Introduction to Systems Biology - Tutorial 3

Program of the lecture "Introduction to Systems Biology" (WS0506)

This tutorial for the module "Control" uses the tools

• CD: CellDesigner is a graphically oriented model construction tool complementary to the text based reaction list for model editing in Copasi (cf. Tutorial 1). The models are exported in SBML format for further analysis with other tools.
• KEGG and BRENDA: Two databases of biochemical knowledge (reactions, enzymes, rate constants derived from the literature) are KEGG in Japan and BRENDA in Köln.
• Copasi: Known from Tutorial 1, we will use Copasi for metabolic control analysis.

The above tools are capable of creating and analyzing models of large biochemical networks (metabolic or signaling or mixed). Contrary to the previous tutorial "Space", here the spatial aspects are only represented by subdivision of the whole volume into compartments. We will start with getting an overall impression of large networks, then see how one could assemble them from information that is retrievable from databases. In the second part some small reaction networks are used to experience Metabolic Control Analysis (the math background of which is given in the next lecture).

• Large network models are interchangable by means of SBML format.
  • Start CellDesigner: > /usr/local/bin/CellDesigner2.5 &
    which opens a single window with several frames therein.
  • Docu: The documentation of the software is downloadable as pdf but not required for the following. The CellDesigner window provides icons on top for the actions, the middle (empty) frame for the network layout and frames at the sides to navigate through the model and species. Next we will browse through one network model. This and several other examples were created in H. Kitano's group in Japan. The CellDesigner software seems to be more suitable than Copasi for creating large models, however via SBML one can then simulate and analyse the models with Copasi (as in tutorial 1).
  • Hepatocyte: Download the SBML file for signaling and metabolic networks in a liver cell, the hepatocyte. Load the file into CellDesigner and answer OK for the conversion. Zoom in and out (zoom fit) with the square-like icons in the middle of the top row. Click on the colored boxes in the main frame and observe species, reactions, compartments. Follow a set of links (signal transduction) from a receptor (looks like a yellow open book) through the nucleus (gene regulation) to enzymes (expressed from genes) of biochemical reactions converting metabolites (green ovals). Reactions in the lower right corner constitute the glycolysis pathway that breaks down Glucose (find this metabolite) into Pyruvate (follow the reactions till you find this) from which many other metabolites are synthesized.
     
• Large network models are collected in KEGG database.
  • SBML models: The 2 most comprehensive repositories for SBML models are KEGG in Japan and Biomodels.net hosted in UK. KEGG has more qualitative information (more than 33000 pathways) relating to the presence or absence of individual components (genes-proteins, enzymes) taken from annotated genomes whereas Biomodels focuses on kinetic models with all parameter values assigned.
  • KEGG: Take a tour through the KEGG database. Then search for the pathway glycolysis. From a long list of different organisms choose Homo sapiens. You arrive at a pathway map with metabolites (dots), reactions (arrows) and enzymes (boxes with enzyme classification numbers=EC). The green enzymes are present in the organism that you have chosen. Locate the dot for Glucose and follow the reactions of glycolysis to Pyruvate. Again you see, Pyruvate is a starting point of many anabolic (synthesis) pathways.
  • PFK: Click on 2.7.1.11. which denotes the enzyme Phosphofructokinase. You the amino acid sequence of the protein, the nucleotide sequence (3 times as long) of the gene and other information on this enzyme in different tissues. To get biochemical details, click "All DBs", then "Enzyme" and the EC number. At the bottom link to BRENDA.
  • BRENDA: This database (hosted in Köln) is specialized on kinetic data. Here you find the specific mechanisms of catalysis and regulation. Click on "Functional Parameters" in the left menu to check the measured Michaelis-Menten constants for different substrates and organisms/tissues/in vitro conditions. All this data has been extracted from the literature which is cited at the bottom. Theoretically, you could now create large models of biochemical networks from information stored in databases. In practice one has to go back to the original literature in many cases.
  • Signaling: Go back to the KEGG database and compare the network models for processes in Alzheimer's disease in Cow and Human. Which key steps lead to the death of neurons? What are the 3 basic processes of axon guidance in the development of human brains? Compare the complexity of the MAPK cascades (under signaling pathways) in yeast (Saccharomyces cerevisiae), fruit fly and human.
     
• Metabolic Control Analysis (MCA).
   
MCA gives you an overview of which model parameters change the steady state concentrations and fluxes stronger or less strong, hence how such control is distributed. Most interesting are the flux control coefficients (derivative of flux versus rate constant) which tell you how strong a change in enzyme concentration eg. by genetic manipulation or regulated gene expression influences a chosen flux eg. the production of a desired product. Often the enzyme changes are compensated by regulatory loops in the biochemical network. Moreover there are elasticities (derivative of flux versus variable) and concentration control coefficients (derivative of steady state concentrations versus rate constant) which provide similar information.
•  
  • Model 1: Start Copasi by
    > setenv COPASIDIR /usr/local/bin/
    > /usr/local/bin/CopasiUI &
    then create model 1 for a two enzyme pathway:
    S -> X -> P
    with simple mass action kinetics and both rate constants = 10. Then click on the Metabolites menu item and set S and P to fixed concentrations = 1 (check box).
    Click Tasks, MCA, tick "perform Steady State Analysis", Run and interprete the flux control. The values are scaled such that for each reaction their sum = 1. Very small values should be interpreted as 0. Imagine you are engineering a novel microorganism that shall have a large product formation flux through the X -> P reaction. Would it be a good idea to increase the corresponding rate constant of the X -> P reaction? Explain the strategy that you would propose.
  • Model 2: Include a third reaction X -> P2 and set P2 fixes=1 and the rate constant =10. Then repeat MCA and interprete the flux control. Change the rate constant of X -> P and check changes in flux control (MCA). Create a plot for the 3 fluxes versus the rate constant of X -> P. Then run a ParameterScan for steady state solution over a range of values for that rate constant. Compare the slopes of the fluxes with the behavior of the flux control coefficients that you had computed for different values of the parameter. If needed, you can download the SBML or Copasi (including plots and tasks) file for the above model.
     
• Modules.
   
In the previous lecture we introduced methods of control theory to determine output responses to time-dependent input signals of modules. These methods (Laplace transformation, transfer function) are implemented in the commercial software MatLab. MatLab is available if you have and use a URZ-login. Some basic steps with transfer functions are illustrated here. Moreover Uri Alon runs a webpage with software for automatic module identification.