There were many interesting talks during these two last days at the First q-bio Conference on cellular information processing. Too many in fact and I can only try to give some excerpts, reflecting more the limitations of my memory than anything else…
John Reinitz presented his work on the quantitative analysis of gene expression in Drosophila early embryo. He first mentioned the FlyEx database, which collects quantitative data on the segmentation gene expression patterns (Poustelnikova, et al 2004) and allows to investigate how errors in segmentation gene expression and cell specification can be corrected. One interesting observation is that the inter-individual variations in the distribution of the bicoid morphogen gradient are significantly larger than the variations observed in the boundary of the downstream gap gene hunchback (Houchmandzadeh et al, 2002). With a model of gap gene expression (Jaeger et al, 2004a, 2004b), it is possible to show that this property of filtering noise from the morphogen’s positional information emerges from the dynamical properties of the system itself. John Reinitz proposed that this phenomenon represents an example of “”http://en.wikipedia.org/wiki/Canalisation_%28genetics%29">canalization" of gene expression, a term first coined by Waddington (1942, Nature 150:563) to metaphorically describe how cell specification corresponds to navigating down the valleys of an “epigenetic landscape” that guides the cell to its final (stable) fate, in spite of initial variations.
Alexander Hoffmann studies the properties of the NF-kB signaling pathway and discussed how its dynamical properties may underly the stimulus-specificity of the response to major inflammatory mediators like TNFa, LPS or IL-1. These stimuli induce different temporal profiles of activation of the component of the pathway (Werner et al, 2005). He showed how the systematical analysis—both by varying the shapes of IKK activation profiles and by parameter sensitivity analysis—of a detailed model could provide insight into the determinants of stimulus-specificity of the response. Experimentally, the use of a specific mutant of NF-kappaB with altered temporal control leads to a loss of stimulus-specificity for 50% of those NF-kappaB target genes that normally exhibit specificity. Alexander Hoffmann also discussed further how the understanding of the dynamical mechanisms underlying stimulus-specificity may enable novel strategies to develop multi-target drugs interfering with only one stimulus-specific branch of the pathway.
In a fast paced talk, Bernhard Palsson presented the last progresses in elucidating E. coli transcriptional regulatory network. Using high-density genome tiling arrays and chromatin immunoprecipitation, his team produced a very impressive series of high-resolution genome-wide location maps of binding sites for the RNA polymerase (RNAP), many transcription factors and several DNA-binding proteins. These experiments were performed under various environmental conditions and in parallel with the respective mRNA expression profiles (also via tiling arrays). In addition, RNAP location was determined in presence and absence of Rifampicin to contrast a static map of bound RNAP with a map of the elongating polymerase. Integration of these various dataset will allow to extract causal relationships between the various binding events and gene expression, providing the blueprint of the global bacterial regulatory transcriptional network. Further integration (Barrett et al, 2005) with the reconstructed genome-wide metabolic network (Feist et al, 2007) will connect transcriptional and metabolic functional states, enabling a top-down approach to identify the regulatory needs of the bacterial cell.
Johan Paulsson uses regulation of plasmid copy number as a simple model system to investigate some fundamental aspects of negative feedback regulation. Plasmids have to keep their copy number low to avoid excessive burden to the cell but have also to suppress the resulting fluctuations, which may cause plasmid loss at cell division. Johan Paulsson showed in a very general way that the time delay inevitably associated with negative feedback imposes a lower limit to fluctuations (essentially the delay provides a window during which noise accumulation cannot be prevented). The first surprise is that these general theoretical limits are relatively high: with 4 copies, plasmid fluctuations are at least 5-15%. The second surprise is that when plasmid copy number is measured experimentally at the single cell level (a difficult task, requiring new techniques developed in Paulsson’s lab), it appears that plasmids suppress fluctuation remarkably close to the theoretical limit.
Again, there were many other great talks and there sure will be tomorrow—unfortunately I will not be able to attend tomorrow’s sessions :-(. I hope that the few examples above illustrate the exceptional quality of this conference and the kind of insight gained from a quantitative study of “small-scale” systems. Only a handful of presentations (BO Palsson, R Linding, B Blagoev) dealt with genome-wide large scale models and datasets—the other end of the spectrum in systems biology. While the organizers told me that this was in part a deliberate choice, it also reminded me that it will probably represent one of the key challenges in systems biology to find ways to bridge the gap between the dynamical properties of small systems and the architecture of large biological networks.