From 2014.igem.org

Revision as of 11:50, 16 October 2014 by EdwardMuir (Talk | contribs)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)

Exeter | ERASE

Project Design

Biological modelling has been used for a variety of purposes. In this project we sought to build a model which would help us to design, analyse and debug our biological system . Furthermore, we wanted a method of visualising and explaining how our bacterium was performing. Specifically, our aims were:

The model should inform the biological design.
The model should inform the choice of experiments.
The model should be receptive to empirical observations.
The model should help explain how our biological system works.

Many iGEM teams have modelled their bacteria at the biochemical level using systems of differential equations. We started in the same manner, and then built a stochastic model on top of this to make the simulation more realistic. We then created a model of the diffusion of TNT across the cell membrane. Another model was created for the growth of the bacteria and the inclusion of a kill switch. Finally, we tried to bring these models together to create as accurate a model of our system as possible.

The Biochemical Level

We reasoned that the primary determinant of TNT degradation within the cell would be the activity of the introduced enzyme, either XenB or NemA (see Old Yellow Enzymes. The degradation of the TNT by these enzymes can be modelled by Michaelis-Menten kinetics. The most basic set of ordinary differential equations which would describe the system would be:

\begin{equation} \frac{d[\text{Cell}]}{dt}=g[\text{Cell}]−d[\text{TNT}] \end{equation} \begin{equation} \frac{d[TNT]}{dt}= \frac{−k_{cat}\text{TNT}[\text{Cell}]}{K_m+ [\text{TNT}]} \end{equation}

where:

[$\text{Cell}$] is the number of cells in the colony.
[$\text{TNT}$] is the number of TNT molecules.
$g$ is the growth rate of the cells
$d$ is the rate at which TNT kills the cells.
$k_{\text{cat}}$, $k_m$ are arbitrary constants of the Michaelis-Menten equations.

In this model the concentration of enzyme is directly proportional to the number of cells. The cells grow at a rate $g$ and are destroyed at a rate proportional to the concentration of TNT in the system. The cells degrade the TNT according to Michaelis-Menten kinetics.

Data for the $k_{\text{cat}}$ and $K_m$ of XenB and NemA was not available, and so we initially ran the simulation using different arbitrary values. These yielded different distinct cases depending on the initial conditions. When the TNT was started at 20, if the amount of cells was started at 5, then the cells would ‘die out’. If the amount of cells was started at 10 for the same amount of TNT, though, the cells would increase indefinitely and the amount of TNT would be reduced to 0.

In order to make this model more accurate it was therefore necessary to gain information about the kinetics of XenB and NemA. We therefore sought to express recombinant XenB and NemA in E. coli, purify the proteins using a His-tag and determine the Kcat and Km directly. The results of this in vitro experimentation can be found in the sections Enzyme Kinetics: HPLC and Enzyme Kinetics: Raman.

The Cellular Level

Another primary determinant of TNT degradation rate will be the concentration of the substrate (TNT) within the cell. The concentration of TNT will be determined partly by the activity of XenB or NemA (see above) but also by the rate of diffusion of the TNT across the cell membrane. We therefore introduced a new set of equations:

\begin{equation} \frac{d[\text{Cell}]}{dt}=g[\text{Cell}]−d[\text{TNT}_{\text{in}}] \end{equation} \begin{equation} \frac{𝑑[\text{TNT}_{\text{in}}]}{dt}= −\frac{V_{\text{max}}\text{TNT}_{\text{in}}[\text{Cell}]}{K_m+ \text{TNT}_{\text{in}}} + m \end{equation} \begin{equation} \frac{𝑑[\text{TNT}_{\text{out}}]}{dt}= −m \end{equation} \begin{equation} m \propto [\text{TNT}][\text{Cell}] \end{equation}

where:

‘[Cell]’ is the number of cells in the colony.
‘[TNT]’ is the number of TNT molecules.
‘[TNTin]’ is the number of TNT molecules inside cells.
‘[TNTout]’ is the number of TNT molecules outside cells.
‘g’ is the growth rate of the cells.
‘d’ is the rate at which TNT kills the cells.
‘kcat’, ‘km’ are arbitrary constants of the Michaelis-Menten equations.
‘m’ represents transport rate across the cell membrane

This system, alog with some arbitrary constnats, was programmed into Matlab and the ode45 nonstiff differential equation solver gave result of the following form:

According to this model, the amount of TNT in the system decreases over time. However, as with our first model, we do not know how well TNT crosses the cell membrane i.e. the value for ‘m’ is unknown. We therefore sought to estimate diffusion rates (m) into the cell using Raman spectroscopy. The results can be seen here Enzyme Kinetics: Raman.

It is possible add more and more elements to our system of differential equations. For instance the rate of transcription and translation of enzymes within the cell could be added. However this would have only added more unknown coefficients to our equations. As such we decided to stop adding terms here as we were confidant that values could be experimentally determined for each of the coefficient in the system.

At the time that the model was created the plan was that coefficients would be sourced from the following experiments:

Coefficient	Experiment
$g$, $d$	The Toxicity of TNT and Nitroglycerin to E. coli.
$V_{\text{max}}$, $k_m$	Validation of Enzyme Function and Activity
$m$	in vivo Activity

A Stochastic Model

Despite making the model more sophisticated, this remains a naïve method of modelling the system. Fundamentally, the system of differential equations is designed so that it behaves as we would expect, and, as with any model, many assumptions are made. The system may not be accurately described by Michaelis-Menten kinetics. It is assumed, for example, that the cells and the TNT are completely mixed and there is no notion of the possibility of spatially variable concentration levels. We also assume that the amount of cells and TNT would be continuous, but in reality both the number of cells and the number of TNT molecules would be integers.

In order to address the problem of the number of cells not being continuous, we created a computer simulation that modelled a population of cells. The program simulated the state of the population over a certain amount of time, split into a number of time intervals.

At each interval, there was a constant probability of each cell splitting. When this happened, the concentrations of the substances inside the cells divide equally between the two cells. The cells could also die at each interval with a constant probability. If this happened, the substances would be distributed evenly across the other cells in the population.

Figure *********: Stochastic model, computationally intensive.

The results of this model were described by a list of parameters such which corresponded to the constants (e.g. Kcat, Km) in the differential equations. By adjusting these parameters different outcomes could be obtained. We ran the simulation with various values of ‘kcat’, and this showed that TNT was degraded faster at higher values (probably show a figure even if it is showing the same result as the simpler model). It should be noted however, this result was clear from the initial equations. Our new model has not provided us with any new information at this point. However, this approach can be used as the basis for more complex models, and so we decided to introduce additional information in the form of the location of each cell in space.

A Spatial Model

To introduce the concept of position into our model, we created a naive simulation of the growth of the bacterial population. Importantly, in this program, each individual bacterium was modelled separately. They bacterium had a position on a grid and each was modelled over a certain period of time.

In addition, the program also simulated a kill switch which worked by producing holin when there was no TNT in the cell, and would kill the cell if it reached a certain amount. Such kill switches are common in the Registry (e.g. ...). In order to implement such a kill switch it would be necessary for the cells to sense the concentration of TNT and, in order to simulate various outcomes, the response to TNT would need to be characterised. We therefore identified and characterised key promoter elements that can respond to TNT and/or nitroglycerin (see Detection of Xenobiotics).

Each cell in the grid has an amount of TNT present and is either empty or occupied by a bacterium. At each time interval, if there was a bacterium in the cell, an amount of TNT proportional to the difference between the amount outside the bacterium and the amount inside the bacterium would diffuse across the membrane i.e.:

\begin{equation} R=k(\text{TNT}_{\text{out}}−\text{TNT}_{\text{in}}) \end{equation}

Where ‘$R$’ is the rate of diffusion and ‘$k$’ is an arbitrary constant.

The bacteria would each have an amount of TNT and an amount of holin associated with them. At each time interval, they had a chance to split into two bacteria, and the concentrations of the substances would be divided equally. The bacterium would only split if there was at least one empty space in the eight cells surrounding it, and the chance that it would split was proportional to the number that were free.

They would also die if the holin level reached a certain threshold or the TNT level reached a certain threshold. This was in order to model the fact that TNT could be toxic to the bacteria E. coli Xenobiotic Tolerance.

At each interval, the amount of TNT in the bacterium would be reduced by a random amount which was proportional to the amount currently in the bacterium. In the visualisation of this model, the colour of each grid cell and each bacterium was represented by the amount of TNT in it. These variation in the colour is proportional to the logarithm of the amount of TNT so that the exponential decrease can be seen more clearly. Finally, the volume of the music is proportional to the number of cells in the population.

Simulating diffusion across the phospholipid membrane

A major problem universal to synthetic biology is determining a method of bringing together products produced by the cell and the target of that product. One of the barriers to the movement of products produced by the cell is the phospholipid membrane. Many proteins are too large to cross this barrier without the help of membrane pumps however smaller molecules can move relatively freely across the boundary.

This model will attempt to determine whether our enzymes or TNT can freely move across the membrane. The time scales associated with this transport will also be ascertained . It will help us decide whether membrane pumps are also needed to to make our construct an effective method of degrading TNT.

The movement of smaller molecules across the lipid bi layer is largely governed by diffusion. Fick's law of diffusion can be used to describe the movement of TNT concentrations in a homogenous fluid,

\begin{equation} J=-DA \frac{dC}{dx}. \end{equation} If a distinction is made for a volume then this equation can be linearized (i.e. the differential is replaced with the difference in concentration inside $C_{\text{in}}$ and outside $C_{\text{out}$ of that volume all over the length between the concentration measurements.): \begin{equation} J_m=-AD_m \frac{C_{\text{out}}-C_{\text{in}}}{l_m}. \end{equation}

This only holds true if the volume to which your concentration is moving has the same properties as that which it came from.

\begin{equation} \frac{dC_1}{dt}=-P_a (C_4-C_1) \end{equation} \begin{equation} \frac{dC_4}{dt}=P_a (C_4-C_1 )-P_b (C_5-C_4) \end{equation} \begin{equation} \frac{dC_5}{dt}=P_b (C_4-C_1 )-P_c (C_8-C_5) \end{equation} \begin{equation} \frac{dC_8}{dt}=P_c (C_8-C_5 ) \end{equation} \begin{equation} P_{a,c}= \frac{D_m \beta}{a,c},P_b=\frac{D_w}{b} \end{equation} β = Partition coefficient, Dw=Diffusion coefficient in water, Dm=Diffusion coefficient in membrane D_m

Figure *********: The diffusion of a small molecule across a double semi-permeable membrane. The red line indicates the concentration level before equilibrium is reached. The small molecule is more soluble within the barrier.

mention lack to cell wall lead onto Enzyme Kinetics: Raman

Summary

Previous: The Problem

Next: The Enzymes