Integrated analysis of regulatoryand metabolic networks revealsnovel regulatory mechanisms inSaccharomyces cerevisiae

Integrated analysis of regulatory and metabolic networks reveals novel regulatory mechanisms in Saccharomyces cerevisiae Speaker: Zhu YaNG 6th step, 2006

Integrated analysis of regulatory and metabolic networks reveals novel regulatory mechanisms in Saccharomyces cerevisiae Speaker: Zhu YANG 6 th step, 2006

Reference · Herrgard,M小.Lee,B.s., Portnoy,V,and Palsson, B.O. 2006. Integrated analysis of regulatory and metabolic networks reveals novel regulatory meChanisms in Saccharomyces cerevisiae Genome Research,16:627-635

Reference • Herrgard, M.J., Lee, B.-S., Portnoy, V., and Palsson, B.O. 2006. Integrated analysis of regulatory and metabolic networks reveals novel regulatory mechanisms in Saccharomyces cerevisiae Genome Research, 16: 627 – 635. 16: 627

Outline Background Approach model-based analysis Data and information Reconstructed transcriptional regulatory network Prediction of gene expression changes Systematic expansion of the regulatory network Prediction of growth phenotypes Discussion ·Conc| usIons

Outline • Background • Approach model-based analysis • Data and information • Reconstructed transcriptional regulatory network • Prediction of gene expression changes • Systematic expansion of the regulatory network • Prediction of growth phenotypes • Discussion • Conclusions

Background Vith the rapidly increasing biological productions, the data integration and interpretation task is made challenging by the incompleteness and noisiness of large-scale data sets literature-derived information has enabled the reconstruction of chemically and biologically consistent mathematical descriptions of biochemical networks in well-studied model organisms. and Furthermore model predictions can be directly compared with experimental data obtained Using a reconstructed genome-scale stoichiometric matrix as a starting point, the constraint-based modeling framework can then be used to make phenotypic predictions that can be compared to experimental data requently used constraint-based approaches include flux balance analysis(FBA) and regulated flux-balance analysis (rF BA) approach

Background • With the rapidly increasing biological productions, the data integration and interpretation task is made challenging by the incompleteness and noisiness of large-scale data sets. • literature-derived information has enabled the reconstruction of chemically and biologically consistent mathematical descriptions of biochemical networks in well-studied model organisms. And Furthermore, model predictions can be directly compared with experimental data obtained. • Using a reconstructed genome-scale stoichiometric matrix as a starting point, the constraint-based modeling framework can then be used to make phenotypic predictions that can be compared to experimental data. Frequently used constraint-based approaches include flux balance analysis (FBA) and regulated flux-balance analysis (rFBA) approach

Approach model-based analysis In vivol Regulatory and rimental metabolic network system model Compare system states Growth rates Expression profiles cat8△ Model Rgt1 ify condition-TF KO pairs with mispredictions Identify missing regulatory mechanisms using interaction data Refine model ChIP-chip Sequence motifs iterative 2 c'C

Approach model-based analysis

Data and information The regulatory network model MMH805/775, is combined with an existing genome-scale metabolic model, iND750 The relevant literature for each metabolic and transcription factor gene was collected through information in the SGD, YPD, and MiPS databases and direct pubmed searches

Data and information • The regulatory network model, iMH805/775, is combined with an existing genome-scale metabolic model, iND750. • The relevant literature for each metabolic and transcription factor gene was collected through information in the SGD, YPD, and MIPS databases and direct PubMed searches

Reconstructed transcriptional regulatory network Starting point: ND750 The regulatory network model part of MH805/775 consists of three layers which were implemented as Boolean rules derived from primary literature of MH805/775 The first layer: activities of 55 TFs in response to 67 extracellular and 15 intracellular metabolite concentrations The second layer: the rules describing the expression of 348 metabolic genes as a function of the transcription factor states and metabolite concentrations in cases in which the direct regulatory mechanisms were unknown. For the remaining metabolic genes, no information on regulation could be found in the literature, and they were assumed to be constitutively expressed in all environmental cond itions The third layer: the gene-protein -reaction associations that encode the relationship between gene expression and presence/absence of a particular reaction in the network

Reconstructed transcriptional regulatory network • Starting point: iND750 • The regulatory network model part of iMH805/775 consists of three layers which were implemented as Boolean rules derived from primary literature of iMH805/775. – The first layer: activities of 55 TFs in response to 67 extracellular and 15 intracellular metabolite concentrations. – The second layer: the rules describing the expression of 348 metabolic genes as a function of the transcription factor states and metabolite concentrations in cases in which the direct regulatory mechanisms were unknown. For the remaining metabolic genes, no information on regulation could be found in the literature, and they were assumed to be constitutively expressed in all environmental conditions. – The third layer: the gene–protein–reaction associations that encode the relationship between gene expression and presence/absence of a particular reaction in the network

Reconstructed transcriptional regulatory network(Contd) iMH805/775 accounts for 805 genes and 775 regulatory interactions, and the network consists of the 750 metabolic genes in iND750 and 55 specific nutrient regulated transcription factors(tFs) The model allows 82 distinct intra-and extracellular metabolites to act as input signals to the regulatory network MMH805/775 also includes rules describing the mode of combinatorial control by different TFs at each promoter This logic-based representation allows in silico prediction of gene expression changes in response to environmental and genetic perturbations and integration of the regulatory network to the metabolic network model as described previous小

Reconstructed transcriptional regulatory network (Cont’d) • iMH805/775 accounts for 805 genes and 775 regulatory interactions, and the network consists of the 750 metabolic genes in iND750 and 55 specific nutrient￾regulated transcription factors (TFs). • The model allows 82 distinct intra- and extracellular metabolites to act as input signals to the regulatory network. • iMH805/775 also includes rules describing the mode of combinatorial control by different TFs at each promoter. • This logic-based representation allows in silico prediction of gene expression changes in response to environmental and genetic perturbations and integration of the regulatory network to the metabolic network model as described previously

Prediction of gene expression changes In silico gene expression change predictions were compared to experimentally measured expression profiles as well as experimentally determined protein DNA interactions(ChIP-chip)and predicted TF-binding motifs to assess the completeness of the MH805/775 network Gene expression data for eight transcription factor knockout strains(rgt1, rox1, gat1, hap1, adr1, gal4, gIn3 cat8 and two overexpression(HAP4, GCN4) strains from previously published reports were used Each of genes was classified as significantly up- regulated, significantly down-regulated, or unchanged in each of the 10 experimental data sets

Prediction of gene expression changes • In silico gene expression change predictions were compared to experimentally measured expression profiles as well as experimentally determined protein– DNA interactions (ChIP-chip) and predicted TF-binding motifs to assess the completeness of the iMH805/775 network. • Gene expression data for eight transcription factor knockout strains (rgt1, rox1, gat1, hap1, adr1, gal4, gln3, cat8) and two overexpression (HAP4, GCN4) strains from previously published reports were used • Each of genes was classified as significantly up￾regulated, significantly down-regulated, or unchanged in each of the 10 experimental data sets

Results of prediction of gene expression changes Gene expression ChlP-ch All TFs GAL4 RGT1 ADR1 6175 18 418 B0191 GLN3 45 247 36 19 19 Model GALA RG ADR1 HAP4 GCN4 New direct Suggested C 5 12 targets direct targets 10 4 420 Suggested Combinatorial regulation targets 8642 642024110123 4202 HAP1 ROXI GLNG GAT1 Verified A Promoter indirect targets Verified direct 420 occupancy score Combinatorial 6 Unverified targets regulation Q 2 a Gene expression 0 -20 2 chang

Results of prediction of gene expression changes