Research in the Dudley lab focuses on two areas: developing new methods to predict an organism’s response to genetic or environmental perturbations and deciphering the novel biology uncovered by applying them. The genotype-phenotype problem is among the most fundamental in biology. Given the complete genome sequence of an individual, can we predict what traits they will exhibit? The emerging field of systems genetics seeks to solve this problem by integrating the principles and technologies of systems biology with genetic analysis. Together with our collaborators, we are applying systems genetics to understand complex traits, such as the predisposition to disease and response to therapeutics. Our work is conducted in the yeast Saccharomyces cerevisiae, which was the first fully sequenced eukaryotic organism and has been key to the development of every major “omic” tool to date. Below are brief descriptions of some of our research interests and ongoing projects.
While yeast is one of the best understood organisms, this knowledge is based on a small set of laboratory strains with limited sequence diversity that makes them poorly suited to model the effects of the numerous alleles and diverse genetic backgrounds that define human populations. To expand the population of strains available for systems genetics research, we assembled a collection of several hundred S. cerevisiae strains. Our collection samples a wide range of geographical locations (38 countries) and ecological niches (plants, insects, soil, industrial processes, and clinical isolates). Because strains from Asia and Africa were substantially underrepresented in publically available repositories, but also likely to be the most genetically diverse, we developed a method for culturing yeast from foodstuffs (e.g. unroasted cocoa and coffee beans) commonly imported from these regions. Using this strategy we have isolated several hundred new strains of from 29 countries across the Americas, Africa, Asia and Oceania.
To characterize the genetic diversity of our strains in a cost effective manner, we used a genome reduction strategy (RAD-seq) that directs sequencing reads to specific restriction sites. This data allowed use to determine the population structure of S. cerevisiae and to demonstrate that our collection contains 4 times the number of polymorphisms as the common laboratory strains. Finally, we have whole genome sequenced a set of 50 new strains that are part of a systems genetics pipeline for the discovery of new biological processes, such as colony morphology (below). We are also using these strains in a collaborative project that integrates genetics, metabolomics, and metabolic modeling to study toxic intermediate accumulation in yeast models of metabolic disease.
High throughput genetics
The field of human genetics is plagued by the problem of “missing heritability”, which might reside in numerous loci of small effect, complex genetic interactions, or somewhere else. Current studies are limited (by power) to detect small effect loci and complex genetic interactions. Yeast crosses performed on very large scales would have the power to detect such loci (if they exist) and inform the methods needed (e.g. depth of mapping) to accurately predict phenotype from genotype in higher organisms, such as mouse and human. While S. cerevisiae is arguably the most genetically tractable eukaryotic organism, yeast genetics still has two major bottlenecks: the generation of large numbers of meiotic progeny and the identification the causative genes within the linked regions. We are developing techniques that marry the power of conventional genetics with ultra-high-throughput DNA sequencing to solve both problems.
Tetrad analysis is a driving force behind the “Awesome Power of Yeast Genetics.” Unfortunately, the manual nature of the process has relegated its application to small-scale studies. Methods that circumvent the laborious process of isolating, disrupting, and placing tetrad spores in an ordered grid have long been sought. We recently published the first high-throughput method (Barcode Enabled Sequencing of Tetrads, BEST) that retains the full information of conventional tetrad analysis and permits the isolation of several hundred tetrads in minutes. BEST also permits the inference of the full genome sequences of non-viable spores, facilitating the study of synthetic lethal interactions that are less limited by strain background and number of interacting genes than current methods.
Even when performed on large scales, complex trait mapping studies yield numerous and relatively large genetic regions that are presumed to contain the gene(s) of interest. Although these studies have the potential of identifying the causative genes underlying a quantitative trait, in practice this is so laborious that only a small number of regions with obvious candidate genes are usually pursued. This limits the discovery of new biology and solving the problem will require new computational and experimental methods. Machine learning methods, including some developed in collaboration with our group, can help prioritize possible causal sequence variants within large chromosomal regions. We are also developing a high throughput, experimental methods for refining genetic regions by isolating rare individuals with informative recombination events.
Colony morphology as a multicellular trait
While microorganisms are often used to study unicellular processes, some wild strains of S. cerevisiae produce complex multicellular features, including folds and channels that form as a colony grows on solid media. These “fluffy” colonies possess properties shared with microbial biofilms, including increased adherence, secretion of an extracellular matrix, and the use of intercellular communication. Fluffy colony formation requires the function and coordination of numerous pathways, and the deletion of key factors produces the “smooth” colony phenotype typical of most laboratory strains. We are applying system genetics, genomics, automated image analysis, and computational modeling to decipher the biological processes underlying this trait.
Our crosses between wild strains of yeast have generated several hundred recombinant progeny that form distinct, genetically heritable colony patterns. That is, while the shapes formed by fluffy colonies differ between strains (“siblings”), clonal isolates of the same strain (“twins”) produce nearly identical shapes. Our goal is to predict the shape of the colony and the temporal dynamics with which it forms based on the strain’s genotype, thereby establishing methods that are able to predict traits that are inherently multicellular. To capture the quantitative parameters needed for this analysis, we developed an automated colony imaging system that captures high resolution, time-lapse images and an automated image analysis pipeline that extracts a rich set of 400 parameters from each image. The trajectory through feature space encodes the spatial and temporal dynamics of colony development and is the basis of an objective system classifying the morphology of each strain (for genetic mapping). We are beginning to use this data (together with RNA expression) to constrain the coarse-grained, grid-based model that we have developed to simulate growth, metabolism, and cellular state distribution within a colony.
ANEUPLOIDY as a PHENOTYPIC SWITCH
While studying colony morphology in the progeny of a cross between a Japanese sake strain and an Ethiopian white tecc strain, we observed that some strains could reversibly switch between the fluffy and smooth state at a frequency (10-3) higher than that expected for spontaneous mutations. Using RAD-seq to assess the molecular karyotype of the cells, we found that the gain of a single chromosome is sufficient to switch a strain from the fluffy to the smooth state, and its subsequent loss to revert the strain back to the fluffy state. We went on to show that copy number imbalance of six of the sixteen S. cerevisiae chromosomes and even a single gene can modulate the switch, suggesting that the state switch is produced by dosage sensitive genes, rather than a general response to altered DNA content. Because aneuploidy is strongly correlated with drug resistance in pathogenic microorganisms, rampant in cancer, and associated with high levels of cancer relapse after drug treatment, aneuploidy itself could be an important target for the development of novel antifungal and cancer therapeutics.