Several problems exist with current methods used to align DNA sequences for comparative sequence analysis. Most dynamic programming algorithms assume that conserved sequence elements are collinear. This assumption appears valid when comparing orthologous protein coding sequences. Functional constraints on proteins provide strong selective pressure against sequence inversions, and minimize sequence duplications and feature shuffling. For non-coding sequences this collinearity assumption is often invalid. For example, enhancers contain clusters of transcription factor binding sites that change in number, orientation, and spacing during evolution yet the enhancer retains its activity. Dot plot analysis is often used to estimate non-coding sequence relatedness. Yet dot plots do not actually align sequences and thus cannot account well for base insertions or deletions. Moreover, they lack an adequate statistical framework for comparing sequence relatedness and are limited to pairwise comparisons. Lastly, dot plots and dynamic programming text outputs fail to provide an intuitive means for visualizing DNA alignments.

Background: Discovery of precise specificity of transcription factors is an important step on the way to understanding the complex mechanisms of gene regulation in eukaryotes. Recently, doublestranded protein-binding microarrays were developed as a potentially scalable approach to tackle transcription factor binding site identification. Results: Here we present an algorithmic approach to experimental design of a microarray that allows for testing full specificity of a transcription factor binding to all possible DNA binding sites of a given length, with optimally efficient use of the array. This design is universal, works for any factor that binds a sequence motif and is not species-specific. Furthermore, simulation results show that data produced with the designed arrays is easier to analyze and would result in more precise identification of binding sites. Conclusion: In this study, we present a design of a double stranded DNA microarray for protein-DNA interaction studies and show that our algorithm allows optimally efficient use of the arrays for this purpose. We believe such a design will prove useful for transcription factor binding site identification and other biological problems.

The regulation of gene expression is thought to play a critical role in the development of life's complexity and has become one of biology's most intensely-studied areas of research. This study has brought us, in a small set of model systems, a catalog of components and the mechanisms by which these work together to activate and repress expression. A variety of genomic approaches hold the promise of both generalizing these mechanisms and, through the use of statistical models, generating new insights. However, this new wealth of genomic information has provided more raw data than new understanding, due in part to the failure of these statistical models to account for the inherent complexity of biological information.

Here I demonstrate three approaches, spanning analysis of binding sites, promoter regions, and developmental enhancers, to create, gain insight from, and, most importantly, emphasize the need for more biologically informed statistical models.

First, I show how measuring the evolutionary properties of a transcription factor's binding sites can inform the differentiation of those sites from other sequences. That differentiation typically requires the interpretation of a score using a p-value, but, contrary to common usage, I find that the optimal such p-value threshold can differ greatly between transcription factors. Second, I develop a graphical model that can describe and exploit trends in the positioning of transcription factor binding sites within promoters. Binding sites are short and degenerate, not specifying by themselves enough information to mediate the organism's task of promoter recognition. However, I show that these positional trends can greatly increase the information available for recognition, further showing how they can be applied to the bioinformatic promoter recognition problem. Third, I use evolutionary simulations to construct a null model for the relative positioning and conservation of binding sites within developmental enhancers. I use this model to show that much of the evidence supporting the importance of overlapping and clustered sites as functional necessities of enhancer organization can be reproduced as artifacts of constraint on binding site composition alone. Finally, I discuss progress towards testing spatially scrambled enhancers generated from these models in transgenic Drosophila embryos.

Animals live in a world teeming with microbial life. Co-existing over evolutionary time, both microbes and animals have evolved methods to influence the other to maximize their respective fitness. It has been the focus of my doctoral work to study how microbes affect animal behavior using the model species Drosophila melanogaster.

Like all other animals so-far encountered, the fruit fly digestive system is occupied by microbes consisting of mostly bacteria. Contrary to my expectations, I did not encounter fly behaviors that changed by varying their associated gut bacteria. In order to understand what changes were taking place in flies with manipulated gut flora, I assayed gene expression in dissected guts or in whole flies. I discovered that associating flies from the embryonic stage with zero, one or three bacterial taxa resulted in the same transcriptional profile in the adult gut, suggesting that adult guts are buffered against the bacteria that passage through them. However, associating flies with the yeast Saccharomyces cerevisiae, whether alone or in combination with these bacteria, was sufficient to recapitulate the transcriptional profile observed in guts of conventionally-reared flies. This suggests that yeast, not bacteria, are key in mediating the gut transcriptional response of adult flies. In contrast, transcription within the entire adult animal varied with exposure to different microbial populations, suggesting that different bacterial taxa can influence fly hosts prior to adulthood.

Yeast, Saccharomyces cerevisiae, is a key food source for D. melanogaster both in the wild and in the laboratory. Previous work by former Eisen laboratory member Kelly Schiabor had demonstrated a correlation between attraction to yeast grown under natural (sugar-replete, nitrogen-limiting) conditions (YVN) and chimeric variant in the gene encoding a component of fly olfaction, odorant receptor 22 (Or22). In conjunction with Alli Quan, I tested the hypothesis that the chimeric Or22 allele in D. melanogaster mediates sensitivity of flies to YVN and therefore makes these flies better suited to detect yeast in the environment. Through sequence analysis, bidirectional crosses between chimeric and non-chimeric lines and ultimately, replacement of a non-chimeric Or22 allele with a chimeric allele within a non-chimeric background, we found that Or22 alone does not mediate sensitivity to YVN. Still, the signs of selection at the Or22 locus suggest that this receptor confers some adaptive function in wild flies.

Entomophthora muscae is a fungal pathogen that infects, alters the behavior of, and then kills dipterans. Predominantly reported in house flies and other large Muscoidea, critically-ill flies summit, extend their proboscis, and raise their wings up and away from their dorsal abdomen in the moments prior to death, dying in an elevated position that appears to benefit fungal dispersal and therefore fitness. Despite being described over 160 years ago, we know little of E. muscae biology, especially the molecular means through which it alters host behavior. Serendipitously, I discovered a strain of E. muscae (CNE1) that infects wild Drosophila species, including D. melanogaster. I have isolated this strain in the laboratory both in vivo, through active propagation between healthy fruit flies, and in vitro, in liquid culture. By observing the isolated fungal culture with modern technology, I have been able to corroborate and add to the series of observations describing how this bizarre organism infects, grows and abandons spent hosts. Crucially, by having isolated a strain that naturally infects a model organism, I have been able to begin testing a variety of molecular hypothesis as to which host machinery is necessary for observed phenotypes, demonstrating that specific neurons and genes are not involved in mediating end-of-life behaviors. With stable E. muscae CNE1 culture, I have been able to show that the fungus first travels to the brain and central nervous system in infected flies before proliferating uncontrollably in the body cavity. I have also, with the aid of Michaels Bronski and Eisen, sequenced the genome of the isolate, and, as a consequence, have been able to assay gene expression of both host and fungus over the course of infection. Ultimately, this work represents only the beginning of what is possible in the E. muscae CNE1-D. melanogaster system, not just for understanding the molecular basis of host behavioral manipulation, but also for studying varied aspects of host-microbe interactions.

Changes in transcriptional regulatory networks are thought to underlie most morphological change observed across taxa, but the general principles of how such networks change and why remain unknown. While many studies have focused on evolutionary changes in cis-acting components of transcriptional networks (enhancers and transcription factor binding sites), changes in the transcription factor proteins which bind these sites have been mostly overlooked. In this thesis I identify the putative Drosophila transcription factor \her her) as having a highly divergent DNA-binding domain, and examine its DNA-binding profile in two fruitfly species, Drosophila melanogaster and Drosophila pseudoobscura. I find evidence for a large-scale divergence in the distribution of HER binding, as well as a possible difference in DNA-binding preference between these two species. This establishes Drosophila her<\italic> as an excellent system for the study of regulatory evolution through changes in transcription factors.

Background

Recognizing regulatory sequences in genomes is a continuing challenge, despite a wealth of available genomic data and a growing number of experimentally validated examples.

Methodology/Principal Findings

We discuss here a simple approach to search for regulatory sequences based on the compositional similarity of genomic regions and known cis-regulatory sequences. This method, which is not limited to searching for predefined motifs, recovers sequences known to be under similar regulatory control. The words shared by the recovered sequences often correspond to known binding sites. Furthermore, we show that although local word profile clustering is predictive for the regulatory sequences involved in blastoderm segmentation, local dissimilarity is a more universal feature of known regulatory sequences in Drosophila.

Conclusions/Significance

Our method leverages sequence motifs within a known regulatory sequence to identify co-regulated sequences without explicitly defining binding sites. We also show that regulatory sequences can be distinguished from surrounding sequences by local sequence dissimilarity, a novel feature in identifying regulatory sequences across a genome.

Enhancers are cis-regulatory elements that regulate when and where genes are expressed. These functions are mediated by transcription factors that bind to sites within the enhancer sequence. It is still an open question how transcription factors find their binding sites in an expanse of genomic DNA or how they facilitate enhancer activity. My doctoral work aims to address these questions in part, through the lens of Drosophila melanogaster embryogenesis. In the early Drosophila embryo, enhancers regulate gene expression in refined spatial patterns, setting a foundation for the adult body plan. In chapter 1, I discuss enhancer based regulation of gene expression - what is currently known about how enhancers function and how they themselves are regulated in the context of embryonic development.

In the nucleus, genomic DNA forms higher-order chromatin structures. However, enhancers and promoters that are transcriptionally active are often highly accessible. In chapter 2, I seek to understand how chromatin accessibility is established around active enhancers and promoters by asking if chromatin accessibility is spatially patterned across the embryo in a way that correlates with transcriptional activity. I used genome-wide methods to assay chromatin accessibility in anterior and posterior halves of Drosophila embryos. I found that genome-wide, chromatin accessibility is remarkably similar in the anterior and posterior half but that enhancers show greater accessibility in the half of their activity. These data are consistent with a model where a uniform chromatin landscape is established first and is later refined by spatially-patterned transcriptional activity.

Zelda is a transcription factor that is necessary for enhancer activation in the early Drosophila embryo. Zelda has been observed to form clusters of high concentration in the nucleus. In chapter 3, I seek to address whether Zelda’s clustering activity is related to its regulatory functions by asking whether these clusters are near transcriptional loci. To answer this question, I developed a method to measure Zelda protein abundance at twenty-three Zelda targets and twenty-five non-targets in the same nucleus. I found that Zelda protein is modestly enriched at targets compared to non-targets. This enrichment is most likely driven by a small percentage of loci that reside in regions of high intensity Zelda protein.

Throughout my PhD, I have sought a method to create large numbers of site-specific mutants in vivo. This is challenging in Drosophila and other organisms where transgenic lines are generated by injection with low transformation rates. Mutating and rearranging binding sites has been largely successful at identifying important enhancer elements, like transcription factor binding sites, in the past. Therefore, a method to create large numbers of site-specific mutants would greatly enhance our ability to comprehensively dissect regions of the genome at greater scale. In chapter 4, I describe a method for generating diverse populations of site-specific enhancer variants where each fly laid in a cross possesses at most one copy of the mutant enhancer. I then show preliminary evidence that this method can efficiently generate diverse mutations. Though this method is not yet fully developed, it is the beginning of a new approach for characterizing enhancers that can be applied to any genomic region.

Understanding the genetic basis underlying the process of speciation is one of the primary goals in the field of evolutionary biology. However, despite recent and exciting progress in the field of speciation genetics, particularly in the area of postzygotic isolating mechanisms, surprisingly little is still known about the genetic basis and evolutionary forces that are important early on in speciation. Notably, molecular mechanisms relating to the evolution of prezygotic isolating barriers are particularly poorly understood. While studies on the genetics of postzygotic isolating barriers are critical to our understanding of how species boundaries are maintained post-speciation, such factors potentially may not have been involved in driving the actual speciation event, and instead may have evolved secondarily. By studying recently diverged populations, we increase the chances that the differences in the genome that we detect are actually directly responsible for driving reproductive isolation and thus speciation. The widespread availability of whole genome sequencing techniques opens up the opportunity to examine speciation at the genomic level in non-model species that may be more applicable to the study of early speciation.

This research takes advantage of whole genome sequencing techniques and a very young semispecies system, Drosophila athabasca. The D. athabasca species complex, which is composed of three overlapping semispecies - Western-Northern, Eastern-A, and Eastern-B, provides a unique system in which to study incipient speciation using population genomics. The three semispecies of D. athabasca are estimated to have diverged less than 25,000 years ago and are morphologically indistinguishable. Individuals will hybridize in the laboratory, but their geographic ranges and distinct courtship songs, which result in prezygotic isolation, differentiate the populations sufficiently for them to be designated as semispecies. This very young divergence time and unique population structure within D. athabasca makes it an ideal system to study the genetics of prezygotic isolation and incipient speciation.

I first generated a de novo reference genome assembly for D. athabasca by sequencing the genome at 30X coverage using Illumina next-generation sequencing technologies and annotated this reference genome using a combination of de novo, comparative, and mRNAseq gene finding methods. In order to examine the genome of D. athabasca at a population genomic level, I established 404 iso-female lines of D. athabasca collected from across the species range, including the previously identified ranges of all three semispecies. I characterized courtship songs from a subset of these lines and sequenced the genomes of 28 individuals, roughly equally distributed geographically and between semispecies, each at 10X coverage.

Using this genome-wide population data, I quantified levels of genome-wide diversity and differentiation within and between semispecies. Despite relatively low levels of divergence within the complex, principal component and phylogenetic analyses using the genomic data clearly separates individuals into distinct genetic groups corresponding to the three behaviorally defined semispecies. Furthermore, phylogenetic analysis places Eastern-A and Eastern-B as sister taxa, confirming previous research indicating that Eastern-A and Eastern-B semispecies are the more closely related semispecies, with Western-Northern being the most anciently diverged of the three semispecies. To infer the speciation history of the D. athabasca complex, I fit the data to demographic models and estimate divergence under a model of isolation with low levels of migration. This model estimates a divergence time of only 6,000 years ago for the Eastern-A/Eastern-B split and 16,000 years for the Western/Eastern split, consistent with a previous hypothesis of population expansion and colonization of North America following the last glacial maximum.

Overall divergence within the semispecies is low, with approximately 2 million sites variable within D. athabasca, and only 1% of these variable sites being private and fixed within semispecies. Furthermore, I find divergence is not evenly distributed across the genome, with the X-chromosome exhibiting increased levels of divergence compared with autosomes. Most interestingly, despite the low levels of overall divergence, genome-wide scans identify a single large spike of differentiation between the two youngest semispecies, which have an estimated divergence time of only 6,000 years. Scans for selection also show strong signatures of a selective sweep within semispecies at this same locus, indicating that divergence in this particular region has likely been driven by selection. Further analysis of this region reveals that it harbors a gene previously identified to be involved in courtship song in other species within the Drosophila genus, suggesting that it may play an important role in the evolution of prezygotic reproductive barriers between the D. athabasca semispecies.

This study provides one of the first genome-wide population genetic investigations of the molecular changes and population parameters important during incipient speciation, contributing important new information to our view of the genetics of speciation. With some of the youngest divergence times used to study speciation genetics on a genome-wide level thus far, examining the patterns of divergence within and between the semispecies of D. athabasca has allowed us to identify a candidate gene that may play a role in the evolution of prezygotic isolation, and thus, reproductive isolation at the earliest stages of speciation.

Many key regulators of early embryogenesis in Drosophila melanogaster are X-linked. However, the canonical, MSL-mediated dosage compensation, which involves hyper-transcription of the genes on the single X chromosome in males, is not active until the post-syncytial stage of development. A separate MSL-independent dosage compensation system active earlier in development has been described, though the mechanism through which the process functions remain unclear. In this study, we quantified transcription in living embryos at single-locus resolution to determine if early dosage compensation of the X-linked genes buttonhead and giant is sensitive to X chromosome dose. We found no evidence for a transcriptionally regulated mechanism of early dosage compensation, suggesting that the previously observed compensation of mRNA levels for these genes is achieved via a post-transcriptional regulatory mechanism.

Drosophila melanogaster and Saccharomyces cerevisiae, two model systems of molecular biology, interact in the wild, but the chemical basis of this interaction is largely unknown. This thesis details an effort to understand the molecular basis of this ecological co-localization.

Results:

While screening a collection of wild and laboratory yeast strains for their ability to attract adult D. melanogaster (Raleigh 437), I noticed a large difference in fly preference for two nearly isogenic strains of S. cerevisiae, BY4741 and BY4742. Using standard genetic analyses, I tracked the difference in preference to lack of mitochondria in BY4742. I used gas chromatography coupled with mass spectrometry (GC-MS) to examine the volatile compounds produced by BY4741 and the mitochondria-deficient BY4742, and found that they differed in their production of many known fly attractants, including ethyl hexanoate and ethyl octanoate, which were produced at much higher levels in yeast strains with mitochondria during aerobic fermentation, a metabolic strategy that distinguishes S. cerevisiae from most other microbes.

Through a detailed investigation of the volatile profiles and RNA expression patterns produced by yeast under multiple nutritional scenarios, I determined that the production of ethyl hexanoate and ethyl octanoate during aerobic fermentation depends on the level and type of nitrogen present in the substrate. Fermentative growth on nitrogen-limited substrates requires mitochondrial engagement and the metabolic configuration employed by S. cerevisiae to contend with this nutritional scenario results in the production of these ester attractants. This nutritional scenario – high sugar but limited nitrogen – matches the composition of fruit, the natural co-localization environment for flies and yeast, suggesting that these volatiles are ecologically relevant cues for D. melanogaster.

D. melanogaster sense ethyl hexanoate and ethyl octanoate via the odorant receptor genes Or22a and Or22b. A single gene version of this locus, which is a chimera of the two ancestral copies, is segregating in natural D. melanogaster populations. I found that D. melanogaster lines harboring the chimeric locus (including Raleigh 437) show more robust preference for ethyl hexanoate and ethyl octanoate-producing S. cerevisiae cultures. This provides an ecological basis, and possibly selective advantage, for the maintenance of this allele in D. melanogaster.