Viewed through the lens of comparative genomics, how have chromosomes and karyotypes evolved, particularly under extreme scenarios of rapid or torpid evolution? To examine this question, I produced and analyzed new high-quality, chromosome-scale genome assemblies, representing cases of extreme chromosome evolution. The interplay between convergent and divergent genomic architectures can further our understanding into the maintenance of chromosome structure and organization as well as the underlying biological mechanisms.

Despite their recent divergence, muntjac deer show striking karyotype differences. In Chapter 2, I produced new chromosome-scale genome assemblies for the Chinese and Indian muntjacs, Muntiacus reevesi (2n=46) and Muntiacus muntjak (2n=6/7), and analyzed their evolution and architecture. I identified six fusion events shared by both species relative to the cervid ancestor and, therefore, present in the muntjac common ancestor, six fusion events unique to the M. reevesi lineage, and 26 fusion events unique to the M. muntjak lineage. One of these M. muntjak fusions reversed an earlier fission in the cervid lineage. Although comparative Hi-C analysis uncovered differences in long-range genome contacts and A/B compartment structures, I discovered widespread conservation of local chromatin contacts between the muntjacs, even near the fusion sites. Analysis of the muntjac genomes revealed new insights into this unique case of rapid karyotype evolution and the resulting biological variation.

In contrast to muntjacs, frogs are a more phylogenetically ancient order, with two prominent species, Xenopus laevis and Xenopus tropicalis, being utilized as vertebrate cell and developmental model systems. In Chapter 3, I reported new chromosome-scale genome assemblies for three distantly related frogs: the common coquí (Eleutherodactylus coqui), the túngara frog (Engystomops pustulosus), and the Zaire dwarf clawed frog (Hymenochirus boettgeri). Through comparative sequence analysis of these and other frog genomes, I identified long-range and even chromosome-scale synteny, denoting the stability and pervasiveness of this type of genomic conservation across the frog phylogeny. These chromosome-scale comparisons also revealed the 13 ancestral chromosomes, with limited Robertsonian translocations and end-to-end fusions explaining the observed chromosome variations. These assemblies, encompassing over a billion years of evolutionary divergence, unmasked insights into torpid karyotype evolution in frogs.

Duplication has long been recognized as an evolutionary source of novelty. The relaxation of purifying selection following duplication allows for normally deleterious mutations to persist long enough to give rise to novel phenotypes. Whole-genome duplications (WGDs) are a specific type of duplication, in which a species suddenly finds itself with two copies of all of its genomic loci. While the fate of most of the duplicated loci is to be lost, those that persist are thought to underlie the innovations seen in groups with a history of polyploidy, such as flowering plants, yeast, Paramecium, and vertebrates. These ancient events give us an idea of how WGDs can drive the radiation of large and diverse phyla, but do not give us any information on the genomic response immediately following polyploidy. This thesis provides insights into the origins of polyploidy and its effects on genome dynamics.

There are two models for the mechanism of polyploidy: autopolyploidy and allopolyploidy. Autopolyploids are formed by doubling the somatic chromosomes in the zygote or early embryo. Allopolyploids are formed by the hybridization of two related, but genetically distinct, species, followed by chromosome doubling. If there are no extant diploid relatives, it can be difficult to distinguish between these two models. One feature of allopolyploids is the lack of recombination between their homeologous chromosomes. The end result is that any markers that were unique to each species while apart, such as transposable element subfamilies, will be asymmetrically distributed on the progenitor chromosomes in an organism that recently underwent a WGD.

Xenopus laevis is an important vertebrate model in developmental and cell biology that has experienced a recent WGD (~40 million years ago [MYA], based on cDNA alignments (Hellsten, 2007). Its diploid cousin Xenopus tropicalis has become a popular genetic model frog. Comparative analysis of these two frog genomes gives us an excellent opportunity to study genome dynamics following whole genome duplication. The discovery of asymmetrically distributed transposon subfamilies supports the model that cross-species hybridization through allotetraploidy is the mechanism underlying the polyploid Xenopus radiation. Thus, the sub-genome sequence divergence of 40 MYA dates the divergence of the progenitor species, not the hybridization event. The asymmetric distribution of these elements between homeologous sequences allows us to assign chromosomes to progenitor species, named “A” and “B”, making X. laevis a unique system to study sub-genome-specific evolution. The wealth of transcriptome and epigenetic data available for Xenopus allows me to assay how these genomic changes affect gene expression as well as gene retention. The combination of these resources with genomic data gives me the resolution needed to date the hybridization both by studying the decay of unitary pseudogenes and by comparative analysis of the transposable elements discussed above.

The sub-genome from progenitor species “A” has more assembled length, longer chromosomes, a higher rate of gene retention, and higher average expression in the adult frog. The B sub-genome has higher synonymous and nonsynonymous mutation rates. The chromosomes orthologous to X. tropicalis 9 and 10 are fused in both sub-genomes of X. laevis, forming homeologous chromosomes 15 and 18, and deviate from the A/B trends discussed above. The regions of these X. laevis chromosomes orthologous to X. tropicalis chromosome 10 have a lower density of diagnostic repeats, no sub-genome bias in gene retention, and have a higher silent substitution rate. This divergence from the rest of the genome is not shared by the regions orthologous to X. tropicalis 9. I hypothesize that the short length of X. tropicalis 10 plays a role in these deviations due to a higher rate of gene conversion on shorter chromosomes.

The ornate arrangement of diverse cells into specialized tissues, organs, and higher structures characteristic of multicellular organisms is all encoded from the same genome sequence. Despite their differences, morphologically distinct cells (e.g. muscle cells and neurons) must transcribe many of the same genes. Morphological indistinguishable cells must often transcribe distinct sets of genes (e.g. different odorant receptor cells). The ensemble of genes expressed in a given cell -- and the relative frequency they are expressed at, give each cell its characteristic identity more so than the presence of individual genes. Therefore understanding the genetic control of development and differentiation is a question not so much of the understanding the gene sequences themselves, but the regulatory structure of the genome which determines how they are deployed.

In order for development to unravel in such a manner that each embryo makes it through the process with all the correct parts in the correct positions at the end, this process must be exceedingly precise. Though often taken for granted, this precision becomes particular impressive if one considers the frequency with which mistakes are made in intelligently designed human built assembly processes. The developing animal must position components correctly on scales of microns (e.g. tissue boundaries) and nanometers (e.g. neuron-junctions), has no external direction of assembly, and requires thermal noise to position many of its components (including essentially all transcription factors - proteins which regulate read access to the genome).

It is not sufficient for the process to be precise. It must also be robust to changes in the conditions in which it operates, such as different thermal environments, nutrient conditions, and chemical environments. This robustness enables a certain degree of plasticity, such that some components of the system can change and evolve new functions, without causing catastrophic failure of the rest of the system.

In my thesis research I have tried to explore some of the molecular mechanisms of gene regulation which support the precise and robust expression of multicellular genomes. Rapid advances in post-genomic technologies have exposed a broad range of fundamental differences in the organization and regulation of multicellular genomes such as Drosophila. I have worked primarily on two phenomena, the use of promoter proximal pausing as a regulatory strategy, and the use of multiple apparently redundant regulatory sequences to drive expression of the same gene. Discovery of both of these phenomena emerged from analysis of whole genome polymerase and transcription factor binding data. Using quantitative high resolution in situ and semi-automated computational image processing I have studied the detailed differences in the transcriptional activation and transcription frequency of genes regulated by these mechanisms. Through this analysis I have shown a strong correlation through more rapid and synchronous gene expression and regulation through release of promoter proximal paused polymerase. Theoretical modeling demonstrates that such an effect can be expected from regulating release of stable downstream state in a general assembly process (such as construction of the RNA Pol II pre-initiation complex).

Analysis of gene expression driven by multiple enhancers with overlapping activity compared to constructs with only a single active enhancer revealed that the process by which an enhancer binds its target transcription factors and activates expression is often limiting enough that having a second independent copy can produce detectable changes in the frequency of transcription. This reduction of natural variation in gene activation is especially important under stress conditions, such as thermal stress or reduced levels of some of the activating factors. Robustness to this sort of variation may be important both for adaptation within a species and the flexibility to allow modification of interacting pathways in the course of evolutionary modification. These investigations also revealed a corrective propensity whereby the simultaneous activity of multiple enhancers, responding to repressors as well as activators, can give rise to correctly restricted gene expression even when the elements taken in isolation drive some degree of ectopic expression.

So far both of these mechanisms have only been reliably documented in multicellular systems, suggesting that the precision and robustness they confer may be an innovation of metazoans in response to increased levels of coordination required to keep many cells functioning in the tight cooperation of a multicellular organism. Doubtless this is but scratching the surface of the mechanisms which ensure such precision and control. However the rapid improvements to both genomic tools and imaging technology make it like to be a promising field for further exploration for years to come.

Genetic variation is the result of DNA sequence differences between individuals or populations. Our understanding of genetic variation has benefited greatly from advances in DNA sequencing, computational power, and statistical tools. I will present four statistical analyses of genetic variation related to genetic mapping, genome editing, evolutionary biology, and de novo genome assembly and annotation. First, I will present a novel methodology for genotype imputation and composite genetic marker construction. This technique, minimum spanning tree imputation, is designed to improve linkage maps in outbred F1 crosses, particularly from genetic markers derived from low confidence sequencing. I then use minimum spanning tree imputation to construct sex-specific genetic maps for Xenopus laevis, Branchiostoma floridae, and Miscanthus sinensis. Second, I will present a case of three mouse lines edited with the CRISPR/Cas9 system in order to delete an enhancer locus in the IL2RA gene. All three lines were confirmed to have the intended deletion. Curiously, one line displayed a severe immune deficient phenotype that persistently bred true. By resequencing mice from these lines, I was able to identify the occurrence of a tandem duplication as an off-target consequence of the editing in the immune compromised individuals that was absent in the other lines. The tandem duplication was then confirmed experimentally. I close by proposing a repair mechanism mediated by microhomology that might have caused the tandem duplication to form. Third, I will compare the chromosomal position of orthologous genes between lancelet amphioxus, Branchiostoma floridae (an early-branching living chordate) and five vertebrates. These comparisons will offer support for the “2R hypothesis,” that early vertebrate organisms underwent two rounds of whole genome duplication. This analysis takes advantage of a novel way of computing and visualizing mutual best hits between previously identified chordate linkage groups (CLGs) and segments of vertebrate chromosomes. Finally, I will present my contributions to the assembly and annotation of the genome of a regenerative model organism, Hofstenia miamia.