Um dos fundamentos teóricos e metodológicos da biologia de sistemas é a busca e interpretação das relações entre biomoléculas que têm lugar no interior da célula e que mantém seu funcionamento. As relações podem existir entre moléculas da mesma natureza, por exemplo entre proteínas, ou entre moléculas de natureza diferente como por exemplo DNA-proteína. De todas as relações possíveis, as relações entre genes têm sido bastante estudadas utilizando dois métodos: correlacionando valores de expressão genética a partir de diferentes medidas de dependência estatística, ou utilizando técnicas biotecnológicas como as interações genéticas que identificam relações entre genes medindo o efeito fenotípico de mutações ou deleções de genes alvo. Sem importar que sejam técnicas diferentes, o que se procura em ambos casos é identificar relações funcionais que pode existir entre um par de genes. Essa semelhança conceitual faz com que seja possível comparar o resultado das duas estratégias a fim de avaliar a proporção de relações funcionais que são identificadas de modo simultâneo pelas medidas de dependência e pelas interações genéticas. Para levar a cabo dita comparação, as medidas de dependência de Pearson e Spearman foram aqui utilizadas para obter as redes de co-expressão de três conjuntos de dados de expressão genética de Saccharomyces cerevisiae. As relações funcionais obtidas no passo anterior foram comparadas com aquelas relações obtidas pela técnica das interações genéticas que se encontram disponíveis nos dois principais bancos de dados. Como resultado dessas comparações, observou-se que apesar das duas técnicas serem desenhadas com o mesmo objetivo (identificar relações funcionais entre genes), o número de relações que são comuns às duas metodologias estudadas é muito baixo. Tanto a diferença nas técnicas de obtenção das relações como a ausência de uma definição específica do que é uma relação funcional, podem ser as principais causas do baixo nível de relação entre as duas estratégias. / One of the theoretical and methodological foundations of systems biology is the search and interpretation of the relationships between biomolecules that take place inside the cell and that maintains its functioning. Those relationships may exist either, between molecules of the same nature, for example between proteins, or between molecules of a different nature such as DNA-protein. Of all possible relationships, gene-gene relationships have been extensively studied using two methods: correlating genetic expression values from different measures of statistical dependence, or using biotechnological techniques such as genetic interactions that identify relationships between genes by measuring the phenotypic effect of mutations or deletions of target genes. Regardless of whether they are different techniques, what is sought in both cases are to identify functional relationships that may exist between a pair of genes. This conceptual similarity makes it possible to compare the results of this two strategies in order to assess the proportion of functional relationships that are simultaneously identified by measures of dependence and genetic interactions. To carry out such a comparison, the Pearson and Spearman dependency measures were used here to obtain the co-expression networks of three sets of Saccharomyces cerevisiae gene expression data. The functional relations obtained in the previous step were compared with those relations obtained by the technique of genetic interactions that are available in the two main databases. As a result of these comparisons, it was observed that although the two techniques are designed with the same objective (to identify functional relations between genes), the number of relations that are common to the two methodologies studied is very low. Both the difference in the techniques of obtaining relationships and the absence of a specific definition about what is a functional relationship can be the main causes of the low level of relationship between this two strategies.

Immediately following a gene duplication event, if both gene copies are to be fixed into a species’ genome there is a period of enhanced selection acting on either one or both duplicates (paralogs) that results in some extent of functional divergence. However, as redundancy among extant duplicates is thought to confer genomic robustness, a consequent question is: how much functional overlap exists between duplicates that are retained over long spans of evolutionary time? To examine this issue I determined the extent of shared protein interactions and protein complex membership for paralogous gene pairs resulting from an ancient Whole Genome Duplication (WGD) event in yeast, finding retained functional overlap to be substantial among this group. Surprisingly however, I found paralogs existing within the same complex tended to maintain greater disparities in expression, suggesting the existence of previously proposed “transcriptional back-up” mechanisms. To test both for existence of such mechanisms and for any phenotypic manifestation of their shared functional overlap I surveyed for the presence of aggravating genetic interactions between 399 WGD-resultant paralog pairs. While these paralogs exhibited a high frequency (~30%) of epistasis, observed genetic interactions were not predictable based on protein interaction overlap. Further, exposure to a limited number of stressors confirmed that additional instances of epistasis were only observable under alternate conditions. As only a small number of stress conditions were tested, the high frequency of genetic interactions reported appears to be a minimum estimate of the true extent of epistasis among WGD paralogs, potentially explaining the lack of overlap with protein interaction data. As it is impossible to survey an infinite condition space, Synthetic Genetic Array (SGA) screening of yeast strains carrying double-deletions of paralog pairs was used to assess functional redundancy among a group of the remaining non-epistatic paralog pairs. The resulting interactions demonstrated functional relationships in non-epistatic paralogs only obvious upon ablation of both duplicates, suggesting that these interactions had initially been masked through redundant function. These findings ultimately suggest an advantage to retained functional overlap among whole genome duplicates that is capable of being stably maintained through millions of years of evolutionary time.

Immediately following a gene duplication event, if both gene copies are to be fixed into a species’ genome there is a period of enhanced selection acting on either one or both duplicates (paralogs) that results in some extent of functional divergence. However, as redundancy among extant duplicates is thought to confer genomic robustness, a consequent question is: how much functional overlap exists between duplicates that are retained over long spans of evolutionary time? To examine this issue I determined the extent of shared protein interactions and protein complex membership for paralogous gene pairs resulting from an ancient Whole Genome Duplication (WGD) event in yeast, finding retained functional overlap to be substantial among this group. Surprisingly however, I found paralogs existing within the same complex tended to maintain greater disparities in expression, suggesting the existence of previously proposed “transcriptional back-up” mechanisms. To test both for existence of such mechanisms and for any phenotypic manifestation of their shared functional overlap I surveyed for the presence of aggravating genetic interactions between 399 WGD-resultant paralog pairs. While these paralogs exhibited a high frequency (~30%) of epistasis, observed genetic interactions were not predictable based on protein interaction overlap. Further, exposure to a limited number of stressors confirmed that additional instances of epistasis were only observable under alternate conditions. As only a small number of stress conditions were tested, the high frequency of genetic interactions reported appears to be a minimum estimate of the true extent of epistasis among WGD paralogs, potentially explaining the lack of overlap with protein interaction data. As it is impossible to survey an infinite condition space, Synthetic Genetic Array (SGA) screening of yeast strains carrying double-deletions of paralog pairs was used to assess functional redundancy among a group of the remaining non-epistatic paralog pairs. The resulting interactions demonstrated functional relationships in non-epistatic paralogs only obvious upon ablation of both duplicates, suggesting that these interactions had initially been masked through redundant function. These findings ultimately suggest an advantage to retained functional overlap among whole genome duplicates that is capable of being stably maintained through millions of years of evolutionary time.

Understanding gene function and genetic relationships is elemental in our efforts to better understand biological systems. Here, I describe a reliable high-throughput approach, Systematic Genetic Interaction analysis (SGI), capable of revealing both weak and strong genetic interactions in the nematode Caenorhabditis elegans. I also present evidence that UNC-73 functions cell-autonomously in an UNC-40 pathway to direct muscle arm extension in C. elegans. Previous efforts to systematically describe genetic interactions between redundant genes on a global scale either have focused on core biological processes in protozoans or have surveyed catastrophic interactions in metazoans. I investigated synthetic genetic interactions between eleven ‘query’ mutants in conserved signal transduction pathways and hundreds of ‘target’ genes compromised by RNAi. A network of 1246 genetic interactions was uncovered through an unbiased global analysis of the interaction matrix, establishing the largest metazoan genetic interaction network to date. To investigate how genetic interactions connect genes on a systems-wide level, the SGI network was superimposed with existing networks of physical, genetic, phenotypic and co-expression interactions. Fifty-six putative functional modules were identified within the superimposed network, one of which regulates fat accumulation and is coordinated by bar-1(ga80)/β-catenin interactions. This led to the discovery that SGI interactions link distinct functional modules on a global scale, which is a previously unappreciated level of organization within metazoan systems. In addition, I present evidence that the properties of genetic networks are conserved between C. elegans and S. cerevisiae, but that the connectivity of the interactions within the current networks is not. Although the buffering between functional modules may differ between species, studying these differences may provide insight into the evolution of divergent form and function. In C. elegans the postsynaptic membrane of the neuromuscular junction reaches its destination through an active process of guided cell extension. The worm has 95 body wall muscles (BWMs) that extend projections called 'muscle arms' to motor axons. The muscle arms harbour the postsynaptic elements of neuromuscular junctions. The stereotypical pattern of muscle arm extension was exploited in a forward genetic screen for new genes required for guided cell migration by looking for mutations that caused a reduction in the number of arms that extend to the motor axons. One of the resulting mutants was tr117, which extended half the number of arms compared to wild type animals. Genetic mapping, complementation tests, and sequencing revealed that tr117 was a mutation in unc-73/Trio, which encodes a guanine nucleotide exchange factor. Expression of UNC-73 specifically in the BWMs rescued the muscle arm development defects of unc-73(e936) mutants, indicating that UNC-73 functions cell-autonomously to regulate muscle arm extension. UNC-73::CFP was localized to muscle arm termini in a pattern similar to that of UNC-40/Dcc, which directs muscle arm extension. UNC-73 over-expression suppressed the Madd phenotype of unc-40 null worms and unc-73(e936) suppressed ectopic myopodia induced by UNC-40 over-expression. These results indicate that UNC-73 functions downstream of UNC-40 in a pathway that regulates muscle arm extension.

The budding yeast, Saccharomyces cerevisiae, serves as an excellent model for identifying fundamental mechanisms of DNA repair. A Local Coherence Detection (LCD) algorithm that uses biclustering to assign genes to multiple functional sub-groups was applied on the chromosome E-MAP containing genetic interactions among genes involved in nuclear processes. Using this method, we found that Asf1 and Rtt109, genes that are together required for histone H3K56 acetylation, cluster together with Ctf4, Ctf18, Ctf8 and Dcc1, genes important for efficient sister chromatid cohesion. It is known that H3K56 acetylation is required for post-repair chromatin reassembly at sites of DNA double-strand breaks (DSBs). The cohesion genes were previously implicated in the repair of some DNA DSBs, but the nature of their involvement has not been reported. The experimental data in my thesis work suggest that Ctf4, Ctf8, Ctf18 and Dcc1 function in the post-repair chromatin reassembly pathway.

The budding yeast, Saccharomyces cerevisiae, serves as an excellent model for identifying fundamental mechanisms of DNA repair. A Local Coherence Detection (LCD) algorithm that uses biclustering to assign genes to multiple functional sub-groups was applied on the chromosome E-MAP containing genetic interactions among genes involved in nuclear processes. Using this method, we found that Asf1 and Rtt109, genes that are together required for histone H3K56 acetylation, cluster together with Ctf4, Ctf18, Ctf8 and Dcc1, genes important for efficient sister chromatid cohesion. It is known that H3K56 acetylation is required for post-repair chromatin reassembly at sites of DNA double-strand breaks (DSBs). The cohesion genes were previously implicated in the repair of some DNA DSBs, but the nature of their involvement has not been reported. The experimental data in my thesis work suggest that Ctf4, Ctf8, Ctf18 and Dcc1 function in the post-repair chromatin reassembly pathway.

Understanding gene function and genetic relationships is elemental in our efforts to better understand biological systems. Here, I describe a reliable high-throughput approach, Systematic Genetic Interaction analysis (SGI), capable of revealing both weak and strong genetic interactions in the nematode Caenorhabditis elegans. I also present evidence that UNC-73 functions cell-autonomously in an UNC-40 pathway to direct muscle arm extension in C. elegans. Previous efforts to systematically describe genetic interactions between redundant genes on a global scale either have focused on core biological processes in protozoans or have surveyed catastrophic interactions in metazoans. I investigated synthetic genetic interactions between eleven ‘query’ mutants in conserved signal transduction pathways and hundreds of ‘target’ genes compromised by RNAi. A network of 1246 genetic interactions was uncovered through an unbiased global analysis of the interaction matrix, establishing the largest metazoan genetic interaction network to date. To investigate how genetic interactions connect genes on a systems-wide level, the SGI network was superimposed with existing networks of physical, genetic, phenotypic and co-expression interactions. Fifty-six putative functional modules were identified within the superimposed network, one of which regulates fat accumulation and is coordinated by bar-1(ga80)/β-catenin interactions. This led to the discovery that SGI interactions link distinct functional modules on a global scale, which is a previously unappreciated level of organization within metazoan systems. In addition, I present evidence that the properties of genetic networks are conserved between C. elegans and S. cerevisiae, but that the connectivity of the interactions within the current networks is not. Although the buffering between functional modules may differ between species, studying these differences may provide insight into the evolution of divergent form and function. In C. elegans the postsynaptic membrane of the neuromuscular junction reaches its destination through an active process of guided cell extension. The worm has 95 body wall muscles (BWMs) that extend projections called 'muscle arms' to motor axons. The muscle arms harbour the postsynaptic elements of neuromuscular junctions. The stereotypical pattern of muscle arm extension was exploited in a forward genetic screen for new genes required for guided cell migration by looking for mutations that caused a reduction in the number of arms that extend to the motor axons. One of the resulting mutants was tr117, which extended half the number of arms compared to wild type animals. Genetic mapping, complementation tests, and sequencing revealed that tr117 was a mutation in unc-73/Trio, which encodes a guanine nucleotide exchange factor. Expression of UNC-73 specifically in the BWMs rescued the muscle arm development defects of unc-73(e936) mutants, indicating that UNC-73 functions cell-autonomously to regulate muscle arm extension. UNC-73::CFP was localized to muscle arm termini in a pattern similar to that of UNC-40/Dcc, which directs muscle arm extension. UNC-73 over-expression suppressed the Madd phenotype of unc-40 null worms and unc-73(e936) suppressed ectopic myopodia induced by UNC-40 over-expression. These results indicate that UNC-73 functions downstream of UNC-40 in a pathway that regulates muscle arm extension.

The remarkable metabolic capacity of the soil-dwelling bacterium Sinorhizobium meliloti is encoded on its three circular replicons: the chromosome and two large megaplasmids, pSymA and pSymB. Despite making up 45% of the genome, the pSymA and pSymB megaplasmids can be cured from S. meliloti. This unique attribute provides an opportunity to study the essentiality of chromosomal genes in the presence or absence of nearly half the genome. By interrogating chromosomal genes via massively parallel transposon insertion sequencing (Tn-seq) in the presence and absence of pSymA and pSymB, we identified 307 genes as being essential for viability regardless of the genomic context and 104 genes as being essential specifically when the megaplasmids are absent. We also found that ten percent of genes encoded on the chromosome genetically interact with genes on pSymA and pSymB. In addition, Tn-seq data were utilized to significantly refine a metabolic model of S. meliloti, facilitating more accurate fitness predictions in user-defined nutrient and genetic contexts. Furthermore, the development of a library of barcoded transposon insertion (BarSeq) mutants has enabled us to identify genes that are essential for robust growth in hundreds of nutrient environments simultaneously. This will greatly assist efforts to assign more specific functions to the ~30% of S. meliloti genes that have remained uncharacterized over the years. S. meliloti has been studied for decades as a model organism for symbiotic communication. Its legume host, Medicago truncatula, provides fixed carbon for the bacteria in order to receive fixed nitrogen in return. The molecular dialogue between S. meliloti and M. truncatula, initiates and controls each stage of symbiotic development. When inside host cells, intracellular bacteria are subjected to an arsenal of plant-derived Nodule-specific Cysteine-Rich (NCR) peptides that induce significant morphological changes prior to nitrogen fixation. It was previously shown that a bacterial peptidase, HrrP, present in about 10% of S. meliloti isolates, could degrade host-derived peptides and give the bacterial symbionts greater fitness at the expense of the host. In a screen through peptidases conserved throughout the core S. meliloti genome, we identified one peptidase (sapA) that, when overexpressed, significantly modulates symbiotic outcome. In a manner similar to HrrP, SapA degrades NCR peptides in vitro. Additionally, expression of sapA seems to occur specifically inside the plant host providing compelling evidence that some rhizobial peptidases may have evolved away from housekeeping and toward symbiotic functions.

This thesis investigates how pairwise combinatorial gene and stimulus perturbation experiments are conducted and interpreted. In particular, I investigate gene perturbation in the form of knockout, which can be achieved in a pairwise manner by SGA or CRISPR/Cas9 methods. In the present literature, I distinguish two approaches to interpretation: the calculation of stimulus and gene interactions, and the identification of equality among phenotypes measured for distinct perturbation conditions. I describe how each approach has been applied to derive hypotheses about gene regulatory networks. I identify conflicts and uncertainties in the assumptions allowing these derivations, and explore theoretically and experimentally approaches to improve the interpretation of genetic interaction data. I apply the approaches to a well-studied gene regulatory branch of the DNA damage checkpoint (DDC) pathway of Saccharomyces cerevisiae, and confirm the known order of genes within this pathway. I also describe observations that seem inconsistent with this pathway structure. I explore this inconsistency experimentally and discover that high concentrations of the DNA alkylating drug methyl methanesulfonate cause a cell division arrest program distinct from a G1 or G2/M checkpoint or from DNA damage adaptation, that resembles an endocycle.

This thesis focuses on multilocus methods designed to detect single nucleotide polymorphisms (SNPs) that are associated with disease using case-control data. I study multilocus methods that allow for interaction in the regression model because epistasis is thought to be pervasive in the etiology of common human diseases. In contrast, the single-SNP models widely used in genome wide association studies (GWAS) are thought to oversimplify the underlying biology. I consider both pairwise interactions between individual SNPs and modular interactions between sets of biologically similar SNPs. Modular epistasis may be more representative of disease processes and its incorporation into regression analyses yields more parsimonious models. My methodological work focuses on strategies to increase power to detect susceptibility SNPs in the presence of genetic interaction. I emphasize the effect of gene-gene independence constraints and explore methods to relax them. I review several existing methods for interaction analyses and present their first empirical evaluation in a GWAS setting. I introduce the innovative retrospective Tukey score test (RTS) that investigates modular epistasis. Simulation studies suggest it offers a more powerful alternative to existing methods. I present diverse applications of these methods, using data from a multi-stage GWAS on prostate cancer (PRCA). My applied work is designed to generate hypotheses about the functionality of established susceptibility regions for PRCA by identifying SNPs that affect disease risk through interactions with them. Comparison of results across methods illustrates the impact of incorporating different forms of epistasis on inference about disease association. The top findings from these analyses are well supported by molecular studies. The results unite several susceptibility regions through overlapping biological pathways known to be disrupted in PRCA, motivating replication study.