45 min read

Systems genetics of DNA damage tolerance

Cisplatin, RAD5 & CRISPR-mediated nonsense

The following post includes the Abstract, Preface and Discussion from my PhD thesis dissertation (Bryant 2019). I have included them here as a convenient starting-point for sharing my dissertation work.

  • Abstract – a high level summary of my thesis.
  • Preface – an historical and personal introduction to each chapter of my thesis.
  • Discussion – informal lessons and insights gained during my dissertation work.
  • Chapter 1Epistasis and cisplatin tolerance – currently available in the published thesis (Bryant 2019).
  • Chapter 2Genomic repercussions of RAD5 overexpression – published and open-access (Bryant et al. 2019).
  • Chapter 3CRISPR-mediated nonsense – published and open-access (Billon et al. 2017).

Abstract 

DNA sequence information is constantly threatened by damage. In the clinic, intentional DNA damage is often used to treat cancer. Cisplatin, a first-line chemotherapy used to treat millions of patients, functions specifically by generating physical links within DNA strands, blocking DNA replication, and killing dividing cells. To maintain genome integrity, organisms have evolved the capacity to repair, respond, or otherwise resist change to the DNA sequence through a network of genetically encoded DNA damage tolerance pathways. In chapter 1, I present advances in experimental design and current progress for a systems genetics approach, using Saccharomyces cerevisiae, to reveal relationships between cisplatin tolerance pathways. Additionally, recent efforts to sequence thousands of cancer genomes have revealed recurrent genetic changes that cause overexpression of specific cisplatin tolerance genes. In chapter 2, I present a submitted manuscript that models overexpression of an essential cisplatin tolerance gene. This study uses a systems genetics approach to reveal the genetic pathways that are essential for tolerating this perturbation, which ultimately led to mechanistic insights for this gene. Convenient genome engineering in Saccharomyces has made this organism an ideal model to develop systems genetics concepts and approaches. In chapter 3, I present a published manuscript that demonstrates a new approach to disrupting genes by making site-specific nonsense mutations. Importantly, this approach does not require cytotoxic double-strand DNA breaks and is applicable to many model organisms for disrupting almost any gene, which may advance systems genetics into new model organisms. Systems genetics provides a framework for determining how DNA damage tolerance pathways act together to maintain cellular fitness and genome integrity. Such insights may one day help clinicians predict which cancers will respond to treatment, potentially sparing patients from unnecessary chemotherapy.

Preface 

Deoxyribonucleic acid (DNA) holds the instructions for life on Earth. Its code consists of only four letters – Adenine (A), Guanine (G), Thymine (T) and Cytosine (C) – the sequence of which determines ribonucleic acids (RNA) to be transcribed, and proteins to be translated in living cells (Dahm 2005; Avery, Macleod, and McCarty 1944; Crick et al. 1961; Nirenberg et al. 1966). Transcribing DNA into RNA and translating RNA into protein forms the foundation for the central dogma of molecular biology (Crick 1970), which emphasizes an important feature of DNA – its sequence can store the information necessary to reproduce the chemistry of life. This feature, along with fortuitous structural properties (Watson and Crick 1953), makes DNA a molecule that can not only reproduce the chemistry of life, but can also be reliably copied and passed to the next generation. However, to reliably and faithfully pass DNA to the next generation, cells must overcome a constant barrage of DNA damage (Lindahl 1993). Fortunately, the genomes of most cells contain instructions to build a large number of proteins that can work together to repair broken DNA, or enable DNA damage tolerance (Cleaver 2016). This genetically encoded capacity for cells to maintain genome integrity and viability in the face of damaged DNA is the focus of my thesis. In particular, I am interested in understanding how cells resist treatment to the DNA damaging agent, cisplatin – one of several platinum-based chemotherapeutic drugs that have been used to treat cancer in millions of patients (Galanski, Jakupec, and Keppler 2005; Simon et al. 2000; Wu et al. 2004; Kelland 2007).

Decades of work by geneticists have revealed DNA damage tolerance to consist of hundreds of genes that operate in numerous pathways that collectively enable cells to resist DNA lesions caused by cisplatin treatment (reviewed by Ciccia and Elledge 2010; Heyer, Ehmsen, and Liu 2010; Rocha et al. 2018). While the mechanisms for many of the major pathways responsible for cisplatin tolerance are well established, how these genetic pathways relate to each other as a system to maintain fitness and DNA integrity when responding to cisplatin treatment remains poorly understood. Moreover, many individual genes known to affect cellular fitness of cisplatin treated cells have no known role in any of the major cisplatin tolerance pathways. And, while much is known about how loss of individual gene function can affect cisplatin tolerance, recent efforts to genomically characterize thousands of tumors has revealed recurrent DNA copy number amplification events that can result in overexpression of specific genes (Cancer Genome Atlas Research Network 2008). The genomic repercussions of gene overexpression, and how this could alter cisplatin tolerance, remains a largely unexplored area in genetics.

The work described here attempts to address some of these unexplored and poorly understood aspects of DNA damage tolerance using systems genetics approaches. Chapter 1, describes the experimental design and current results for an ongoing collaborative effort to map all cisplatin specific DNA damage tolerance pathway relationships by systematically measure fitness of double-mutants to identify genetic interactions (i.e. double-mutant combinations that cause an unexpected change in fitness). Chapter 2, is a submitted manuscript that utilizes a genetic interaction profile to reveal the genomic repercussions of RAD5 overexpression, one of the most important cisplatin tolerance genes, whose human homolog is frequently overexpressed in specific cancers. Chapter 3, is a published manuscript that describes a new approach to efficiently disrupt the function of genes, which will open new avenues for studying the systems genetics of DNA damage tolerance in a diverse range of organisms. An introduction to the scientific and thesis context for each chapter is provided below in Chapter 1 introduction, Chapter 2 introduction & Chapter 3 introduction. Ultimately, understanding the fundamental genetic determinants of cisplatin tolerance may one day help clinicians predict which cancers will respond to cisplatin treatment, thus informing treatment decisions, and sparing patients from unnecessary chemotherapy (O’Connor 2015).

Epistasis and cisplatin tolerance 

The field of genetics seeks to answer one simple question – how does DNA sequence, combined with environmental exposure, result in the observable features of an organism? In other words, how does genotype, plus environment, produce phenotype? Historically, geneticists have approached the “genotype to phenotype” question systematically using genetic screens. In such screens, changes to an organism’s DNA sequence would be made randomly using some kind of DNA damaging agent to produce mutants. Mutants that gained the phenotype of interest would be cataloged, and the tedious process of characterizing each mutant allele to identify the causative DNA sequence change would commence. The classic genetic screen began to lose prominence with the advent of high-throughput sequencing technologies that allowed scientists to produce the first complete sequence assemblies for the human genome and several model organisms (Lander et al. 2001). Having sequenced a genome, scientists could computationally predict all of an organism’s protein coding genes – no known phenotype was necessary. Researchers could now perform their genetic screens in reverse. Instead of asking, “given a phenotype, what are the alleles?” scientists could now ask, “given the alleles, what are the phenotypes?” Thus, the reverse genetic screen was born.

Geneticists studying the budding yeast, Saccharomyces cerevisiae, were early adopters of the reverse genetic screen, which first required the construction of a large collection of defined alleles. Only a few short years after sequencing the yeast genome, every gene in yeast had been individually deleted to produce the non-essential gene disruption collection (Winzeler et al. 1999; Giaever et al. 2002). Remarkably, of the 6,000 protein coding genes in the yeast genome, approximately 5,000 were not essential for the organism to survive, which left a big open question – are these genes important for anything? The reverse genetic screen was the perfect tool for quickly answering this question. After exposing this strain collection to over 350 different chemicals, a group of researchers reassuringly found that indeed these genes were important, as 97% of genes showed a growth defect, or advantage, in at least one treatment condition (Hillenmeyer et al. 2008). This reverse genetic screening strategy, known as chemogenomic profiling, when used with a comprehensive collection of gene deletions, produces a list, or “profile”, of genes that contribute to fitness when treated with a particular chemical. Importantly, information gained from chemogenomic profiling is a two-way street; as we learn more about gene function, a chemogenomic profile can tell us something about the biological impact of the chemical or treatment itself. For example, early genetic screens in yeast from the 1960s and 1970s had already identified genes that are essential to tolerate exposure to ultraviolet, x-ray, and gamma radiation (Snow 1967; Cox and Parry 1968; Lemontt 1971; Game and Mortimer 1974). These genes were given names starting with “RAD” and given a number (e.g. RAD3, RAD6, RAD52, etc.). Later work had determined that many of the RAD genes produced proteins that performed various DNA repair processes to combat DNA damage (Haynes and Kunz 1981). With this knowledge of RAD gene function in mind, when we generate a new chemogenomic profile for an uncharacterized compound and find that its profile includes many RAD genes, we can infer that this new compound likely causes DNA damage. Thus, the more knowledge we have of gene function, the more powerful chemogenomic profiling becomes.

Cisplatin, being a critically important and clinically relevant drug, received plenty of attention during the heyday of chemogenomic profiling (Birrell et al. 2002; Wu et al. 2004; Lee et al. 2005; Huang et al. 2005; Hastie et al. 2006; Liao et al. 2007). However, while chemogenomic profiling has identified over 100 genes in yeast that are important for cisplatin tolerance, this list of genes does not tell us the biological process, or pathway, in which each gene participates. Moreover, a chemogenomic profile alone cannot tell us how the collection of pathways that affect fitness in response to cisplatin relate to each other as a complete system. For this, we leverage the power of double-mutants to identify which genes have shared contributions, or compensatory contributions to cisplatin sensitivity. This strategy was elegantly used by Brendel and Haynes (1973) to group genes known at the time to be involved in DNA repair into three major pathways, later defined as the RAD3, RAD6 and RAD52 epistasis groups. An epistasis group, in this context, refers to a set of mutations that share a contribution to the phenotype of DNA damage tolerance and can thus be grouped into shared damage tolerance pathways. Epistasis itself is a class of genetic interaction, a relationship between two genetic perturbations, that geneticists use to determine pathway membership. In the context of fitness, epistasis is most often observed when two deleterious genetic perturbations result in better fitness than expected when the two perturbations are combined. Thus, for clarity, I often describe such epistasis as a positive fitness interaction. In addition to identifying positive fitness interactions, double-mutant analysis can also identify negative fitness interactions – pairs of mutations that cause worse than expected fitness and indicate that each gene contributes to fitness in parallel compensatory pathways.

The major limitation to double-mutant analysis is time and energy. For example, in yeast, with a collection of 5,000 non-essential gene deletions, a researcher would have to perform approximately 12.5 million crosses to identify all possible fitness interactions among these gene deletions for a single treatment condition. Remarkably, after developing the Synthetic Genetic Array (SGA) method, researchers managed to use cleverness and robotics to produce a genetic interaction profile for nearly every gene deletion in budding yeast (Tong et al. 2001; Tong and Boone 2006; Pan et al. 2006; Costanzo et al. 2010; Baryshnikova et al. 2010). The combined interaction profiles generate an interaction network that represents the genetic landscape of Saccharomyces cerevisiae (Usaj et al. 2017; Costanzo et al. 2016).

Unfortunately, this genetic landscape only reveals genetic relationships for pathways that contribute to fitness in standard laboratory conditions. To produce this network for the same drug treatment conditions used by Hillenmeyer et al. (2008) to reveal a phenotype for nearly every single gene deletion, one would have to phenotype all 12.5 million crosses 350 times, generating over 4 billion measurements. Even worse, gene deletions are only one type of genetic perturbation. To generate an interaction landscape for every single nucleotide change in yeast, and for 350 treatment conditions, one would need to generate more than a quadrillion fitness measurements (>1015). Needless to say, a more focused approach for characterizing pathway relationships for genes involved in cisplatin tolerance is warranted. The strategy here is simple – focus on a set of genes that share a phenotype as single mutants and only generate crosses among this small set of genes. This strategy, when combined with SGA, is known as epistatic mini-array profiling, or E-MAP (Schuldiner et al. 2005, 2006; Collins et al. 2006, 2007; Fiedler et al. 2009; Braberg et al. 2013). And, when combined with a treatment it is referred to as a differential E-MAP, or dE-MAP (St. Onge et al. 2007; Bandyopadhyay et al. 2010; Guénolé et al. 2013; Srivas et al. 2013; Hustedt et al. 2015)

Chapter 1 presents an advancement in the experimental design for a cisplatin dE-MAP. Currently published dE-MAPs typically use only one, or two, fixed concentrations of drug treatment. However, detectable sensitivity for cisplatin tolerance mutants ranges across nearly two orders of magnitude of cisplatin concentration. For example, the most sensitive mutants, rad5∆, rad6∆ and rad18∆, have a measurable phenotype at 2 µM cisplatin, and are completely dead at 25 µM. Whereas, psy4∆, a gene named for cisplatin sensitivity, has a barely detectable phenotype at 80 µM. To measure a genetic interaction, ideally the experiment uses a treatment condition where at least one of the two mutations has a phenotype, and neither mutation causes death. Therefore, to optimize detection of genetic interactions, we use a common “low” concentration where most strains will remain viable, and an optimized “high” concentration that produces a phenotype of the query mutation – the gene deletion being crossed to the mini-array of cisplatin tolerance genes. While this project remains ongoing, we have successfully used this strategy to produce a high-resolution set of genetic interactions covering 70% of the cisplatin tolerance network.

Genomic repercussions of RAD5 overexpression 

When I began my PhD, I spent a few months meeting with professors looking for scientific inspiration. Having a background in de novo gene synthesis, I was naturally drawn to the topic of DNA recombination and repair, so I set up a meeting with Professor Rodney Rothstein – a leader in this field. During our meeting, Rodney described a project that had nothing to do with DNA repair. Instead, he described an ongoing project in his lab that aimed to identify genetic interactions that could be used to treat cancer. Fundamentally, the project was an extension of the genetic screening for cancer drug targets approach described by Hartwell et al. (1997). The described approach attempts to address a major problem in the current strategy for treating cancers, which is that typical chemotherapy and radiation treatment programs have many dangerous side-effects because they do not specifically kill cancer cells, rather they preferentially kill all dividing cells. The logic of the genetic approach described by Hartwell, begins with the observation that there are numerous recurring genetic alterations that are observed in cancer, which represent an an attack surface for selectively killing cancer cells. Geneticists had long observed a phenomenon they called synthetic lethality, wherein combining two otherwise harmless mutations results in cell death (i.e. an extreme example of a negative fitness genetic interaction). So the theory goes, if a cancer has a mutation and we know the synthetic lethal interactions of that mutation, then we can selectively target the cancer cell by inhibiting a synthetic lethal interactor with a drug. This drug would selectively kill only cancer cells harboring the interacting mutation, while leaving normal cells unharmed. Remarkably, this approach has received some traction with the development of clinically approved PARP inhibitors, which exploit a synthetic lethal genetic interaction between PARP and commonly observed mutations in the breast cancer susceptibility genes, BRCA1 and BRCA2 (Ashworth and Lord 2018; Lord, Tutt, and Ashworth 2015). Needless to say, I found the synthetic lethality approach to targeted cancer therapy to be incredibly interesting, so I asked to join the Rothstein lab.

The first major challenge to using synthetic lethality for treating cancer is to identify common genetic alterations in cancer. Then, by pairing these common genetic alterations with a collection of mutant strains one can identify synthetic lethal interactions. When I started in the Rothstein lab, the project I joined was attempting to leverage yeast genetics to identify synthetic lethal interactions for genes that are commonly overexpressed in PTEN deficient cancer. After performing a number of these genetic screens, it became clear that the data we were using to identify these overexpressed genes could not distinguish between genes that are overexpressed in individual cells, and genes that are just overrepresented in the population of cells used to measure expression (e.g. S-phase regulated genes will be overrepresented in a sample that has more dividing cells). Sadly, this issue undermined the initial motivation for performing my first synthetic lethal screens. Undeterred, I looked to other sources to find data for frequent genetic alterations in cancer. Around this time, The Cancer Genome Atlas consortium (Cancer Genome Atlas Research Network 2008) had released data for thousands of genomically characterized tumors. Importantly, these data included not only gene expression information, but also copy number information. After diving into the data, it became clear that each cancer cohort had distinct patterns of recurrent focal and chromosomal copy number amplifications resulting in overexpression of specific genes. Importantly, because gene copy number amplification is a physical change that occurs within individual cells, the change in expression could be attributed to an actual increase in transcript levels within cells rather than just a population level change.

In Chapter 2, I present a submitted manuscript that was motivated by the observation that the RAD5 human homolog, HLTF, incurs frequent copy number amplification and corresponding overexpression in several types of human cancer (typically squamous cell carcinoma). RAD5 is a member of the most essential cisplatin tolerance pathway, and, when overexpressed, we found that Rad5 bypasses canonical regulatory signals to cause cisplatin sensitivity and genome instability by driving hyperactive recombination at replication forks.

CRISPR-mediated nonsense 

Yeast has been an ideal model organism for studying genetic interactions at scale to identify genetic determinants of phenotype, and to reveal pathway relationships between genes. One reason yeast has achieved such traction as a genetic model arises from the ease with which scientists can modify the yeast genome in the lab. The ability to make targeted changes in an organism’s genome emerged from pioneering work in the field of genetic recombination, which led to the development of one-step gene disruption – a fast and efficient approach to knockout a gene (Rothstein 1983; Baudin et al. 1993; Wach et al. 1994). Eventually, with the release of the S. cerevisiae genome (Goffeau et al. 1996; Goffeau and Vassarotti 1991) – the first complete DNA sequence assembly of any eukaryotic organism – work began to construct a comprehensive collection of yeast knockout mutants representing nearly all of the roughly 5,000 non-essential genes found in budding yeast (Winzeler et al. 1999; Giaever et al. 2002). This collection paved the way for the development of reverse genetic screening approaches, which utilize a large set of defined mutants to identify those that give rise to a particular phenotypic feature, such as cisplatin sensitivity (see Chapter 1), or a genetic interaction profile (see Chapter 2). In principle, one-step gene disruption should be feasible in any recombination proficient organism, which includes most of the tree of life. In fact, adaptation of the one-step gene disruption approach in mice was the basis of the 2007 Nobel Prize in Physiology or Medicine (Vogel 2007; Koller et al. 1990; Zijlstra et al. 1990, 1989; Koller and Smithies 1989), and the mouse genetics community has followed suit in sequencing the mouse genome (Waterston et al. 2002), and constructing their own large collections of knockout mice (Grimm 2006). However, one-step gene disruption, and other recombination mediated genome engineering methods, rely on rare DNA repair events, rendering these approaches inefficient in most organisms – entire PhDs could be dedicated to constructing a single knockout mouse. Thus, at the beginning of my PhD in the fall of 2011, yeast remained a dominant model in the field of systems genetics and was the focus of my research on the genetic relationships between cisplatin tolerance pathways and the genetic repercussions of RAD5 overexpression.

However, in the spring of 2013, when I was working as a teaching assistant for Professor Ron Prywes and Professor Songtao Jia’s course in molecular biology, a fellow teaching assistant, Ling Ye, informed me of a new advancement in the field of genome engineering that he was sure would change the future of genetics research. This new advancement in genome engineering technology that Ling described to me is the now famous CRISPR-Cas9 system, which has indeed changed the face of genetics and biomedical research (Jinek et al. 2012; Gasiunas et al. 2012; Deltcheva et al. 2011; Sapranauskas et al. 2011; Garneau et al. 2010; Cong et al. 2013; Mali et al. 2013). The elegance of the CRISPR-Cas9 system for genome engineering is that it consists of a single protein (Cas9) and a short programmable guide RNA that targets Cas9 to a specific location in the genome where, upon recognition of the target sequence, it will generate a DNA double-strand break (DSB).

Suddenly, that rare DNA repair event necessary for recombination mediated genome engineering could be efficiently induced, opening up new avenues for modifying the genomes of almost any organism. Moreover, this simple system was amenable to simple modifications which allowed scientists to localize other proteins to specific DNA sequences without causing DNA breaks. Using this approach, Komor et al. (2016) and others, modified the CRISPR system to localize an enzyme that would directly convert cytosines (C) to thymines (T) in a small window of five nucleotides near the guide targeted sequence without having to create a DSB – an attractive feature for those interested in modifying cells that might already be deficient in DNA repair. When this system, known as CRISPR-mediated base editing, was first described, I had recently begun a collaboration with Dr. Alberto Ciccia’s lab to help with sequence analysis for a pilot run of a CRISPR-Cas9 based reverse genetic screen. Dr. Ciccia and a postdoctoral research associate in his lab, Dr. Pierre Billon, realized that CRISPR-mediated base editing of C to T could be used to disrupt gene function by converting four specific codons to premature stop codons in protein coding genes. What was unclear though, was whether this approach for knocking out gene function would be generalizable given the limited number of targetable codons and other technical restrictions dictated by the CRISPR system.

Chapter 3 is a published manuscript I co-authored with Dr. Billon and Dr. Ciccia addressing the scope and feasibility of CRISPR-mediated base editing for disrupting gene function (Billon et al. 2017). The title of the original publication – CRISPR-mediated base editing enables efficient disruption of eukaryotic genes through induction of STOP codons (iSTOP) – sums up the ultimate conclusion; CRISPR-mediated C to T base editors allow DSB free induction of stop codons at 23 different codons per gene on average (i.e. 1 iSTOP targetable codon for every 26 codons in the human genome). Thus, iSTOP provides a new tool alongside CRISPR-Cas9, and classic one-step gene disruption, that no longer relies on DSBs and recombination to efficiently knockout gene function. iSTOP, and other gene editing tools, will enable scientists to look beyond the humble baker’s yeast, to many other branches of life, to gain insight from systems genetics into the genetic determinants of life.

Discussion 

Genome integrity is essential for organisms to faithfully pass genetic information to future generations. While genomic DNA is constantly subject to damage, organisms have evolved numerous mechanisms to maintain sequence information and allow replication of their genomes despite this damage. Defects in DNA damage tolerance pathways, either inherited from birth, or acquired during the life of an organism, can result in numerous diseases such as Fanconi anemia, xeroderma pigmentosum, ataxia telangiectasia, Nijmegen breakage, Cockayne, Werner & Bloom syndromes (Knoch et al. 2012). In addition to these diseases, defects in DNA damage tolerance often arise during the development of cancer. Perhaps the best known examples of this are mutations in the breast cancer susceptibility genes, BRCA1 and BRCA2, which are both thought to be involved in homology directed repair (Roy, Chun, and Powell 2011; Zhang 2013). By leveraging knowledge of genetic interactions, new drugs have been developed to selectively kill cancers harboring mutations in BRCA genes by exploiting synthetic lethality between PARP and BRCA (Lord, Tutt, and Ashworth 2015; Ashworth and Lord 2018). Long before it became clear that many cancers harbor mutations in DNA repair pathways, oncologists have treated patients with DNA damage through radiation and chemotherapy. As DNA sequencing costs have dropped, ever more tumors are being sequenced to identify common mutations and changes in gene expression, which has led to an increased appreciation that the status of DNA damage tolerance pathways in tumors can have a profound effect on a tumor’s response to our most common cancer therapies (Alexandrov et al. 2013; Dietlein, Thelen, and Reinhardt 2014). As insight into the DNA damage tolerance network develops, strategies for selectively killing repair deficient tumors with drug combinations designed to target redundancies in the repair network are beginning to emerge (Cheng et al. 2013; Nickoloff et al. 2017). While some tumors respond well to radiation and DNA damaging chemotherapies, many tumors either do not respond, or acquire resistance to such treatments. This phenomenon underscores the importance of determining how the genetic landscape of a tumor impacts the overall damage tolerance network. The emerging field of systems genetics (Baliga et al. 2017), and specifically the study of genetic interaction networks, offers an important perspective for researchers and clinicians trying to understand how the DNA damage tolerance network enables tumor evolution while maintaining cancer cell viability. The structure of this network can enable precision medicine by providing critical insights into pathway organization that can reveal tumor specific genetic weaknesses (Hartwell et al. 1997). Below, I will discuss some lessons and insights from my work on the systems genetics of DNA damage tolerance – a stepping stone on the long path to precision medicine in oncology.

Epistasis: undetected ≠ nonexistent 

In Chapter 1, I presented progress toward constructing a genetic landscape for the cisplatin tolerance network in Saccharomyces. The design of this experiment advances the dE-MAP approach for detecting treatment specific pathway membership, and overall pathway relationships (Bandyopadhyay et al. 2010; Ideker and Krogan 2012). Specifically, we took advantage of the null model for genetic interactions of colony fitness to account for the fact that detection of genetic relationships between two alleles critically depends on the phenotype of each allele (Chapter 1, Figure 2). While this fact has been understood by geneticists for decades, it has some unappreciated and important consequences for interpreting existing genetic interaction networks. Perhaps the most important consequence of requiring a phenotype to detect epistasis is that an undetected interaction is not equivalent to no interaction. In fact, conspicuous absence of an expected synergistic interaction may be weak evidence for epistasis in an as yet unknown treatment condition. Two epistatic alleles will not show any genetic interaction if neither allele has a phenotype in the tested condition. To detect the genetic relationship between these alleles necessitates identification of a condition that generates a phenotype for at least one allele. This limitation does not exist for synergistic alleles, which can account for the fact that the number of negative genetic interactions far exceed the number of positive genetic interactions in the genetic landscape of S. cerevisiae (Costanzo et al. 2016). For our dE-MAP, we treated mutants with optimized concentrations of cisplatin to enhance detection of both positive and negative genetic interactions. The current network supports the notion that nucleotide excision repair (NER) is the primary pathway responsible for repairing cisplatin lesions. However, the lesion bypassing postreplication repair (PRR) pathway appears to coordinate at least three critical nodes in the cisplatin tolerance network, and is thus the most essential pathway for maintaining fitness in the presence of cisplatin. While our network was able to reveal these insights, it is clear that, even with optimized treatment conditions, detecting epistasis remains challenging using the dE-MAP approach alone. An approach to overcome this limitation may be to use the conspicuous absence of genetic interactions between otherwise highly correlated interaction profiles to pick pairs of alleles for further fitness analysis using a quantitative spot assay and a full cisplatin titration.

TCGA, BioGRID & The Cell Map: insights from public data 

Rad5, a member of PRR, was the focus of Chapter 2. During a visit with Dr. Karlene Cimprich, she mentioned that there was some evidence that the RAD5 homolog, HLTF, was overexpressed in cancer (also noted in the discussion by Kile et al. (2015)). We therefore turned to publicly available sequencing data from The Cancer Genome Atlas (TCGA) using software that enabled programmatic access to gene-level summaries of this data (Cancer Genome Atlas Research Network 2008; Jacobsen and Questions 2018; Cerami et al. 2012; Gao et al. 2013). Indeed, the RAD5 homolog, was frequently amplified and overexpressed in several classes of squamous cell carcinomas. This data-driven insight motivated the study of the genomic repercussions of RAD5 overexpression. By screening the yeast gene disruption and temperature-sensitive mutant collections we identified a clear signature of genetic requirements for RAD5 overexpression. This signature led us to the discovery that, when overexpressed, Rad5 can bypass canonical PRR regulatory signals to cause hyperactive recombination at replication forks. This hyperactive recombination is also accompanied by genome instability and cisplatin sensitivity.

What was especially fascinating about the RAD5 story was that both deletion and overexpression caused cisplatin sensitivity, but only overexpression could drive recombination. This distinction was revealed by a careful comparison of the RAD5 overexpression and deletion genetic interaction profiles. To compare interaction profiles, I used publicly available data from the BioGRID interaction database (Stark et al. 2006), and I developed a landscape enrichment analysis that uses easily accessible genetic interaction data from The Cell Map (Usaj et al. 2017; Costanzo et al. 2016; van der Maaten and Hinton 2008) (Chapter 2, Figure 1 & Chapter 2, Methods: Landscape enrichment analysis). This landscape enrichment analysis proved to be an intuitive way to visualize pathway enrichment for a genetic interaction profile, and allowed facile exploration of the genetic landscape to address questions such as, “what interactions are conspicuously absent?” In the case of rad5∆, interactions with crossover resolution (e.g. top3 & sgs1∆) were clearly absent even though many of the mutant alleles in that region of the landscape had been identified as having negative fitness interactions with rad5∆ (Chapter 2, Figure 2). Here, I would like to highlight that, in constructing our cisplatin dE-MAP, we also observed no interaction between rad5∆ and crossover resolution in the untreated network, but, when we treated with cisplatin, we observed epistasis – an example of conspicuous absence of synergy as weak evidence for epistasis (Chapter 1, Figure 4)! Revealing the rad5∆ sgs1∆ epistasis interaction in untreated conditions was possible, but only when using a very sensitive low-throughput method (Chapter 2, Figure 2).

Landscape enrichment: embracing perplexity 

As mentioned above, the landscape enrichment analysis from Chapter 2, Figure 1C enabled facile exploration of publicly available genetic interaction networks. This analysis is similar to Systematic Functional Annotation and Visualization of Biological Networks (SAFE), described by Baryshnikova (2016), but differs in its approach to dimension reduction (i.e. the two-dimensional clustering approach used to generate the “landscape”). The implementation presented in the SAFE publication uses a spring-embedded network layout, whereas the landscape presented for the RAD5 chapter was generated using t-SNE (van der Maaten and Hinton 2008). The t-SNE method for dimension reduction is best summarized by Laurens van der Maaten himself (https://lvdmaaten.github.io/tsne), which can take an \(A \times N\) matrix and reduce it to an \(A \times 2\) matrix where the two dimensions attempt to preserve local similarities between the profiles of \(A\). For generating the genetic landscape of Saccharomyces cerevisiae, rows in the matrix (\(A\)) are genes, and columns in the matrix (\(N\)) are genetic profile correlation values. Importantly, \(N\) can be any number of features that might provide information for generating clusters of genes or alleles. Thus, the landscape enrichment approach can be easily adapted to integrate other sources of data, and could be used to integrate existing biological information to generate a genetic landscape for human genes. t-SNE is lauded for its ability to reveal structure at many different scales. An example of this can be seen in Chapter 2, Figure 1C middle and right, where zooming in on the dominant cluster, reveals additional local clustering of HDR, DNA replication, and crossover resolution. When viewing a t-SNE map it is important to be aware of the limitations of the approach, and avoid some common misinterpretations, nicely described by Wattenberg, Viégas, and Johnson (2016). Notably, relative distances between clusters on the landscape are largely meaningless. Clusters may appear on opposite sides of the landscape, but this does not mean they are more dissimilar than a distinct cluster present half-way across the landscape. t-SNE also has a parameter, called “perplexity”, which loosely corresponds to the expected number of members in a major cluster. This parameter can have a major impact on clustering and must be lower than the number of rows in the matrix (i.e. the size of \(A\), or the number of genes). Recommended values are between 5 and 50. The default value of perplexity for the implementation of t-SNE in the R programming language is 30 (Krijthe 2015), and, unaware of its meaning, my first attempt to run t-SNE on a matrix with only 29 rows resulted in the following message, “Error: Perplexity is too large.” After a brief existential crisis, I lowered the perplexity, which resulted in beautiful clustering similar to that observed in Chapter 1, Figure 6. Paraphrasing a comment from Dr. Eugene Koonin on the quality of clusters, “you know them when you see them.”

Quantifying spots: replacing the drop assay 

When working with high-throughput genetics, validating results is essential. This often involves strain verification, back-crossing, and low-throughput recapitulation of high-throughput observations using a gold-standard assay. For yeast colony growth, owing mostly to simplicity and familiarity, the serial dilution drop assay reigns supreme. The serial dilution drop assay certainly works well for detecting obvious fitness defects. However, when combining multiple alleles to confirm potentially modest genetic interactions, the qualitative nature of the drop assay quickly becomes a burden, leading to more eye-squinting and head-scratching than insight. For the RAD5 manuscript we developed a quantitative colony growth assay for generating high-resolution colony fitness estimates (Chapter 2, Figure 1D & Chapter 2, Methods: Colony growth and fitness measurements). In my experience, this assay is able to resolve fitness differences as small as 5%. A good example of this resolution can be seen in Chapter 2, Figure 2B. While the effect is small, a significant positive interaction was detected between rad5∆ and sgs1∆ in standard laboratory growth conditions – additional evidence for epistasis between rad5∆ and crossover resolution without requiring cisplatin treatment! The quantitative colony growth assay, while low in throughput, is essential for validating our genetic observations from high-throughput approaches like SGA and SPA. With the CRISPR revolution beginning to produce combinatorial reverse genetic screens, it will be greatly beneficial for the systems genetics community to develop high-resolution validation approaches that are not subject to some of the biases that can occur in these pooled screening approaches (Shen et al. 2017; Sack et al. 2016).

The CRISPR revolution: systems genetics in new model systems 

Reverse genetic screens in mammalian cell lines have been possible through high-throughput RNA interference experiments for nearly twenty years (Carpenter and Sabatini 2004). Work in this field has progressed methods for performing pooled screening experiments to score fitness changes from a single transfected culture, wherein loss or gain of library representation is detected through next-generation sequencing (Sims et al. 2011). Initially, such experiments incurred high sequencing costs, and were plagued by off-target effects and many irreproducible results (Bhinder and Djaballah 2013). While these limitations have largely been overcome through reduced sequencing costs, formalized analysis pipelines, and improved experimental design (Schaefer et al. 2018), the arrival of CRISPR appears to have taken functional genomics in mammalian cells a major step forward (Evers et al. 2016). The favored approach for CRISPR based screens currently appears to be using Cas9 to induce double-strand breaks that, when improperly repaired, can result in small insertions and deletions, potentially resulting in frameshift mutations that may cause nonsense and gene disruption. In Chapter 3, I presented a new CRISPR method, called iSTOP, that allows direct generation of programmable nonsense without requiring a double-strand break and mutagenic repair. This feature may enable iSTOP to overcome the high levels of cell lethality caused by Cas9-induced double-strand breaks (Haapaniemi et al. 2018; Ihry et al. 2018). Importantly, for studying the DNA damage tolerance network in model systems beyond yeast, iSTOP may avoid confounding effects caused by requiring repair machinery to disrupt genes while simultaneously knocking out repair pathways. For this reason, I look forward to analyzing results from experiments comparing pooled screens using standard Cas9 and iSTOP.

References 

Alexandrov, Ludmil B, Serena Nik-Zainal, David C Wedge, Samuel A J R Aparicio, Sam Behjati, Andrew V Biankin, Graham R Bignell, et al. 2013. “Signatures of Mutational Processes in Human Cancer.” Nature 500 (7463): 415–21. https://doi.org/10.1038/nature12477.

Ashworth, Alan, and Christopher J Lord. 2018. “Synthetic Lethal Therapies for Cancer: What’s Next After PARP Inhibitors?” Nature Reviews. Clinical Oncology 15 (9): 564–76. https://doi.org/10.1038/s41571-018-0055-6.

Avery, O T, C M Macleod, and M McCarty. 1944. “Studies on the Chemical Nature of the Substance Inducing Transformation of Pneumococcal Types: Induction of Transformation by a Desoxyribonucleic Acid Fraction Isolated from Pneumococcus Type III.” The Journal of Experimental Medicine 79 (2): 137–58. https://www.ncbi.nlm.nih.gov/pubmed/19871359.

Baliga, Nitin S, Johan L M Björkegren, Jef D Boeke, Michael Boutros, Nigel P S Crawford, Aimée M Dudley, Charles R Farber, et al. 2017. “The State of Systems Genetics in 2017.” Cell Systems 4 (1): 7–15. https://doi.org/10.1016/j.cels.2017.01.005.

Bandyopadhyay, Sourav, Monika Mehta, Dwight Kuo, Min-Kyung Sung, Ryan Chuang, Eric J Jaehnig, Bernd Bodenmiller, et al. 2010. “Rewiring of Genetic Networks in Response to DNA Damage.” Science (New York, N.Y.) 330 (6009): 1385–9. https://doi.org/10.1126/science.1195618.

Baryshnikova, Anastasia. 2016. “Systematic Functional Annotation and Visualization of Biological Networks.” Cell Systems 2 (6): 412–21. https://doi.org/10.1016/j.cels.2016.04.014.

Baryshnikova, Anastasia, Michael Costanzo, Yungil Kim, Huiming Ding, Judice Koh, Kiana Toufighi, Ji-Young Youn, et al. 2010. “Quantitative Analysis of Fitness and Genetic Interactions in Yeast on a Genome Scale.” Nature Methods 7 (12): 1017–24. https://doi.org/10.1038/nmeth.1534.

Baudin, A, O Ozier-Kalogeropoulos, A Denouel, F Lacroute, and C Cullin. 1993. “A Simple and Efficient Method for Direct Gene Deletion in Saccharomyces cerevisiae.” Nucleic Acids Research 21 (14): 3329–30. https://www.ncbi.nlm.nih.gov/pubmed/8341614.

Bhinder, Bhavneet, and Hakim Djaballah. 2013. “Systematic Analysis of RNAi Reports Identifies Dismal Commonality at Gene-Level and Reveals an Unprecedented Enrichment in Pooled shRNA Screens.” Combinatorial Chemistry & High Throughput Screening 16 (9): 665–81. https://www.ncbi.nlm.nih.gov/pubmed/23848309.

Billon, Pierre, Eric E Bryant, Sarah A Joseph, Tarun S Nambiar, Samuel B Hayward, Rodney Rothstein, and Alberto Ciccia. 2017. “CRISPR-Mediated Base Editing Enables Efficient Disruption of Eukaryotic Genes Through Induction of STOP Codons.” Molecular Cell 67 (6): 1068–1079.e4. https://doi.org/10.1016/j.molcel.2017.08.008.

Birrell, Geoff W, James A Brown, H Irene Wu, Guri Giaever, Angela M Chu, Ronald W Davis, and J Martin Brown. 2002. “Transcriptional Response of Saccharomyces cerevisiae to DNA-Damaging Agents Does Not Identify the Genes That Protect Against These Agents.” Proceedings of the National Academy of Sciences of the United States of America 99 (13): 8778–83. https://doi.org/10.1073/pnas.132275199.

Braberg, Hannes, Huiyan Jin, Erica A Moehle, Yujia A Chan, Shuyi Wang, Michael Shales, Joris J Benschop, et al. 2013. “From Structure to Systems: High-Resolution, Quantitative Genetic Analysis of Rna Polymerase Ii.” Cell 154 (4): 775–88. https://doi.org/10.1016/j.cell.2013.07.033.

Brendel, M, and R H Haynes. 1973. “Interactions Among Genes Controlling Sensitivity to Radiation and Alkylation in Yeast.” Molecular & General Genetics: MGG 125 (3): 197–216. https://www.ncbi.nlm.nih.gov/pubmed/4359562.

Bryant, Eric E. 2019. “Systems Genetics of DNA Damage Tolerance – Cisplatin, RAD5 &Amp; CRISPR-Mediated Nonsense.” Columbia University. https://doi.org/10.7916/d8-k1d0-kb09.

Bryant, Eric E, Ivana Šunjevarić, Luke Berchowitz, Rodney Rothstein, and Robert J D Reid. 2019. “Rad5 Dysregulation Drives Hyperactive Recombination at Replication Forks Resulting in Cisplatin Sensitivity and Genome Instability.” Nucleic Acids Research 47 (17): 9144–59. https://doi.org/10.1093/nar/gkz631.

Cancer Genome Atlas Research Network. 2008. “Comprehensive Genomic Characterization Defines Human Glioblastoma Genes and Core Pathways.” Nature 455 (7216): 1061–8. https://doi.org/10.1038/nature07385.

Carpenter, Anne E, and David M Sabatini. 2004. “Systematic Genome-Wide Screens of Gene Function.” Nature Reviews. Genetics 5 (1): 11–22. https://doi.org/10.1038/nrg1248.

Cerami, Ethan, Jianjiong Gao, Ugur Dogrusoz, Benjamin E Gross, Selcuk Onur Sumer, Bülent Arman Aksoy, Anders Jacobsen, et al. 2012. “The cBio Cancer Genomics Portal: An Open Platform for Exploring Multidimensional Cancer Genomics Data.” Cancer Discovery 2 (5): 401–4. https://doi.org/10.1158/2159-8290.CD-12-0095.

Cheng, Haiying, Zhenfeng Zhang, Alain Borczuk, Charles A Powell, Adayabalam S Balajee, Howard B Lieberman, and Balazs Halmos. 2013. “PARP Inhibition Selectively Increases Sensitivity to Cisplatin in ERCC1-Low Non-Small Cell Lung Cancer Cells.” Carcinogenesis 34 (4): 739–49. https://doi.org/10.1093/carcin/bgs393.

Ciccia, Alberto, and Stephen J Elledge. 2010. “The DNA Damage Response: Making It Safe to Play with Knives.” Molecular Cell 40 (2): 179–204. https://doi.org/10.1016/j.molcel.2010.09.019.

Cleaver, James E. 2016. “Profile of Tomas Lindahl, Paul Modrich, and Aziz Sancar, 2015 Nobel Laureates in Chemistry.” Proceedings of the National Academy of Sciences of the United States of America 113 (2): 242–5. https://doi.org/10.1073/pnas.1521829112.

Collins, Sean R, Kyle M Miller, Nancy L Maas, Assen Roguev, Jeffrey Fillingham, Clement S Chu, Maya Schuldiner, et al. 2007. “Functional Dissection of Protein Complexes Involved in Yeast Chromosome Biology Using a Genetic Interaction Map.” Nature 446 (7137): 806–10. https://doi.org/10.1038/nature05649.

Collins, Sean R., Maya Schuldiner, Nevan J. Krogan, and Jonathan S. Weissman. 2006. “A Strategy for Extracting and Analyzing Large-Scale Quantitative Epistatic Interaction Data.” Genome Biology 7 (7): R63. https://doi.org/10.1186/gb-2006-7-7-r63.

Cong, Le, F Ann Ran, David Cox, Shuailiang Lin, Robert Barretto, Naomi Habib, Patrick D Hsu, et al. 2013. “Multiplex Genome Engineering Using CRISPR/Cas Systems.” Science 339 (6121): 819–23. https://doi.org/10.1126/science.1231143.

Costanzo, Michael, Anastasia Baryshnikova, Jeremy Bellay, Yungil Kim, Eric D Spear, Carolyn S Sevier, Huiming Ding, et al. 2010. “The Genetic Landscape of a Cell.” Science 327 (5964): 425–31. https://doi.org/10.1126/science.1180823.

Costanzo, Michael, Benjamin VanderSluis, Elizabeth N Koch, Anastasia Baryshnikova, Carles Pons, Guihong Tan, Wen Wang, et al. 2016. “A Global Genetic Interaction Network Maps a Wiring Diagram of Cellular Function.” Science 353 (6306). https://doi.org/10.1126/science.aaf1420.

Cox, B S, and J M Parry. 1968. “The Isolation, Genetics and Survival Characteristics of Ultraviolet Light-Sensitive Mutants in Yeast.” Mutation Research 6 (1): 37–55. https://www.ncbi.nlm.nih.gov/pubmed/5708072.

Crick, F. 1970. “Central Dogma of Molecular Biology.” Nature 227 (5258): 561–3. https://www.ncbi.nlm.nih.gov/pubmed/4913914.

Crick, F H, L Barnett, S Brenner, and R J Watts-tobin. 1961. “General Nature of the Genetic Code for Proteins.” Nature 192 (December): 1227–32. https://www.ncbi.nlm.nih.gov/pubmed/13882203.

Dahm, Ralf. 2005. “Friedrich Miescher and the Discovery of DNA.” Developmental Biology 278 (2): 274–88. https://doi.org/10.1016/j.ydbio.2004.11.028.

Deltcheva, Elitza, Krzysztof Chylinski, Cynthia M Sharma, Karine Gonzales, Yanjie Chao, Zaid A Pirzada, Maria R Eckert, Jörg Vogel, and Emmanuelle Charpentier. 2011. “CRISPR RNA Maturation by Trans-Encoded Small RNA and Host Factor RNase III.” Nature 471 (7340): 602–7. https://doi.org/10.1038/nature09886.

Dietlein, Felix, Lisa Thelen, and H Christian Reinhardt. 2014. “Cancer-Specific Defects in DNA Repair Pathways as Targets for Personalized Therapeutic Approaches.” Trends in Genetics 30 (8): 326–39. https://doi.org/10.1016/j.tig.2014.06.003.

Evers, Bastiaan, Katarzyna Jastrzebski, Jeroen P M Heijmans, Wipawadee Grernrum, Roderick L Beijersbergen, and Rene Bernards. 2016. “CRISPR Knockout Screening Outperforms shRNA and Crispri in Identifying Essential Genes.” Nature Biotechnology 34 (6): 631–3. https://doi.org/10.1038/nbt.3536.

Fiedler, Dorothea, Hannes Braberg, Monika Mehta, Gal Chechik, Gerard Cagney, Paromita Mukherjee, Andrea C Silva, et al. 2009. “Functional Organization of the S. cerevisiae Phosphorylation Network.” Cell 136 (5): 952–63. https://doi.org/10.1016/j.cell.2008.12.039.

Galanski, Markus, Michael A Jakupec, and Bernhard K Keppler. 2005. “Update of the Preclinical Situation of Anticancer Platinum Complexes: Novel Design Strategies and Innovative Analytical Approaches.” Current Medicinal Chemistry 12 (18): 2075–94. https://www.ncbi.nlm.nih.gov/pubmed/16101495.

Game, J C, and R K Mortimer. 1974. “A Genetic Study of X-Ray Sensitive Mutants in Yeast.” Mutation Research 24 (3): 281–92. https://www.ncbi.nlm.nih.gov/pubmed/4606119.

Gao, Jianjiong, Bülent Arman Aksoy, Ugur Dogrusoz, Gideon Dresdner, Benjamin Gross, S Onur Sumer, Yichao Sun, et al. 2013. “Integrative Analysis of Complex Cancer Genomics and Clinical Profiles Using the cBioPortal.” Science Signaling 6 (269): pl1. https://doi.org/10.1126/scisignal.2004088.

Garneau, Josiane E, Marie-Ève Dupuis, Manuela Villion, Dennis A Romero, Rodolphe Barrangou, Patrick Boyaval, Christophe Fremaux, Philippe Horvath, Alfonso H Magadán, and Sylvain Moineau. 2010. “The CRISPR/Cas Bacterial Immune System Cleaves Bacteriophage and Plasmid DNA.” Nature 468 (7320): 67–71. https://doi.org/10.1038/nature09523.

Gasiunas, Giedrius, Rodolphe Barrangou, Philippe Horvath, and Virginijus Siksnys. 2012. “Cas9-crRNA Ribonucleoprotein Complex Mediates Specific DNA Cleavage for Adaptive Immunity in Bacteria.” Proceedings of the National Academy of Sciences of the United States of America 109 (39): E2579–86. https://doi.org/10.1073/pnas.1208507109.

Giaever, Guri, Angela M Chu, Li Ni, Carla Connelly, Linda Riles, Steeve Véronneau, Sally Dow, et al. 2002. “Functional Profiling of the Saccharomyces cerevisiae Genome.” Nature 418 (6896): 387–91. https://doi.org/10.1038/nature00935.

Goffeau, A., B. G. Barrell, H. Bussey, R. W. Davis, B. Dujon, H. Feldmann, F. Galibert, et al. 1996. “Life with 6000 Genes.” Science 274 (5287): 546, 563–67.

Goffeau, A., and A. Vassarotti. 1991. “The European Project for Sequencing the Yeast Genome.” Research in Microbiology 142 (7): 901–3. https://doi.org/10.1016/0923-2508(91)90071-H.

Grimm, D. 2006. “A Mouse for Every Gene.” Science 312 (5782): 1862–6. https://doi.org/10.1126/science.312.5782.1862.

Guénolé, Aude, Rohith Srivas, Kees Vreeken, Ze Zhong Wang, Shuyi Wang, Nevan J Krogan, Trey Ideker, and Haico van Attikum. 2013. “Dissection of DNA Damage Responses Using Multiconditional Genetic Interaction Maps.” Molecular Cell 49 (2): 346–58. https://doi.org/10.1016/j.molcel.2012.11.023.

Haapaniemi, Emma, Sandeep Botla, Jenna Persson, Bernhard Schmierer, and Jussi Taipale. 2018. “CRISPR-Cas9 Genome Editing Induces a p53-Mediated DNA Damage Response.” Nature Medicine 24 (7): 927–30. https://doi.org/10.1038/s41591-018-0049-z.

Hartwell, L H, P Szankasi, C J Roberts, A W Murray, and S H Friend. 1997. “Integrating Genetic Approaches into the Discovery of Anticancer Drugs.” Science 278 (5340): 1064–8. https://www.ncbi.nlm.nih.gov/pubmed/9353181.

Hastie, C James, Cristina Vázquez-Martin, Amanda Philp, Michael J R Stark, and Patricia T W Cohen. 2006. “The Saccharomyces cerevisiae Orthologue of the Human Protein Phosphatase 4 Core Regulatory Subunit R2 Confers Resistance to the Anticancer Drug Cisplatin.” The FEBS Journal 273 (14): 3322–34. https://doi.org/10.1111/j.1742-4658.2006.05336.x.

Haynes, Robert, and Bernard Kunz. 1981. “DNA Repair and Mutagenesis in Yeast.” Cold Spring Harbor Monograph Archive 11A (0). https://cshmonographs.org/index.php/monographs/article/view/4235/3411.

Heyer, Wolf-Dietrich, Kirk T Ehmsen, and Jie Liu. 2010. “Regulation of Homologous Recombination in Eukaryotes.” Annual Review of Genetics 44: 113–39. https://doi.org/10.1146/annurev-genet-051710-150955.

Hillenmeyer, Maureen E, Eula Fung, Jan Wildenhain, Sarah E Pierce, Shawn Hoon, William Lee, Michael Proctor, et al. 2008. “The Chemical Genomic Portrait of Yeast: Uncovering a Phenotype for All Genes.” Science 320 (5874): 362–5. https://doi.org/10.1126/science.1150021.

Huang, Ruea-Yea, Martha Eddy, Marija Vujcic, and David Kowalski. 2005. “Genome-Wide Screen Identifies Genes Whose Inactivation Confer Resistance to Cisplatin in Saccharomyces cerevisiae.” Cancer Research 65 (13): 5890–7. https://doi.org/10.1158/0008-5472.CAN-04-4093.

Hustedt, Nicole, Andrew Seeber, Ragna Sack, Monika Tsai-Pflugfelder, Bhupinder Bhullar, Hanneke Vlaming, Fred van Leeuwen, et al. 2015. “Yeast PP4 Interacts with ATR Homolog Ddc2-Mec1 and Regulates Checkpoint Signaling.” Molecular Cell 57 (2): 273–89. https://doi.org/10.1016/j.molcel.2014.11.016.

Ideker, Trey, and Nevan J Krogan. 2012. “Differential Network Biology.” Molecular Systems Biology 8 (January): 565. https://doi.org/10.1038/msb.2011.99.

Ihry, Robert J, Kathleen A Worringer, Max R Salick, Elizabeth Frias, Daniel Ho, Kraig Theriault, Sravya Kommineni, et al. 2018. “p53 Inhibits CRISPR-Cas9 Engineering in Human Pluripotent Stem Cells.” Nature Medicine 24 (7): 939–46. https://doi.org/10.1038/s41591-018-0050-6.

Jacobsen, Anders, and cBioPortal Questions. 2018. cgdsr: R-Based API for Accessing the MSKCC Cancer Genomics Data Server (CGDS). https://CRAN.R-project.org/package=cgdsr.

Jinek, Martin, Krzysztof Chylinski, Ines Fonfara, Michael Hauer, Jennifer A Doudna, and Emmanuelle Charpentier. 2012. “A Programmable Dual-RNA-Guided DNA Endonuclease in Adaptive Bacterial Immunity.” Science 337 (6096): 816–21. https://doi.org/10.1126/science.1225829.

Kelland, Lloyd. 2007. “The Resurgence of Platinum-Based Cancer Chemotherapy.” Nature Reviews. Cancer 7 (8): 573–84. https://doi.org/10.1038/nrc2167.

Kile, Andrew C, Diana A Chavez, Julien Bacal, Sherif Eldirany, Dmitry M Korzhnev, Irina Bezsonova, Brandt F Eichman, and Karlene A Cimprich. 2015. “HLTF’s Ancient HIRAN Domain Binds 3’ DNA Ends to Drive Replication Fork Reversal.” Molecular Cell 58 (6): 1090–1100. https://doi.org/10.1016/j.molcel.2015.05.013.

Knoch, Jennifer, York Kamenisch, Christian Kubisch, and Mark Berneburg. 2012. “Rare Hereditary Diseases with Defects in DNA-Repair.” European Journal of Dermatology 22 (4): 443–55. https://doi.org/10.1684/ejd.2012.1654.

Koller, B H, P Marrack, J W Kappler, and O Smithies. 1990. “Normal Development of Mice Deficient in Beta 2M, MHC Class I Proteins, and CD8+ T Cells.” Science 248 (4960): 1227–30. https://www.ncbi.nlm.nih.gov/pubmed/2112266.

Koller, B H, and O Smithies. 1989. “Inactivating the Beta 2-Microglobulin Locus in Mouse Embryonic Stem Cells by Homologous Recombination.” Proceedings of the National Academy of Sciences of the United States of America 86 (22): 8932–5. https://www.ncbi.nlm.nih.gov/pubmed/2682666.

Komor, Alexis C, Yongjoo B Kim, Michael S Packer, John A Zuris, and David R Liu. 2016. “Programmable Editing of a Target Base in Genomic DNA Without Double-Stranded DNA Cleavage.” Nature 533 (7603): 420–4. https://doi.org/10.1038/nature17946.

Krijthe, Jesse H. 2015. Rtsne: T-Distributed Stochastic Neighbor Embedding Using Barnes-Hut Implementation. https://github.com/jkrijthe/Rtsne.

Lander, E S, L M Linton, B Birren, C Nusbaum, M C Zody, J Baldwin, K Devon, et al. 2001. “Initial Sequencing and Analysis of the Human Genome.” Nature 409 (6822): 860–921. https://doi.org/10.1038/35057062.

Lee, William, Robert P St Onge, Michael Proctor, Patrick Flaherty, Michael I Jordan, Adam P Arkin, Ronald W Davis, Corey Nislow, and Guri Giaever. 2005. “Genome-Wide Requirements for Resistance to Functionally Distinct DNA-Damaging Agents.” PLoS Genetics 1 (2): e24. https://doi.org/10.1371/journal.pgen.0010024.

Lemontt, J F. 1971. “Mutants of Yeast Defective in Mutation Induced by Ultraviolet Light.” Genetics 68 (1): 21–33. https://www.ncbi.nlm.nih.gov/pubmed/17248528.

Liao, Chunyan, Bin Hu, Matthew J Arno, and Barry Panaretou. 2007. “Genomic Screening in vivo Reveals the Role Played by Vacuolar H+ ATPase and Cytosolic Acidification in Sensitivity to DNA-Damaging Agents Such as Cisplatin.” Molecular Pharmacology 71 (2): 416–25. https://doi.org/10.1124/mol.106.030494.

Lindahl, T. 1993. “Instability and Decay of the Primary Structure of DNA.” Nature 362 (6422): 709–15. https://doi.org/10.1038/362709a0.

Lord, Christopher J, Andrew N J Tutt, and Alan Ashworth. 2015. “Synthetic Lethality and Cancer Therapy: Lessons Learned from the Development of PARP Inhibitors.” Annual Review of Medicine 66: 455–70. https://doi.org/10.1146/annurev-med-050913-022545.

Mali, Prashant, Luhan Yang, Kevin M Esvelt, John Aach, Marc Guell, James E DiCarlo, Julie E Norville, and George M Church. 2013. “RNA-Guided Human Genome Engineering via Cas9.” Science 339 (6121): 823–6. https://doi.org/10.1126/science.1232033.

Nickoloff, Jac A, Dennie Jones, Suk-Hee Lee, Elizabeth A Williamson, and Robert Hromas. 2017. “Drugging the Cancers Addicted to DNA Repair.” Journal of the National Cancer Institute 109 (11). https://doi.org/10.1093/jnci/djx059.

Nirenberg, M., T. Caskey, R. Marshall, R. Brimacombe, D. Kellogg, B. Doctor, D. Hatfield, et al. 1966. “The RNA Code and Protein Synthesis.” Cold Spring Harbor Symposia on Quantitative Biology 31: 11–24.

O’Connor, Mark J. 2015. “Targeting the DNA Damage Response in Cancer.” Molecular Cell 60 (4): 547–60. https://doi.org/10.1016/j.molcel.2015.10.040.

Pan, Xuewen, Ping Ye, Daniel S Yuan, Xiaoling Wang, Joel S Bader, and Jef D Boeke. 2006. “A DNA Integrity Network in the Yeast Saccharomyces cerevisiae.” Cell 124 (5): 1069–81. https://doi.org/10.1016/j.cell.2005.12.036.

Rocha, Clarissa Ribeiro Reily, Matheus Molina Silva, Annabel Quinet, Januario Bispo Cabral-Neto, and Carlos Frederico Martins Menck. 2018. “DNA Repair Pathways and Cisplatin Resistance: An Intimate Relationship.” Clinics 73 (suppl 1): e478s. https://doi.org/10.6061/clinics/2018/e478s.

Rothstein, Rodney J. 1983. “One-Step Gene Disruption in Yeast.” Methods in Enzymology 101 (101). Elsevier: 202–11. https://doi.org/10.1016/0076-6879(83)01015-0.

Roy, Rohini, Jarin Chun, and Simon N Powell. 2011. “BRCA1 and BRCA2: Different Roles in a Common Pathway of Genome Protection.” Nature Reviews. Cancer 12 (1): 68–78. https://doi.org/10.1038/nrc3181.

Sack, Laura Magill, Teresa Davoli, Qikai Xu, Mamie Z Li, and Stephen J Elledge. 2016. “Sources of Error in Mammalian Genetic Screens.” G3 6 (9): 2781–90. https://doi.org/10.1534/g3.116.030973.

Sapranauskas, Rimantas, Giedrius Gasiunas, Christophe Fremaux, Rodolphe Barrangou, Philippe Horvath, and Virginijus Siksnys. 2011. “The Streptococcus thermophilus CRISPR/Cas System Provides Immunity in Escherichia coli.” Nucleic Acids Research 39 (21): 9275–82. https://doi.org/10.1093/nar/gkr606.

Schaefer, Christiane, Nikhil Mallela, Jochen Seggewiß, Birgit Lechtape, Heymut Omran, Uta Dirksen, Eberhard Korsching, and Jenny Potratz. 2018. “Target Discovery Screens Using Pooled shRNA Libraries and Next-Generation Sequencing: A Model Workflow and Analytical Algorithm.” PloS One 13 (1): e0191570. https://doi.org/10.1371/journal.pone.0191570.

Schuldiner, Maya, Sean R Collins, Natalie J Thompson, Vladimir Denic, Arunashree Bhamidipati, Thanuja Punna, Jan Ihmels, et al. 2005. “Exploration of the Function and Organization of the Yeast Early Secretory Pathway Through an Epistatic Miniarray Profile.” Cell 123 (3): 507–19. https://doi.org/10.1016/j.cell.2005.08.031.

Schuldiner, M, S R Collins, J S Weissman, and N J Krogan. 2006. “Quantitative Genetic Analysis in Saccharomyces cerevisiae Using Epistatic Miniarray Profiles (E-MAPs) and Its Application to Chromatin Functions.” Methods 40 (4): 344–52. https://doi.org/10.1016/j.ymeth.2006.07.034.

Shen, John Paul, Dongxin Zhao, Roman Sasik, Jens Luebeck, Amanda Birmingham, Ana Bojorquez-Gomez, Katherine Licon, et al. 2017. “Combinatorial CRISPR-Cas9 Screens for de novo Mapping of Genetic Interactions.” Nature Methods 14 (6): 573–76. https://doi.org/10.1038/nmeth.4225.

Simon, J A, P Szankasi, D K Nguyen, C Ludlow, H M Dunstan, C J Roberts, E L Jensen, L H Hartwell, and S H Friend. 2000. “Differential Toxicities of Anticancer Agents Among DNA Repair and Checkpoint Mutants of Saccharomyces cerevisiae.” Cancer Research 60 (2): 328–33. https://www.ncbi.nlm.nih.gov/pubmed/10667584.

Sims, David, Ana M Mendes-Pereira, Jessica Frankum, Darren Burgess, Maria-Antonietta Cerone, Cristina Lombardelli, Costas Mitsopoulos, et al. 2011. “High-Throughput RNA Interference Screening Using Pooled shRNA Libraries and Next Generation Sequencing.” Genome Biology 12 (10): R104. https://doi.org/10.1186/gb-2011-12-10-r104.

Snow, R. 1967. “Mutants of Yeast Sensitive to Ultraviolet Light.” Journal of Bacteriology 94 (3): 571–5. https://www.ncbi.nlm.nih.gov/pubmed/6035260.

Srivas, Rohith, Thomas Costelloe, Anne-Ruxandra Carvunis, Sovan Sarkar, Erik Malta, Su Ming Sun, Marijke Pool, et al. 2013. “A UV-Induced Genetic Network Links the RSC Complex to Nucleotide Excision Repair and Shows Dose-Dependent Rewiring.” Cell Reports 5 (6): 1714–24. https://doi.org/10.1016/j.celrep.2013.11.035.

Stark, Chris, Bobby-Joe Breitkreutz, Teresa Reguly, Lorrie Boucher, Ashton Breitkreutz, and Mike Tyers. 2006. “BioGRID: A General Repository for Interaction Datasets.” Nucleic Acids Research 34 (Database issue): D535–9. https://doi.org/10.1093/nar/gkj109.

St. Onge, Robert P, Ramamurthy Mani, Julia Oh, Michael Proctor, Eula Fung, Ronald W Davis, Corey Nislow, Frederick P Roth, and Guri Giaever. 2007. “Systematic Pathway Analysis Using High-Resolution Fitness Profiling of Combinatorial Gene Deletions.” Nature Genetics 39 (2): 199–206. https://doi.org/10.1038/ng1948.

Tong, A H, M Evangelista, A B Parsons, H Xu, G D Bader, N Pa é, M Robinson, et al. 2001. “Systematic Genetic Analysis with Ordered Arrays of Yeast Deletion Mutants.” Science 294 (5550): 2364–8. https://doi.org/10.1126/science.1065810.

Tong, Amy Hin Yan, and Charles Boone. 2006. “Synthetic Genetic Array Analysis in Saccharomyces Cerevisiae.” Methods in Molecular Biology 313: 171–92. https://www.ncbi.nlm.nih.gov/pubmed/16118434.

Usaj, Matej, Yizhao Tan, Wen Wang, Benjamin VanderSluis, Albert Zou, Chad L Myers, Michael Costanzo, Brenda Andrews, and Charles Boone. 2017. “TheCellMap.Org: A Web-Accessible Database for Visualizing and Mining the Global Yeast Genetic Interaction Network.” G3 7 (5): 1539–49. https://doi.org/10.1534/g3.117.040220.

van der Maaten, Laurens, and Geoffrey Hinton. 2008. “Visualizing Data Using t-SNE.” Journal of Machine Learning Research 9 (Nov): 2579–2605. http://www.jmlr.org/papers/v9/vandermaaten08a.html.

Vogel, Gretchen. 2007. “Nobel Prizes. A Knockout Award in Medicine.” Science 318 (5848): 178–9. https://doi.org/10.1126/science.318.5848.178.

Wach, A, A Brachat, R Pöhlmann, and P Philippsen. 1994. “New Heterologous Modules for Classical or PCR-Based Gene Disruptions in Saccharomyces cerevisiae.” Yeast 10 (13): 1793–1808. https://www.ncbi.nlm.nih.gov/pubmed/7747518.

Waterston, Robert H, Kerstin Lindblad-Toh, Ewan Birney, Jane Rogers, Josep F Abril, Pankaj Agarwal, Richa Agarwala, et al. 2002. “Initial Sequencing and Comparative Analysis of the Mouse Genome.” Nature 420 (6915): 520–62. https://doi.org/10.1038/nature01262.

Watson, J D, and F H Crick. 1953. “Molecular Structure of Nucleic Acids; a Structure for Deoxyribose Nucleic Acid.” Nature 171 (4356): 737–8. https://www.ncbi.nlm.nih.gov/pubmed/13054692.

Wattenberg, Martin, Fernanda Viégas, and Ian Johnson. 2016. “How to Use T-Sne Effectively.” Distill. https://doi.org/10.23915/distill.00002.

Winzeler, E A, D D Shoemaker, A Astromoff, H Liang, K Anderson, B Andre, R Bangham, et al. 1999. “Functional Characterization of the S. cerevisiae Genome by Gene Deletion and Parallel Analysis.” Science 285 (5429): 901–6. https://www.ncbi.nlm.nih.gov/pubmed/10436161.

Wu, H Irene, James A Brown, Mary J Dorie, Laura Lazzeroni, and J Martin Brown. 2004. “Genome-Wide Identification of Genes Conferring Resistance to the Anticancer Agents Cisplatin, Oxaliplatin, and Mitomycin C.” Cancer Research 64 (11): 3940–8. https://doi.org/10.1158/0008-5472.CAN-03-3113.

Zhang, Junran. 2013. “The Role of BRCA1 in Homologous Recombination Repair in Response to Replication Stress: Significance in Tumorigenesis and Cancer Therapy.” Cell & Bioscience 3 (1): 11. https://doi.org/10.1186/2045-3701-3-11.

Zijlstra, M, M Bix, N E Simister, J M Loring, D H Raulet, and R Jaenisch. 1990. “Beta 2-Microglobulin Deficient Mice Lack CD4-8+ Cytolytic T Cells.” Nature 344 (6268): 742–6. https://doi.org/10.1038/344742a0.

Zijlstra, M, E Li, F Sajjadi, S Subramani, and R Jaenisch. 1989. “Germ-Line Transmission of a Disrupted Beta 2-Microglobulin Gene Produced by Homologous Recombination in Embryonic Stem Cells.” Nature 342 (6248): 435–8. https://doi.org/10.1038/342435a0.