Bivalent genes are frequently associated with developmental and lineage specification processes. Resolving their bivalency enables fast changes in their expression, which potentially can trigger cell fate decisions. Here, we provide a theoretical model of bivalency that allows for predictions on the occurrence, stability and regulatory capacity of this prominent modification state. We suggest that bivalency enables balanced gene expression heterogeneity that constitutes a prerequisite of robust lineage priming in somatic stem cells. Moreover, we demonstrate that interactions between the histone and DNA methylation machineries together with the proliferation activity control the stability of the bivalent state and can turn it into an unmodified state. We suggest that deregulation of these interactions underlies cell transformation processes as associated with acute myeloid leukemia (AML) and provide a model of AML blast formation following deregulation of the Ten-eleven Translocation (TET) pathway

Organoids retain the morphological and molecular patterns of their tissue of origin, are self-organizing, relatively simple to handle and accessible to genetic engineering. Thus, they represent an optimal tool for studying the mechanisms of tissue maintenance and aging. Long-term expansion under standard growth conditions, however, is accompanied by changes in the growth pattern and kinetics. As a potential explanation of these alterations, epigenetic drifts in organoid culture have been suggested. Here, we studied histone tri-methylation at lysine 4 (H3K4me3) and 27 (H3K27me3) and transcriptome profiles of intestinal organoids derived from mismatch repair (MMR)-deficient and control mice and cultured for 3 and 20 weeks and compared them with data on their tissue of origin. We found that, besides the expected changes in short-term culture, the organoids showed profound changes in their epigenomes also during the long-term culture. The most prominent were epigenetic gene activation by H3K4me3 recruitment to previously unmodified genes and by H3K27me3 loss from originally bivalent genes. We showed that a long-term culture is linked to broad transcriptional changes that indicate an ongoing maturation and metabolic adaptation process. This process was disturbed in MMR-deficient mice, resulting in endoplasmic reticulum (ER) stress and Wnt activation. Our results can be explained in terms of a mathematical model assuming that epigenetic changes during a long-term culture involve DNA demethylation that ceases if the metabolic adaptation is disturbed.

Microfluidics is a versatile tool with many applications in biology. Its ability to manipulate small volumes of liquid precisely has led to the development of many microfluidic assay platforms. They could handle small amounts of samples and carry out analysis with high sensitivity and throughput. Microfluidic assays have provided new insights into scarce biological samples at higher resolution. In this thesis, we developed microfluidic tools to conduct low input ChIP-seq and ChIRP-seq. We applied them to a variety of samples profiling different targets. The native MOWChIP-seq platform was developed to map RNA polymerase II, transcription factors and histone deacetylase binding in 1,000-50,000 cells. We examined mouse prefrontal cortex and cerebellum using this technology. We found extensive differences that correlated with distinct neurological functions of the brain regions. The same platform and workflow were used to profile five key histone modifications in human lung tumor and normal tissue samples. Integrative analysis with gene expression data revealed extensive chromatin remodeling in lung tumor. Spatial histone modification mapping was conducted in mouse neocortex in a similar fashion. We generated an epigenomic tomography that demonstrated the molecular state of the brain in 3D. Lastly, we developed a microfluidic version of the ChIRP-seq process which successfully conducted the assay using only 500K cells. This improvement makes ChIRP-seq in tissue samples feasible. / Doctor of Philosophy / Microfluidics is a type of technology that can control small volumes of liquid in a miniature system. It can carry out reactions on very small scales with higher precision and sensitivity than conventional methods. Microfluidics has found many uses in the field of biology, especially dealing with samples available in limited quantities. These low input microfluidic platforms have helped researchers gain new knowledge on many complex questions. In this thesis, we developed microfluidic tools to carry out low input ChIP-seq and ChIRP-seq. These are two established techniques used to map where certain targets are located on the genome of an organism. These targets include specific chemical modifications to the wrapper protein of DNA (histone modification), proteins that take part in transcription and expression of genes (RNA polymerase II, transcription factors) and other molecules. Our nMOWChIP-seq system removed the need for fixation by chemicals. It was able to examine RNA polymerase II, transcription factors and other enzymes using 1,000-50,000 cells. Traditional ChIP-seq requires more than 10 million cells and time-consuming chemical treatment steps. Our technology greatly improved sensitivity and ease of use. We also used this platform to test five important histone modifications in human lung tumors and healthy tissues. We constructed a spatial map of histone modification in mouse brain by analyzing slices of the cortex. Finally, we developed a microfluidic version of ChIRP-seq process to map locations of long non-coding RNAs in cultured human cells. The cells needed for a successful test were reduced to 500K from 20 million of the original workflow.

DNA is wrapped around proteins called histones, whose modification regulates numerous cellular processes. Therefore it is not surprising that mutations in the genes that modify the histones are frequently associated with human cancer. For example, mutations in SETD2, encoding the sole enzyme that catalyses histone H3 lysine 36 trimethylation (H3K36me3), occur frequently in multiple cancer types. This identifies H3K36me3 loss as an important event in cancer development, and also as a potential therapeutic target. This thesis investigates the following questions: (1) how does the loss of H3K36me3 contribute to cancer development; and (2) what therapy can be used to kill cancers that have already lost H3K36me3. To answer the first question, this thesis shows that H3K36me3 facilitates the accurate repair of DNA double-stranded breaks (DSBs) by homologous recombination (HR). H3K36me3 promotes HR by recruiting CtIP to the site of DSBs to carry out resection, allowing the binding of HR proteins (such as RPA and RAD51) to the damage sites. Thus it is proposed that error-free HR repair within H3K36me3-decorated transcriptionally active genomic regions suppresses genetic mutations which could promote tumourigenesis. To answer the second question, this thesis reveals a clinically relevant synthetic lethal interaction between H3K36me3 loss and WEE1 inhibition. WEE1 inhibition selectively kills H3K36me3-deficient cells by inhibiting DNA replication, and subsequent fork stalling results in MUS81 endonuclease-dependent DNA damage and cell death. The mechanism is found to be synergistic depletion of RRM2 (ribonucleotide reductase small subunit), the enzyme that generates deoxyribonucleotides (dNTPs). This work reveals two pathways that regulate RRM2: one involves transcriptional activation of RRM2 by H3K36me3, and the other involves RRM2 degradation regulated by Cyclin-Dependent Kinase, CDK1 (which is controlled by WEE1, CHK1 and ATR). Based on this mechanism, the synthetic lethal interaction is expanded, from between two genes, to between two pathways. Supported by in vivo experiments, the study suggests that patients with cancers that have lost H3K36me3 could benefit from treatment with the inhibitors of WEE1, CHK1 or ATR.

In dieser Dissertation wurde eine isolierte Brachydaktylie vom Typ E (BDE) untersucht. Grundlage war eine Familie mit autosomal-dominanten Erbgang BDE. Der genetische Hintergrund ist eine balancierte Translokation t(8;12)(q13;p11.2). Der Bruchpunkt auf derivativem Chromosom der(8) liegt 86 kb strangaufwärts des chondrogenetisch essentiellen Kandidatengens PTHLH (Parathyroid hormone like hormone). PTHLH ist für die Differenzierungsrate von proliferativen Chondrozyten verantwortlich. Positiv oder negativ reguliertes Pthlh führen zu einer Dysbalance mit Brachydaktylie-ähnlichen Phänotypen in murinen Tiermodellen. Der Leserahmen des Kaliumkanals KCNB2 auf Chromosom 8 wurde durch die Translokation in Intron 2 getrennt. Chrondrogenetische KCNB2 Funktionen konnten durch in situ Hybridisierungen ausgeschlossen werden. Der Translokationsbruchpunkt auf der(8) liegt in einer in Mammalia hochkonservierten Region und beeinhaltet ein Bindungsmotiv für AP1 Transkriptionsfaktoren. Durch die Translokation befindet sich in unmittelbarer Nähe eine Kernkonsensussequenz für ETS Transkriptionsfaktoren. AP1 und ETS Transkriptionsfaktoren interagieren und wurden auf eine potentielle PTHLH Regulation untersucht. Epigenetische Histonmodifizierungen, charakteristisch für cis-regulatorische Elemente, sowie Reportergenassays mit AP1 und ETS1 Bindungsmotiven zeigten einen Bezug zur PTHLH Regulation. Bindungsassays mit AP1 und ETS1 Transkriptionsfaktoren an den Bruchpunktsequenzen, sowie funktionelle in vitro Experimente mit Chondrozyten verifizierten die Hypothese, dass der Translokationsbruchpunkt strangaufwärts von PTHLH regulatorische Eigenschaften besitzt. Die AP1 und ETS1 Transkriptionsfaktoren regulierten PTHLH positiv in ATDC5 und C28/I2 Chondrozyten. In chondrogeninduzierten Patientenfibroblasten war die PTHLH Expression inhibiert. Die molekulare Pathogenese der BDE wurde durch die bisher unbekannte chondrogene PTHLH Fehlregulation dargestellt. / We studied a 3-generation family with Brachydactyly Type E (BDE) and identified a t(8;12)(q13;p11.2) translocation. We identified PTHLH (Parathyroid hormone like hormone) on chromosome 12p11.2 and the ionchannel KCNB2 on chromosome 8q13 as candidate genes. KCNB2 was disrupted in intron 2, while the chromosome 12 breakpoint is localized 86 kb upstream of PTHLH; only the latter gene is involved in chondrogenesis. The 12p11.2 breakpoint is conserved and features an AP1 binding site 86 kb upstream of PTHLH. Due to the translocation, an ETS binding site from 8q13 resided near the AP1 site. Since both transcription factors interact, we tested if AP1 and ETS1 can activate PTHLH in ATDC5 and C28/I2 chondrocytes. We used the breakpoint sequences of the derivative chromosomes 8 and 12 and the nonaffected chromosome 8 and 12 allele sequences in reporter-gene assays. Reporter-gene constructs containing the der(8) breakpoint revealed activation in murine and human chondrocytes. The enrichment of histone modifications, implicating cis-regulatory effects were investigated in the breakpoint area. We found the enriched histone H3K4me1 modification at the chromosome 12 breakpoint position in murine and human chondrocytes, while affected fibroblasts showed higher H3K4me1 enrichment at the der(8) breakpoint compared to wt(12) allele. Furthermore, the breakpoint sequence bound to AP1 and C-ets-1 in EMSA. Western blotting after PMA-stimulated AP1 and ETS1 activation and overexpression of different AP1 and ETS1 combinations showed activated PTHrP expression in chondrocytes. In chondrogenic induced BDE fibroblasts PTHLH was inhibited, while IHH was upregulated. We suggest that PTHLH was dysregulated by the translocation in BDE chondrocytes. This could lead to BDE. We highlight the impact to characterize genomic breakpoints in detail and demonstrate a novel AP1- and ETS1-directed chondrogenic PTHLH regulation in wild-type chondrocytes and dysregulation in the pathogenesis of BDE.

La transition épithélio-Mésenchymateuse (TEM) est un processus de plasticité cellulaire qui existe dans le développement embryonnaire et qui permet la formation des tissus et organes. Dans la cancérogénèse, ce processus est réactivé par des facteurs de transcription dont l’action implique très probablement un remodelage de la chromatine. La cartographie exacte de ces changements épigénétiques est peu connue à l’échelle du génome entier, même si il y a eu quelques études antérieures explorant les changements de quelques loci de façon bien ciblée. Ce mémoire traite du remodelage épigénétique médié par le facteur de transcription Twist1 dans un modèle de lignée mammaire immortalisée. L’architecture de ce remodelage a été cartographiée grâce à l’utilisation des techniques de haut-Débit pour analyser la méthylation de l’ADN (DREAM) et les modifications des histones (ChIPseq). Nos résultats montrent un changement majeur du méthylome pendant la TEM avec une hyperméthylation focale et une hypométhylation globale des corps des gènes prédominant au niveau des « domaines partiellement méthylés »; ces domaines sont déjà connus dans le développement pour gagner de façon concomitante à leur hypométhylation des marques d’histone répressives. Nous avons aussi observé un remodelage des domaines de l’histone répressive H3K27me3 avec une réduction de leur taille, et surtout le quasi doublement du nombre de gènes bivalents qui accompagne la transition. Le couplage de la méthylation de l’ADN avec le profil des microRNA nous a permis d’identifier le miR-203 comme l’unique microRNA régulé par méthylation de l’ADN durant la TEM; nous avons aussi montré que l’extinction épigénétique du miR-203 est requise pour la TEM et l’acquistion des propriétés de cellules souches. Enfin, nous avons réalisé une caractérisation génétique et/ou épigénétique de deux cancers rares, les carcinomes fibrolamellaires du foie et les carcinomes du rein à translocation. Pour les carcinomes fibrolamellaires du foie, nous avons décrit la nature endocrine de cette tumeur et établi une signature épigénétique basée sur la méthylation de l’ADN pouvant servir à différencier les formes histologiques appelées « pures » des formes « mixtes ». Pour les cancers du rein à translocation, nous avons montré les bases génétiques et épigénétiques de la différence entre les formes pédiatriques et adultes, avec la découverte fréquente du gain du bras chromosomique 17q dans les formes adultes. Nous avons aussi identifié une mutation récurrente dans le gène qui remodèle la chromatine INO80D appartenant à la famille INO80. En conclusion, ce travail explore le rôle de l’étude de l’épigénome pour comprendre la reprogrammation pendant les processus physiologiques comme la TEM d’une part et le cancer d’autre part. / The epithelial-Mesenchymal transition (EMT) is a process of cellular plasticity that exists in embryonic development and which allows the formation of tissues and organs. In carcinogenesis, the process is reactivated by transcription factors whose action probably involves chromatin remodeling. The exact mapping of these epigenetic changes is poorly understood genome-Wide, although there have been some previous studies exploring changes in so few well-Targeted loci. This thesis deals with the epigenetic remodeling mediated by the transcription factor Twist1 in a model of human mammary immortalized cell line. The architecture of this remodeling has been mapped through the use of high-Throughput techniques to analyze DNA methylation (DREAM) and histone modifications (ChIPseq). Our results suggest a major change in the EMT methylome with focal hypermethylation and gene body hypomethylation predominantly within "partially methylated domains"; these areas are already known in development to gain repressive histone marks concomitantly with DNA hypomethylation. We also observed landscape remodeling of repressive histone mark H3K27me3 with a reduction in domains size, and especially the almost doubling of the number of bivalent genes. The coupling of DNA methylation with the profile of microRNA has allowed us to identify miR-203 as single microRNA regulated by DNA methylation during EMT; we have also shown that epigenetic suppression of miR-203 is both required for EMT and acquisition of stem cell properties. Finally, we performed a genetic and/or epigenetic characterization of two rare cancers, named fibrolamellar hepatocellular carcinomas and translocation renal cell carcinomas. In fibrolamellar hepatocellular carcinoma, we described the endocrine nature of this tumor and established a signature based on DNA methylation which can be used to distinguish histological forms called "pure" from "mixed" fibrolamellar hepatocellular carcinomas. Regarding translocation renal cell carcinomas, we established the genetic and epigenetic basis of differences between pediatric and adult forms, characterized by frequent gain of 17q gain chromosomal arm in adults. We also identified recurrent mutations in the chromatin remodeling gene INO80D which belongs to INO80 family. In conclusion, this work explores the impact of analyzing the epigenome to understand reprogramming during physiological processes such as EMT and cancer.

Les modifications post-traductionelles des histones contribuent à l’expression génique et à la stabilité du génome. La méthylation de la lysine 4 de l’histone H3 (H3K4me), une marque associée aux promoteurs de gènes transcrits, est déposé par les methyltransferases hautement conservées de la famille SET1, dans le contexte du complexe COMPASS. SET-2, l’homologue de SET1 chez Caenorhabditis elegans, est responsable de la déposition de H3K4me dans la lignée germinale, et son inactivation provoque une perte progressive de la fertilité. Le but de mon travail de thèse a été d’étudier comment SET-2 et la méthylation de H3K4 contribuent au maintien de la lignée germinale. J’ai montré que l’absence de SET-2 provoque une sensibilité accrue aux dommages à l’ADN. Cependant, les voies de signalisation et de réparation de ces dommages sont fonctionnelles dans le mutant set-2. Par séquençage de l’ADN, j’ai par ailleurs montré que la stérilité progressive observée en l’absence de set-2 n’est pas due à une capacité de réparation réduite. L’ensemble de mes résultats suggère que H3K4me pourrait agir en aval de la signalisation de dommages à l’ADN, en influençant l’organisation de la chromatine aux sites des cassures double brin. J’ai d’autre part mis en évidence une nouvelle fonction pour la méthylation de H3K4 dans l’organisation de la chromatine en montrant que set-2 interagit génétiquement avec le complexe Condensine II et la Topoisomérase II, facteurs clefs de l’organisation mitotique des chromosomes. Des expériences de microscopie par FLIM-FRET ont d’ailleurs validé une fonction de H3K4 méthylée dans l’organisation de la chromatine dans la lignée germinale. Enfin, j’ai montré par analyses transcriptomiques que la protéine CFP-1 du complexe COMPASS est impliquée dans la régulation du programme transcriptionnel de la lignée germinale et que cette fonction est indépendante de SET-2. L’ensemble de mes résultats montre comment la régulation chromatinienne impacte le maintien d’une lignée germinale fonctionnelle à plusieurs niveaux. / Post-translational modifications of histones contribute to gene expression and genome stability. Methylation of lysine 4 of histone H3 (H3K4me), a mark associated with actively transcribed genes, is deposited by the highly conserved SET1 family methyltransferases acting in COMPASS related complexes. SET-2, the SET1 homologue in Caenorhabditis elegans, is responsible for the deposition of H3K4me in the germ line, and its inactivation causes progressive loss of fertility. The purpose of my PhD work was to study how SET-2 and the methylation of H3K4 contribute to the maintenance of the germ line. I have shown that the absence of SET-2 causes increased sensitivity to DNA damage. However, the DNA damage-induced signaling and repair pathways are functional in the set-2 mutant. By DNA sequencing, I have also shown that the progressive sterility observed in the absence of set-2 is not due to a reduced repair capacity. Together, my results suggest that H3K4 methylation may act downstream of DNA damage signaling, potentially by influencing the organization of chromatin at the sites of double-strand breaks. I have also described a new function for H3K4 methylation in the organization of chromatin by showing that set-2 genetically interacts with the Condensitin II complex and Topoisomerase II, key factors in mitotic chromosome organization. Moreover, FLIM-FRET microscopy experiments have validated a role for H3K4 methylation in germline chromatin organization. Finally, using transcriptomic analyses, I have described a function for CFP-1, a component of the COMPASS complex, in the regulation of the germline transcriptional program independent of SET-2. Altogether, my results show how chromatin regulation affects the maintenance of a functional germline through multiple mechanisms.

No description available.

<p>Lung cancer is the third most prevalent cancer in the world; however it is the leading cause of cancer related deaths worldwide. Non-small cell lung cancer (NSCLC) accounts for ~85% of the lung cancer cases. The current strategies to treat NSCLC patients with frequent causal genetic mutations is through targeted therapeutics. Approximately 10-35% of NSCLC patient tumors have activated mutations in the Epidermal Growth Factor Receptor (EGFR) resulting in uncontrolled cellular proliferation. The standard-of care for such patients is EGFR-Tyrosine Kinase Inhibitors (EGFR-TKIs), a class of targeted therapeutics that specifically inhibit EGFR activity. One such EGFR-TKI used in this study is erlotinib. Following erlotinib treatment, tumors rapidly regress at first; however, over 50% of patients develop erlotinib resistance within a year post treatment. Development of resistance remains to be the major challenge in treatment of NSCLC using EGFR-TKIs such as erlotinib. </p> <p>In approximately 60% of cases, acquired erlotinib resistance in patients is attributed to a secondary mutation in EGFR, whereas in about 20% of cases, activation of alternative signaling pathways is the reported mechanism. For the remaining 15-20% of <a>cases</a> the mechanism of resistance remains unknown. Therefore, it can be speculated that the common methods used to identify genetic mutations in tumors post erlotinib treatment, such as histologic analysis and genetic screening may fail to identify alterations in epigenetic mediators of erlotinib resistance, also including microRNAs (miRNAs). MiRNAs are short non-coding RNAs that post-transcriptionally negatively regulate their target transcripts. Hence, in this study two comprehensive screens were simultaneously conducted in erlotinib sensitive cells: 1) a genome-wide knock-out screen, conducted with the hypothesis that loss of function of certain genes drive erlotinib resistance, 2) a miRNA overexpression screen, conducted with the hypothesis that certain miRNAs drive the development of erlotinib resistance when overexpressed. The overreaching goal of the study was to identify novel drivers of erlotinib resistance such as microRNAs or other epigenetic factors in NSCLC.</p><p>The findings of this study led to the identification of a tumor suppressive protein and an epigenetic regulator, SUV420H2 (KMT5C) that has never been reported to be involved in erlotinib resistance. On the other hand, the miRNA overexpression screen identified five miRNAs that contribute to erlotinib resistance that were extensively analyzed using multiple bioinformatic tools. It was predicted that the miRNAs mediate erlotinib resistance via multiple pathways, owing to the ability of each miRNA to target multiple transcripts via partial complementarity. Importantly, a correlation between the two screens was identified clearly supporting the use of two simultaneous screens as a reliable technique to determine highly significant miRNA-target interactions. Overall, the findings from this study suggest that epigenetic factors, such as histone modifiers and miRNAs function as critical mediators of erlotinib resistance, possibly belonging to the 15-20% of NSCLC cases with unidentified mechanisms involved in erlotinib resistance.</p><p></p>

L'épigénome est défini par l’ensemble de l’information chromatinienne autre que celle fournie par la séquence ADN. Au sein d'une même espèce et pour un type cellulaire donné, chaque individu présente des caractéristiques particulières de l'épigénome. Les épi-polymorphismes, définis comme étant les différences inter-individus de marques chromatiniennes, sont encore partiellement caractérisés et peuvent être liés aux phénotypes de chacun. La première partie de mon travail a été d'identifier et d'interpréter chez S.cerevisiae l'impact des épi-polymorphismes de modification des queues d'histones. Pour y parvenir, j'ai cartographié les épigénomes de cinq modifications différentes (3 acétylations et 2 méthylations) chez trois souches de levures issues de différents isolats naturels. Par une méthode de ChIP-seq et le développement d'un outil informatique, j'ai comparé les épigénomes de ces souches à l'échelle de nucléosomes individuels. L'étude des propriétés génomiques des épi-polymorphismes m'a alors permis de découvrir certaines caractéristiques encore inconnues et décrites dans ce manuscrit.Par ailleurs, j'ai voulu aborder le lien entre épi-polymorphismes et réponse transcriptionnelle à l'environnement. Pour cela, j'ai construit un jeu de souches mutantes dérivées de souches naturelles, où certains épi-polymorphismes ne peuvent plus être maintenus. J'ai analysé par RNA-seq les transcriptomes de certaines de ces souches avant et après un changement environnemental. Malheureusement, l'analyse des résultats a révélé que la qualité des données ne permettent pas d'établir le lien recherché mais les outils mis en place sont désormais disponibles.J'ai enfin étudié la dynamique d'évolution d'un épigénome en présence ou en l'absence de pression de sélection. Pour cela, j'ai suivi une modification d'histone (l'acétylation de la lysine 14 de l'histone H3) chez la levure pendant 1.000 générations dans deux conditions d'évolution expérimentale différentes : l'une sélective, l'autre neutre. J'ai mis en évidence des différences remarquables et inattendues entre ces deux régimes évolutifs. Des études mécanistiques détaillées restent à faire pour caractériser la nature et les propriétés de ces différences. / Epigenome is defined as the entire chromatin information other than the DNA sequence. Within a given species and for a given cell type, each indivual has specific epigenomic characteristics. Epigenomic differences between individuals (refered to as 'epi-polymorphisms') remain poorly characterized, although cases were reported where they could be linked to phenotypic differences. In my thesis, I used the model organism S. cerevisiae to identify histone modification epi-polymorphisms and study their biological impact. I profiled the epigenome of five different histone modifications (3 acetylations and 2 methylations) in three natural yeast strains. By ChIP-seq methods and software developments, I compared these strains at single-nucleosome resolution and discovered novel characteristics of these epi-polymorphisms which are described in this manuscript.Furthermore, I constructed a research framework to investigate the link between epi-polimorphisms and response to environmental cues. For this, I built a set of mutant strains derived from natural strains but where some epi-polymorphisms can no longer be maintained. I analyzed by RNA-seq the transcriptomes of some of these mutant strains before and after an environmental shift. Unfortunately, the quality of this initial data produced was not sufficient to link epi-polymorphisms to differntial responses, but the strain resources remain available for further investigations. Finally, I studied the evolutionary dynamics of epi-polymorphisms in the presence or absence of selection pressure. To do so, I followed the evolution of H3K14ac for 1.000 generations under two conditions of yeast experimental evolution ( selective or neutral). Marked differences were observed between the two regimes, revealing unexpected consequences of the presence of selection. Further mechanistic studies will be needed to elucidate the full properties of these differences.