome activity. The most common TFT implementation consists of a faster-maturing green fluorescent protein (GFP) and 130 a slower-maturing red fluorescent protein (RFP)81,82,85,130. Because the two fluorescent proteins mature at different rates, the RFP / GFP ratio changes over time. If the TFT’s degradation rate is faster than the RFP’s maturation rate, the negative log RFP / GFP ratio is directly proportional to the TFT’s degradation rate81,85. The RFP / GFP ratio is also independent of the TFT’s expression level,81,85, enabling high-throughput, quantitative measurements of TFT turnover in genetically diverse cell populations25,85. All TFTs in the present study contained superfolder GFP (sfGFP)83 and the RFP mCherry84 separated by an unstructured 35 amino acid peptide sequence to minimize fluorescence resonance energy transfer85. To measure proteasome activity with our TFTs, we fused the ubiquitin-independent degrons from the mouse ornithine decarboxylase (ODC) and yeast Rpn4 proteins to our sfGFP-mCherry TFTs. ODC, an enzyme involved in polyamine biosynthesis, contains a ubiquitin-independent degron in its C-terminal 37 amino acids70,71,79,146. Rpn4, a transcription factor for proteasome genes, contains a ubiquitin-independent degron in its N-terminal 80 amino acids72,73,77. Both degrons are recognized and bound by the 19S regulatory particle without ubiquitin conjugation and function as unstructured initiation regions46 for 20S core particle degradation. Attaching either degron to a heterologous protein converts it into a short-lived proteasomal substrate. ODC degron-protein fusions have half-lives of approximately 6 minutes147. While the half-life of Rpn4 degron-protein fusions has not been precisely determined, previous results suggest it is between 10 and 20 minutes73,148. The ODC and 131 Rpn4 degron sfGFP-mCherry TFTs thus provide direct, quantitative, substrate-specific readouts of proteasome activity. We used a previously described approach25 to construct TFT reporters and yeast strains harboring TFTs. Each TFT contained the constitutively active TDH3 promoter, the ADH1 terminator, sfGFP, mCherry, and the KanMX selection module that confers resistance to the antibiotic G418149. TFTs were constructed so that the ubiquitin-independent degron was immediately adjacent to mCherry (Figure 2C), consistent with established guidelines for optimizing TFT function82. We used BFA0190 as the plasmid backbone for all TFT plasmids. BFA0190 contains 734 bp of sequence upstream and 380 bp of sequence downstream of the LYP1 ORF separated by a SwaI restriction site. We inserted TFT reporters into BFA0190 by digesting the plasmid with SwaI and inserting TFT components between the LYP1 flanking sequences using isothermal assembly cloning (Hifi Assembly Cloning Kit; New England Biolabs [NEB], Ipswich, MA, USA). The 5’ and 3’ LYP1 flanking sequences in each TFT plasmid contain natural SacI and BglII restriction sites, respectively. We produced linear DNA transformation fragments by digesting TFT-containing plasmids with SacI and BglII and gel purifying the fragments (Monarch Gel Purification, NEB). Genomic integration of each linear transformation fragment results in deletion of the LYP1 gene, allowing selection for TFT integration at the LYP1 locus using the toxic amino acid analogue thialysine (S-(2-aminoethyl)-L-cysteine hydrochloride)150–152 and G418149. All plasmids used in this study are listed in Supplementary Table 2. 132 We constructed plasmids to overexpress RPT6 via the gene’s native promoter or the strong, constitutively active ACT1 promoter. To do so, we digested the backbone plasmid BFA001 (Addgene plasmid #35121 - a gift from John McCusker) with HindIII and NdeI (NEB). We then PCR amplified the low copy CEN origin of replication and the ACT1 promoter from plasmid BFA0129. The RPT6 promoter and open reading frame were amplified from genomic DNA extracted from the BY strain using the “10 minute prep protocol”153. We PCR amplified the natMX149 cassette, which confers resistance to the antibiotic clonNAT, from BFA001. The mNeon154 GFP was amplified from plasmid BFA0254. Plasmids were assembled using isothermal assembly cloning (Hifi Assembly Cloning Kit; NEB) and their sequence identity was verified by whole-plasmid sequencing (Plasmidsaurus, Eugene, OR, USA). The overexpression plasmids thus contain the low copy CEN replication origin, RPT6 under the control of the RPT6 or ACT1 promoter, and the natMX resistance cassette. For experiments quantifying Rpt6 abundance, we used plasmids BFA0263 and BFA0264 in which Rpt6 is tagged with mNeon, while for experiments measuring proteasome activity during RPT6 overexpression, we used plasmids BFA0267 and BFA0268 in which Rpt6 is not tagged with a fluorophore. In parallel, we also constructed an empty vector control plasmid (BFA0271), which contains the HindIII and NdeI digested BFA001 backbone, the CEN replication origin, and the natMX resistance cassette. To tag the chromosomal and plasmid RPT6 gene copies, we appended mNeon to the protein’s N-terminus to avoid interference with the protein’s C-terminal tail, which 133 serves an important role in proteasome 19S regulatory particle assembly142–144. Previous studies have previously used N-terminal Rpt6 tagging to measure the protein’s abundance and proteasome activity155–158. We also tagged the chromosomal ALG13 open reading frame using a similar procedure. Alg13 contains a C-terminal degron that is necessary for the protein’s proteasomal degradation159. Because failure to degrade excess ALG13 produces glycosylation defects159, we appended the mNeon tag to the protein’s N-terminus. Yeast Strains and Handling Yeast Strains We used two genetically divergent Saccharomyces cerevisiae yeast strains for characterizing our proteasome activity TFTs and mapping genetic influences on proteasome activity. The haploid BY strain used here (genotype: MATa his3 ho) is a laboratory strain that is closely related to the S. cerevisiae S288C reference strain. The haploid RM strain used here is a vineyard isolate with genotype MAT can1::STE2pr-SpHIS5 his3::NatMX AMN1-BY ho::HphMX URA3-FY. BY and RM differ, on average, at 1 nucleotide per 200 base pairs, such that approximately 45,000 single nucleotide variants (SNVs) between the strains can serve as markers in a genetic mapping experiment87,88,92,93. We also engineered a BY strain lacking the RPN4 gene (hereafter “BY rpn4”) to characterize the sensitivity and dynamic range of our TFT reporters. We replaced the RPN4 gene with the NatMX cassette, which confers resistance to the antibiotic nourseothricin149. To 134 do so, we transformed BY with a DNA fragment created by PCR amplifying the NatMX cassette from plasmid from Addgene plasmid #35121 (a gift from John McCusker) using primers with 40 bp of homology to the 5’ upstream and 3’ downstream sequences of RPN4 using the transformation procedure described below. To create strains in which the chromosomal copy of RPT6 is N-terminally tagged with mNeon, we PCR amplified the RPT6 promoter from BY or RM genomic DNA, the mNeon open reading frame from plasmid BFA0129, the RPT6 open reading frame from BY genomic DNA, and the kanMX resistance cassette from plasmid BFA0254. We then created transformation fragments containing these elements using splicing overlap extension PCR160, which were transformed into the BY strain using the procedure described below. Strain genotypes are presented in Table 2. Supplementary Table 3 lists the full set of strains used in this study. Strain genotypes Short Name Genotype Antibiotic Resistance Auxotrophies BY MATa his3 ho histidine RM MAT can1::STE2pr-Sp HIS5 clonNAT, hygromycin histidine his3::NatMX ho::HphMX BY rpn4 MATa his3 ho rpn4::NatMX clonNAT histidine The media formulations for all experiments are listed in Table 3. Synthetic complete media powders (SC -lys and SC -his -lys -ura) were obtained from 135 Sunrise Science (Knoxville, TN, USA). We added the following reagents at the following concentrations to yeast media where indicated: G418, 200 mg / mL (Fisher Scientific, Pittsburgh, PA, USA); clonNAT (nourseothricin sulfate, Fisher Scientific), 50 mg / L; thialysine (S-(2-aminoethyl)-L-cysteine hydrochloride; MilliporeSigma, St. Louis, MO, USA), 50 mg / L; canavanine (L-canavanine sulfate, MilliporeSigma), 50 mg / L. Media Formulations Media Name Abbreviation Formulation Yeast-Peptone-Dextrose YPD 10 g / L yeast extract 20 g / L peptone 20 g / L dextrose Synthetic Complete SC 6.7 g / L yeast nitrogen base 1.96 g / L amino acid mix -lys 20 g / L dextrose Haploid Selection SGA 6.7 g / L yeast nitrogen base 1.74 g / L amino acid mix -his -lys -ura 20 g / L dextrose Sporulation SPO 1 g / L yeast extract 10 g / L potassium acetate 0.5 g / L dextrose Yeast Transformations We used the lithium acetate / single-stranded carrier DNA / polyethyline glycol (PEG) method for all yeast transformations161. In brief, yeast strains were 136 inoculated into 5 mL of YPD liquid medium for overnight growth at 30 °C. The next day, we diluted 1 mL of each saturated culture into 50 mL of fresh YPD and grew cells for 4 hours. Cells were washed in sterile ultrapure water and then in transformation solution 1 (10 mM Tris HCl [pH 8.0], 1 mM EDTA [pH 8.0], and 0.1 M lithium acetate). After each wash, we pelleted the cells by centrifugation at 3,000 rpm for 2 minutes in a benchtop centrifuge and discarded supernatants. After washing, cells were suspended in 100 L of transformation solution 1 along with 50 g of salmon sperm carrier DNA and approximately 1 g of each linear transforming DNA fragment or approximately 300 ng of each transforming plasmid and incubated at 30 °C for 30 minutes with rolling. Subsequently, 700 L of transformation solution 2 (10 mM Tris HCl [pH 8.0], 1 mM EDTA [pH 8.0], and 0.1 M lithium acetate in 40% PEG) was added to each tube, followed by a 30 minute heat shock at 42 °C. Transformed cells were then washed in sterile, ultrapure water, followed by addition of 1 mL of liquid YPD medium to each tube. Cells were incubated in YPD for 90 minutes with rolling at 30 °C to allow for expression of antibiotic resistance cassettes. We then washed the cells with sterile, ultrapure water and plated 200 L of cells on solid SC -lys medium with G418 and thialysine for genomic integration of the TFTs, YPD plus G418 for chromosomal RPT6 mNeon tagging, and YPD plus clonNAT for RPT6 overexpression plasmids. We single-colony purified multiple independent colonies (biological replicates) from each transformation plate for further analysis as indicated in the text. Transformation at each targeted genomic locus was verified by colony PCR162 using the primers listed in Supplementary Table 4. 137 Yeast Mating and Segregant Populations We used a modified synthetic genetic array (SGA) methodology151,152 to create populations of genetically variable, recombinant cells (“segregants”) for genetic mapping. BY strains with either ODC or Rpn4 TFTs were mixed with the RM strain on solid YPD medium and grown overnight at 30 °C. We selected for diploid cells (successful BY / RM matings) by streaking mixed BY / RM cells onto solid YPD medium containing G418, which selects for the KanMX cassette in the TFT in the BY strain, and clonNAT, which selects for the NatMX cassette in the RM strain. Diploid cells were inoculated into 5 ml of liquid YPD and grown overnight at 30 °C. The next day, cultures were washed with sterile, ultrapure water, and resuspended in 5 mL of SPO liquid medium (Table 3). We induced sporulation by incubating cells in SPO medium at room temperature with rolling for 9 days. After confirming sporulation by brightfield microscopy, we pelleted 2 mL of cells, which were then washed with 1 mL of sterile, ultrapure water, and resuspended in 300 L of 1 M sorbitol containing 3 U of Zymolyase lytic enzyme (United States Biological, Salem, MA, USA) to degrade ascal walls. Asci were digested for 2 hours at 30 °C with rolling. Spores were then washed with 1 mL of 1 M sorbitol, vortexed for 1 minute at the highest intensity setting, and resuspended in sterile ultrapure water. We confirmed the release of cells from asci by brightfield microscopy and plated 300 l of cells onto solid SGA medium containing G418 and canavanine. This media formulation selects for haploid cells with (1) a TFT via G418, (2) the MATa mating type via the Schizosaccharomyces pombe HIS5 gene under the control of the STE2 promoter (which is only active in 138 MATa cells), and (3) replacement of the CAN1 gene with S. pombe HIS5 via the toxic arginine analog canavanine151,152. Haploid segregants were grown for 2 days at 30 °C and harvested by adding 10 mL of sterile, ultrapure water and scraping the cells from each plate. Each segregant population cell suspension was centrifuged at 3000 rpm for 10 minutes and resuspended in 1 mL of SGA medium. We added 450 L of 40% (v / v) sterile glycerol solution to 750 L to each segregant culture and stored this mixture in screw cap cryovials at -80 °C. We stored 2 independent sporulations each of the ODC and Rpn4 degron TFT-containing segregants (derived from our initial matings) as independent biological replicates. Flow Cytometry and Fluorescence-Activated Cell Sorting Flow Cytometry We characterized our proteasome activity TFTs using flow cytometry. For all flow cytometry experiments, we inoculated yeast strains into 400 L of liquid SC -lys medium with G418 for overnight growth in 2 mL 96 well plates at 30 °C with 1000 rpm mixing on a MixMate (Eppendorf, Hamburg, Germany). The next day, 4 L of each saturated culture was inoculated into a fresh 400 L of G418-containing SC -lys media and cells were grown for an additional 3 hours prior to flow cytometry. We performed all flow cytometry experiments on an LSR II flow cytometer (BD, Franklin Lakes, NJ, USA) equipped with a 20 mW 488 nm laser with 488 / 10 and 525 / 50 filters for measuring forward and side scatter and sfGFP fluorescence, 139 respectively, as well as a 40 mW 561 nm laser and a 610 / 20 filter for measuring mCherry fluorescence. Table 4 lists the parameters and settings for all flow cytometry and fluorescence-activated cell sorting (FACS) experiments. Flow cytometry and FACS settings. Parameter Laser Line (nm) Laser Setting (V) Filter forward scatter (FSC) 488 500 488/10 side scatter (SSC) 488 275 488/10 sfGFP / mNeon 488 500 525/50 mCherry 561 615 610/20 All flow cytometry data was analyzed using R163 and the flowCore R package164. We filtered each flow cytometry dataset to exclude all events outside of 10% the median forward scatter (a proxy for cell size). This gating approach captured the central peak of cells in the FSC histogram and removed cellular debris, aggregates of multiple cells, and restricted our analyses to cells of the same approximate size25. For flow cytometry experiments related to reporter characterization, we recorded 10,000 cells each from 8 independent biological replicates per strain for the ODC and Rpn4 degron TFTs. We extracted the median from each independent biological replicate and used these values for statistical analyses. The statistical significance of between strain differences for the ODC and Rpn4 degron TFTs was assessed using a two-tailed t-test without correction for multiple testing. We used an ANOVA with strain (BY or RM) and reporter (ODC or Rpn4 degron TFT) 140 as fixed factors to assess the statistical significance of the interaction of genetic background with reporter. For flow cytometry experiments related to fine-mapping the chromosome VIIb QTL, we used the following procedures. We recorded 10,000 cells each from 12 independent biological replicates per strain (BY RPT6 -175 BY and BY RPT6 -175 RM) per guide RNA per reporter (ODC and Rpn4 TFTs, as well as proline, serine, threonine, and tryptophan N-degron TFTs). We observed that, consistent with previous results25, the output of the TFTs changed over the course of each flow cytometry experiment. We used a previously-described approach in which the residuals of a regression of the TFT’s output on time were used to correct for this effect25,88. We then Z-score normalized the sets of median values for each reporter, setting the mean equal to the median of the BY RPT6 -175 BY allele strain. The effect of the RPT6 -175 genotype was assessed using a linear mixed model implemented in the R packages ’lme4’165 and ’lmertest’166 using RPT6 -175 genotype and guide RNA as fixed effects and plate as a random effect. We used a similar approach to measure mNeon tagged Rpt6 or Alg13 abundance in strains engineered to contain either the BY or RM allele at RPT6 -175, except that the statistical significance of the difference between strains was assessed using a t-test uncorrected for multiple testing. For experiments measuring proteasome activity during RPT6 overexpression, we evaluated statistical significance between strains using a linear mixed model with plasmid (empty vector, RPT6 overexpression via the RPT6 promoter, or RPT6 overexpression 141 via the ACT1 promoter) as a fixed factor and plate as a random effect with Benjamini Hochberg correction of p values167. Fluorescence-Activated Cell Sorting (FACS) We used FACS to collect pools of segregant cells for genetic mapping by bulk segregant analysis87,88. We thawed and inoculated segregant populations into 5 mL of SGA medium containing G418 and canavanine for overnight growth at 30 °C with rolling. The following morning, we diluted 1 mL of cells from each segregant population into a fresh 4 mL of SGA medium containing G418 and canavanine. Diluted segregant cultures were grown for 4 hours prior to sorting on a FACSAria II cell sorter (BD). Plots of side scatter (SSC) height by SSC width and forward scatter (FSC) height by FSC width were used to remove doublets from each sample and cells were further filtered to contain cells within 7.5% of the central FSC peak. We empirically determined that this filtering approach excluded cellular debris and aggregates while retaining the primary haploid cell population. We also defined a fluorescence-positive population by retaining only those TFT-containing cells with sfGFP fluorescence values higher than negative control BY and RM strains without TFTs. We collected pools of 20,000 cells each from the 2% high and low proteasome activity tails (Figure 2B / C) from two independent biological replicates for each TFT. Pools of 20,000 cells were collected into sterile 1.5 mL polypropylene tubes containing 1 mL of SGA medium that were grown overnight at 30 °C with rolling. After overnight growth, 142 we mixed 750 L of cells with 450 L of 40% (v / v) glycerol and stored this mixture in 2 mL 96 well plates at -80 °C. Genomic DNA Isolation, Library Preparation, and Whole-Genome Sequencing To isolate genomic DNA from sorted segregant pools, we first pelleted 800 L of each pool by centrifugation at 3,700 rpm for 10 minutes. Supernantants were discarded and cell pellets were resuspended in 800 L of a 1 M sorbitol solution containing 0.1 M EDTA, 14.3 mM -mercaptoethanol, and 500 U of Zymolyase lytic enzyme (United States Biological) to digest cell walls. Zymolyase digestions were carried out by resuspending cell pellets with mixing at 1000 rpm for 2 minutes followed by incubation for 2 hours at 37 °C. After completing the digestion reaction, we pelleted and resuspended cells in 50 L of phosphate-buffered saline. We then used the Quick-DNA 96 Plus kit (Zymo Research, Irvine, CA, USA) to extract genomic DNA according to the manufacturer’s protocol, including an overnight protein digestion in a 20 mg / mL proteinase K solution at 55 °C prior to loading samples onto columns. DNA was eluted from sample preparation columns using 40 L of DNA elution buffer (10 mM Tris-HCl [pH 8.5], 0.1 mM EDTA). DNA concentrations for each sample were determined with the Qubit dsDNA BR assay kit (Thermo Fisher Scientific, Waltham, MA, USA) in a 96 well format using a Synergy H1 plate reader (BioTek Instruments, Winooski, VT, USA). 143 We used genomic DNA from our segregant pools to prepare a short-read library for whole-genome sequencing on the Illumina Next-Seq platform using a previously-described approach25,87,88. The Nextera DNA library kit (Illumina, San Diego, CA, USA) was used according to the manufacturer’s instructions with the following modifications. We fragmented and added sequencing adapters to genomic DNA by adding 5 ng of DNA to a master mix containing 4 L of Tagment DNA buffer, 1 L of sterile molecular biology grade water, and 5 L of Tagment DNA enzyme diluted 1:20 in Tagment DNA buffer and incubating this mixture on a SimpliAmp thermal cycler using the following parameters (Thermo Fisher Scientific): 55 °C temperature, 20 L reaction volume, 10 minute incubation. We PCR amplified libraries prior to sequencing by adding 10 L of the tagmentation reaction to a master mix containing 1 L of an Illumina i5 and i7 index primer pair mixture, 0.375 L of ExTaq polymerase (Takara), 5 L of ExTaq buffer, 4 L of a dNTP mixture, and 29.625 L of sterile molecular biology grade water. To multiplex samples for sequencing, we generated all 96 possible index oligo combinations using 8 i5 and 12 i7 index primers. Libraries were PCR amplified on a SimpliAmp thermal cycler using the following parameters: initial denaturation at 95 °C for 30 seconds, then 17 cycles of 95 °C for 10 seconds (denaturation), 62 °C for 30 seconds (annealing), and 72 °C for 3 minutes (extension). The DNA concentration of each reaction was quantified using the Qubit dsDNA BR assay kit (Thermo Fisher Scientific). We pooled equimolar amounts of each sample, ran this mixture on a 2% agarose gel, and extracted and purified DNA in the 400 bp 144 to 600 bp region using the Monarch Gel Extraction Kit (NEB) according to the manufacturer’s instructions. The pooled library was submitted to the University of Minnesota Genomics Center (UMGC) for quality control assessment and Illumina sequencing. UMGC staff performed three quality control (QC) assays prior to sequencing. The PicoGreen dsDNA quantification reagent (Thermo Fisher Scientific) was used to determine library concentration, with a concentration 1 ng/L required to pass. The Tapestation electrophoresis system (Agilent Technologies, Santa Clara, CA, USA) was used to determine library size, with libraries in the range of 200 to 700 bp passing. Finally, the KAPA DNA Library Quantification kit (Roche, Basel, Switzerland) was used to determine library functionality, with libraries requiring a concentration 2 nM to pass. The submitted library passed each QC assay. The library was sequenced on a Next-Seq 550 instrument in mid-output, 75 bp paired-end mode, generating 153,887,828 reads across all samples, with a median of 9,757,090 and a range of 5,994,921 to 14,753,319 reads per sample. The mean read quality for all samples was 30. The median read coverage of the genome was 21, with a range of 16 to 25 across all samples. Data will be deposited into the NIH Sequence Read Archive following publication. QTL Mapping We used a previously-described approach to identify QTLs from our whole-genome sequencing data25,87,88. We initially filtered our raw reads to retain only those with a mean base quality score greater than 30. Filtered reads were 145 aligned to the S. cerevisiae reference genome (sacCer3) with the Burroughs-Wheeler alignment tool168. We used samtools169 to first remove unaligned reads, non-uniquely aligned reads, and PCR duplicates, and then to produce VCF files containing coverage and allelic read counts at each of 18,871 high-confidence, reliable SNPs63,93, with BY alleles as reference and RM alleles as alternative alleles. QTLs were called from allele counts using the MULTIPOOL algorithm170. MULTIPOOL estimates a logarithm of the odds (LOD) score by calculating a likelihood ratio from two models. In the noncausal model, the locus is not associated with the trait and the high and low proteasome activity pools have the same frequency of the BY and RM alleles. In the causal model, the locus is associated with the trait, such that the BY and RM allele frequencies differ between pools. QTLs were defined as loci with a LOD 4.5. In a previous study25, we empirically determined that this threshold produces a 0.5% false discovery rate (FDR) for TFT-based genetic mapping by bulk segregant analysis. We used the following MULTIPOOL settings: bp per centiMorgan = 2,200, bin size = 100 bp, effective pool size = 1,000. As in previous studies87,88, we excluded variants with allele frequencies higher than 0.9 or lower than 0.125,87,88. QTL confidence intervals were defined as a 2-LOD drop from the QTL peak (the QTL position with the highest LOD value). We computed the RM allele frequency difference (AF) between the high and low proteasome activity pools at each allele to visualize QTLs. We also used AF at each QTL peak to determine the magnitude and direction of the QTL’s effect. When the RM allele frequency difference at a 146 QTL is positive, the RM allele of the QTL is associated with higher proteasome activity. Negative RM allele frequency differences indicate QTLs where the RM allele is associated with lower proteasome activity. Because allele frequencies are affected by random counting noise, we smoothed allele frequencies along the genome using loess regression prior to calculating AF for each sample. QTL Fine-Mapping By Allelic Engineering We used CRISPR-Cas9 to edit the RPT6 -175 locus in the BY strain. Guide RNAs (gRNAs) targeting RPT6 were obtained from the CRISPR track of the UCSC Genome Browser171. To control for potential off-target edits by CRISPR-Cas9, we used two unique guide RNAs to engineer each allelic edit. We selected two gRNAs in the RPT6 open-reading frame (ORF) based on their proximity to the RPT6 -175 variant (PAM sequences 226 and 194 bp from RPT6 -175), their CRISPOR specificity scores172 (100 each, where 100 is the highest possible predicted specificity), and their predicted cleavage scores173 (66 and 56, where 55 indicates high predicted cleavage efficiency). We inserted each gRNA into a plasmid that expresses Cas9 under the control of the constitutively active TDH3 promoter as follows. We digested backbone plasmid BFA022425 with the restriction enzymes HpaI and BsmBI (New England Biolabs) to remove the backbone vector’s existing gRNA. The cut vector was gel purified using the Monarch Gel Purification kit (New England Biolabs) according to the manufacturer’s instructions. We then performed isothermal assembly cloning using the HiFi Assembly Kit with the gel purified vector backbone and oligos 147 encoding each gRNA (OFA1198 or OFA1199; Supplementary Table 4) to create plasmids BFA0242 and BFA0243 (Supplementary Table 2). Plasmids were miniprepped from DH5 E. coli cells using the Monarch Plasmid Miniprep kit. The sequence identities of BFA0242 and BFA0243 were confirmed by Sanger sequencing. We created repair templates for co-transformation with BFA0242 and BFA0243 as follows. We first extracted genomic DNA from BY and RM using the “10 minute prep” protocol153. Genomic DNA from each strain was used as a template for PCR amplification of the RPT6 promoter using oligos OFA1204 and OFA1207 (Supplementary Table 4). To prevent Cas9 cutting after editing of the RPT6 -175 locus, we introduced two synonymous substitutions into the RPT6 ORF by converting the serine codons GGA and TCA to AGT at base pairs 22-24 and 49-51. Synonmous substitutions were introduced using splicing overlap by extension PCR160 with primers OFA1208 and OFA1209. Full repair templates were then amplified using either the BY or RM UBR1 promoter and the BY RPT6 ORF as templates in a splicing overlap extension by PCR reaction with primers OFA1204 and OFA1205 (Supplementary Table 4). The sequence identify of all repair templates was verified by Sanger sequencing. To create BY strains with edited RPT6 alleles, we co-transformed 150 ng of either plasmid BFA0242 or BFA0243 with 1.5 g of repair template using the transformation protocol above. The transformation reaction was streaked onto solid SC medium lacking histidine to select for the HIS3 selectable marker in 148 BFA0242 or BFA0243. Colonies from transformation plates were single-colony purified on solid medium lacking histidine, then patched onto solid YPD medium. To verify allelic edits, we performed colony PCR using oligos 1204 and 1206 (Supplementary Table 4). Reaction products were gel purified using the Monarch Gel Purification kit (New England Biolabs) and Sanger sequenced using oligos OFA1204 and OFA1206 to confirm both the sequence of the RPT6 promoter and the synonymous substitutions in the RPT6 ORF. Strains with the desired edits were then transformed to contain TFT reporters as indicated above. We tested 12 independent biological replicates per strain per guide RNA per TFT. For subsequent statistical analyses, we pooled strains with the same allelic edit engineered with unique guide RNAs. Data and Statistical Analysis All data and statistical analyses were performed using R163. In all boxplots, the center line shows the median, the box bounds the first and third quartiles, and the whiskers extend to 1.5 times the interquartile range. DNA binding motifs in the RPT6 promoter were assessed using the Yeast Transcription Factor Specificity Compendium database174. We inferred the allelic status of RPT6 -175 by comparing the BY and RM alleles to a likely-ancestral Taiwanese strain. The frequency of the RM allele at RPT6 -175 was calculated across and within clades of a global panel of 1,011 S. cerevisiae isolates115. Final figures and illustrations were made using Inkscape (version 0.92; Inkscape Project). 149 Data and Materials Availability Computational scripts used to process data, for statistical analysis, and to generate plots are available at: http://www.github.com/mac230/proteasome_QTL_paper Whole-genome sequencing data is available through the NIH Sequence Read Archive under Bioproject accession PRJNA885116. Yeast strains and plasmids used in this study are available on request. Correspondence should be addressed to FWA. Author Contributions Conceptualization: MAC, FWA Formal Analysis: MAC Funding Acquisition: MAC, FWA Investigation: MAC, RRA Methodology: MAC, FWA Resources: FWA Supervision: MAC, FWA Validation: MAC, RRA Visualization: MAC Writing - Original Draft: MAC Writing - Review and Editing: MAC, FWA 150 http://www.github.com/mac230/proteasome_QTL_paper Acknowledgements We thank the members of the Albert laboratory for feedback on the project and manuscript. We thank the University of Minnesota’s Flow Cytometry Resource and Genomics Center for their contributions to the project. Competing Interests The authors declare that they have no competing interests. Financial Disclosure Statement This work was supported by NIH grants F32-GM128302 to MAC and R35-GM124676 to FWA from the National Institute of General Medical Sciences (https://www.nigms.nih.gov/). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. 151 https://www.nigms.nih.gov/ References 1. Finley, D., Ulrich, H. D., Sommer, T. & Kaiser, P. The ubiquitin-proteasome system of Saccharomyces cerevisiae. Genetics 192, 319–360 (2012). 2. Hershko, A. & Ciechanover, A. The ubiquitin system. Annu Rev Biochem 67, 425–479 (1998). 3. Schwartz, A. L. & Ciechanover, A. The ubiquitin-proteasome pathway and pathogenesis of human diseases. Annu Rev Med 50, 57–74 (1999). 4. Collins, G. A. & Goldberg, A. L. The Logic of the 26S Proteasome. Cell 169, 792–806 (2017). 5. Varshavsky, A. Naming a targeting signal. Cell 64, 13–15 (1991). 6. Ciechanover, A., Orian, A. & Schwartz, A. L. Ubiquitin-mediated proteolysis: biological regulation via destruction. Bioessays 22, 442–451 (2000). 7. Abell, N. S. et al. Multiple causal variants underlie genetic associations in humans. Science 375, 1247–1254 (2022). 8. Hanna, J. & Finley, D. A proteasome for all occasions. FEBS Lett 581, 2854–2861 (2007). 9. Inobe, T. & Matouschek, A. Paradigms of protein degradation by the proteasome. Curr Opin Struct Biol 24, 156–164 (2014). 10. Coux, O., Tanaka, K. & Goldberg, A. L. Structure and functions of the 20S and 26S proteasomes. Annu Rev Biochem 65, 801–847 (1996). 11. Kisselev, A. F., Akopian, T. N., Woo, K. M. & Goldberg, A. L. The sizes of peptides generated from protein by mammalian 26 and 20 S proteasomes. Implications for understanding the degradative mechanism and antigen presentation. J Biol Chem 274, 3363–3371 (1999). 12. Zhao, J., Zhai, B., Gygi, S. P. & Goldberg, A. L. mTOR inhibition activates overall protein degradation by the ubiquitin proteasome system as well as by autophagy. Proc Natl Acad Sci U S A 112, 15790–15797 (2015). 13. Christiano, R. et al. A Systematic Protein Turnover Map for Decoding Protein Degradation. Cell Rep 33, 108378 (2020). 14. Kong, K. E. et al. Timer-based proteomic profiling of the ubiquitin-proteasome system reveals a substrate receptor of the GID ubiquitin ligase. Mol Cell 81, 2460–2476 (2021). 15. Solomon, V., Lecker, S. H., Goldberg, A. L. & Goldberg, A. L. The N-end rule pathway catalyzes a major fraction of the protein degradation in skeletal muscle. J Biol Chem 273, 25216–25222 (1998). 16. Smith, S. E., Koegl, M. & Jentsch, S. Role of the ubiquitin/proteasome system in regulated protein degradation in Saccharomyces cerevisiae. Biol Chem 377, 437–446 (1996). 17. Kornitzer, D. & Ciechanover, A. Modes of regulation of ubiquitin-mediated protein degradation. J Cell Physiol 182, 1–11 (2000). 152 18. Bett, J. S. Proteostasis regulation by the ubiquitin system. Essays Biochem 60, 143–151 (2016). 19. Marshall, R. S. & Vierstra, R. D. Dynamic Regulation of the 26S Proteasome: From Synthesis to Degradation. Front Mol Biosci 6, 40 (2019). 20. Schmidt, M. & Finley, D. Regulation of proteasome activity in health and disease. Biochim Biophys Acta 1843, 13–25 (2014). 21. Shringarpure, R. & Davies, K. J. Protein turnover by the proteasome in aging and disease. Free Radic Biol Med 32, 1084–1089 (2002). 22. Zheng, C., Geetha, T. & Babu, J. R. Failure of ubiquitin proteasome system: risk for neurodegenerative diseases. Neurodegener Dis 14, 161–175 (2014). 23. Dantuma, N. P. & Bott, L. C. The ubiquitin-proteasome system in neurodegenerative diseases: precipitating factor, yet part of the solution. Front Mol Neurosci 7, 70 (2014). 24. Varshavsky, A. N-degron and C-degron pathways of protein degradation. Proc Natl Acad Sci U S A 116, 358–366 (2019). 25. Collins, M. A., Mekonnen, G. & Albert, F. W. Variation in ubiquitin system genes creates substrate-specific effects on proteasomal protein degradation. bioRxiv (2021) doi:10.1101/2021.05.05.442832. 26. Ligt, J. de et al. Diagnostic exome sequencing in persons with severe intellectual disability. N Engl J Med 367, 1921–1929 (2012). 27. Agarwal, A. K. et al. PSMB8 encoding the B5i proteasome subunit is mutated in joint contractures, muscle atrophy, microcytic anemia, and panniculitis-induced lipodystrophy syndrome. Am J Hum Genet 87, 866–872 (2010). 28. Liu, Y. et al. Mutations in proteasome subunit B type 8 cause chronic atypical neutrophilic dermatosis with lipodystrophy and elevated temperature with evidence of genetic and phenotypic heterogeneity. Arthritis Rheum 64, 895–907 (2012). 29. Kröll-Hermi, A. et al. Proteasome subunit PSMC3 variants cause neurosensory syndrome combining deafness and cataract due to proteotoxic stress. EMBO Mol Med 12, e11861 (2020). 30. Brehm, A. et al. Additive loss-of-function proteasome subunit mutations in CANDLE/PRAAS patients promote type I IFN production. J Clin Invest 125, 4196–4211 (2015). 31. Tomaru, U. et al. Decreased proteasomal activity causes age-related phenotypes and promotes the development of metabolic abnormalities. Am J Pathol 180, 963–972 (2012). 32. Ozaki, K. et al. A functional SNP in PSMA6 confers risk of myocardial infarction in the Japanese population. Nat Genet 38, 921–925 (2006). 33. Heckman, M. G. et al. Genetic variants associated with myocardial infarction in the PSMA6 gene and Chr9p21 are also associated with ischaemic stroke. Eur J Neurol 20, 300–308 (2013). 34. Sjakste, T. et al. Association of microsatellite polymorphisms of the human 14q13.2 region with type 2 diabetes mellitus in Latvian and Finnish populations. Ann Hum Genet 71, 772–776 (2007). 153 https://doi.org/10.1101/2021.05.05.442832 35. Wing, S. S. The UPS in diabetes and obesity. BMC Biochem 9 Suppl 1, S6 (2008). 36. Webb, E. L. et al. Search for low penetrance alleles for colorectal cancer through a scan of 1467 non-synonymous SNPs in 2575 cases and 2707 controls with validation by kin-cohort analysis of 14 704 first-degree relatives. Hum Mol Genet 15, 3263–3271 (2006). 37. Zeng, C. et al. Identification of Susceptibility Loci and Genes for Colorectal Cancer Risk. Gastroenterology 150, 1633–1645 (2016). 38. Shameer, K. et al. A genome- and phenome-wide association study to identify genetic variants influencing platelet count and volume and their pleiotropic effects. Hum Genet 133, 95–109 (2014). 39. Stuart, P. E. et al. Genome-wide association analysis identifies three psoriasis susceptibility loci. Nat Genet 42, 1000–1004 (2010). 40. Iio, E. et al. Genome-wide association study identifies a PSMD3 variant associated with neutropenia in interferon-based therapy for chronic hepatitis C. Hum Genet 134, 279–289 (2015). 41. Song, X. et al. ). Front Plant Sci 12, 695503 (2021). 42. Belle, A., Tanay, A., Bitincka, L., Shamir, R. & O’Shea, E. K. Quantification of protein half-lives in the budding yeast proteome. Proc Natl Acad Sci U S A 103, 13004–13009 (2006). 43. Christiano, R., Nagaraj, N., Frohlich, F. & Walther, T. C. Global proteome turnover analyses of the Yeasts S. cerevisiae and S. pombe. Cell Rep 9, 1959–1965 (2014). 44. Schwanhausser, B. et al. Global quantification of mammalian gene expression control. Nature 473, 337–342 (2011). 45. Kraut, D. A. & Matouschek, A. Proteasomal degradation from internal sites favors partial proteolysis via remote domain stabilization. ACS Chem Biol 6, 1087–1095 (2011). 46. Prakash, S., Tian, L., Ratliff, K. S., Lehotzky, R. E. & Matouschek, A. An unstructured initiation site is required for efficient proteasome-mediated degradation. Nat Struct Mol Biol 11, 830–837 (2004). 47. Martinez-Fonts, K. et al. The proteasome 19S cap and its ubiquitin receptors provide a versatile recognition platform for substrates. Nat Commun 11, 477 (2020). 48. Kraut, D. A. et al. Sequence- and species-dependence of proteasomal processivity. ACS Chem Biol 7, 1444–1453 (2012). 49. Yu, H. et al. Conserved Sequence Preferences Contribute to Substrate Recognition by the Proteasome. J Biol Chem 291, 14526–14539 (2016). 50. Hoyt, M. A. et al. Glycine-alanine repeats impair proper substrate unfolding by the proteasome. EMBO J 25, 1720–1729 (2006). 51. Koodathingal, P. et al. ATP-dependent proteases differ substantially in their ability to unfold globular proteins. J Biol Chem 284, 18674–18684 (2009). 52. Kish-Trier, E. & Hill, C. P. Structural biology of the proteasome. Annu Rev Biophys 42, 29–49 (2013). 53. Chau, V. et al. A multiubiquitin chain is confined to specific lysine in a targeted short-lived protein. Science 243, 1576–1583 (1989). 154 54. Finley, D. Recognition and processing of ubiquitin-protein conjugates by the proteasome. Annu Rev Biochem 78, 477–513 (2009). 55. Thrower, J. S., Hoffman, L., Rechsteiner, M. & Pickart, C. M. Recognition of the polyubiquitin proteolytic signal. EMBO J 19, 94–102 (2000). 56. Baugh, J. M., Viktorova, E. G. & Pilipenko, E. V. Proteasomes can degrade a significant proportion of cellular proteins independent of ubiquitination. J Mol Biol 386, 814–827 (2009). 57. Hsieh, L. S., Su, W. M., Han, G. S. & Carman, G. M. Phosphorylation regulates the ubiquitin-independent degradation of yeast Pah1 phosphatidate phosphatase by the 20S proteasome. J Biol Chem 290, 11467–11478 (2015). 58. Kumar Deshmukh, F., Yaffe, D., Olshina, M. A., Ben-Nissan, G. & Sharon, M. The Contribution of the 20S Proteasome to Proteostasis. Biomolecules 9, (2019). 59. Dange, T. et al. Blm10 protein promotes proteasomal substrate turnover by an active gating mechanism. J Biol Chem 286, 42830–42839 (2011). 60. Ustrell, V., Hoffman, L., Pratt, G. & Rechsteiner, M. PA200, a nuclear proteasome activator involved in DNA repair. EMBO J 21, 3516–3525 (2002). 61. Stadtmueller, B. M. & Hill, C. P. Proteasome activators. Mol Cell 41, 8–19 (2011). 62. Morozov, A. V. & Karpov, V. L. Proteasomes and Several Aspects of Their Heterogeneity Relevant to Cancer. Front Oncol 9, 761 (2019). 63. Bloom, J. S., Ehrenreich, I. M., Loo, W. T., Lite, T. L. & Kruglyak, L. Finding the sources of missing heritability in a yeast cross. Nature 494, 234–237 (2013). 64. Yen, H. C., Xu, Q., Chou, D. M., Zhao, Z. & Elledge, S. J. Global protein stability profiling in mammalian cells. Science 322, 918–923 (2008). 65. Geffen, Y. et al. Mapping the Landscape of a Eukaryotic Degronome. Mol Cell 63, 1055–1065 (2016). 66. Ella, H., Reiss, Y. & Ravid, T. The Hunt for Degrons of the 26S Proteasome. Biomolecules 9, (2019). 67. Lutz, S., Van Dyke, K., Feraru, M. A. & Albert, F. W. Multiple epistatic DNA variants in a single gene affect gene expression in trans. Genetics 220, (2022). 68. Renganaath, K. et al. Systematic identification of cis-regulatory variants that cause gene expression differences in a yeast cross. Elife 9, (2020). 69. Collins, M. A., Mekonnen, G. & Albert, F. W. Variation in ubiquitin system genes creates substrate-specific effects on proteasomal protein degradation. Elife 11, (2022). 70. Zhang, M., Pickart, C. M. & Coffino, P. Determinants of proteasome recognition of ornithine decarboxylase, a ubiquitin-independent substrate. EMBO J 22, 1488–1496 (2003). 71. Takeuchi, J., Chen, H., Hoyt, M. A. & Coffino, P. Structural elements of the ubiquitin-independent proteasome degron of ornithine decarboxylase. Biochem J 410, 401–407 (2008). 72. Xie, Y. & Varshavsky, A. RPN4 is a ligand, substrate, and transcriptional regulator of the 26S proteasome: a negative feedback circuit. Proc. Natl. Acad. Sci. U.S.A. 98, 3056–3061 (2001). 155 73. Ha, S. W., Ju, D. & Xie, Y. The N-terminal domain of Rpn4 serves as a portable ubiquitin-independent degron and is recognized by specific 19S RP subunits. Biochem Biophys Res Commun 419, 226–231 (2012). 74. Ju, D. & Xie, Y. Proteasomal degradation of RPN4 via two distinct mechanisms, ubiquitin-dependent and -independent. J Biol Chem 279, 23851–23854 (2004). 75. Erales, J. & Coffino, P. Ubiquitin-independent proteasomal degradation. Biochim Biophys Acta 1843, 216–221 (2014). 76. Hoyt, M. A., Zhang, M. & Coffino, P. Ubiquitin-independent mechanisms of mouse ornithine decarboxylase degradation are conserved between mammalian and fungal cells. J Biol Chem 278, 12135–12143 (2003). 77. Mannhaupt, G., Schnall, R., Karpov, V., Vetter, I. & Feldmann, H. Rpn4p acts as a transcription factor by binding to PACE, a nonamer box found upstream of 26S proteasomal and other genes in yeast. FEBS Lett 450, 27–34 (1999). 78. Morozov, A. V., Spasskaya, D. S., Karpov, D. S. & Karpov, V. L. The central domain of yeast transcription factor Rpn4 facilitates degradation of reporter protein in human cells. FEBS Lett 588, 3713–3719 (2014). 79. Hoyt, M. A., Zhang, M. & Coffino, P. Probing the ubiquitin/proteasome system with ornithine decarboxylase, a ubiquitin-independent substrate. Methods Enzymol 398, 399–413 (2005). 80. Momose, I. et al. In vivo imaging of proteasome inhibition using a proteasome-sensitive fluorescent reporter. Cancer Sci 103, 1730–1736 (2012). 81. Khmelinskii, A. & Knop, M. Analysis of protein dynamics with tandem fluorescent protein timers. Methods Mol. Biol. 1174, 195–210 (2014). 82. Khmelinskii, A. et al. Incomplete proteasomal degradation of green fluorescent proteins in the context of tandem fluorescent protein timers. Mol. Biol. Cell 27, 360–370 (2016). 83. Pedelacq, J. D., Cabantous, S., Tran, T., Terwilliger, T. C. & Waldo, G. S. Engineering and characterization of a superfolder green fluorescent protein. Nat. Biotechnol. 24, 79–88 (2006). 84. Shaner, N. C. et al. Improved monomeric red, orange and yellow fluorescent proteins derived from Discosoma sp. red fluorescent protein. Nat. Biotechnol. 22, 1567–1572 (2004). 85. Khmelinskii, A. et al. Tandem fluorescent protein timers for in vivo analysis of protein dynamics. Nat. Biotechnol. 30, 708–714 (2012). 86. Albert, F. W., Bloom, J. S., Siegel, J., Day, L. & Kruglyak, L. Genetics of trans-regulatory variation in gene expression. Elife 7, (2018). 87. Albert, F. W., Treusch, S., Shockley, A. H., Bloom, J. S. & Kruglyak, L. Genetics of single-cell protein abundance variation in large yeast populations. Nature 506, 494–497 (2014). 88. Brion, C., Lutz, S. M. & Albert, F. W. Simultaneous quantification of mRNA and protein in single cells reveals post-transcriptional effects of genetic variation. Elife 9, (2020). 156 89. Hoyt, M. A. et al. A genetic screen for Saccharomyces cerevisiae mutants affecting proteasome function, using a ubiquitin-independent substrate. Yeast 25, 199–217 (2008). 90. Takeuchi, J., Chen, H. & Coffino, P. Proteasome substrate degradation requires association plus extended peptide. EMBO J 26, 123–131 (2007). 91. Ju, D., Xu, H., Wang, X. & Xie, Y. The transcription activation domain of Rpn4 is separate from its degrons. Int J Biochem Cell Biol 42, 282–286 (2010). 92. Ehrenreich, I. M., Gerke, J. P. & Kruglyak, L. Genetic dissection of complex traits in yeast: insights from studies of gene expression and other phenotypes in the BYxRM cross. Cold Spring Harb Symp Quant Biol 74, 145–153 (2009). 93. Ehrenreich, I. M. et al. Dissection of genetically complex traits with extremely large pools of yeast segregants. Nature 464, 1039–1042 (2010). 94. Michelmore, R. W., Paran, I. & Kesseli, R. V. Identification of markers linked to disease-resistance genes by bulked segregant analysis: a rapid method to detect markers in specific genomic regions by using segregating populations. Proc Natl Acad Sci U S A 88, 9828–9832 (1991). 95. Varshavsky, A. The N-end rule pathway and regulation by proteolysis. Protein Sci. 20, 1298–1345 (2011). 96. Hwang, C. S., Shemorry, A. & Varshavsky, A. N-terminal acetylation of cellular proteins creates specific degradation signals. Science 327, 973–977 (2010). 97. Bachmair, A., Finley, D. & Varshavsky, A. In vivo half-life of a protein is a function of its amino-terminal residue. Science 234, 179–186 (1986). 98. Baker, R. T. & Varshavsky, A. Yeast N-terminal amidase. A new enzyme and component of the N-end rule pathway. J Biol Chem 270, 12065–12074 (1995). 99. Hu, R. G. et al. Arginyltransferase, its specificity, putative substrates, bidirectional promoter, and splicing-derived isoforms. J Biol Chem 281, 32559–32573 (2006). 100. Turner, G. C., Du, F. & Varshavsky, A. Peptides accelerate their uptake by activating a ubiquitin-dependent proteolytic pathway. Nature 405, 579–583 (2000). 101. Wickner, R. B. MKT1, a nonessential Saccharomyces cerevisiae gene with a temperature-dependent effect on replication of M2 double-stranded RNA. J Bacteriol 169, 4941–4945 (1987). 102. Tadauchi, T., Inada, T., Matsumoto, K. & Irie, K. Posttranscriptional regulation of HO expression by the Mkt1-Pbp1 complex. Mol Cell Biol 24, 3670–3681 (2004). 103. Sinha, H., Nicholson, B. P., Steinmetz, L. M. & McCusker, J. H. Complex genetic interactions in a quantitative trait locus. PLoS Genet 2, e13 (2006). 104. Deutschbauer, A. M. & Davis, R. W. Quantitative trait loci mapped to single-nucleotide resolution in yeast. Nat Genet 37, 1333–1340 (2005). 105. Tanaka, K. et al. IRA2, a second gene of Saccharomyces cerevisiae that encodes a protein with a domain homologous to mammalian ras GTPase-activating protein. Mol Cell Biol 10, 4303–4313 (1990). 106. Smith, E. N. & Kruglyak, L. Gene-environment interaction in yeast gene expression. PLoS Biol 6, e83 (2008). 157 107. Howell, L. A., Peterson, A. K. & Tomko, R. J. Proteasome subunit a1 overexpression preferentially drives canonical proteasome biogenesis and enhances stress tolerance in yeast. Sci Rep 9, 12418 (2019). 108. Padmanabhan, A., Vuong, S. A. & Hochstrasser, M. Assembly of an Evolutionarily Conserved Alternative Proteasome Isoform in Human Cells. Cell Rep 14, 2962–2974 (2016). 109. Vilchez, D. et al. Increased proteasome activity in human embryonic stem cells is regulated by PSMD11. Nature 489, 304–308 (2012). 110. Wang, Y. et al. CKIP-1 couples Smurf1 ubiquitin ligase with Rpt6 subunit of proteasome to promote substrate degradation. EMBO Rep 13, 1004–1011 (2012). 111. Zemoura, K. & Benke, D. -aminobutyric acidB receptors is mediated by the interaction of the GABAB2 C terminus with the proteasomal ATPase Rtp6 and regulated by neuronal activity. J Biol Chem 289, 7738–7746 (2014). 112. Owsianik, G., Balzi l, L. & Ghislain, M. Control of 26S proteasome expression by transcription factors regulating multidrug resistance in Saccharomyces cerevisiae. Mol Microbiol 43, 1295–1308 (2002). 113. Wang, X., Yen, J., Kaiser, P. & Huang, L. Regulation of the 26S proteasome complex during oxidative stress. Sci Signal 3, ra88 (2010). 114. Salin, H. et al. Structure and properties of transcriptional networks driving selenite stress response in yeasts. BMC Genomics 9, 333 (2008). 115. Peter, J. et al. Genome evolution across 1,011 Saccharomyces cerevisiae isolates. Nature 556, 339–344 (2018). 116. Flick, K. & Kaiser, P. Protein degradation and the stress response. Semin Cell Dev Biol 23, 515–522 (2012). 117. Hanna, J., Meides, A., Zhang, D. P. & Finley, D. A ubiquitin stress response induces altered proteasome composition. Cell 129, 747–759 (2007). 118. Gomes, A. V. Genetics of proteasome diseases. Scientifica (Cairo) 2013, 637629 (2013). 119. Arima, K. et al. Proteasome assembly defect due to a proteasome subunit beta type 8 (PSMB8) mutation causes the autoinflammatory disorder, Nakajo-Nishimura syndrome. Proc Natl Acad Sci U S A 108, 14914–14919 (2011). 120. Liu, X., Li, Y. I. & Pritchard, J. K. Trans Effects on Gene Expression Can Drive Omnigenic Inheritance. Cell 177, 1022–1034 (2019). 121. Komander, D. & Rape, M. The ubiquitin code. Annu Rev Biochem 81, 203–229 (2012). 122. Ohtake, F., Saeki, Y., Ishido, S., Kanno, J. & Tanaka, K. The K48-K63 Branched Ubiquitin Chain Regulates NF-KB Signaling. Mol Cell 64, 251–266 (2016). 123. French, M. E., Koehler, C. F. & Hunter, T. Emerging functions of branched ubiquitin chains. Cell Discov 7, 6 (2021). 124. Murakami, Y. et al. Ornithine decarboxylase is degraded by the 26S proteasome without ubiquitination. Nature 360, 597–599 (1992). 125. Deveraux, Q., Ustrell, V., Pickart, C. & Rechsteiner, M. A 26 S protease subunit that binds ubiquitin conjugates. J Biol Chem 269, 7059–7061 (1994). 158 126. Husnjak, K. et al. Proteasome subunit Rpn13 is a novel ubiquitin receptor. Nature 453, 481–488 (2008). 127. Brem, R. B. & Kruglyak, L. The landscape of genetic complexity across 5,700 gene expression traits in yeast. Proc Natl Acad Sci U S A 102, 1572–1577 (2005). 128. Brem, R. B., Yvert, G., Clinton, R. & Kruglyak, L. Genetic dissection of transcriptional regulation in budding yeast. Science 296, 752–755 (2002). 129. Yvert, G. et al. Trans-acting regulatory variation in Saccharomyces cerevisiae and the role of transcription factors. Nat Genet 35, 57–64 (2003). 130. Khmelinskii, A. et al. Protein quality control at the inner nuclear membrane. Nature 516, 410–413 (2014). 131. Kredel, S. et al. mRuby, a bright monomeric red fluorescent protein for labeling of subcellular structures. PLoS ONE 4, e4391 (2009). 132. Matz, M. V. et al. Fluorescent proteins from nonbioluminescent Anthozoa species. Nat Biotechnol 17, 969–973 (1999). 133. Albert, F. W., Muzzey, D., Weissman, J. S. & Kruglyak, L. Genetic influences on translation in yeast. PLoS Genet. 10, e1004692 (2014). 134. Nguyen Ba, A. N. et al. Barcoded bulk QTL mapping reveals highly polygenic and epistatic architecture of complex traits in yeast. Elife 11, (2022). 135. Nguyen, N. N. et al. Proteasome B5 subunit overexpression improves proteostasis during aging and extends lifespan in Drosophila melanogaster. Sci Rep 9, 3170 (2019). 136. Chondrogianni, N. et al. Overexpression of proteasome beta5 assembled subunit increases the amount of proteasome and confers ameliorated response to oxidative stress and higher survival rates. J Biol Chem 280, 11840–11850 (2005). 137. Vilchez, D. et al. RPN-6 determines C. elegans longevity under proteotoxic stress conditions. Nature 489, 263–268 (2012). 138. Fabre, B. et al. Label-free quantitative proteomics reveals the dynamics of proteasome complexes composition and stoichiometry in a wide range of human cell lines. J Proteome Res 13, 3027–3037 (2014). 139. Sahu, I. et al. The 20S as a stand-alone proteasome in cells can degrade the ubiquitin tag. Nat Commun 12, 6173 (2021). 140. Tsvetkov, P. et al. Compromising the 19S proteasome complex protects cells from reduced flux through the proteasome. Elife 4, (2015). 141. Tomko, R. J. & Hochstrasser, M. Molecular architecture and assembly of the eukaryotic proteasome. Annu Rev Biochem 82, 415–445 (2013). 142. Park, S. et al. Hexameric assembly of the proteasomal ATPases is templated through their C termini. Nature 459, 866–870 (2009). 143. Park, S. et al. Reconfiguration of the proteasome during chaperone-mediated assembly. Nature 497, 512–516 (2013). 144. Sokolova, V., Li, F., Polovin, G. & Park, S. Proteasome Activation is Mediated via a Functional Switch of the Rpt6 C-terminal Tail Following Chaperone-dependent Assembly. Sci Rep 5, 14909 (2015). 159 145. Padovani, C., Jevtic, P. & Rape, M. Quality control of protein complex composition. Mol Cell 82, 1439–1450 (2022). 146. Zhang, M., MacDonald, A. I., Hoyt, M. A. & Coffino, P. Proteasomes begin ornithine decarboxylase digestion at the C terminus. J Biol Chem 279, 20959–20965 (2004). 147. Henderson, A., Erales, J., Hoyt, M. A. & Coffino, P. Dependence of proteasome processing rate on substrate unfolding. J Biol Chem 286, 17495–17502 (2011). 148. Reichard, E. L. et al. Substrate Ubiquitination Controls the Unfolding Ability of the Proteasome. J Biol Chem 291, 18547–18561 (2016). 149. Goldstein, A. L. & McCusker, J. H. Three new dominant drug resistance cassettes for gene disruption in Saccharomyces cerevisiae. Yeast 15, 1541–1553 (1999). 150. Zwolshen, J. H. & Bhattacharjee, J. K. Genetic and biochemical properties of thialysine-resistant mutants of Saccharomyces cerevisiae. J Gen Microbiol 122, 281–287 (1981). 151. Baryshnikova, A. et al. Synthetic genetic array (SGA) analysis in Saccharomyces cerevisiae and Schizosaccharomyces pombe. Methods Enzymol 470, 145–179 (2010). 152. Kuzmin, E., Costanzo, M., Andrews, B. & Boone, C. Synthetic Genetic Array Analysis. Cold Spring Harb Protoc 2016, pdb.prot088807 (2016). 153. Hoffman, C. S. & Winston, F. A ten-minute DNA preparation from yeast efficiently releases autonomous plasmids for transformation of Escherichia coli. Gene 57, 267–272 (1987). 154. Shaner, N. C. et al. A bright monomeric green fluorescent protein derived from Branchiostoma lanceolatum. Nat Methods 10, 407–409 (2013). 155. Gauss, R., Trautwein, M., Sommer, T. & Spang, A. New modules for the repeated internal and N-terminal epitope tagging of genes in Saccharomyces cerevisiae. Yeast 22, 1–12 (2005). 156. Glickman, M. H., Rubin, D. M., Fried, V. A. & Finley, D. The regulatory particle of the Saccharomyces cerevisiae proteasome. Mol Cell Biol 18, 3149–3162 (1998). 157. Russell, S. J., Steger, K. A. & Johnston, S. A. Subcellular localization, stoichiometry, and protein levels of 26 S proteasome subunits in yeast. J Biol Chem 274, 21943–21952 (1999). 158. Marquez-Lona, E. M., Torres-Machorro, A. L., Gonzales, F. R., Pillus, L. & Patrick, G. N. Phosphorylation of the 19S regulatory particle ATPase subunit, Rpt6, modifies susceptibility to proteotoxic stress and protein aggregation. PLoS One 12, e0179893 (2017). 159. Averbeck, N., Gao, X. D., Nishimura, S. & Dean, N. Alg13p, the catalytic subunit of the endoplasmic reticulum UDP-GlcNAc glycosyltransferase, is a target for proteasomal degradation. Mol Biol Cell 19, 2169–2178 (2008). 160. Horton, R. M., Hunt, H. D., Ho, S. N., Pullen, J. K. & Pease, L. R. Engineering hybrid genes without the use of restriction enzymes: gene splicing by overlap extension. Gene 77, 61–68 (1989). 161. Gietz, R. D. & Schiestl, R. H. High-efficiency yeast transformation using the LiAc/SS carrier DNA/PEG method. Nat Protoc 2, 31–34 (2007). 160 162. Ward, A. C. Rapid analysis of yeast transformants using colony-PCR. Biotechniques 13, 350 (1992). 163. R Core Team. R: A language and environment for statistical computing. (R Foundation for Statistical Computing, 2021). 164. Hahne, F. et al. flowCore: a Bioconductor package for high throughput flow cytometry. BMC Bioinformatics 10, 106 (2009). 165. Bates, D., Machler, M., Bolker, B. & Walker, S. Fitting linear mixed-effects models using lme4. Journal of Statistical Software 67, 1–48 (2015). 166. Kuznetsova, A., Brockhoff, P. B. & Christensen, R. H. B. lmerTest package: Tests in linear mixed effects models. Journal of Statistical Software 82, 1–26 (2017). 167. Benjamini, Y. & Hochberg, Y. Controlling the false discovery rate: A practical and powerful approach to multiple testing. Journal of the Royal Statistical Society Series B (Methodological) 57, 289–300 (1995). 168. Li, H. & Durbin, R. Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics 25, 1754–1760 (2009). 169. Li, H. et al. The Sequence Alignment/Map format and SAMtools. Bioinformatics 25, 2078–2079 (2009). 170. Edwards, M. D. & Gifford, D. K. High-resolution genetic mapping with pooled sequencing. BMC Bioinformatics 13 Suppl 6, S8 (2012). 171. Kent, W. J. et al. The human genome browser at UCSC. Genome Res 12, 996–1006 (2002). 172. Haeussler, M. et al. Evaluation of off-target and on-target scoring algorithms and integration into the guide RNA selection tool CRISPOR. Genome Biol 17, 148 (2016). 173. Doench, J. G. et al. Optimized sgRNA design to maximize activity and minimize off-target effects of CRISPR-Cas9. Nat Biotechnol 34, 184–191 (2016). 174. Boer, C. G. de & Hughes, T. R. YeTFaSCo: a database of evaluated yeast transcription factor sequence specificities. Nucleic Acids Res 40, D169–179 (2012). 161 Chapter IV. Quantitative Trait Gene discovery by reciprocal hemizygote scanning Randi R. Avery and Frank W. Albert* Department of Genetics, Cell Biology, & Development, University of Minnesota, Minneapolis, MN, USA * To whom correspondence should be addressed: falbert@umn.edu Introduction Genetic variation among individuals leads to phenotypic variation in most complex traits, including many common human diseases. One way to elucidate how genetic variation shapes a specific phenotype is through quantitative trait locus (QTL) mapping. QTLs are areas of the genome that contain variation that affects a trait of interest. Genetic effects are numerous and tend to have small effect sizes. Therefore, large sample sizes are needed to achieve statistical power. Studies in humans have revealed the polygenic nature of many traits, such as height and risk of schizophrenia (The International Schizophrenia Consortium, 2009; Yang et al., 2010). High-powered studies in yeast have found up to hundreds of QTLs for various growth conditions (Bloom et al., 2013, 2015; Nguyen Ba et al., 2022). As genetic studies become more statistically powerful, it has been shown that complex, quantitative traits tend to be regulated by many genes in diverse biological processes (Duveau et al., 2021). A gene that contains DNA variants that affect a phenotype is known as a quantitative trait gene (QTG). Most QTLs are wide and can contain dozens of 162 https://www.zotero.org/google-docs/?ovQ9cr https://www.zotero.org/google-docs/?ovQ9cr https://www.zotero.org/google-docs/?rzDfjC https://www.zotero.org/google-docs/?rzDfjC https://www.zotero.org/google-docs/?J5cDSh genes. Therefore, fine-mapping approaches are necessary in order to determine the causative QTG or QTGs within a QTL. The reciprocal hemizygosity test and allele engineering via homologous recombination are useful ways to fine-map QTLs by testing a single gene or small genomic region for causality (Lutz et al., 2019; Steinmetz et al., 2002; Stern, 2014). The reciprocal hemizygosity (RH) test has been used to test individual genes for causality of a measurable phenotype (Steinmetz et al., 2002; Stern, 2014). In an RH test, two genetically different strains of a model organism, such as yeast, are crossed to form a hybrid. Knocking out one allele of a candidate gene creates a “hemizygous” genotype for that gene. This strain is compared to the “reciprocal” strain, where the corresponding allele on the homologous chromosome is knocked out. The phenotypes of these two reciprocal hemizygote strains are then compared. A phenotypic difference between the reciprocal strains reveals the candidate gene to be causal (i.e., a QTG) (Stern, 2014). Because the RH test compares two hemizygous strains, it is not confounded by haploinsufficiency (i.e. trait differences due to carrying just one copy of a gene in an otherwise diploid organism). However, because QTLs are wide, determining causative QTGs by using these methods to test one gene at a time is laborious and time-consuming. Furthermore, testing genes based on their known or speculated function may miss important genes that a priori appear to not be involved in the phenotype (Boyle et al., 2017). Therefore, it is important to assay the multitude of genes that may affect a phenotype of interest in a high-throughput, unbiased approach. 163 https://www.zotero.org/google-docs/?45kurR https://www.zotero.org/google-docs/?45kurR https://www.zotero.org/google-docs/?O36lw7 https://www.zotero.org/google-docs/?op6tKg https://www.zotero.org/google-docs/?xEaIqN Knowledge of causal genes within QTLs would provide insight into the mechanisms through which DNA variants shape complex traits, such as those described in Chapter II, where I discuss variation affecting protein degradation via the ubiquitin-proteasome system. Traditional RH tests only examine one gene per assay, limiting throughput and potentially resulting in biases towards “obvious” candidate genes an investigator might choose to test. Therefore, attempts have been made to apply the RH test genome-wide (henceforth referred to as “RH scanning”), enabling the testing of many genes at once (Weiss et al., 2018; Wilkening et al., 2014). One attempt to perform RH scanning in yeast was through crossing a deletion strain to a genetically distant, intact strain to create each hemizygote strain individually; reciprocal hemizygote strains were created for 75% of yeast open reading frames, which were then pooled (Wilkening et al., 2014). However, the approach produced many false positive results due to the deletion strains being prone to chromosomal abnormalities and suppressor mutations that accumulated during long growth periods (Teng et al., 2013). RH scanning has been successfully implemented to study evolution of thermotolerance in a yeast inter-species hybrid by altering the technique that creates the hemizygote pool (Weiss et al., 2018). Weiss et al. used transposon insertional mutagenesis to create a hemizygote pool in the diploid hybrid (Weiss et al., 2018). Creating a pool of cells with one transposon insertion per cell enabled testing thousands of genes in one assay with limited opportunity for compensatory mutations to arise thereby elucidating QTGs with a high true 164 https://www.zotero.org/google-docs/?y5TP4i https://www.zotero.org/google-docs/?gH4ZY4 https://www.zotero.org/google-docs/?KFmWI7 https://www.zotero.org/google-docs/?wVaIz3 https://www.zotero.org/google-docs/?fuUIkT https://www.zotero.org/google-docs/?fuUIkT positive rate without using laborious fine-mapping approaches. Here, I applied the genome-wide RH scanning approach to determine causal genes for growth in YPD. I used the piggyBac transposon to mutagenize a hybrid between two genetically different Saccharomyces cerevisiae strains to yield a large reciprocal hemizygote pool. Illumina sequencing of transposon insertion sites revealed 4,260 open reading frames (ORFs) contained at least one insertion in both alleles, which comprises ~65% of all S. cerevisiae ORFs. I grew replicates of the hemizygote pool in nutrient-rich media at 30°C for ~70 cell divisions and tracked insertion frequencies at multiple timepoints. Genes with a significant allelic difference in change in insertion frequency over time were considered candidate QTGs. Using a custom computational pipeline and linear modeling, I identified 265 genes with at least a nominally significant (p < 0.05) allelic effect on growth. However, validation experiments using individually engineered strains hemizygous for each of the most significant genes did not recapitulate the results of the scan. Currently, we are improving the sequencing library preparation to increase the number of insertions we can count, thereby increasing statistical power, sensitivity, and precision. Future improvements to the approach include measuring phenotypes that produce stronger effects, such as growth in stress conditions. A successful approach revealing QTGs would aid in understanding how genetic variation affects important, complex cellular traits and would be a key advancement for the field. 165 Results Genome-Wide Reciprocal Hemizygote Pool I used transposon mutagenesis to apply the reciprocal hemizygosity test genome-wide in a single experiment (RH scanning) to determine quantitative trait genes (QTGs) for growth in YPD (Fig. 1). A plasmid containing the piggyBac transposon (Weiss et al., 2018) was transformed into a diploid hybrid of two genetically divergent strains of S. cerevisiae, BY and RM (Methods). I performed 20 separate transformations and pooled the resulting cultures to reduce the instances of jackpot insertions by the transposon. Each of the 20 transformations was performed on 50 mL of cells grown to an O.D. of 0.47. I aimed to create an RH pool such that each cell contained a single insertion so that trait measurements would be accurately attributed to the gene that was disrupted and not confounded by genetic interactions that may be caused by multiple insertions. Fig. 1: Experimental overview of RH scanning approach. 166 https://www.zotero.org/google-docs/?YUlwX5 In order to determine the proportion of cells in my final RH pool that indeed contained insertions I counted the number of colonies that grew on selection media vs. non-selective nutrient rich media as a proxy for transformation efficiency. A subset of transformed cells were diluted via serial dilution to 1:10,000 and 1:100,000. One hundred µL of these dilutions were then plated onto three plates each of 5FOA+G418 and YPD. The G418 selects for cells that carry the transposon, which contains the KanMX cassette. The 5FOA selects for cells that have lost the plasmid, which contains URA3. The plates were incubated at 30°C for three days and colonies were counted. On average, YPD plates contained 53 colonies, while section plates contained 19 colonies. This assay revealed that over one-third of the final RH pool received, and subsequently lost, the plasmid and maintained the transposon (Fig. 2). Fig 2: Colony counting assay to determine frequency of colonies with the transposon. The transposon contains the KanMX cassette, which confers resistance to G418. When cells from the same sample were plated on nutrient rich and selection media, about one-third of cells grew on selection media, suggesting about one-third of cells contained the transposon with the KanMX cassette. 167 In order to ensure that each cell only contained one insertion, I performed single colony streaking to isolate six different clones for analysis (Fig. 3). Each clone was prepared for whole genome sequencing following the approach in (Collins et al., 2023) and sequenced on the Illumina iSeq platform at the University of Minnesota Genomics Center (UMGC). Analysis of sequencing revealed that, for each clone, the genomic portion of any reads with transposon sequence mapped to the same location in the genome. Each of the six clones contained insertions that mapped to different loci (Table 1), supporting that the transposon inserts into loci randomly. Taken together, these data show that I obtained a genome-wide reciprocal hemizygote pool with one transposon insertion per cell. Fig 3: Approach for determining how many insertions each transformant receives. 168 https://www.zotero.org/google-docs/?XbX0e9 Table 1: Number of insertions and the loci that were observed for each colony tested. The piggyBac transposon preferentially inserts into TTAA sites in the genome, via a cut and paste mechanism (Fraser et al., 1995; Mitra et al., 2008). Transposon insertion sites were sequenced on the Illumina NovaSeq S1 chip 2x150bp using paired-end sequencing (Methods). This resulted in an average of 45 million reads per library. Transposon insertion site sequencing revealed that in my RH pool, 84% of the 77,123 TTAA sites in the BY reference genome received at least one transposon insertion. Visualization of the sequencing reads per insertion site reveals insertions across the entire genome (Fig. 4). The majority (87%) of 6,675 open reading frames in the S. cerevisiae genome were found to have at least one unique transposon insertion (Fig. 5). 169 https://www.zotero.org/google-docs/?hv8ZmP Fig 4: Number of reads aligning to each transposon insertion site along the genome. Plots are not adjusted for chromosome length. Fig. 5: Most genes (87%) contain at least one unique transposon insertion, with longer genes containing more. The genes with the most unique insertions are indicated. 170 Phenotypic selection of the reciprocal hemizygote pool I sought to determine the QTGs for growth in YPD. In order to maintain complex hemizygote pools, I executed the selection in large cultures of 250 mL YPD+G418 in 500 mL flasks. I inoculated each of the three flasks with aliquots of glycerol stocks of the RH pools to an O.D. of 0.05 (approximately 140 million cells) to serve as three biological replicates. The cultures were grown at 30°C with shaking. Every twelve hours I diluted the cultures to maintain the cells in log phase. I took 30 mL samples of cells from the cultures after 12, 60, and 108 hours, to use for sequencing. I spun down these samples in a tabletop centrifuge and froze the cell pellets for DNA extraction. Additionally, I took a sample of cells directly from the glycerol stocks (time point 0) for sequencing. Sequencing library creation and read processing To determine transposon insertion sites, and therefore frequency of reciprocal hemizygotes over time, I extracted DNA from each of the time points mentioned above, and performed transposon-sequencing following (Weiss et al., 2018) (see Methods). Briefly, DNA was extracted using Zymo Research Quick-DNA Fungal/Bacterial Miniprep Kit, sheared to 300 bp fragments, cleaned, and ligated with common adaptors. The transposon insertion sites in these fragments were amplified via PCR using one primer specific to one adaptor and another primer that is complementary to a segment near the end of the transposon. After the initial rounds of PCR, only fragments that contain the sequence at the end of the transposon are amplified. The final libraries were size 171 https://www.zotero.org/google-docs/?aqUqrA https://www.zotero.org/google-docs/?aqUqrA selected and cleaned using a KAPA Pure Bead kit. The libraries were then submitted to UMGC for quality control assays and sequencing on the NovaSeq S1 chip with paired-end sequencing. The twelve libraries resulted in an average of 50 million reads (Fig. 6A). To analyze transposon insertion sequencing results, I developed a pipeline using a combination of established command line tools and original Python and R scripts (Methods) (Fig. 6B). Briefly, reads without transposon insertions were excluded, and the transposon sequence was trimmed from remaining reads. This step retained on average 85% of the reads per library. The genomic portion was aligned to the S288C reference genome; 90% of remaining reads aligned to the genome. PCR deduplication excluded the most reads: only 14% of reads passed deduplication. This suggests that the 25 cycles of PCR used here in order to obtain enough material for sequencing may be a step where improvements can be made in the future. Next, non-allelic reads were removed by excluding alignments that had a mapping quality of zero. On average, only 22% of reads passed this filter (Fig. 6). Allelic reads were counted at positions where known variants between the BY and RM genomes are located. Finally, I wrote a Python script to extract BY and RM allelic counts for statistical analysis. These filtering steps of the 12 libraries resulted in an average of 118,000 total insertion counts per library across 4,260 open reading frames (ORFs). 172 Fig. 6: Loss of reads along sequencing analysis pipeline. A. Average number of reads of 12 libraries from starting Illumina reads, reads containing the transposon insertion, reads that mapped to the BY reference genome, and finally mapped reads that the pipeline could distinguish belonging to the BY or RM allele. B. Analysis tools used at each step of the pipeline, and the percent of reads kept at each step compared to the previous step. Variants in reads that were distinguishable between BY and RM alleles were counted, resulting in an average total of 118,000 variant counts across the 12 libraries. The insertion counts across all of the libraries were combined into one data frame for statistical analysis across all four time points and three replicates. ORFs with 0 insertions in either BY or RM were excluded from further analysis. The resulting total number of insertions per ORF ranged greatly from a minimum of two (1 in BY and 1 in RM) to 16,447 insertions, with a median of 149 (Fig. 7). The 10 ORFs with the most insertions are displayed in the table in Fig. 7. The distribution of insertions per ORF (Fig. 7) is right skewed, showing that some ORFs contain extremely high insertions, while the majority of ORFs contain relatively few insertions. This wide range could be due to multiple factors. The transposon itself could have a preference for certain sites, and jump into the same locus in separate cells multiple times. Or there may be an issue with PCR deduplication while processing the reads. The ORF with the most insertions 173 across the growth experiment was YLR334C, which is a dubious ORF, and does not likely encode for a functional protein. However, the ORF overlaps a long terminal repeat sequence (Fisk et al., 2006). Therefore, the transposon could have homology with this sequence and may have jumped there at a high frequency. Another explanation for the high counts in YLR334C is that there is not a true transposon insertion at this locus and reads that map to this locus are being counted as insertions due to sequence similarity to the transposon, which can cause reads to be maintained in our pipeline. Fig 7: Total number of transposon insertions (‘counts’) per ORF. Table inset lists the top 10 ORFs for insertion counts and the number of TTAA sites for the ORF. The plot shows the rest of ORFs with insertions. YOL081W is IRA2. ORF: open reading frame. Candidate Quantitative Trait Genes In order to determine QTGs, a negative binomial linear model was used to test for a significant difference in the rate at which BY and RM insertions per ORF 174 https://www.zotero.org/google-docs/?CSA7iO changed over time. Each ORF that contained at least one transposon insertion in both alleles was tested for statistical significance. The total number of insertion counts per library was used as an offset in the model to account for variation in the total number of counts per library (Fig. 8). Time point 0 has the least amount of counts, but this is unsurprising since these sequencing libraries were performed on cells that came from glycerol stocks, which produces a lower yield of DNA during extraction, as expected. Furthermore, the cultures were grown in the presence of G418 over the course of the experiment. This selected for cells that contained transposon insertions, therefore yielding higher transposon insertion counts at later time points. Model: total insertions in gene ~ time point + allele + time point:allele + offset(log(Offset) Fig 8: Average total number of counts for the three libraries at each of the four time points. The total number of counts for each library was used as the offset for the data from that library in the model. 175 In total, the model resulted in 265 genes at nominal significance (p < 0.05), which were considered candidate QTGs for growth in YPD. Two genes passed Bonferroni correction (p < 0.05 for 4,260 ORFs tested): BUB1 and VPS53 (Table 2). Insertions in the BY allele of BUB1 decreased over time, while insertions in the RM allele increased over time (Fig. 9A). This suggests that the BY allele of BUB1 confers a growth advantage for S. cerevisiae, relative to the RM allele. The gene product of BUB1 is a protein kinase involved in the spindle checkpoint during cell division that phosphorylates Bub3p, which is another spindle checkpoint protein (Roberts et al., 1994). Perhaps the BY allele produces Bub1p that is more efficient in passing the spindle checkpoint leading to faster cell division, and therefore a growth advantage. On the other hand, the BUB1 null mutant has been shown to lead to chromosome instability (Yuen et al., 2007), which can sometimes lead to a growth advantage (Dunham et al., 2002). Growth advantage due to copy number variation has been reported under stress conditions, whereas I grew my RH pool in nutrient-rich media. However, the cultures were not kept in perfect log phase, and therefore could have been affected by nutrient-poor states when the cultures were close to saturation. Therefore, it could be that the BUB1 hemizygote leads to copy number variation in other parts of the genome, not captured in our sequencing, giving these strains a growth advantage. The second strongest effect was seen for VPS53, a protein in the complex Golgi-associated retrograde protein, which is required for transport of Golgi membrane proteins (Conibear & Stevens, 2000). Insertions in the BY allele of 176 https://www.zotero.org/google-docs/?akPrzq https://www.zotero.org/google-docs/?t5ynvk https://www.zotero.org/google-docs/?NE0tYC https://www.zotero.org/google-docs/?qxWW6s VPS53 increased over time, while insertions in the RM allele decreased over time (Fig. 9B). This suggests that the RM allele of VPS53 confers a growth advantage compared to the BY allele. Null vps53 mutants cause issues with the transportation of Golgi membrane proteins to the vacuole, and null vps52 and vps54 mutants cause similar phenotypes (Conibear & Stevens, 2000). In the data from my scan, VPS54 had a p-value of 0.44, and VPS52 had no insertions. The relationship between VPS53 and a growth advantage in YPD is not clear. Perhaps a variant in the RM allele of VPS53 leads to increased efficiency in the Golgi-associated retrograde protein complex, causing a growth advantage compared to strains with the BY allele. BUB1 and VPS53 plus four additional genes with the strongest p-values (Table 2) were chosen for further analysis because they had relatively high insertion counts and none were essential genes, which simplified validation experiments described in the next section. These top six genes are involved in a variety of pathways and vary in gene length, number of variants, TTAA sites, and total insertion counts. They also do not have a clear influence on the growth phenotype, so if these were true QTGs, my results could provide new information about the functions of these genes. Of note, ALT1 (Fig. 9C), the third strongest effect from the scan (Table 2), does not contain any missense variants between BY and RM. However, the allelic effect could be caused by a cis-acting variant, such a promoter variant that could affect the expression of ALT1. Fig. 9C shows that insertions in the BY allele slightly increased over time, while insertions in the RM allele had a consistent frequency over time. 177 https://www.zotero.org/google-docs/?EGdfpm Table 2: Top six most significant genes based on the p-value resulting from the linear model. Source for Functions: SGD Fig 9: Normalized transposon insertion counts per allele over time for the top three most significant genes based on the p-value resulting from the linear model. A. BUB1 B. VPS53 C. ALT1. Grey shading indicates standard error. Because the total number of insertions varied widely across ORFs (Fig. 7), I tested whether insertion count was correlated with statistical significance. If they were indeed correlated, this would suggest that non-causal genes may become statistically significant due to stochastic differences between the two hemizygotes of an RH pair with high counts, and not because they had a biological effect on growth (i.e. resulting in many false positives). There was a weak, but significant 178 correlation between log(total insertions/ORF) and -log(p-value): Pearson’s r = 0.24, p < 2.2e-16 (Fig. 10). This suggests that genes with higher counts are slightly more likely to have a lower p-value. However, since the correlation was weak, I reasoned that a high number of insertions was not the main determinant of statistical significance and the top hits were still worth investigating. Fig 10: Log of total transposon insertions per ORF compared to the -log(p-value) from the linear model for each ORF. Nominally significant (p < 0.05) p-values highlighted in blue. I analyzed genes that have been found to affect quantitative traits in many studies, specifically for BY and RM: HAP1, MKT1, and IRA2 (Albert et al., 2018; Brem et al., 2002; Brion et al., 2013; Nguyen Ba et al., 2022; Smith & Kruglyak, 2008; Yeh et al., 2022). None of these three genes had significant p-values, especially after Bonferroni correction, in my RH scan (Table 3, Fig. 11) suggesting my results contain false negatives. 179 https://www.zotero.org/google-docs/?pkt1py https://www.zotero.org/google-docs/?pkt1py https://www.zotero.org/google-docs/?pkt1py Table 3: Genes I expected to produce significant results from the RH scan were not significant based on the linear model. Fig 11: Transposon insertion counts per allele over time for the three genes expected to produce significant results in the RH scan. A. HAP1 B. MKT1 C. IRA2. Grey shading indicates standard error. Validation via direct RH tests showed the RH scan did not reveal true QTGs I directly tested the effects of the alleles of the top genes via individual reciprocal hemizygosity tests that directly compare growth rates of engineered hemizygote strains. I transformed BY strain YFA0017 and RM strain YFA1611 (Methods) with the KanMX cassette amplified via PCR using primers to create hangovers homologous to the desired locus. Each gene was knocked out via a separate transformation, such that each new strain had one gene knocked out. VPS53 and COQ5 could not be successfully knocked out in BY and / or RM, in multiple attempts, and these genes were excluded from further analysis. Haploid 180 knock-outs were crossed with the corresponding intact haploid to create diploid BY-RM hybrids that were hemizygotic for each gene of interest (Fig. 12). I measured a total of nine hemizygotes per strain per gene, resulting in 18 reciprocal hemizygote strains per gene (Methods). Fig 12: Experimental approach overview to create directed RH strains for validation of the top hits from the RH scan. The growth rate of each of the 18 reciprocal hemizygote strains was measured on a plate reader three times. Fig. 13 shows the growth rate of eight reciprocal hemizygotes: four genes that were expected to affect growth rate based on the RH scan (candidate QTGs): BUB1, ALT1, FRE8, and KTD1; three genes that were expected to affect growth rate based on prior studies: IRA2, MKT1, and HAP1; and one gene that was expected to not be significant based on a high p-value (p = 0.99) from the RH scan: ERV29. Only IRA2 and MKT1 resulted in p-values less than 0.05. All four candidate QTGs had a p-value greater than 0.2. The gene we expected to have a high p-value (ERV29) actually had a lower p-value than the four candidate QTGs. From these results, we concluded that the scan produced false positives 181 and false negatives, and I was unable to confirm any true positives. However, the effects of the three known causal genes, IRA2, MKT1, and HAP1 were not very strong, and HAP1 was not significant. These results suggest that the effect of a variant(s) on the measured phenotype must be strong in order to be detected via a reciprocal hemizygote test, further indicating that obtaining true positives via a genome-wide RH scan may be difficult to achieve. 182 Fig 13: Growth rate of directed reciprocal hemizygotes for each gene tested in the validation assay. Black indicates genes from the RH scan with the lowest p-values. None of the four genes recapitulated the results of the RH scan. Purple indicates genes expected to show a significant difference in growth rate between the hemizygotes based on the literature. Blue indicates a gene tested expected to not show a significant difference in growth rate between hemizygotes based on its p-value from the RH scan. Discussion I created a genome-wide reciprocal hemizygote pool in S. cerevisiae to test for causal genes for growth in YPD. I used transposon mutagenesis to disrupt ORFs across the genome in a diploid hybrid to form the pool. Three biological replicate pools were each exposed to a competitive growth selection in YPD. At four time points, cells were collected from each culture and DNA was extracted. Transposon sequencing was performed to locate the insertion sites across the BY and RM genomes for each replicate and timepoint. Top hit genes from this RH scan were tested directly using individual RH tests to determine if these engineered strains recapitulated the results of the genome-wide scan. However, the difference in growth rate between the directed reciprocal hemizygote strains did not support the results of the genome-wide scan. This lack of recapitulation could be due to multiple reasons, each with possible solutions. First, we lose many reads during the PCR deduplication step (Fig. 6). This is likely due to the fact that out of the entire genome of each cell (approximately 12 Mb), we only need to sequence the transposon insertion site. We therefore need to use many (25) rounds of PCR to amplify enough DNA for sequencing. PCR does not equally amplify each DNA molecule, so certain 183 molecules may be over amplified, taking up sequencing reads, which we filter out in the deduplication step. Further, deduplication is based on reads identical in sequence and length. The length of the molecule depends on the transposon insertion site and where the molecule is sheared. If the insertion sites from unique insertion events that occurred at the same locus in different cells happen to be sheared at the exact same base, these would be counted as one insertion event because we currently could not distinguish between them. Therefore, the addition of unique molecular identifiers (UMIs) can be added during sequencing library preparation and used to deduplicate reads. We are currently testing this approach. Under my guidance, Kevin Zhan, an MD/PhD student in our lab, updated our library preparation protocol to include UMIs, created new libraries on cells from the same set of samples described above, and they have been sequenced by UMGC. Megan Lawler, a technician in our lab, is currently analyzing the sequencing results. Second, many reads are lost because they do not align to a locus of the genome where there is enough variation to determine if the insertion landed in the BY versus RM allele, which is imperative for our analysis. In the same new libraries, in addition to UMIs, we also sheared the DNA to a longer average length (500 bp versus 300 bp) and used a longer read length (2x300 bp versus 2x150 bp). Thus far, however, we have not observed a noticeable improvement in the number of insertions per gene. Perhaps using long-read sequencing could be a solution, but thus far has been cost prohibitive. 184 Third, the selection of growth in YPD might not have been strong enough to produce allele frequency differences in causal genes over time that were greater than stochastic changes in allele frequencies of insertions in other genes in the pools. A recent study found that fewer variants with fitness effects for growth in a nutrient rich medium than in stress conditions (Chen et al., 2023). Growth rate may be affected by many genes of small effect, making it less likely that each individual hemizygote for true causal genes could change in frequency to a great enough extent to become statistically significant despite technical noise. The approach may work in the BY-RM hybrid for stronger phenotypic selection, such as thermotolerance, as was originally used in (Weiss et al., 2018). In order to ensure that the RH scan works in the BY-RM hybrid, testing a phenotype with fewer causal genes and a greater selection pressure, such as growth in copper (Buzby et al., 2024), should be a future direction. Finally, if mutations arose over the course of the selection, we would not be able to see this from our sequencing data. Just as in (Wilkening et al., 2014), copy number variants may be arising, creating off-target effects, and skewing our results. Therefore, performing a stronger selection over a shorter period of time could reduce the chance of off-target mutations arising by decreasing the number of cell divisions. The very next step that should be taken to determine the viability of this approach is to perform computational simulations. The simulations should include the number of insertions, difference in insertions counts between alleles, and counts per gene needed to reach statistical significance based on the 185 https://www.zotero.org/google-docs/?YhpET1 https://www.zotero.org/google-docs/?QijB59 https://www.zotero.org/google-docs/?TW1UJF https://www.zotero.org/google-docs/?1yj9mn number of variants between BY and RM. This would reveal if the RH scanning approach is experimentally feasible for the BY-RM hybrid. If the RH scanning approach can be successfully applied to the BY-RM hybrid, this would impact the field of yeast complex trait genetics as we could elucidate causal genes directly and reveal the genetic architecture that contributes to phenotypic variation at a finer resolution. Methods Strains The S. cerevisiae strains, BY and RM, were used for this study. BY is laboratory strain and a close relative of the genome reference strain S288C. RM is a vineyard isolate and these two strains differ at an average of one single nucleotide per 200 base pairs (bp), providing abundant genetic variation for the dissection of complex traits. I crossed the BY strain YFA0017 (genotype: MATa his3∆1 leu2∆0 met15∆0 ura3∆0 YGR192C-GFP-HIS3MX) to RM strain EFA0001 (genotype: MATalpha ura3∆ his3∆::NATMX ho∆::HphMX AMN1-BY) by inoculating the strains on top of each other on solid YPD media. After overnight growth on YPD, confirmed presence of shmoos visually via microscopy, and single colony streaked the diploids on media that selects against the BY strain. I then took eight colonies and inoculated both a YPD plate and solid sporulation media. Five colonies were determined to have a diploid genotype if they formed tetrads on sporulation media. The corresponding colonies were taken from the YPD plate and saved as glycerol stocks to use for further experimentation. The resulting diploid used for further experimentation was YFA1234, genotype: 186 his3∆1/his3∆::NATMX leu2∆0/LEU2 met15∆0/MET15 ura3∆0/ura3∆ HO/ho∆::HphMX AMN1/AMN1-BY YGR192C-GFP-HIS3MX. Transposon mutagenesis I used the plasmid pJR487 (Fig. 14), gifted by the Rachel Brem lab (Weiss et al., 2018) for transposon mutagenesis. The plasmid contains a transposase, driven by a the highly expressive TDH3 promoter, and the piggyBac transposon containing the KanMX cassette, which confers resistance to G418. The plasmid does not contain a centromere, and therefore cannot replicate. This means that as cells divide, the plasmid will be diluted out, which helps prevent multiple transposon insertions per cell. To create the large RH pool via transposon mutagenesis, I used a large sample (0.2676 mg) of the plasmid for each of 20 transformations following the protocol in (Weiss et al., 2018). Briefly, diploid YFA1234 was single colony streaked and grown on YPD at 30°C for two days. One colony was used to inoculate 50 mL of YPD + G418 (G418 was always used at a concentration of 300 µg/mL) and incubated overnight at 30°C with shaking. The next day two one-liter flasks containing 500 mL YPD were inoculated with nine mL of overnight growth. After four hours of growth, the two cultures were combined and then separated into 20 50-mL tubes and centrifuged. These cells were used for the lithium acetate / single-stranded carrier DNA / poly-ethylene glycol (PEG) transformation method (Gietz & Schiestl, 2007) according to modifications in (Weiss et al., 2018). Each of the 20 tubes was used for an individual transformation to help reduce jackpot insertions. 187 https://www.zotero.org/google-docs/?TF2qtl https://www.zotero.org/google-docs/?TF2qtl https://www.zotero.org/google-docs/?sLAm4s https://www.zotero.org/google-docs/?lekwAC https://www.zotero.org/google-docs/?Y6onIn The transformed cells were plated onto nine 145 mm 5FOA plates and incubated at 30°C overnight. A subset of these cells were saved for a transformation efficiency assay. The next day, cells were scraped from the plates and saved as glycerol stocks at -80°C for further experimentation. Fig 14: Map of pJR487 plasmid gifted from the Rachel Brem lab that contains the transposon used for mutagenesis in the RH scan (Weiss et al., 2018) Transposon sequencing In order to determine transposon insertion sites, the approach from (Weiss et al., 2018) was adapted for our RH scan. First, genomic DNA from a given pool of hemizygotes was extracted using Zymo Research Quick-DNA Fungal/Bacterial Miniprep Kit (catalog #: D6005). One prep was used per sample to yield 21 – 49 ng/µL in 100 µL according to Qubit™ dsDNA BR Assay (catalog #: Q32850). Fifty microliters of each sample were sent to the University of Minnesota Genomics Center (UMGC) for shearing on the Covaris S220 to the average length of 300 bp. 188 https://www.zotero.org/google-docs/?WHqo5H https://www.zotero.org/google-docs/?fWeyUz https://www.zotero.org/google-docs/?fWeyUz Sonicated samples underwent a double bead cleanup using KAPA Pure Beads (Rouche #: KK8001). The first cut was 0.7X bead volume and the second was 0.9X. This double bead cleanup captured fragments between ~250 – 450bp. Samples were eluted in 10 mM Tris-HCl. The NEBNext® Ultra™ II DNA Library Prep (catalog #: E7103S) was used for end repair and dA-tailing of the DNA fragments as well as ligating custom adaptors according to the manufacturer’s protocol. The custom adaptors (Wetmore et al., 2015) consisted of two oligos of differing lengths that assist in preferentially amplifying fragments that contain the transposon sequence (see Table 1 for sequences and modifications.) Oligos were ordered from IDT with HPLC purification and combined to create a 50 uM stock. Seventy-five microliters of this stock was annealed using a thermocycler, via boiling and slowly cooling, and then diluted to 15 uM using ice-cold duplex buffer from IDT. The adaptor-ligated fragments were again cleaned with beads, this time with beads provided with the NEBNext Library Prep kit, first at 0.27X bead volume then 0.2X. Samples were frozen at -20°C until the next step was performed. In order to obtain a sequencing library of insertion sites I used selective PCR amplification of genome-transposon junction sites (Fig. 15). The forward primer contains a sequence that is complementary to a segment near the end of the transposon. This transposon-specific primer was ordered from IDT with PAGE-purification. The reverse primer is complementary to the custom adaptor and contains barcodes for multiplexing on Illumina sequencers. The reverse primers were ordered from IDT with HPLC purification. Each PCR reaction 189 https://www.zotero.org/google-docs/?9eKAnp contained 3-6 ng/µL cleaned, adaptor-ligated DNA; 2 µL at 10 uM each of forward and reverse primers; 25 µL of JumpStart™ Taq ReadyMix™ (SKU# P2893); and molecular grade water up to 50 µL. I found it to be imperative to use JumpStart Taq instead of the Q5 provided in the NEBNext Library Prep Kit. The Q5 polymerase has 3-5’ exonuclease activity, whereas JumpStart does not. The exonuclease activity can “chew back” on the transposon-specific primer, which leads to decreased specificity and essentially will amplify the entire genome, instead of transposon-genome junctions preferentially. The PCR program from (Weiss et al., 2018; Wetmore et al., 2015) was adapted as follows: 94°C at 2 minutes and 25 cycles of 94°C for 30 seconds, 60°C for 20 seconds, and 72°C for 30 seconds. A final extension was performed at 72°C for 10 minutes. Reactions were held at 8°C until the next step. The heated lid was set to 105°C. Fig 15: Schematic of transposon insertion site amplification used in sequencing library preparation. Finally, the completed library was once again size selected with a double KAPA bead cleanup (0.6X and 0.9X, respectively), according to the KAPA Pure Bead protocol. The final library was eluted in 27 µL of 10 mM Tris-HCl and the resulting concentration was measured with Qubit (ranging from 2.8-4.9 ng/µL per library). Twenty-five microliters of each library was submitted to UMGC for 190 https://www.zotero.org/google-docs/?Ozxugj Illumina Sequencing in an Eppendorf DNA LoBind® Plate (catalog #0030603303). UMGC performed three quality control checks before sequencing: sizing on an Agilent Bioanalyzer, concentration with Picogreen, and KAPA qPCR of the pooled libraries. All 16 libraries were pooled and sequenced on the Illumina NovaSeq S1 chip 2x150bp paired-end sequencing for a total of 887,434,379 read pairs. This included four libraries that were not used for data analysis in this work. All data shown are based on the other 12 libraries. Analysis of sequencing results To analyze transposon insertion sequencing results, I developed a pipeline using a combination of established command line tools and original Python and R scripts. First, Cutadapt (Martin, 2011) was used to search for the end of the transposon sequence that gets inserted into the genome. This sequence is then trimmed from the reads, leaving the genomic sequence where the transposon inserted. Reads that do not contain a transposon sequence are excluded. The remaining trimmed reads were aligned to the BY reference genome using the Burrows-Wheeler Aligner (Li & Durbin, 2009). Alignments were then sorted and PCR duplicates were removed using SAMtools (Li et al., 2009). SAMtools was also used to remove alignments that had a mapping quality of zero. Next, BCFtools (Danecek et al., 2021) was used to count allelic reads at positions where known variants between the BY and RM genomes are located. I wrote a Python script that extracts BY and RM allelic counts for statistical analysis. 191 https://www.zotero.org/google-docs/?5FQ0jy https://www.zotero.org/google-docs/?NOCPny https://www.zotero.org/google-docs/?xsamdY https://www.zotero.org/google-docs/?gN4BS2 Directed, individual reciprocal hemizygotes First, I knocked out each gene I wanted to test for causality in the BY and RM haploids. I knocked out the genes by replacing the coding sequence plus around 600 bp upstream with the KanMX cassette, which confers resistance to G418. The BY strain used for creating directed hemizygotes was YFA0017, the same strain I used to create the diploid for the RH pool. However, the RM strain used for the diploid in the RH pool, EFA0001, contained a hygromycin-resistance cassette (HphMX). The KanMX and HphMX cassettes contain sequence similarities, so to ensure that while inserting KanMX cassette into desired alleles it did not replace the HphMX cassette instead, I decided to remove the HphMX cassette from EFA0001. In order to do this, I utilized “CRISPR-Swap,” a high-efficiency genome engineering method developed in the Albert Lab (Lutz et al., 2019) to replace the HphMX cassette in RM with the HO allele, amplified from the BY strain, which prevents the strain from switch mating types. I then used colony PCR to confirm absence of the HphMX cassette and presence of the BY HO allele (Fig. 16). The resulting RM strain was named YFA1611 (genotype: MATalpha ura3∆ his3∆::NATMX HO-BY AMN1-BY) and used for the validation experiments. (Fig.16 shows three additional strains that were saved, but not used for further experimentation). 192 https://www.zotero.org/google-docs/?oGiE5i https://www.zotero.org/google-docs/?oGiE5i Fig 16: Electrophoresis gel image showing colony PCR results used to confirm absence of the Hyg cassette and presence of the HO sequence in each of four colonies. I grew the transformants on solid media with G418 so that strains that did not contain KanMX strains could not grow. I then used colony PCR to confirm that 1) the KanMX cassette inserted into the desired locus and 2) that the endogenous locus was removed. Once I obtained three confirmed knock-outs in each of the BY and RM haploid strains I crossed them to intact diploids (Fig. 12) to create diploid BY-RM hybrids, hemizygous for the given gene. I collected three colonies from each mating, to result in nine hemizygous BY and RM strains each, or 18 strains for each reciprocal hemizygote. I then measured growth rate in YPD of the reciprocal hemizygotes using a 96-well plate reader. A volume of 800 µL of YPD was inoculated with the strains from a glycerol stock stored at -80°C and grown overnight at 30°C with shaking. The next day, 15 µL of the overnight growth was used to inoculate 85 µL of YPD in a 96 well plate, and placed on a Synergy H1 193 microplate reader (BioTek Instruments, Winooski, VT, USA). The O.D. of the hemizygotes was measured every six minutes for 18 hours, which was enough time for all samples to reach saturation. The growth rate was defined as the inflection point of logistic growth. Data from the plate reader was analyzed using R code written by Maggie Kliebhan and the R package Growthcurver (Sprouffske & Wagner, 2016). Acknowledgements I thank the members of the Albert laboratory for feedback, especially Sheila Lutz in the beginning stages of the project. We thank the University of Minnesota’s Genomics Center for their contributions to the project. Author Contributions Conceptualization: RRA, FWA Formal Analysis: RRA Funding Acquisition: FWA Investigation: RRA Methodology: RRA, FWA Resources: FWA Supervision: FWA Validation: RRA Visualization: RRA Writing - Original Draft: RRA Writing - Review and Editing: RRA, FWA 194 https://www.zotero.org/google-docs/?uMH9Ck https://www.zotero.org/google-docs/?uMH9Ck References Albert, F. W., Bloom, J. S., Siegel, J., Day, L., & Kruglyak, L. (2018). Genetics of trans-regulatory variation in gene expression. eLife, 7, e35471. https://doi.org/10.7554/eLife.35471 Bloom, J. S., Ehrenreich, I. M., Loo, W. T., Lite, T.-L. V., & Kruglyak, L. (2013). Finding the sources of missing heritability in a yeast cross. Nature, 494(7436), 234–237. https://doi.org/10.1038/nature11867 Bloom, J. S., Kotenko, I., Sadhu, M. J., Treusch, S., Albert, F. W., & Kruglyak, L. (2015). Genetic interactions contribute less than additive effects to quantitative trait variation in yeast. Nature Communications, 6(1), 8712. https://doi.org/10.1038/ncomms9712 Boyle, E. A., Li, Y. I., & Pritchard, J. K. (2017). An Expanded View of Complex Traits: From Polygenic to Omnigenic. Cell, 169(7), 1177–1186. https://doi.org/10.1016/j.cell.2017.05.038 Brem, R. B., Yvert, G., Clinton, R., & Kruglyak, L. (2002). Genetic Dissection of Transcriptional Regulation in Budding Yeast. Science, 296(5568), 752–755. https://doi.org/10.1126/science.1069516 Brion, C., Ambroset, C., Sanchez, I., Legras, J.-L., & Blondin, B. (2013). Differential adaptation to multi-stressed conditions of wine fermentation revealed by variations in yeast regulatory networks. BMC Genomics, 14(1), 681. https://doi.org/10.1186/1471-2164-14-681 Buzby, C., Plavskin, Y., Sartori, F. M. O., Tong, Q., Vail, J. K., & Siegal, M. L. (2024). Epistasis and cryptic QTL identified using modified bulk segregant analysis of copper resistance in budding yeast. Genomics. https://doi.org/10.1101/2024.10.28.620582 Chen, S.-A. A., Kern, A. F., Ang, R. M. L., Xie, Y., & Fraser, H. B. (2023). Gene-by-environment interactions are pervasive among natural genetic variants. Cell Genomics, 3(4). https://doi.org/10.1016/j.xgen.2023.100273 Collins, M. A., Avery, R., & Albert, F. W. (2023). Substrate-specific effects of natural genetic variation on proteasome activity. PLOS Genetics, 19(5), e1010734. https://doi.org/10.1371/journal.pgen.1010734 Conibear, E., & Stevens, T. H. (2000). Vps52p, Vps53p, and Vps54p Form a Novel Multisubunit Complex Required for Protein Sorting at the Yeast Late Golgi. Molecular Biology of the Cell, 11(1), 305–323. https://doi.org/10.1091/mbc.11.1.305 Danecek, P., Bonfield, J. K., Liddle, J., Marshall, J., Ohan, V., Pollard, M. O., Whitwham, A., Keane, T., McCarthy, S. A., Davies, R. M., & Li, H. (2021). Twelve years of SAMtools and BCFtools. GigaScience, 10(2), giab008. https://doi.org/10.1093/gigascience/giab008 Dunham, M. J., Badrane, H., Ferea, T., Adams, J., Brown, P. O., Rosenzweig, F., & Botstein, D. (2002). Characteristic genome rearrangements in experimental evolution of Saccharomyces cerevisiae. Proceedings of the National Academy of 195 https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL Sciences, 99(25), 16144–16149. https://doi.org/10.1073/pnas.242624799 Duveau, F., Vande Zande, P., Metzger, B. P., Diaz, C. J., Walker, E. A., Tryban, S., Siddiq, M. A., Yang, B., & Wittkopp, P. J. (2021). Mutational sources of trans-regulatory variation affecting gene expression in Saccharomyces cerevisiae. eLife, 10, e67806. https://doi.org/10.7554/eLife.67806 Fisk, D. G., Ball, C. A., Dolinski, K., Engel, S. R., Hong, E. L., Issel-Tarver, L., Schwartz, K., Sethuraman, A., Botstein, D., Cherry, J. M., & Saccharomyces Genome Database Project. (2006). Saccharomyces cerevisiae S288C genome annotation: A working hypothesis. Yeast (Chichester, England), 23(12), 857–865. https://doi.org/10.1002/yea.1400 Fraser, M. J., Cary, L., Boonvisudhi, K., & Wang, H.-G. H. (1995). Assay for Movement of Lepidopteran Transposon IFP2 in Insect Cells Using a Baculovirus Genome as a Target DNA. Virology, 211(2), 397–407. https://doi.org/10.1006/viro.1995.1422 Gietz, R. D., & Schiestl, R. H. (2007). High-efficiency yeast transformation using the LiAc/SS carrier DNA/PEG method. Nature Protocols, 2(1), 31–34. https://doi.org/10.1038/nprot.2007.13 Li, H., & Durbin, R. (2009). Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics, 25(14), 1754–1760. https://doi.org/10.1093/bioinformatics/btp324 Li, H., Handsaker, B., Wysoker, A., Fennell, T., Ruan, J., Homer, N., Marth, G., Abecasis, G., Durbin, R., & 1000 Genome Project Data Processing Subgroup. (2009). The Sequence Alignment/Map format and SAMtools. Bioinformatics, 25(16), 2078–2079. https://doi.org/10.1093/bioinformatics/btp352 Lutz, S., Brion, C., Kliebhan, M., & Albert, F. W. (2019). DNA variants affecting the expression of numerous genes in trans have diverse mechanisms of action and evolutionary histories. PLOS Genetics, 15(11), e1008375. https://doi.org/10.1371/journal.pgen.1008375 Martin, M. (2011). Cutadapt removes adapter sequences from high-throughput sequencing reads. EMBnet.Journal, 17(1), 10. https://doi.org/10.14806/ej.17.1.200 Mitra, R., Fain-Thornton, J., & Craig, N. L. (2008). piggyBac can bypass DNA synthesis during cut and paste transposition. The EMBO Journal, 27(7), 1097–1109. https://doi.org/10.1038/emboj.2008.41 Nguyen Ba, A. N., Lawrence, K. R., Rego-Costa, A., Gopalakrishnan, S., Temko, D., Michor, F., & Desai, M. M. (2022). Barcoded bulk QTL mapping reveals highly polygenic and epistatic architecture of complex traits in yeast. eLife, 11, e73983. https://doi.org/10.7554/eLife.73983 Roberts, B. T., Farr, K. A., & Hoyt, M. A. (1994). The Saccharomyces cerevisiae Checkpoint Gene BUB1 Encodes a Novel Protein Kinase. Molecular and Cellular Biology, 14(12), 8282–8291. https://doi.org/10.1128/mcb.14.12.8282-8291.1994 Smith, E. N., & Kruglyak, L. (2008). Gene–Environment Interaction in Yeast Gene Expression. PLoS Biology, 6(4), e83. https://doi.org/10.1371/journal.pbio.0060083 196 https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL Sprouffske, K., & Wagner, A. (2016). Growthcurver: An R package for obtaining interpretable metrics from microbial growth curves. BMC Bioinformatics, 17(1), 172. https://doi.org/10.1186/s12859-016-1016-7 Steinmetz, L. M., Sinha, H., Richards, D. R., Spiegelman, J. I., Oefner, P. J., McCusker, J. H., & Davis, R. W. (2002). Dissecting the architecture of a quantitative trait locus in yeast. Nature, 416(6878), 326–330. https://doi.org/10.1038/416326a Stern, D. L. (2014). Identification of loci that cause phenotypic variation in diverse species with the reciprocal hemizygosity test. Trends in Genetics, 30(12), 547–554. https://doi.org/10.1016/j.tig.2014.09.006 Teng, X., Dayhoff-Brannigan, M., Cheng, W.-C., Gilbert, C. E., Sing, C. N., Diny, N. L., Wheelan, S. J., Dunham, M. J., Boeke, J. D., Pineda, F. J., & Hardwick, J. M. (2013). Genome-wide Consequences of Deleting Any Single Gene. Molecular Cell, 52(4), 485–494. https://doi.org/10.1016/j.molcel.2013.09.026 The International Schizophrenia Consortium. (2009). Common polygenic variation contributes to risk of schizophrenia and bipolar disorder. Nature, 460(7256), 748–752. https://doi.org/10.1038/nature08185 Weiss, C. V., Roop, J. I., Hackley, R. K., Chuong, J. N., Grigoriev, I. V., Arkin, A. P., Skerker, J. M., & Brem, R. B. (2018). Genetic dissection of interspecific differences in yeast thermotolerance. Nature Genetics, 50(11), 1501–1504. https://doi.org/10.1038/s41588-018-0243-4 Wetmore, K. M., Price, M. N., Waters, R. J., Lamson, J. S., He, J., Hoover, C. A., Blow, M. J., Bristow, J., Butland, G., Arkin, A. P., & Deutschbauer, A. (2015). Rapid Quantification of Mutant Fitness in Diverse Bacteria by Sequencing Randomly Bar-Coded Transposons. mBio, 6(3), e00306-15. https://doi.org/10.1128/mBio.00306-15 Wilkening, S., Lin, G., Fritsch, E. S., Tekkedil, M. M., Anders, S., Kuehn, R., Nguyen, M., Aiyar, R. S., Proctor, M., Sakhanenko, N. A., Galas, D. J., Gagneur, J., Deutschbauer, A., & Steinmetz, L. M. (2014). An Evaluation of High-Throughput Approaches to QTL Mapping in Saccharomyces cerevisiae. Genetics, 196(3), 853–865. https://doi.org/10.1534/genetics.113.160291 Yang, J., Benyamin, B., McEvoy, B. P., Gordon, S., Henders, A. K., Nyholt, D. R., Madden, P. A., Heath, A. C., Martin, N. G., Montgomery, G. W., Goddard, M. E., & Visscher, P. M. (2010). Common SNPs explain a large proportion of the heritability for human height. Nature Genetics, 42(7), 565–569. https://doi.org/10.1038/ng.608 Yeh, C.-L. C., Jiang, P., & Dunham, M. J. (2022). High-throughput approaches to functional characterization of genetic variation in yeast. Current Opinion in Genetics & Development, 76, 101979. https://doi.org/10.1016/j.gde.2022.101979 Yuen, K. W. Y., Warren, C. D., Chen, O., Kwok, T., Hieter, P., & Spencer, F. A. (2007). Systematic genome instability screens in yeast and their potential relevance to cancer. Proceedings of the National Academy of Sciences of the United States of America, 104(10), 3925–3930. https://doi.org/10.1073/pnas.0610642104 197 https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL https://www.zotero.org/google-docs/?4UtvNL Chapter V. Conclusion Randi R. Avery Department of Genetics, Cell Biology, & Development, University of Minnesota, Minneapolis, MN, USA My research converges on the question of how many and which loci contain natural genetic variation that affects complex traits in yeast. I addressed this question through two approaches: QTL mapping in various environments for the quantitative trait of ubiquitin-proteasome system (UPS) activity (Chapters II and III) and attempting to map causal genes directly for growth in YPD (Chapter IV). Chapters II and III, show that UPS activity is a highly polygenic trait, and the loci that affect UPS activity are dependent upon the environment, showing great complexity to this important cellular trait. The work in Chapter II evolved greatly over the course of the project. My first goal was to map QTLs for two new UPS reporters: UFD and 4xUb. The UFD reporter is a substrate that is recognized by the ubiquitin system, specifically the E3 ligase Ufd4p (Johnson et al., 1995), and subsequently degraded by the proteasome. The 4xUb reporter binds to the proteasome directly (Thrower et al., 2000), bypassing the ubiquitin system before being degraded. The additional cellular processes needed to complete UFD degradation compared to 4xUb is 198 https://www.zotero.org/google-docs/?57DGQ0 https://www.zotero.org/google-docs/?hk5iES https://www.zotero.org/google-docs/?hk5iES reflected in the genetic complexity affecting UPS activity measured by these two reporters. The UFD reporter had five QTLs in baseline medium, while the 4xUb reporter only had two. This suggests that the more cellular components and processes involved in the trait being measured, the more likely it is for the trait to have greater genetic complexity. I found that when more components are needed for degradation (i.e. the ubiquitin system) the genetic architecture is more complex than when fewer components are needed (i.e. the proteasome alone). Furthermore, although pathways that involve the ubiquitin system are more complex, the complexity is different (i.e. different loci influence the trait) between pathways, measured by different reporters. Because the UPS is highly environmentally-dependent (Bajorek et al., 2003; Finley & Prado, 2020; Grimm et al., 2012; Laporte et al., 2008; Sontag et al., 2014; Waite et al., 2016), we then decided to expose these two reporters to various environments that we expected to affect UPS activity. We additionally included four reporters that Dr. Collins used in previous work that showed the UPS is genetically complex in baseline medium (Collins et al., 2022, 2023). The addition of multiple environments transformed the project’s main focus to be on genotype-by-environment interactions. This is a contribution to the field of yeast complex trait genetics as the majority of GxE studies have solely measured transcript abundance (Boye et al., 2024; Grishkevich & Yanai, 2013; Smith & 199 https://www.zotero.org/google-docs/?THsbXB https://www.zotero.org/google-docs/?THsbXB https://www.zotero.org/google-docs/?THsbXB https://www.zotero.org/google-docs/?oeBIVA https://www.zotero.org/google-docs/?WJIX63 Kruglyak, 2008) or growth (Bloom et al., 2013, 2015; Nguyen Ba et al., 2022). From our results, we see that sources of GxE that affect the UPS are different for each component of the UPS and environment we tested. This reveals a daunting task for the field, that the genetic architecture will change for every strain, environment, or trait measured. However, we began to notice patterns in the loci that affect the UPS, as well as loci that contribute to GxE. Known hotspots, namely MKT1, HAP1, and IRA2 (Albert et al., 2014, 2018; Deutschbauer & Davis, 2005; Steinmetz et al., 2002; Wilkening et al., 2014; Yeh et al., 2022) contained a large portion of QTLs found, as well as comparisons of QTLs that exhibited GxE (about 20% in both cases). These genes have no obvious or direct link to the function of the UPS, and therefore likely affect the UPS through indirect means. This begs the question if these hotspots are also major contributors to GxE for other traits or other environments. If hotspots are a main contributor in many cases, we can narrow our focus of study to these loci to predict phenotype from genotype, even across environments. Focusing on these hotspots could help us elucidate the mechanisms that drive GxE. In the near future, directly measuring the extent of GxE at QTLs on additional quantitative traits using model systems will help solidify how generalizable or unique/nuanced characteristics of GxE are across species. This would facilitate the long term goal of predicting phenotypes from genotypes. 200 https://www.zotero.org/google-docs/?WJIX63 https://www.zotero.org/google-docs/?SgAk4V https://www.zotero.org/google-docs/?SgAk4V https://www.zotero.org/google-docs/?Z0xv0y https://www.zotero.org/google-docs/?Z0xv0y https://www.zotero.org/google-docs/?Z0xv0y There are two main limitations to this study. First, we chose a very conservative approach for quantifying GxE. GxE generally takes three forms: presence / absence, sign change, and magnitude change (Yadav & Sinha, 2018). Due to the bulk-segregant mapping approach we used, differences in magnitude may not directly correspond to differences in magnitude of effect from the locus. We therefore excluded GxE due to differences in magnitude of effect and counted QTLs with the same direction of effect as non-GxE comparisons, therefore likely missing many cases of GxE (Cubillos et al., 2014; Smith & Kruglyak, 2008; Yadav & Sinha, 2018). Second, we did not perform fine-mapping experiments to find the causal gene or variant(s) within the QTLs. It could be the case that sign change QTLs are due to two presence / absence cases with opposite direction of effect, in close proximity. If this is the case, this would actually increase our GxE count, as one sign change case would be counted as two presence / absence cases. But because we do not know the causal genes in the loci, it is difficult to find patterns of causal loci with similar modes of action, limiting our ability to use these results to describe biological mechanisms or for prediction beyond those of the involvement of hotspots. Because this work revealed dozens of loci and current fine-mapping approaches are time-consuming, laborious, and often focus on candidate genes with obvious related functions (therefore likely missing many causal genes with indirect mechanisms), an approach to directly map loci to gene-level resolution is advantageous. This was the goal of my work in Chapter IV. I used transposon 201 https://www.zotero.org/google-docs/?ccrSo7 https://www.zotero.org/google-docs/?ccrSo7 https://www.zotero.org/google-docs/?Cy2dJw https://www.zotero.org/google-docs/?Cy2dJw mutagenesis to create a genome-wide reciprocal hemizygosity pool to directly map causal genes for the BY-RM hybrid (termed “RH scanning”). This project also evolved greatly over the course of my PhD. After creating the RH pool, I first attempted to discover quantitative trait genes (QTGs) for the expression of TDH3, a highly expressed gene that has been extensively studied in S. cerevisiae (Albert et al., 2014; Duveau et al., 2021; Metzger & Wittkopp, 2019). I used fluorescence activated cell sorting (FACS) to sort for hemizygotes with extremely high and extremely low TDH3 expression, similar to the sorting approaches described in Chapters II and III. However, upon analysis of Illumina reads that covered transposon insertion sites, the results did not make sense based on our expectations. Therefore, in order to increase the number of cells we could sequence, and therefore reciprocal hemizygotes, we transitioned from TDH3 expression to growth in YPD, which allowed for the collection of large volumes of cells. After experimental validation of the top hits from the RH scan for growth in YPD, it was clear that the approach still produced false positive and false negative results, with no confirmed true positives. In Chapter IV I discuss multiple future directions for improving upon this approach. If successful, we would no longer need to perform laborious and time-consuming fine-mapping strategies to determine causal genes for phenotypes (Collins et al., 2022, 2023; Lutz et al., 2019; Steinmetz et al., 2002; Stern, 2014). We would be able to determine the genetic architecture of complex traits to a finer resolution, compared to QTL mapping. This would be greatly advantageous as most regulatory variation occurs at multiple loci throughout the 202 https://www.zotero.org/google-docs/?gFXH6z https://www.zotero.org/google-docs/?gFXH6z https://www.zotero.org/google-docs/?DkmTlS https://www.zotero.org/google-docs/?DkmTlS genome (Albert et al., 2018). We also would be more likely to discover novel cellular mechanisms because the approach would not be based on previously-known functions of genes, as is common with current fine-mapping approaches. This would give us a clearer picture of indirect effects. We will be able to more accurately predict phenotype from genotype if we know causal loci, in addition to our understanding of genetic architecture based on QTL mapping alone. This dissertation contributes to the understanding of the genetic architecture of UPS activity and how that architecture changes across environments. I have also built a foundation for developing an RH scanning approach in the BY-RM hybrid. This work contributes to the ultimate goal of predicting phenotype from genotype. It is clear that hotspots should be a major focus in pursuing that goal. However, with a successful RH scanning approach, one need not limit their searches to particular loci, and we therefore can gain a comprehensive understanding about how the entirety of the genome works together to result in the phenotypes we see. 203 https://www.zotero.org/google-docs/?F0kpwo References Albert, F. W., Bloom, J. S., Siegel, J., Day, L., & Kruglyak, L. (2018). Genetics of trans-regulatory variation in gene expression. eLife, 7, e35471. https://doi.org/10.7554/eLife.35471 Albert, F. W., Treusch, S., Shockley, A. H., Bloom, J. S., & Kruglyak, L. (2014). Genetics of single-cell protein abundance variation in large yeast populations. Nature, 506(7489), 494–497. https://doi.org/10.1038/nature12904 Bajorek, M., Finley, D., & Glickman, M. H. (2003). Proteasome disassembly and downregulation is correlated with viability during stationary phase. Current Biology: CB, 13(13), 1140–1144. https://doi.org/10.1016/s0960-9822(03)00417-2 Bloom, J. S., Ehrenreich, I. M., Loo, W. T., Lite, T.-L. V., & Kruglyak, L. (2013). Finding the sources of missing heritability in a yeast cross. Nature, 494(7436), 234–237. https://doi.org/10.1038/nature11867 Bloom, J. S., Kotenko, I., Sadhu, M. J., Treusch, S., Albert, F. W., & Kruglyak, L. (2015). Genetic interactions contribute less than additive effects to quantitative trait variation in yeast. Nature Communications, 6(1), 8712. https://doi.org/10.1038/ncomms9712 Boye, C., Nirmalan, S., Ranjbaran, A., & Luca, F. (2024). Genotype × environment interactions in gene regulation and complex traits. Nature Genetics, 56(6), 1057–1068. https://doi.org/10.1038/s41588-024-01776-w Collins, M. A., Avery, R., & Albert, F. W. (2023). Substrate-specific effects of natural genetic variation on proteasome activity. PLOS Genetics, 19(5), e1010734. https://doi.org/10.1371/journal.pgen.1010734 Collins, M. A., Mekonnen, G., & Albert, F. W. (2022). Variation in ubiquitin system genes creates substrate-specific effects on proteasomal protein degradation. eLife, 11, e79570. https://doi.org/10.7554/eLife.79570 Cubillos, F. A., Stegle, O., Grondin, C., Canut, M., Tisné, S., Gy, I., & Loudet, O. (2014). Extensive cis -Regulatory Variation Robust to Environmental Perturbation in Arabidopsis. The Plant Cell, 26(11), 4298–4310. https://doi.org/10.1105/tpc.114.130310 Deutschbauer, A. M., & Davis, R. W. (2005). Quantitative trait loci mapped to single-nucleotide resolution in yeast. Nature Genetics, 37(12), 1333–1340. https://doi.org/10.1038/ng1674 Duveau, F., Vande Zande, P., Metzger, B. P., Diaz, C. J., Walker, E. A., Tryban, S., Siddiq, M. A., Yang, B., & Wittkopp, P. J. (2021). Mutational sources of trans-regulatory variation affecting gene expression in Saccharomyces cerevisiae. eLife, 10, e67806. https://doi.org/10.7554/eLife.67806 Finley, D., & Prado, M. A. (2020). The Proteasome and Its Network: Engineering for Adaptability. Cold Spring Harbor Perspectives in Biology, 12(1), a033985. https://doi.org/10.1101/cshperspect.a033985 Grimm, S., Höhn, A., & Grune, T. (2012). Oxidative protein damage and the proteasome. Amino Acids, 42(1), 23–38. https://doi.org/10.1007/s00726-010-0646-8 Grishkevich, V., & Yanai, I. (2013). The genomic determinants of genotype × environment interactions in gene expression. Trends in Genetics, 29(8), 479–487. https://doi.org/10.1016/j.tig.2013.05.006 Johnson, E. S., Ma, P. C. M., Ota, I. M., & Varshavsky, A. (1995). A Proteolytic Pathway That Recognizes Ubiquitin as a Degradation Signal. Journal of Biological Chemistry, 270(29), 17442–17456. https://doi.org/10.1074/jbc.270.29.17442 204 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 Laporte, D., Salin, B., Daignan-Fornier, B., & Sagot, I. (2008). Reversible cytoplasmic localization of the proteasome in quiescent yeast cells. The Journal of Cell Biology, 181(5), 737–745. https://doi.org/10.1083/jcb.200711154 Lutz, S., Brion, C., Kliebhan, M., & Albert, F. W. (2019). DNA variants affecting the expression of numerous genes in trans have diverse mechanisms of action and evolutionary histories. PLOS Genetics, 15(11), e1008375. https://doi.org/10.1371/journal.pgen.1008375 Metzger, B. P. H., & Wittkopp, P. J. (2019). Compensatory trans ‐regulatory alleles minimizing variation in TDH3 expression are common within Saccharomyces cerevisiae. Evolution Letters, 3(5), 448–461. https://doi.org/10.1002/evl3.137 Nguyen Ba, A. N., Lawrence, K. R., Rego-Costa, A., Gopalakrishnan, S., Temko, D., Michor, F., & Desai, M. M. (2022). Barcoded bulk QTL mapping reveals highly polygenic and epistatic architecture of complex traits in yeast. eLife, 11, e73983. https://doi.org/10.7554/eLife.73983 Smith, E. N., & Kruglyak, L. (2008). Gene–Environment Interaction in Yeast Gene Expression. PLoS Biology, 6(4), e83. https://doi.org/10.1371/journal.pbio.0060083 Sontag, E. M., Vonk, W. I. M., & Frydman, J. (2014). Sorting out the trash: The spatial nature of eukaryotic protein quality control. Current Opinion in Cell Biology, 26, 139–146. https://doi.org/10.1016/j.ceb.2013.12.006 Steinmetz, L. M., Sinha, H., Richards, D. R., Spiegelman, J. I., Oefner, P. J., McCusker, J. H., & Davis, R. W. (2002). Dissecting the architecture of a quantitative trait locus in yeast. Nature, 416(6878), 326–330. https://doi.org/10.1038/416326a Stern, D. L. (2014). Identification of loci that cause phenotypic variation in diverse species with the reciprocal hemizygosity test. Trends in Genetics, 30(12), 547–554. https://doi.org/10.1016/j.tig.2014.09.006 Thrower, J. S., Hoffman, L., Rechsteiner, M, & Pickart, C. M. (2000). Recognition of the polyubiquitin proteolytic signal. The EMBO Journal, 19(1), 94–102. https://doi.org/10.1093/emboj/19.1.94 Waite, K. A., Mota-Peynado, A. D.-L., Vontz, G., & Roelofs, J. (2016). Starvation Induces Proteasome Autophagy with Different Pathways for Core and Regulatory Particles. Journal of Biological Chemistry, 291(7), 3239–3253. https://doi.org/10.1074/jbc.M115.699124 Wilkening, S., Lin, G., Fritsch, E. S., Tekkedil, M. M., Anders, S., Kuehn, R., Nguyen, M., Aiyar, R. S., Proctor, M., Sakhanenko, N. A., Galas, D. J., Gagneur, J., Deutschbauer, A., & Steinmetz, L. M. (2014). An Evaluation of High-Throughput Approaches to QTL Mapping in Saccharomyces cerevisiae. Genetics, 196(3), 853–865. https://doi.org/10.1534/genetics.113.160291 Yadav, A., & Sinha, H. (2018). Gene–gene and gene–environment interactions in complex traits in yeast. Yeast, 35(6), 403–416. https://doi.org/10.1002/yea.3304 Yeh, C.-L. C., Jiang, P., & Dunham, M. J. (2022). High-throughput approaches to functional characterization of genetic variation in yeast. Current Opinion in Genetics & Development, 76, 101979. https://doi.org/10.1016/j.gde.2022.101979 205 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4 https://www.zotero.org/google-docs/?jKrhQ4