Natural variation at pathogen resistance gene drives incompatibility within Arabidopsis thaliana accessions

Rubén Alcázar, Jane Parker and Matthieu Reymond

Accumulation of genetic incompatibilities within species can lead to reproductive isolation and potentially, speciation. In this study, we show that allelic variation at pathogen resistance gene conditions the occurrence of incompatibility between Arabidopsis thaliana accessions. Geographical partitioning of incompatible alleles correlates with the occurrence of incompatibilities. Variation in disease susceptibility also correlates with a signature for selection at this locus. We provide a first example of a local adaptation process evolving in incompatibility within plant species.

Fitness and structure landscapes for pre-miRNA processing

Ralf Bundschuh, Juliette de Meaux, and Michael Lässig

The processing from pre-miRNA to mature miRNA in plants involves a mechanism, which depends on an extended stem in the secondary structure of the pre-miRNA. Here, we show how natural selection acts on this secondary structure to produce evolutionary conservation of the processing mechanism together with modularity of the pre-miRNA molecules, making this molecular function independent of others. Our main results are:

1. Selection on miRNA processing can be described by a fitness landscape which depends on the free energy ΔG of the processing stem as quantitative molecular phenotype. Stem structures with ΔG>17kcal/mol have a constant fitness value; less stable stems incur a relative fitness cost. We infer this fitness landscape by a genome-wide analysis of Arabidopsis thaliana, comparing the stem free energy distribution of 130 functional pre-miRNAs with a background ensemble of random RNA molecules.

2. This fitness landscape (together with the overall sequence divergence) predicts the divergence of the free energy phenotype, as well as the selection coefficients of pre-miRNA sequence changes. These predictions are in excellent quantitative agreement with the results of a genome-wide cross-species comparison of orthologous pre-miRNAs between A. thaliana and A. lyrata. We conclude that there is a strong evolutionary constraint on processing phenotype and function, although individual sequence changes are only under weak selection. Our analysis shows how natural selection interacts with the complex genotype-phenotype mapping induced by RNA folding.

3. Actual pre-miRNA structures are modular: their stem free energy is significantly less affected by deleterious mutations in the remainder of the molecule than for random RNA molecules. This modularity is driven by mutational load favoring structures with more stable stems. This suggests a general evolutionary mechanism by which selection, mutations, and genetic drift generate modularity, a feature that is important for the independence of molecular functions in RNA, proteins, and extended molecular networks.

Deep sequencing, mutual information and the thermodynamics of gene regulation

Curtis Callan

The regulation of gene expression is thought to be largely governed by the binding thermodynamics of regulatory proteins. I will argue that high-throughput sequencing, combined with a novel analysis technique based on mutual information, can give a precise thermodynamic account of gene regulatory function. I will describe the application of this method to the bacterial lac promoter, showing how one can infer the in vivo interaction energy between two proteins in the cell from sequence data alone. This approach to experiment design and data analysis is inspired by modern practice in the physical sciences.

Dysregulation of gene expression and allelic imbalance in mammalian interspecific hybrids

Andrew G. Clark and Xu Wang

Interspecific hybrids provide an excellent opportunity to study the effects of a profound but well-defined and replicable genome-wide perturbation.  To the extent that gene expression regulation has evolved in an adaptive manner, hybrid dysregulation may reveal attributes of the adaptive process. The successful physiological function of interspecific hybrids is a testament to the robustness of many biological processes, and is surprising given the profound dysregulation at the transcriptional level (e.g. in Drosophila).  In this study we obtain a quantitative genome-wide picture of the degree of dysregulation in the horse, donkey and reciprocal hybrids, we sequenced the transcriptomes of invasive trophoblast cells (placenta) from each. This tissue was chosen because it represents a focal point of interaction between the mother and fetus in the developing placenta.  A total of 11.37 Gbps of trophoblast sequences were obtained from the four animals, and approximately 70% of the short-read sequences were uniquely mapped to the horse RefSeq database. The expression profile of both mule and hinny placenta strongly resemble the horse more than the donkey, and we found many instances of genes whose expression level in mule and hinny was outside the range of the parental horse and donkey (transgressive variation). With the novel SNPs we discovered between horse and donkey, we have been able for the first time to quantify the allelic imbalance in these F1 hybrid mammals. Our data provide a detailed quantitative assessment of the dysregulation of gene expression in interspecific hybrids, and identify many instances of cis-acting SNPs that drive the species-specific imbalance.  Whole biological pathways generally express alleles from both species, but significant skews in horse vs. donkey allelic expression are rampant.

Environmental gene regulation of an expansive aquatic arthropod genome

John Colbourne

Understanding the ecological context of genome structure, function and evolution begins by sequencing the genomes of species that are sentinels within ecosystems. The recent investigations by the Daphnia Genomics Consortium are uncovering how this micro-crustacean's gene inventory and regulation are products of the many challenges common in aquatic environments. A surprising result from the D. pulex genome annotation project is its impressive catalog of genes; more than 1/3 of genes have no detectable homologs in any other available proteome. The large gene count is due to numerous gene duplications that yield an abundance of tandem gene clusters that are specific to the Daphnia lineage.

Using tiling-path and multi-plex high-throughput microarrays that probe the entire sequenced genome, we conduct experiments that sample the transcriptome during development and reproduction, and during distinct ecological challenges. These experiments simultaneously annotate the boundaries of gene models, identify alternative transcripts, increase the gene count by over 40% and reveal the functional diversity of orphaned and expanded gene families. Our investigation of gene expression under varying environmental conditions shows that the lineage-specific genes are the most responsive to ecological challenges.

These studies contribute to and benefit from the Daphnia Genomics Consortium.

On the evolutionary theory of cost and benefit

Erez Dekel

Protein production is a costly process that reduces cell’s fitness. Proteins produced in an appropriate environment confer a benefit to the cell. Here we investigate the cost and benefit of protein production in bacteria. Cost is measured as the reduction in growth rate when bacteria produce un-needed proteins, and benefit is measured by the relative increase in growth. We show that bacteria can evolutionary tune their protein production level as a function of the environment. We use full genome sequence to find the mutations that confer optimal cost-benefit. We also find that cost is transient, and that bacteria can reduce cost of protein production to almost zero. We show that bacteria reduce the cost of protein production by using the ppGpp system of ribosomal gene regulation.

Emerging viruses: changes in viral genome result in a better manipulation of host regulatory networks

Santiago F. Elena

Understanding the underlying mechanisms by which viruses are able to jump species barriers and adapt to new hosts has been a challenging problem for virologists because the large number of cellular factors involved in the process and the complexity of interactions established during of viral infectious cycle. The classic approach has been the identification, after tremendous efforts, of one or a few host genes involved in the interaction. However, the generalization of the use of “omic” techniques for the analysis of host-virus interaction has open for the first time the possibility of producing a whole picture of the interaction. In this contribution, I will present results from an evolution experiment in which we artificially adapted Tobacco etch potyvirus (TEV) to a new host, Arabidopsis thaliana. We found that a few changes in the viral genome had a deep effect on the way the virus interacts with the regulatory transcriptional network of the new host. Next, we sought for specificities and commonalities in the way different viruses interact with a common host. We use A. thaliana as model host to analyze the sets of over/under-expressed genes from infection experiments with the potyviruses TEV, Turnip mosaic virus and Plum pox virus, and the phylogenetically unrelated Turnip crinkle virus. We firstly analyze data in terms of biological functions (GO/KEGG). Then, taking advantage of the published regulatory networks of A. thaliana, we dissect the viral mode of action showing a directed mechanism by altering the expression of key genes on the interactome. The set of genes specifically responding for phylogenetically related viruses represents interactions that have been acquired during the evolutionary diversification of a viral family. While those interactions shared by phylogenetically unrelated viruses should represent a non-specific response of the plant to virus infection.

Why there is more to gene evolution than protein function: splicing, nucleosomes and dual-coding sequence

Laurence Hurst

There is considerable variation in the rate at which different genes evolve.  Why is this?  Classically it has been considered that the density of functionally important sites must predict rates of protein evolution.  Likewise, amino acid choice is usually assumed to reflect optimal protein function and codon choice is random.  Here I briefly present evidence suggesting that this view is too simplistic.  In particular I concentrate on how selection acting during the protein’s production history can also affect gene evolutionary rates as well as amino acid and codon choice. Exploring the role of selection at the DNA and RNA level, I specifically address how the need 1) to specify exonic splice enhancer motifs in pre-mRNA and 2) to ensure nucleosome positioning on DNA, impacts amino acid choice, codon choice and rates of evolution at both synonymous and non-synonymous sites.

Molecular innovations at the origin of eukaryotes

Eugene V. Koonin

The origin of eukaryotes is linked to numerous and striking molecular innovations such as the endomembrane system including the nucleus with the elaborate nuclear pore complex, the distinct type of cytoskeleton, the RNA intereference system, ubiquitin signaling, and more. The emergence of these novel structures and functional systems involved elaboration of pre-existing analogous systems of archaea and bacteria, exaptation, and reassortment of diverse components, in some cases, of distinct prokaryotic origins. I will present an overview of molecular novelties associated with eukaryogenesis and argue that many of these innovations can be linked into a single causal chain of events triggered by mitochondrial endosymbiosis.

Viability and Development

Martin Kreitman

In Drosophila, as in many other organisms, mutations that disrupt developmental processes, such as embryogenesis, also reduce the ability an individual to survive to adulthood, i.e., viability. Null alleles for every gene in the segmentation pathway, which establishes segmental identity in the developing fly embryo, for example, are classic recessive lethal mutations, as might be expected. Mild mutations in these same genes, however, can have both subtle effects on both gene expression and viability. But with reduced viability, some individuals survive while others die, and survivors often appear to be phenotypically “normal”. Reduced viability, in this case, might be more appropriately described as “incompletely penetrant lethality”. In this talk I explore incompletely penetrant lethality in a developmental context, building on experimental results produced by M. Ludwig and Manu with site-directed mutations in the Drosophila even-skipped (eve) stripe 2 enhancer. The Eve protein is fused to a yellow fluorescent reporter protein (YFP), which allows us to visualize and measure the formation of Eve stripes in live embryos, including Eve stripe 2 misexpression. We also investigate downstream consequences of these mild eve regulatory mutations, both on the segmentation process and also on viability. From these experiments, I suggest that reduced viability, a.k.a. incompletely penetrant lethality, is not property of mean Eve stripe 2 performance, but rather is a property of its variance. Based on this idea, I discuss the role of stochasticity and limits of canalization in the propagation of mutant phenotypes in complex genetic traits.

Accessibility of mutational pathways in random and empirical fitness landscapes

J. Franke, A. Kloezer, J.A.G.M. de Visser and J. Krug

After decades of work devoted to the mathematical modeling of fitness landscapes, empirical data on epistatic interactions in multi-locus systems are now becoming available for single genes as well as on the organismic level. Such studies have shown that adaptive processes are often remarkably constrained, in the sense that only a small fraction of the possible evolutionary pathways connecting two genotypes are accessible to an adaptive dynamics which requires fitness to increase in each mutational step. Here we ask whether this behavior is specific to the considered examples, or whether it reflects a typical feature of multidimensional genotype spaces. To address this question, we investigate the statistical properties of selectively accessible paths for different models of random fitness landscapes. We identify several generic features of the distribution of the number of accessible paths between a genotype and its reversal sequence and study its variation with the number of loci. The theoretical predictions are compared to an ensemble of landscapes derived from an empirical 8-locus data set of fitness values for the asexual fungus Aspergillus niger. We conclude that most features of the data set are consistent with a model of random epistasis in which each locus interacts with approximately 4 other loci. We also elucidate the structure of lethal genotypes in this system.

Clonal interference in the evolution of influenza

Natalja Strelkowa and Michael Lässig

The seasonal influenza A virus undergoes rapid evolution to escape human immune response. Adaptive changes occur primarily in the viral epitope, the antibody- binding domain of the surface protein haemagglutinin. Here, we show that the genome of influenza A (H3N2) evolves by strong clonal interference. This mode of evolution explains its recurrent selective sweeps with clusters of simultaneous fixations occurring about every 4 years. We find an average of at least 1.3 strongly beneficial aminoacid substitutions (driving mutations) per year, and a given selective sweep has about 4 driving mutations on average. Influenza’s tempo of evolution is shown to be governed not only by positive selection within the epitope, but also by background selection outside the epitope: antigenic adaptation and conservation of other viral functions interfere with each other. Our analysis is based on frequency time-series of polymorphisms, which are obtained from a coalescence tree of influenza sequences over 39 years. From these data, we estimate strength and time-dependence of natural selection acting on influenza by means of a multi-strain sequence evolution model. Our results imply that a quantitative understanding of influenza's antigenic and epidemiological dynamics will have to be based on a broader spectrum of genetically linked viral functions.

Cooperation in bacteria: a synthetic perspective

Stanislas Leibler

Synthetic microbial systems present a unique opportunity for a quantitative study of selection in dynamic populations. I will present a short review of some classical arguments in theory of natural selection, in particular of those connected with origins of cooperation. I will show how simple experiments with bacteria could help to make these arguments precise and to "demystify" somewhat the subject.

This work has been done in collaboration with John Chuang, Olivier Rivoire and Edo Kussell.

Statistical models for predicting HIV phenotypes and effectiveness of antiretroviral therapies

Thomas Lengauer

The prime problem in applying antiviral drug therapy to AIDS patients is the variability of the AIDS virus HIV. Presented with a given drug therapy the virus follows evolutionary escape paths to drug resistance, which renders the drug therapy ineffective. Statistical computer models can help estimate the level of resistance of a given viral variant against a given antiviral drug. Furthermore, such models can estimate the evolutionary path to resistance followed by the virus.

We describe the science behind the geno2pheno Server which offers such models freely via the internet: Given the relevant portion of an HIV genome, geno2pheno predicts the resistance of HIV to any of a number of antiviral drugs that are in clinical use. Furthermore the server ranks combination drug therapies with respect to their expected effectiveness against the given HIV variant

Entry inhibitors aim at preventing HIV from entering the host immune cell. Here, we are confronted with a different computational challenge: When entering the human host cell, HIV uses one of two coreceptor molecules on the cell surface. We present a statistical model offered via the geno2pheno server that predicts which of the two coreceptors the virus uses. Determining this “viral tropism” is essential for disease prognosis and in the context of applying the new entry inhibitors.

Both model classes are trained using various linear and nonlinear statistical learning procedures. The training data are carefully assembled databases comprising relevant genotypic, phenotypic and clinical parameters.

The geno2pheno server has been developed in the context of the Arevir consortium, a German National research consortium targeted at the bioinformatical analysis of HIV resistance data, and is currently in prototypical use for research purposes. Members of the consortium and their associated practices treat about two thirds of the AIDS patients in Germany. In recent years, data collection and method development has been lifted to a European level in the EuResist project.

Systematic characterization of transcription factor function in vivo and in vitro

Sebastian Maerkl

We have developed several high-throughput microfluidic platforms for the quantitative analysis of transcription factors both in vitro and in vivo. Using our recently developed MITOMI platform we are able to obtain absolute binding affinities of transcription factors to hundreds of target DNA sequences. More recently we have been developing a kinetic MITOMI version capable of taking 768 on- and off-rate measurements in parallel. For in vivo analysis of transcription factor function we have employed synthetic biology to synthesize hundreds of defined promoter constructs and developed a microfluidic platform capable of measuring protein expression dynamics of over 1000 yeast strains in parallel with high spatio-temporal resolution. Our current experiments are geared towards bridging the gap in our understanding between in vitro measurements and how those quantitatively translate into in vivo promoter outputs in the yeast S. cerevisiae.

Diminishing returns epistasis between genes during adaptation

Christopher Marx

Genetic interactions have proven extremely useful in inferring gene function or physiological connections between cellular functions but little is known regarding epistasis between beneficial mutations. We generated all possible combinations of alleles that arose during the rapid adaptation of a strain of Methylobacterium whose growth on methanol depended upon a newly-introduced pathway. The proportional selective benefit for three of the four loci consistently decreased when introduced upon more fit backgrounds. These same three alleles were all shown to reduce morphological defects that were caused by expression of a foreign pathway. A simple model segregating the apparent contribution of individual alleles to costs and benefits effectively predicted the interactions between them. These results suggest that diminishing returns epistasis may be a major component underlying the consistent observation that the rate of fitness increase slows down dramatically following introduction into new environments.

Maintenance of allelic polymorphism in miRNA824 processing

Jin-Yong Hu, Li Lei and Juliette de Meaux

In plants and animals, gene expression can be down-regulated at the post transcriptional level by microRNAs (miRNAs), a class of small endogenous RNA. Comparative analysis of miRNA content across species indicates continuous birth and death of these loci in the course of evolution. However, little is known about the microevolutionary dynamics of these genetic elements, especially in plants. We examined polymorphism at the miRNA encoding locus in Arabidopsis thaliana, miR824, which is found only in the Brassicacae. We compare their diversity to other miRNA-encoding loci conserved across distant taxa. We find that this locus presents a pattern of diversity deviating from neutrality: miR824 displays signs of balancing selection. A detailed examination of structural variation among alleles suggests non random evolution of a thermo-resistant substructure in the precursor. Expression analysis of pre-miR824 and its target, AGL16, indicates that these structural differences impact the processing of mature miR824. In order to elucidate the cause for maintenance of miRNA824 alleles, we analyzed the phenotypic consequences of this polymorphism, both in the greenhouse and in the field. We find that this polymorphism impacts multiple phenotypes presumably subject to fluctuating selection in the field.

Evolutionary innovation and selection in cancer development

Christopher McFarland and Leonid Mirny

TThe development of cancer can be considered as an evolutionary process within an organism: cells acquire mutations, compete for resources, and are subject of selection. Somatic evolution of cancer proceeds at an elevated mutation rate and leads to accumulation of two classes of mutations: a few driver mutations, which constitute innovations essential for cancer development; and many more random passenger mutations. Our hypothesis is that passenger mutations can be deleterious to cancer, but nevertheless hitchhike to fixation making a population of cancer cells more vulnerable for certain types of treatments.

We (i) develop a computational model of cancer somatic evolution that considers both driver and passenger mutations and is amenable to analytical treatment; (ii) and characterize mutations found in sequenced cancers. Simulations demonstrate that deleterious passenger mutations can evade purifying selection and accumulate in cancer by hitchhiking. Analysis of cancer genomes supports this finding by demonstrating that sequenced passenger mutations are damaging to harboring proteins. Theory and simulations show that competition between passenger and driver mutation lead to the emergence of the two regimes of cancer progression: one characterized by a long period of dormancy and possible regression, and the other showing exponential growth of a tumor. Simulated treatment, which increases deleterious effect of passengers, leads to switching form the growth to shrinkage and rapid cancer meltdown. Possible molecular mechanisms that could elevate deleterious effect of passenger mutations via inhibition of assisted protein folding are discussed.

Analysis of variation in molecular phenotypes

Ville Mustonen

The main objective of our group's work is to understand functional and evolutionary properties of naturally occurring variationIn this talk I discuss two model systems which enable us to address this objective:

1) Several high-throughput technologies are available to measure interaction energies between transcription factor (TF) molecules and their target sequences, e.g. protein binding microarrays (Badis et al., 2009) and mechanically induced trapping of molecular interactions (Maerkl and Quake, 2007). We have extracted the sequence specific part of the interaction for two TFs from such data and used them to analyse inter- and intra-specific variation in yeast S.cerevisiae and S.paradoxus.

2) Protein kinases are a large and important protein family controlling many signaling pathways and other cellular processes e.g. transcription and apoptosis. A vast body of functional information on, for example, domain structure of kinases exists and mutations in kinases are known to cause disease. We have analysed somatic and germline variation in human cancer samples, (data set from (Greenman et al. 2007)), based on evolutionarily grounded scoring of mutations combined with the existing functional annotation of these proteins.

Causes and evolution of gene dispensability

Csaba Pal

Perhaps one of the most striking discoveries of modern molecular genetics was the extent by which organisms appear to tolerate mutations or even complete loss of their genes. Systematic single gene deletion studies on microbes have revealed that 70-80% of the single mutant strains are viable with no apparent phenotypic deformation. Our research concentrates largely on yeast (S. cerevisiae) and E. coli, and we seek to understand the physiological and evolutionary mechanisms behind this pattern. The following questions sum up our research: (i) Are these seemingly dispensable genes redundant or do they have important contribution under special environmental conditions not yet tested in the laboratory? (ii) How far the deleterious impacts of gene deletions can be mitigated during evolution, and what factors limit the extent of compensatory evolution? (iii) Is it likely that some of these genes increase the rate of evolutionary adaptation? To address these issues, we combine evolutionary genomics with systems biology and laboratory experimental evolution protocols.

Evidence that mutation is universally biased towards AT in bacteria

Dmitri Petrov

Mutation is the engine that drives evolution and adaptation forward in that it generates the variation on which natural selection acts. Mutation is a random process that nevertheless occurs according to certain biases. Elucidating mutational biases and the way they vary across species and within genomes is crucial to understanding evolution and adaptation. Here we demonstrate that clonal pathogens that evolve under severely relaxed selection are uniquely suitable for studying mutational biases in bacteria. We estimate mutational patterns using sequence datasets from five such clonal pathogens belonging to four diverse bacterial clades that span most of the range of genomic nucleotide content. We demonstrate that across different types of sites and in all four clades mutation is consistently biased towards AT. This is true even in lineages that have high genomic GC content. In all studied cases the mutational bias towards AT is primarily due to the high rate of C/G to T/A transitions. These results suggest that bacterial mutational biases are far less variable than previously thought. They further demonstrate that variation in nucleotide content cannot stem entirely from variation in mutational biases and that natural selection and/or a natural selection-like process such as biased gene conversion strongly affect nucleotide content.

Recombination, interaction and the rate of adaptation

Boris Shraiman

The talk will examine the effect of recombination in three closely related contexts: 1)Acceleration of the rate of adaptation in large populations, 2) Allele versus genotype selection in the presence of genetic interactions, 3) The hitch-hiking effect and "quasi-neutrality" in genetically diverse adapting populations.

Learning about interactions and selection from sequence data

Shamil Sunyaev

Data from the 1000 Genomes project enables a new method to characterize deleterious genetic variation in humans . The method uses allelic age conditioned on frequency. The effect described by Maruyama in 1974 predicts that, at a given frequency, neutral variants are older than either deleterious or advantageous variants. Because low frequency non-synonymous variation is expected to be enriched in deleterious alleles, the effect predicts that non-synonymous variants are, on average, younger than synonymous or non-coding variants of the same frequency. Sequencing data enable new approaches for estimating relative allelic age: mutation and recombination "clocks". The mutation clock estimates, for each variant, the number of likely younger (less frequent and completely linked) variants. The recombination clock estimates the length of the haplotype block on which the variant exists. The two clocks can be combined to improve power. Analysis of data generated by the 1000 Genomes project agrees with theory and computer simulations. The mutation and recombination clocks show that non-synonymous alleles are younger than synonymous alleles. Consistently with simulations, the mutation clock is more sensitive, and the effect is stronger for damaging variants than for benign non-synonymous variants (we used PolyPhen-2 for predictions of the functional effect of non-synonymous variants). The effect disappears at higher frequencies, because few deleterious variants reach higher frequencies. Consistently with the Maruyama effect, population-specific variants are more damaging than variants shared between populations (thus older), when conditioned on frequency. As expected, however, benign variants are not significantly more population specific than synonymous variants. We also tested presence of epistatic selection using non-monotonic LD decay. Early results suggest that epistasis is an important feature of natural selection in humans

Phylostratigraphic approaches to trace evolutionary innovations

Tomislav Domazet-Lošo and Diethard Tautz

Phylostratigraphy is a method to correlate the evolutionary origin of founder genes of gene families with particular macroevolutionary transitions (1, 2). It is based on a model of genome evolution that suggests that the origin of complex phenotypic innovations will be accompanied by the emergence of such founder genes, the descendants of which can still be traced in extant organisms. For example, the origin of multicellularity can be considered to be a macroevolutionary transition, for which new gene functions would have been required. Cancer should be tightly connected to multicellular life since it can be viewed as a malfunction of interaction between cells in a multicellular organism. A phylostratigraphic tracking of the origin of cancer genes should therefore also provide insights into the origin of multicellularity. There are indeed two strong peaks of emergence of cancer related genes, one at the time of the origin of the first cell and the other around the time of the evolution of the multicellular metazoan organisms. These peaks correlate with two major classes of cancer genes, the "caretakers", which are involved in general functions that support genome stability and the "gatekeepers", which are involved in cellular signaling and growth processes. Interestingly, this phylogenetic succession mirrors the ontogenetic succession of tumor progression, where mutations in caretakers are thought to precede mutations in gatekeepers (3). I will also show new data where we use phylostratigraphy to calculate a transcriptomic age index. Plotting the age index along ontogenetic development shows that the phylotypic stage of different organisms is characterized by particularly old genes, while new genes have arisen mainly in the context of juvenile and adult development. I will discuss these findings with respect to the conrasting theories of "opportunities of selection" versus "developmental constraint".

1. Domazet-Lošo T, Brajković J, Tautz D: A phylostratigraphy approach to uncover the genomic history of major adaptations in metazoan lineages. Trends Genet 2007, 23:533–539.
2. Domazet-Lošo T, Tautz D: An ancient evolutionary origin of genes associated with human genetic diseases. Mol Biol Evol 2008, 25:2699-2707.
3. Domazet-Lošo T, Tautz D: Phylostratigraphic tracking of cancer genes suggests a link to the emergence of multicellularity in metazoa. BMC Biology (in press)