Team:Lyon-INSA/modelling

From 2012.igem.org

Revision as of 13:57, 26 September 2012 by Beryl (Talk | contribs)

Modelling

An interesting question when you are more biologist than mathematician (too many complicated equations!!!). And as most of our team members are biologists/biochemists… Thus, we tried to explain modelling and our model in a comprehensive way for everyone.


This is our Biological Modelling for Dummies !


Click on the title to show/hide the text.

What is modelling ?

Definitions:

A model is a symbolic representation of an object’s or phenomenon’s aspects in the real world.

All models are false. Some are useful.” Georges Box


Modeling is the process that allows the development of a model. It’s taking into account [1]:
  • The phenomenon to represent
  • A specific formal system (equation, diagram..)
  • Objectives (what we want to do with the model)
  • Data (for variables) and knowledge (relation between variables) available or accessible by experimentation or observation


The tasks to obtain and use the model depend on the biological situation and the formal system chosen. Nevertheless, it must:
  • Have a formalization work, which is the model writing
  • Manipulate the model in the formal system to describe its properties (theoretical main behaviour independantly on the values of the parameters)
  • Establish relationships with other representations (computer program, graph function)
  • Interpret and compare different representations obtained in the formal world with the biological reality (often that reality is seen through experimental data)

References:
  • [1] Alain Pavé, Modélisation en biologie et en écologie, Aléas, 1994

Biological System description

Situation:

After the destruction of the biofilm by “Biofilm Killer” bacteria, we want to have the choice to either create a surfactant or establish a positive biofilm, both to prevent the recolonization of the surface by deleterious organisms. The switch is done by environmental conditions: two inducers can be added to select one behaviour or another.

Biological system to model:

For this, we have created the following construction, with a double regulation:


Figure 1: The construction of the biological model, with the following elements: 2 promoters (Pxyl and Plac), 2 repressors (LacI and XylR proteins)
and 2 inducers (IPTG and Xylose), and also sfp and abrB genes for Sfp and AbrB proteins.


This system is a gene-regulatory network, where two different states are possible:
  • Formation of a naturally toxic bio-surfactant through sfp gene, which has antimicrobial properties that prevents the recolonization of the surface. The surfactant used is surfactine, whose production is activated by the sfp gene. This is the COAT option.
  • Establishment of a positive biofilm by the inhibition of the main biofilm repressor abrB gene. This is the STICK option.


Figure 2: The two possible states: surfactant formation for the top construction or biofilm formation for the second construction

LacI and XylR proteins are called repressors. They bind to their respective promoters (Plac and Pxyl) that disable the ARN polymerase to bind to the promoters and so proteins under the inactivated promoter are not produced.
There are also two inducers in the system, IPTG (isopropyl β-D-1-thiogalactopyranoside) and Xylose (monosaccharide of the aldopentose type). In the absence of these inducers, both constructions are inhibited. If only one of them is present, the inhibition disappears and the corresponding construction is expressed.



For example, in the presence of Xylose, XylR proteins will form an enzymatic complex with their Xylose sugar. Thus, the inhibition of Pxyl caused by XylR will disappear and there will be a bigger production of Sfp, AbrB and LacI proteins. Sfp production induces surfactine production, and AbrB production involves the repression of the biofilm formation. Eventually, LacI production will inhibit XylR production, so there will be stabilisation of Pxyl activation. In opposition, in the presence of IPTG, LacI proteins will bind to their ligand, and Plac promoter will be free. So XylR proteins will be overproduced, limiting Sfp and AbrB productions. Thus, there will be no surfactine in the environment, biofilm formation can begin.

Aim of the model

With this model, we want to verify the biological system, to be sure that the switch is possible. We also want to predict the behaviour of this biological system depending on the quantity of inducers present in the environment.

However....

We are working in a Bacillus subtilis strain and some parameters such as XylR values on binding/unbinding kinetic, association constants... cannot be found in the literature and most of the existing values come from a E. coli strain. Furthermore, we are finishing the final construction and its characterization is underway. Parameters will be measured very soon.

Because of this lack of information, we will create a theoretical model in order to identify the main system behaviour.

Furthermore, as we are mainly biologists in the team, we thought interesting to explain how we can easily obtain a mathematical model from a biological system.

Biological modelling for dummies !


Basic knowledge:

We want to transform the biologic system into mathematical equations in order to be able to determine the quantity of inducers (input) needed to obtain a particular behaviour (output).


Figure 3: The black box model:
there will be the STICK or COAT option depending on the inducers concentration


Ordinary differential equation (ODE):
  • mathematical equation
  • format:
  • explanation: used in biology and physics to represent the growth or evolution of a quantity dx (i.e. population or concentration) proportional to the population size/effective concentration x during a period of time t
  • x is called a variable


Elements of the model

First list of variables:

We want to have as output the concentration of LacI and XylR depending of the input of inducers concentrations. We know that the repressors can bind either to their promoter (Plac and Pxyl respectively) or to their inducer (IPTG and xylose). Thus, first of all, we have the following variables in the system:


Figure 4: The model variables at first glimpse


Binding and unbinding kinetics:

We decided to analyse the relation between repressors, promoters and inducers depending on the law of mass action. It is a branch of chemical kinetics, which states that the speed of a chemical reaction is proportional to the quantity of the reacting substances. These substances will bind with an association kinetic k and unbind with a dissociation kinetic km.


Figure 5: Binding and unbinding kinetics in the model


It is working either for LacI binding to its Plac promoter than for XylR to Pxyl and also the inducers and LacI and XylR. This binding creates a new complex.

Figure 6: The model variables

Equations of the model

Now, we can find the equations, based on the behaviour of each element. There will be 3 types of equations, each of them related to the nature of the variable, i.e. a promoter, an inducer or a repressor.


Promoters:
As described above, there are Plac and Pxyl promoters and each of them can be free (with no repressor binds on it) or occupied (with the corresponding repressor associated).
Thanks to the law of mass action and binding and unbinding kinetics, we obtain the equations like this:

So the equations for the promoters are this:


Figure 7: :Promoters equations

Inducers:

With the same method based on law of mass action and binding/unbinding kinetic, we obtain the inducers equations.


Figure 8: Inducers equations

Repressors:

Now, for the repressors, the method is quite similar as before. However, we have to take into account that the proteins have a degradation rate (δ) depending on their nature and the environment. The quantity of protein produced at each time depends of the promoter under control (α) because of the different expression levels.


LacI Equation


We obtain XylR equation exactly as LacI's.

Figure 9: The repressors equations


Parameters of the model

For this model, we need at least 12 parameters that characterize the variables and their relationship between each other.

Name Description Unit Value Reference
Prod_Plac Production rate from Plac promoter mol.s-1 1.66E23 1
Prod_Pxyl Production rate from Pxyl promoter mol.s-1 *** **
k1 binding kinetic of LacI and IPTG mol-1.s-1 1.2E5 2
km1 unbinding kinetic of LacI_IPTG s-1 2.1E-1 2
k2 binding kinetic of XylR and Xylose mol-1.s-1 *** **
km2 unbinding kinetic of XylR_Xylose s-1 *** **
k3 binding kinetic of LacI and Plac mol-1.s-1 5.1E6 2
km3 unbinding kinetic of PlacO s-1 3.7E-2 2
k4 binding kinetic of XylR and Plac mol-1.s-1 *** **
km4 unbinding kinetic of XylR and PlacO s-1 *** **
δ_LacI degradation rate of LacI s-1 3
δ_XylR degradation rate of XylR s-1 *** **

Model parameters. *** for no value and ** for no reference

The available values have been measure mainly in E. coli strain.

Hypotheses

The following hypotheses have been made for this model.
  • we just need LacI concentration for surfactant production and not Sfp and AbrB concentration because there is a proportional link between them. If there are LacI proteins produced, there will be also Sfp and AbrB protein.
  • we assumed that there will be no degradation of IPTG and xylose due to their high stability. [4]
  • we are aware of LacI[5] and XylR[6] dimerisation as fondamental fonctional unit but they are not taking into account in this model.

References:
  • [1] Nature. 2000 Jan 20;403(6767):335-8. A synthetic oscillatory network of transcriptional regulators. Elowitz MB, Leibler S.
  • [2] Xu H.,Moraitis M., Reedstrom R. J., Matthews K. S. 1998. Kinetic and thermodynamic studies of purine repressor binding to corepressor and operator DNA. J. Biol. Chem. 273:8958–8964.
  • [3] Tuttle et al. Model-Driven Designs of an Oscillating Gene Network., Biophys J 89(6):3873-3883, 2005
  • [4] Herzenberg, L.A., Studies on the induction of beta-galactosidase in a cryptic strain of Escherichia coli. Biochim. Biophys. Acta, 31, 525 (1959)
  • [5] Ramot, R. et al, Lactose Repressor Experimental Folding Landscape: Fundamental Functional Unit and Tetramer Folding Mechanisms. Biochemistry (2012)
  • [6] Song S., Park C. Organization and regulation of the D-xylose operons in Escherichia coli K-12: XylR acts as a transcriptional activator. J Bacteriol. 1997 Nov;179(22):7025-32.

Results


Expected results:
Finally, we translated the biological system into mathematical equations. As we can see, there is a lot of paramaters for the binding and unbinding kinetics, degradation rates and productions from promoters. However, values for only a few of them are available. This is why we needed to simulate the behaviour of our system.
We want to have an overproduction of XylR in the presence of IPTG, for the STICK option. And an overproduction of LacI when there is xylose in the environment, for the COAT option.
According to the above theoretical model, we should generate the expected toggle switch (figure below), using the appropriate values for all parameters.


Figure 10: Expected results of the model




Experimentations:
So far, we can consider the values for the Plac promoter, LacI and IPTG that have been measured in E. coli to apply to our B. subtilis model. We have performed experiments with the Pxyl, xylose and XylR to evaluate whether this promoter can be modelled using similar values. When this step is performed, our model will allow us to determine two important concentration limits:
  • 1) the lowest IPTG concentration for the induction of the COAT option
  • 2) the lowest xylose concentration for the induction of the STICK option.


However, we did some tests to compare some parameters such as the production rate of each promoter. Indeed, it is an important element in the protein regulation, and can make a big difference between the two constructions.
RESULTATS de Carine


We also thought of a way to obtain binding and unbinding kinetics of the proteins. Some methods such as Isothermal titration calorimetry (ITC) permit to determine the thermodynamic parameters of interactions in solution

Retrieved from "http://2012.igem.org/Team:Lyon-INSA/modelling"