Evolutionary Systems Biology
Abstracts of the 15 Oct 2010, Edinburgh, UK |
Supported by Cologne University's Collaborative Research Centre
Please see the main workshop webpage for all other information. Program overview 12:55 Welcome by Laurence Loewe 13:00 Invited Talk by Prof. John Yin: 13:30 Talk by Pedro Beltrao: 13:50 Talk by Nicholas K. Priest: 14:10 Talk by Craig Maclean: 14:30 Talk by Laurence Loewe: Break + Poster viewing 15:30 Invited Talk by Prof. Juliette de Meaux: 16:00 Talk by Chris Knight: 16:20 Talk by David Alvarez-Ponze: 16:40 Talk by Aurelien Mazurie: 17:00 Talk by Marco J. Morelli: Discussion + Poster viewing Poster 1 by Tamas Korcsmaros: Poster 2 by Sam Farrell: Poster 3 by Giovanni Marco Dall'Olio: Poster 4 by Aurelien Mazurie: Poster 5 by Marco J. Morell:
_________________________________________________
Invited Talks
13:00 Invited Talk by Prof. John Yin: Abstract: The dynamics of a virus infection within its host is governed at its earliest stages by virus growth within infected cells. The kinetics of virus growth in cells reflects molecular processes associated with the regulated decoding of the viral genome: transcription, translation, and genome replication, followed by assembly and release of progeny virus particles. We are developing computational models and cell-culture measurements to better understand how these and other processes contribute to the synchronous one-step growth behavior of viruses in cells. As a model system we study vesicular stomatitis virus (VSV), a negative single-stranded RNA virus, growing on BHK cells. Here we highlight recent advances on two fronts: (i) computational models of gene-order variants of VSV, and (ii) wet-lab measures of virus production by single cells infected by single VSV particles. For viruses, changes in gene order may inhibit or enhance their growth, advancing their use as live attenuated vaccines, gene-therapy vectors or anti-tumor therapeutics. We have developed a kinetic model for the intracellular growth of VSV that accounts for all 120 possible permutations of these five genes in the VSV genome, 3´-N-P-M-G-L-5´. Gene-order is important of VSV because genes closer to the 3' end are expressed earlier and to higher levels than those at the 5' end. Computed VSV variants showed 6,000-fold variation in virus production in BHK cells. Wild-type VSV was predicted to grow with the third and the second highest progeny yields out of the 120 gene-order permuted strains in BHK and DBT cells, respectively, suggesting VSV is a generalist; it achieves high fitness in different host cellular environments. Interestingly, by ranking the effects of gene position on growth we found that wild-type VSV growth is most sensitive to gene-order permutations that increase production of the lowest expressed gene (L) or reduce production of the most highly expressed gene (N). These studies begin to reveal how virus growth depends not only on gene functions encoded in the virus genome, but also on their developmental context, defined by their dynamic patterns of expression. Established single-cycle wet-lab measures of virus growth within infected cells provide population averages, which mask potential cell-to-cell variation. We used fluorescence-activated cell sorting to isolate single cells infected by single particles of a recombinant VSV expressing green fluorescent protein. Measured virus yields spanned a broad range from 8000 to below the detection limit of 10 infectious virus particles per cell. Viral genetic variation and host-cell cycle differences were unable to fully account for the observed yield differences. Computer simulations of the VSV dynamics within an infected cell suggest a potentially key role for stochastic gene expression to the observed yield variation. These examples begin to show how interactions between computational models and wet-lab measures can contribute to a deeper understanding of viral fitness. We thank the National Institutes of Health (USA) for supporting our studies: AI-071197, AI-077296, RR-023167.
15:30 Invited Talk by Prof. Juliette de Meaux: Authors: Fei He (1), Xu Zhang (2), U. Goebel (1), F. Turck (1), J. Borevitz (2) and J. de Meaux (1) (1) Max Planck Institute for Plant Breeding Research, Abstract: The contribution of cis-regulation to adaptive evolutionary change is believed to be essential, yet little is known about the evolutionary rules that govern regulatory sequences. In the genus Arabidopsis, species display markedly different ecological preferences. What is the role played by cis-regulatory DNA during ecological speciation? We generated F1 interspecific hybrids, in which each cell contains a copy of the A. thaliana genome and a copy of the A. lyrata genome. In this context, the two genomes have the same transcriptional background and cis-regulatory variation can be inferred by monitoring species specific allele expression levels. We used Affymetrix whole-genome SNP arrays, featuring 250 000 SNPs distributed genome-wide, to monitor allele-specific expression in flowers at the genome-wide level. We predict that over 3000 genes display cis-regulatory variation at an FDR of 0.01%. Using pyrosequencing as an alternative technology, we confirm the accuracy of our predictions on independent biological samples. We observe a striking asymmetry in the distribution of cis-regulatory modifications between genomes. These differences can be associated to both variation in regulatory elements and extensive epigenetic differences. Our work emphasizes the diversity of molecular mechanism controlling inter-specific regulatory differences in plant species.
Contributed Talks
13:30 Talk by Pedro Beltrao: (1) Department of Cellular and Molecular Pharmacology, University of California San Francisco, San Francisco, California, USA Abstract: Protein phosphorylation is one of the most well characterized post-translational modifications. Yet, very little is known about the evolutionary dynamics of kinase-substrate interactions and its functional consequences. We have used a mass-spectrometry approach to characterize the in-vivo phosphoproteomes of three fungal species (S. cerevisiae, S. pombe, C. albicans) and have combined this data with cross-species genetic interaction data for comparative studies. We observed that kinase-substrate interactions change at a fast rate although the average level of phosphorylation of functional groups (i.e. complexes, signaling pathways) is well conserved. In order to address this apparent contradiction we have developed methods to predict the function of known phosphorylation sites using structural information as well as data on other post-translational modifications. This analysis indicates that the positional conservation of phosphorylation sites significantly under-predicts the conservation of function. The observation that phosphorylation sites can change during evolution without affecting the function suggest that neutral evolution plays a big role in shaping kinase-substrate networks.
13:50 Talk by Nicholas K. Priest: Abstract: The fate of a deleterious mutation depends, in part, on its mutational context. A mutation that would otherwise be deleterious can have a compensatory effect when it occurs in a genetic pathway that has been made nonfunctional by other deleterious mutations. Compensatory mutations have been largely ignored because they are thought to be rare. As a consequence, we do not understand their basic properties. Under what mutational contexts are compensatory mutations likely to occur? Relative to the site of the initial deleterious mutation, where in the gene pathway are compensatory mutations likely to occur? Does mutational load and/or mating system affect the fixation of compensatory mutation? Here, we address these questions using a common paradigm in evolutionary systems biology, the gene regulatory network. Similar to previous approaches, we examined the consequences of mutation on the stability of thousands of randomly sampled gene regulatory networks; but, we did not disallow the possibility that some mutations – deleterious on the own – may have compensatory effects. We found that the rate of deleterious mutation increases with the size of the regulatory network. While, in contrast, the rate of compensatory mutation is largely insensitive to the size of the regulatory network. By selecting random nonfunctional networks and sampling mutations at each node in the network, we found that compensatory mutations are more likely to occur in genes that directly interact with mutant gene. And, we show, through evolutionary simulations, that compensatory mutations are much more likely to be maintained when they occur in species with asexually reproduction and low levels of genetic load. These findings reveal compensatory mutations are likely to have important evolutionary consequences in specific mutational contexts.
14:10 Talk by Craig Maclean: Abstract: The distribution of fitness effects of spontaneous beneficial mutations is crucial to our understanding of adaptation by natural selection. Population genetic theory predicts that this distribution is exponential whenever fitness is high, suggesting that a mechanistic understanding of the fitness effects of beneficial mutations derived from systems biology may contribute little to our understanding of the properties of beneficial mutations. To test this idea, we used an experimental evolution approach involving adaptation to the antibiotic rifampicin in the opportunistic human pathogen Pseudomonas aeruginosa. As predicted by population genetic theory, the fitness effects of beneficial mutations are exponentially distributed when the fitness of the wild-type is high. However, when the fitness of the wild type is low, the fitness effects of beneficial mutations are no longer exponentially distributed because of a bias towards mutations of large effect. We show that this non-exponential distribution can be explained by a detailed structural understanding of the interactions that occur between rifampicin and RNA polymerase. This work shows how a detailed mechanistic understanding of genes under selection, in this case derived from structural biology, can be integrated with statistical population genetics to understand how beneficial mutations impact fitness. At a more applied level, our results suggest that systems biology approaches are critical for being able to predict the evolution of antibiotic resistance.
14:30 Talk by Laurence Loewe: Abstract: An important goal of evolutionary systems biology is to use current systems biology models to help quantify crucial evolutionary parameters. Here this approach is used to investigate the fraction of adaptive mutations that can occur in the circadian clock of the green alga Ostreococcus tauri. A stochastic model of the biochemical reactions of this circadian clock is simulated using kinetic parameters estimated from experimentally determined time courses of various molecules under different light-dark regimes. For this model, several different fitness correlates are defined and observed. These fitness correlates measure aspects of the quality of the circadian clock and are hence expected to correlate with fitness under the wide range of realistic circumstances where circadian clocks have an impact on fitness. Here random changes of up to 10% around the original value are explored for each of the 20 low-level kinetic properties of the model. The resulting high-level effects on fitness correlates are computed for different settings. The fractions of random changes that result in improved fitness correlates are reported for various definitions of fitness correlates. If random DNA changes cause equivalent random changes in reaction rates and the fitness correlates defined here indeed affect fitness, then this approach predicts the fraction of adaptive mutations for this system. Besides mutation accumulation experiments and inferences from DNA diversity data, this approach provides a third way for arriving at estimates of the frequency of adaptive mutations, a parameter of paramount importance for understanding both the evolution and robustness of biological systems.
16:00 Talk by Chris Knight: Abstract: The networks of genes required for dealing with different stresses and metabolic environments are not only complex in themselves, but are regulated in particular, frequently asymmetric, ways. For instance, pre-exposure to heat stress has been shown to prepare a yeast population for an oxidative stress, but not the other way around. This is presumably due to the co-ordination of cellular responses (e.g. in gene expression) in a way that is hypothesised to be advantageous in the environment in which the organism has evolved, for instance environments where oxidative stresses typically follow heat stresses. We have characterised the network of such preexposure interactions for a wild strain of the yeast Saccharomyces cerevisiae, focusing on the metabolism of different sugars known to be present in their natural environment, and environmentally relevant stresses. We identify a range of known and previously unknown relationships, many asymmetric. We go on to show how this network has evolved in a strain of the sister species, S. paradoxus, isolated from the same location. We find striking differences in the network architecture, indicative of major changes in the coordination of environmental responses between the species. We are seeking both to understand the significance these relationships and their evolution more clearly and to find correlates at the level of gene sequence evolution
16:20 Talk by David Alvarez-Ponze: Departament de Genètica, Facultat de Biologia, Universitat de Barcelona, Av. Diagonal 645, 08028 Barcelona, Spain Abstract: The complexity of the biological function relies on large networks of interacting molecules. The evolutionary properties of these networks, however, are far from fully understood. It has been shown that selective pressures depend on the position of genes in the network. We have previously shown that in the Drosophila insulin/TOR signal transduction pathway there is a correlation between pathway position and the strength of purifying selection, being downstream genes the most constrained (Alvarez-Ponce et al. 2009). Here we have studied the evolutionary dynamics of this well-characterized pathway in vertebrates. More specifically, we have studied the impact of natural selection on the evolution of 72 genes of this pathway. We have found that in vertebrates there is also a similar gradient on the selective constraint levels along the insulin/TOR pathway. This feature is neither the result of a polarity in the impact of positive selection nor of a series of factors affecting selective constraint (gene expression level and breadth, codon bias, protein length, and connectivity). We also found that pathway genes encoding physically interacting proteins tend to evolve under similar selective constraints. The results indicate that the architecture of the vertebrate insulin/TOR pathway constrains the molecular evolution of its components. Therefore, the polarity detected in Drosophila is neither specific nor incidental of this genus. Hence, although the underlying biological mechanisms remain unclear, they may be similar in both vertebrates and Drosophila. References: Alvarez-Ponce D, Aguadé M, Rozas J. 2009. Network-level molecular evolutionary analysis of the insulin/TOR signal transduction pathway across 12 Drosophila genomes. Genome Res. 19:234-242. 16:40 Talk by Aurelien Mazurie: Abstract: Macromolecular mechanisms of the cell are shaped by evolutionary processes, and so are biological networks. These networks, sum of all the interactions between cellular components, are used as a convenient representation of the cell's internal organization. From the structure of these networks one can derive general principles of this organization. What is missing, however, is the precise knowledge of how this organization is linked to the phylogeny of species, and how this organization is affected by environmental constraints. I will present here results of an ongoing project to characterize this relationship. First will be introduced an intuitive, robust and high-level description of the metabolic capabilities of species, termed network of interacting pathways or NIP. I will show that NIPs capture sufficient information about the underlying evolutionary events leading to the formation of metabolic networks to permit accurate prediction of the phylogenetic position of species. Finally, I will demonstrate that the metabolism of species subject to distinct evolutionary paths or environmental constraints show subtle but significant differences in their high-level organization. These changes are unevenly distributed among metabolic pathways, with specific categories of pathways being promoted to more central locations as an answer to environmental constraints. References:
17:00 Talk by Marco J. Morelli: Abstract: Single-stranded viruses are characterised by high mutation rates (around 10^-4 /nt/replication) and within host population sizes are often very large, suggesting viral populations should exhibit high levels of genetic heterogeneity. However, conventional sequencing techniques suggest that viral populations sampled from different individuals over the course of a single outbreak often differ by a few nucleotides only. This could result from the failure of conventional sequencing technologies to detect minority variants (i.e. genomes containing mutations whose frequency is below 50% in the population). We used Next-Generation Sequencing (NGS) to sequence within-host viral populations of Foot-and-Mouth Disease Virus (FMDV). We developed a novel pipeline to analyse NGS data and determine the population genetic structure, accounting for artefactual variations due to various sources of error. We applied our methodology to three samples obtained from a single host in the context of an infection experiment. Results indicated that most nucleotide sites throughout the genome are polymorphic, but appear only in a very small minority of the population (<0.5%). This variability observed at the larger viral population level, is ultimately generated by the process of viral replication, which takes place in cells. We designed a multi-scale model to investigate the evolutionary processes linking cellular viral replication with the generation of genetic diversity within the larger-scale viral population within the host. The model implements biochemical details of the replication process at the cell level and a range of selection scenarios that are expected to operate during the infection of a host. With a careful combination of experiments and modelling, we were able to infer possible evolutionary processes operating on viral populations that are difficult to observe or measure directly. Future directions include the study of host-to-host transmission bottlenecks and spreading of viruses in a group of hosts.
_____________________________________________________
Posters
Poster 1 by Tamas Korcsmaros: Abstract: The integration of biochemical and genetic information to signaling networks and their examination using advanced tools of network analysis is a recent, novel trend. Signaling pathways, e.g., Hedgehog or Notch, control a large variety of cellular processes, and their defects may lead to the development of diseases. Reliable analyses of these pathways need uniform pathway definitions and curation rules. However, current signaling databases rarely apply such rules and thus underrepresent pathway cross-talks and multi-pathway proteins, i.e., proteins having function in more than one pathway through cross-talks. Previously, we compiled a network of signaling pathways of three metazoans (Caenorhabditis elegans, Drosophila melanogaster and humans). Data were manually collected using bioinformatics tools and more than 1000 publications. The created network contains directed connections with exact references (PubMed IDs) of 8 major signaling pathways from C. elegans, D. melanogaster, and humans (http://SignaLink.org). We examined how the complexity of intracellular signaling increases with a growing complexity of the organisms. In worm only 6 of the 8 curated pathways are active, and only 5 of the 30 ( = 6×5 ) possible cross-talk types are present. In fly all 8 curated pathways are active, and the cross-talk network is significantly denser than in the worm. In humans all 8 curated signaling pathways are active and 53 of the 56 possible directed cross-talk types are active. The ubiquity of cross-talks expands both the repertoire of possible phenotypes and the system-level responses to environmental and pathological changes. The number of cross-talks relative to all signaling interactions is an additional indicator of signaling complexity. In the worm 4.6% of all signaling interactions are cross-talks, in the fly 10.5%, and in humans 30.3%. The presence of cross-talks in many pathways is a sign of the efficient utilization of resources: (1) expanding the functions of an already existing pathway protein is more efficient than evolving a novel protein or a whole pathway; (2) forming new input-output combinations gives novel regulatory and decision possibilities and can increase robustness and adaptation. Analysis of the evolutionary rate of signaling proteins revealed that human multi-pathway proteins evolve at a slower rate than the other signaling proteins (dN/dS 0.04 vs 0.09, on average). The evolutionary conservation of multi-pathway proteins can be similar to the constrained evolution of bottleneck and hub proteins detected earlier in yeast. Conversely, the mutation of a multi-pathway protein is common in many cancer types (www.sanger.ac.uk/genetics/CGP/Census). A possible explanation for this that cancer cells live in a stressful environment, where stress induced mutation can occur. It can increase the mutational rate of many proteins, including the crucial, multi-pathway proteins. Their altered function can have a much larger impact on the cellular processes than the alteration of a single pathway protein.
Poster 2 by Sam Farrell: Abstract: Experimental evolution can be a powerful tool for addressing questions on the genetic basis of evolutionary adaptation, allowing us to study evolution directly on manageable timescales in a controlled environment. The repeatable evolution of Pseudomonas fluorescens in a spatially structured environment provides just such a tool. P.fluorescens diversifies rapidly into niche specialists, one of which is a class of biofilm-forming strains named “wrinkly spreaders” (WS). Around half of WS strains studied previously contained loss-of-function mutations in the wspF gene, part of the Wsp signalling pathway. The Wsp pathway is analogous to the Che chemotaxis system in Escherichia coli. Such mutations cause downstream overproduction of a cellulose-like polymer, which in turn is essential in WS biofilm formation. Mutations in wspF do not simply destroy the protein's effectiveness; instead, different mutations in wspF lead to different WS fitnesses. Previous research was unable to establish clear causal links from different genetic changes in wspF, via the Wsp pathway, to different effects on fitness. We are addressing this question both using in silico models of the Wsp pathway and further characterising wspF mutations in vivo. We have evolved large numbers of novel WS strains, with a view to building a comprehensive collection of beneficial WS-causing wspF mutations. We have constructed ODE-based models of the Wsp pathway. Analysing a large number of samples of possible steady-state fluxes gives a broad view of the system's behavioural capabilities. These approaches, combined with assays to measure Wsp pathway output and strain fitness, will help bridge the gap between genotypic and phenotypic evolution in this model system.
Poster 3 by Giovanni Marco Dall'Olio: Abstract: In our group we study how the selective forces that act on a set of genes are related to the structure of the biological pathways they belong. Is it true that genes with a higher number of interactions are more conserved? How positive selection varies between the genes in the upstream positions of a metabolic pathway, and the ones downstream? The N-linked glycosylation pathway is one of the most interesting models for this type of studies. The earliest part of the pathway, usually referred to as N-linked precursor biosynthesis, is composed by a series of similar and linear reactions catalyzed by highly conserved genes, that could be considered as having a similar 'age', as being exposed to similar selective constraints, and have a single 'output' or product. The latter part of the pathway corresponds to one of the most complex networks of reactions in multicellular organisms, in which a few genes are organized in a complex network of reactions which can lead to more than 15 millions of different products. Here we will describe a study on the positive selection forces on this pathway, comparing several human populations from the Human Genetic Diversity Panel (HGDP) genotyped on the 650k Illumina chip.
Poster 4 by Aurelien Mazurie:
Poster 5 by Marco J. Morelli: |
