Multiple rounds of whole genome duplication (WGD) have played a pivotal role in the expansion, elaboration, and evolutionary diversification of vertebrate genomes. In addition to sharing two rounds of whole genome duplication with all other vertebrates, a teleost-specific genome duplication (TGD) occurred in the stem of the teleost lineage ~350 million years ago (MYA) and is thus a genomic synapomorphy shared by all ~26,000 extant species. The TGD has variously been implicated in accelerated speciation, evolution of morphological complexity, increased rates of molecular evolution, and the evolution of novelty, and therefore is therefore of significant interest for its impact on teleost evolution and also as a model for understanding the evolutionary patterns and processes which accompany WGDs more generally. Investigation of the TGD has contributed extensively to the general understanding of WGDs however, until the present work, a relatively narrow taxonomic sampling of species within a single teleost subdivision, Clupeocephala, have been investigated. This taxonomic bias has left potentially relevant evolutionary changes to the teleost genome in the immediate wake of the TGD obscured. Due to their deeply branching ancestry, species belonging to the two other major teleost subdivisions, Osteoglossomorpha and Elopomorpha, are well positioned for deeper comparative genomic analyses of the TGD and the accompanying phenomenon of diploidization. The focus of the present work has been to develop the first genomic resources specifically for osteoglossomorphs and to investigate the evolutionary patterns and processes which accompanied diploidization prior the deep divergence of the three extant teleost subdivisions. To this end, I have generated de novo genome and transcriptome data from four osteoglossomorph taxa (Pantodon buchholzi, Osteoglossum bicirrhosum, Chitala ornata, and Gnathonemus petersii) and conducted comparative genomic and phylogenetic analysis with other teleosts and pre-TGD vertebrates including the gar Lepisosteus oculeatus. With a focus on Hox and other ANTP class homeobox-containing transcription factor families I provide evidence that speciation of the major teleost subdivisions occurred prior to the termination of the diploidization process following TGD and discuss the evolutionary implications of this model. Beginning with an analysis of the Hox clusters in P. buchholzi I show that divergent resolution of TGD-generated Hox duplicates occurred both at the individual gene level as well as at the level of whole cluster losses. Detailed phylogenetic analyses of the P. buchholzi Hox clusters further revealed that the transition from polyploid alleles to full paralogs during the diploidization process can occur independently in different lineages when speciation rapidly follows WGDs, causing duplicated genes to exhibit a special case of four-way gene homology which I have termed 'tetralogy'. A genome-wide survey of ANTP class homeobox genes in a de novo assembly of the P. buchholzi genome revealed that ancient TGD duplicates of at least 14 subfamilies were preserved uniquely in the P. buchholzi genome and lost from clupeocephalan teleosts. Finally, by comparing the Hox complements in gar and P. buchholzi with three additional osteoglossomorphs I show that the diversity in potential duplicate resolution patterns is also highly variable between osteoglossomorph families. Overall, this work highlights the importance of considering not only the relative timing of gene duplication and speciation in comparative genomic analyses but also their timing relative to diploidization. Going forward, the research community will need to carefully evaluate the effects differences in diploidization rate and pattern, both between lineages and across the genome, have had in influencing the fate of individual gene duplicates as well as upon the macroevolutionary phenomena frequently correlated with WGDs more generally.

With the proliferation of plant genomes being sequenced, assembled, and annotated, duplicate gene loss from whole genome duplication events, also known in plants as frac- tionation, has shown to have a different pattern from the classic gene duplication models described by Ohno in 1970. Models proposed more recently, the Gene Balance and Gene Dosage hypotheses, try to model this pattern. These models, however, disagree with each other on the relative importance of gene function and gene expression. In this thesis we explore the effects of gene function and gene expression on duplicate gene loss and retention. We use gene sequence similarity and gene order conservation to construct our gene fam- ilies. We applied multiple whole genome comparison methods across various plants in rosids, asterids, and Poaceae in looking for a general pattern. We found that there is great consistency across different plant lineages. Genes categorized as metabolic genes with low level of expression have relatively low fractionation resistance, losing duplicate genes readily, while genes categorized as regulation and response genes with high level of expression have relatively high fractionation resistance, retaining more duplicate gene pairs or triples. Though both gene function and gene expression have important effects on retention pattern, we found that gene function has a bigger effect than gene expression. Our results suggest that both the Gene Balance and Gene Dosage models account to some extent for fractionation resistance.

We propose a new, continuous model of the fractionation process (duplicate gene deletion after polyploidization) on the real line. The aim is to infer how much DNA is deleted at a time, based on segment lengths for alternating deleted (invisible) and undeleted (visible) regions. After deriving a number of analytical results for "one-sided" fractionation, we undertake a series of simulations that help us identify the distribution of segment lengths as a gamma with shape and rate parameters evolving over time. This leads to an inference procedure based on observed length distributions for visible and invisible segments. We suggest extensions of this mathematical and simulation work to biologically realistic discrete models, including two-sided fractionation.

Le Dimethylsulfoniopropionate (DMSP) est une molécule à fort impact écologique couramment produite par le phytoplancton marin, mais très rarement chez les plantes à fleurs: seulement chez quelques genres (dont Spartina chez les Poacées). Bien que les étapes enzymatiques impliquées dans la voie de biosynthèse du DMSP soient connues chez les spartines, son origine ainsi que les gènes impliqués restent encore à découvrir chez les plantes. Cette étude s’est fixée pour objectif de contribuer à élucider les mécanismes à l’origine de cette fonction chez les spartines. Cette question a été appréhendée à travers différentes approches : biochimique, métabolomique, transcriptomique, génomique comparative et phylogénétique. Les résultats ont montré que la capacité à synthétiser le DMSP a une origine unique au sein du genre Spartina et se serait mise en place il y a 3-10 millions d’années. Cette capacité est intervenue chez l’ancêtre d’un des deux principaux clades (hexaploïde) de spartines, puis a été héritée chez toutes les espèces dérivant de ce clade (hexaploïdes à dodécaploïdes). Les espèces de l’autre clade (tétraploïde) et leurs descendants (quel que soit leur niveau de ploïdie) n’accumulent pas de DMSP. En utilisant les génomes séquencés des espèces de Poacées ainsi que les ressources génomiques et transcriptomiques disponibles chez les spartines, les gènes candidats intervenant dans les 4 étapes de la voie de biosynthèse proposée dans la littérature ont été explorés. L’identification des gènes intervenant dans les deux étapes intermédiaires, supposées spécifiques de la capacité de synthèse du DMSP représente un véritable défi dans la mesure où seules des activités enzymatiques putatives ont été proposées à ce jour (sans connaissance des enzymes spécifiques ni de leur séquence protéique). Nous avons pu identifier une série de gènes candidats pour chacune des deux fonctions concernées (décarboxylase et amine oxydase), comparer leur niveau de transcription entre les espèces DMSP+ et DMSP-, et prédire leur localisation cellulaire. De plus, des analyses d’activités enzymatiques ont permis de formuler de nouvelles hypothèses et pistes de recherches sur l’émergence de cette nouvelle voie de biosynthèse chez les spartines. / Dimethylsulfoniopropionate (DMSP) is an ecologically important molecule produced by most marine phytoplankton species, but very rarely by flowering plants: only in a few genera (including Spartina in Poaceae). Despite the different enzymatic steps involved in DMSP biosynthesis are well known, the origin of the function and the genes encoding the different enzymes are yet to be discovered. To explore the evolutionary mechanisms involved in the DMSP accumulation in Spartina, we used various approaches, including biochemical analyses, metabolomics, transcriptomics, comparative genomics and phylogenetics. Notably, we demonstrate that the ability to synthesize DMSP evolved once in the Spartina genus, sometimes 3-10 million years ago. This functional innovation occurred following the emergence of the hexaploid clade, and was inherited by all Spartina species deriving from this hexaploid ancestor. Spartina species belonging to the tetraploid clade and their deriving species do not accumulate DMSP (whatever their ploidy level). Using Poaceae sequenced genomes as well as Spartina genomic and transcriptomic resources obtained in our laboratory, candidate genes involved in the four different enzymatic steps of the DMSP biosynthesis pathway were searched. Identifying genes involved in the intermediate (2nd and 3rd) steps that are specific to this pathway was particularly challenging as only putative enzymatic activities have been proposed so far (corresponding protein sequences and genes are unknown). A set of candidate genes potentially involved in these two steps (with decarboxylase and amine oxydase activities) were identified and their transcription levels were compared among DMSP producing (DMSP+) and non-producing (DMSP-) Spartina species. Their putative cellular localization was also predicted. Moreover, enzymatic activity assays open new hypotheses and research perspectives regarding this enigmatic biosynthesis pathway in Spartina.

The ParaHox cluster contains three Hox‐related homeobox genes. The evolution of this sister of the Hox‐gene clusters has been studied extensively in metazoans with a focus on its early evolution. Its fate within the vertebrate lineage, and in particular following the teleost‐specific genome duplication, however, has not received much attention. Three of the four human ParaHox loci are linked with PDGFR family tyrosine kinases. We demonstrate that these loci arose as duplications in an ancestral vertebrate and trace the subsequent history of gene losses. Surprisingly, teleost fishes have not expanded their ParaHox repertoire following the teleost‐specific genome duplication, while duplicates of the associated tyrosine kinases have survived, supporting the hypothesis of a large‐scale duplication followed by extensive gene loss.

Like many angiosperms, Brassica rapa underwent several rounds of whole genome duplication during its evolutionary history. Brassica rapa is particularly valuable for studying genome evolution because it also experienced whole genome triplication shortly after it diverged from the common ancestor it shares with Arabidopsis thaliana about 17-20 million years ago. While many B. rapa genes appear resistant to paralog retention, close to 50% of B. rapa genes have retained multiple, paralogous loci for millions of years and appear to be multi-copy tolerant. Based on previous studies, gene function may contribute to the selective pressure driving certain genes back to singleton status. It is suspected that other factors, such as gene expression patterns, also play a role in determining the fate of genes following whole genome triplication. Published RNA-seq data was used to determine if gene expression patterns influence the retention of extra gene copies. It is hypothesized that retention of genes in duplicate and triplicate is more likely if those genes are expressed in a tissue-specific manner, as opposed to being expressed globally across all tissues. This study shows that genes expressed specifically in flowers and roots in B. rapa are more resistant to fractionation than globally expressed genes following whole genome triplication. In particular, there appears to have been selection on genes expressed specifically in flower tissues to retain higher copy numbers and for all three copies to exhibit the same flower-specific expression pattern. Future research to determine if these observations in Brassica rapa are consistent with other angiosperms that have undergone recent whole genome duplication would confirm that retention of flower-specific-expressed genes is a general feature in plant genome evolution and not specific to B. rapa.

Whole genome duplication, or polyploidy, is a common process in plants by which failures in meiosis or fertilization result in offspring with twice the number of chromosomes. This doubles the number of copies of every gene, an effect thought to generate new ‘raw material' upon which natural selection can act. Few studies exist examining the consequences of polyploidy for plant physiological traits. Doubling the number of gene copies may have unknown effects on leaf structure and function. In this study, I compare diploid, tetraploid, and hexaploid species within the genus Helianthus (wild sunflowers). Forty different accessions of wild sunflowers were grown under standardized greenhouse conditions and phenotyped for both leaf functional traits and leaf hyperspectral reflectance. Interestingly, I find that whole genome duplication can have effects on leaf functional traits relevant to both size and ecophysiology, and thus that polyploidy may lead to functional trait differentiation between polyploids and their diploid progenitors.

Background: The Atlantic salmon, Salmo salar L., is a prominent member of the Salmonidae family, and has been the focus of intense research because of its environmental and economic significance as an iconic sporting species and its global importance as an aquaculture species. Furthermore, salmonids constitute ideal organisms for the study of evolution by gene duplication as they are pseudotetraploid descendants of a common ancestor whose genome was duplicated some 25 to 100 million years ago. Whole-genome duplication is considered a major evolutionary force capable of creating vast amounts of new genetic material for evolution to act upon, promoting speciation by acquisition of new traits. Recently, large-scale comparison of paralogous genes in Atlantic salmon suggested that asymmetrical selection was acting on a significant proportion of them. However, to elucidate the physiological consequences of gene and genome duplications, studies integrating molecular evolution and functional biology are crucial. To this end, sequence and molecular analyses were performed on duplicated Elovl5 fatty-acyl elongases of Atlantic salmon, as they are responsible for a rate-limiting reaction in the elongation process of long-chain polyunsaturated fatty acids (LC-PUFA), critical components of all vertebrates. The aim of the research presented here was to investigate the role of gene duplication as an evolutionary process capable of creating genetic novelty, and to identify the potential ecological and physiological implications. Results: Linkage analyses indicated that both fatty-acyl elongases segregated independently and located elovl5 duplicates on different linkage groups. Genetic mapping using microsatellites identified in each elovl5 locus assigned elovl5a and elovl5b to chromosomes ssa28 and ssa13, respectively. In silico sequence analysis and selection tests indicated that both salmon Elovl5 proteins were subject to purifying selection, in agreement with previous results showing indistinguishable substrate specificities. Gene expression and promoter analysis indicated that Elovl5 duplicates differed in response to dietary lipids and tissue expression profile. Lipid biosynthesis and metabolic gene expression profiling performed in Atlantic salmon SHK-1 cells, suggested that the control of lipid homeostasis in fish is similar to that described in higher vertebrates, and revealed the particular importance of Lxr and Srebp transcription factors (TFs) in the regulation of LC-PUFA biosynthetic enzymes. Sequence comparison of upstream promoter regions of elovl5 genes showed intense differences between duplicates. Promoter functional analysis by co-transfection and transcription factor transactivation showed that both elovl5 duplicates were upregulated by Srebp overexpression. However, elovl5b exhibited a higher response and its promoter contained a duplication of a region containing response elements for Srebp and NF-Y cofactors. Furthermore, these studies indicated an Lxr/Rxr dependant response of elovl5a, which was not observed in elovl5b. Analysis of the genomic sequences of elovl5 duplicates by comparison to various sequence databases showed an asymmetrical distribution of transposable elements (TEs) in both introns and promoter regions. Further comparison to introns of the single elovl5 gene in pike indicated much higher TE distribution in salmon genes compared to the pike. Conclusions: Although not conclusive, the most parsimonious origin for the salmon elovl5 duplicates is that they are derived from a WGD event. This conclusion is also supported by the close similarity of two elovl5 paralogs in the recently available rainbow trout genome. Regardless of their origin, Atlantic salmon elovl5 genes have been efficiently retained in the genome under strong functional constraints indicating a physiological requirement for both enzymes to be functionally active. In contrast, upstream promoter regions have strongly diverged from one another, indicating a relaxation of purifying selection following the duplication event. This divergence of cis-regulatory regions has resulted in regulatory diversification of the elovl5 duplicates and regulatory neofunctionalisation of elovl5a, which displayed a novel Lxr/Rxr-dependant response not described in sister or other vertebrate lineages. Promoter analysis indicated that the observed elovl5 differential response to dietary variation could be partly attributed to varying transcriptional regulation driven by lipid-modulated TFs. The distribution of TEs in elvol5 genes of Atlantic salmon shows a clear increase in TE mobilisation after the divergence of esocids and salmonids. This must have occurred after the elongase duplication and thus the salmonid WGD event and contributes to the observed regulatory divergence of elovl5 paralogs.

Neopterygian fishes represent over half of vertebrate richness in the Recent and display staggering phenotypic variety, yet little is known about the first 150 million years of their evolution. Furthermore, neopterygian richness and disparity is highly unevenly partitioned between teleost fishes, with ~29,000 species expressing a plethora of phenotypes, and holostean fishes, with 8 species and just two morphological styles. Fossil phenotypes have the unique ability to illuminate the assembly of neopterygian disparity, and can reveal the pattern by which the uneven partitioning of disparity arose. Morphology and function were quantified with landmarks and six functional traits, respectively, for 356 neopterygian species known globally throughout the first 150 million years of their history. The main axes of morphological and functional variation were derived and used to examine a series of evolutionary questions. Pertinently, they revealed how disparity was accumulated for 60% of the neopterygian radiation; morphological disparity increased through time, whereas functional disparity remained stable. The morphological dataset was expanded to include shape data for 398 species and size data for 471 species. Time scaled supertrees containing 671 mostly Mesozoic, but also living neopterygian species, were created. Together, the trees and traits were used to quantify evolutionary rates and innovation and test the predictions of genome duplication enhanced morphological diversification in teleosts, and the presence of 'living fossil' characteristics in holosteans. The analyses revealed higher rates and greater innovation in teleosts guaranteed to possess duplicated genomes, consistent with the predictions of genome duplication enhanced diversification. The only 'living fossil' characteristic of holosteans is their poor capacity for size innovation, yet they possess relatively high rates of shape evolution. However, estimates of rates and innovation are heavily influenced by timescale choice, emphasising the need for workers to perform their analyses on a variety of plausible timescales to determine the limits of their conclusions.

<p>The melanocortin (MC) receptors are G-protein coupled receptors thatparticipate in several important physiological functions such as the regulation of the energy balance. This thesis focuses on the evolutionary aspect of the MC receptors and their pharmacology.</p><p>One MC4 receptor and two MC5 receptor subtypes were found in a teleost fish, zebrafish. This indicates that the MC receptor subtypes arose very early in vertebrate evolution. Important pharmacological and functional properties, as well as gene structure and syntenic relationships have been highly conserved over a period of more than 400 million years implying that these receptors participate in vital physiological functions. Moreover, we found a MC4 receptor from a shark, spiny dogfish that represents the most distant MC receptor gene cloned to date. We also characterized the pharmacology of a MC4 receptor in goldfish. The conserved central expression pattern and physiological role in regulation of food intake of the MC4 receptor suggests that neuronal pathways of the melanocortin system may be important for regulation of energy homeostasis in most vertebrates. We determined the chromosomal position of the chicken MC receptors genes and found conserved synteny of the MC2, MC5, and MC4 receptor genes. These results suggest that there exist a clustering of these genes that is ancient. Analysis of conserved synteny with mammalian genomes and paralogon segments prompted us to predict an ancestral gene organization that may explain how this family has been formed through both local duplication and tetraploidization processes.</p><p>There are several common point mutations in the human MC1 receptor that are over represented in North European red-heads, and in individuals with pale skin. We pharmacologically characterised four naturally occurring human MC1 receptor variants providing molecular explanation to the respective phenotype.</p><p>The MC receptor subtypes have highly diverse physiological functions despite having relative high similarities in their primary structure. Our studies on the structural and functional properties of the MC receptor subtypes have provided insight into the molecular mechanism of how the specification of these receptors may have occurred.</p>