Difference between revisions of "Team:TU Darmstadt/Model/Kill Switch Modeling"

Revision as of 21:00, 25 October 2020

Introduction Model One Model Two Model Three

Introduction

This year presented us with several challenges. Due to the coronavirus outbreak and the resulting lockdown, we were not able to work in the wet lab and had to focus our work on the dry lab. Therefore, we decided to focus our resources in an all-around modeling effort. We used homology modeling to calculate a 3D structure of our laccase to improve our enzymes. The biofilm modeling simulated the growth of the biofilm and the third modeling branch was built around a MATLAB^[1] model of a kill switch, based on the ComXQPA ^[2] system . The latter will be described in the following.
The kill switch model consists of three independent models which build upon each other. The first one describes the quorum sensing part, by simulating five individual cells and their interaction. The second one was built after the first model and describes one cell, to test different other parameters in the cell, whilst reducing its complexity. The third model improved the second one by describing two additional parameters.
The circumstances allowed us to put more time into refining and testing our models, so we were able to build a more in-depth model of the kill switch. We also hypothesized that a good working model may theoretically prove our kill switch. With matching data from the lab, the model should be able to precisely fine-tune the kill-switch.
Due to limited lab access, we opted for a new strategy and devised the work to a conceptual model of our kill switch with MATLAB SimBiology^[3] and MATLAB SimBiology Analyzer. The models are based on ordinary differential equations which are solved with a numerical analyser. We developed three iterations of our models.
In the following text, we would like to show you the process of the three models we created.

Model One

Questions

1. Is it possible to simulate different cells in the model?
2. Is it realistic to model the quorum sensing part of our kill switch?
3. What rate of change values do we need to look for?

Assumptions

1. The pathways of the quorum-sensing system can be described as rate of changes
2. The reactions are simple mass-action reactions

Model description

In the first model, we focused on the quorum sensing part. For this purpose, we simulated five individual cells that share a common extracellular space.
First, the cells are equally producing ComX from pre_ComX and release it into the extracellular space. From there it can be transported back into the cell via unspecific membrane transporters. Inside of the cells, ComX activates ComP which phosphorylates ComA to ComA-P.

Goal

The goal of the first model was to experiment with MATLAB SimBiology and produce the first searchable rate of change values. It should be a starting point to refine the model. Also, we wanted to show that it is possible to simulate the quorum sensing part of our model.

Results and future development

This initial model helped us to confine first values and to gain an impression of how the software and the modeling works. Also we extrapolated one critical problem: The problem with this model is that the ComX is distributed in a uniform manner, which is unrealistic because the cells are neither equally distributed nor in an equal development stage. Due to the number of cells, there will be some that are more efficient than others. So, we decided to break it up and only simulate one cell. With lab access, it should be possible to determine the needed values by measuring the extracellular ComX concentration of different colonies or we could try to find the minimal number of cells so that the colony survives and the kill switch is not triggered.
For the further development of this model, it is essential to know how high the intracellular concentration of ComX must be in the cell to phosphorylate ComA. This task needs to be performed in the future, for instance by another iGEM team.

GraphicalAbstract — **Figure 1:** Here you can see a scheme of the first model. The green ellipses represent the species while the boxes represent the enzymes which catalyzes the reaction. Along the black lines the reactions take place.

Equations Model One

A) Fluxes

$$1. ComQ=ComQ.kf*Cell01.PreComX $$
$$2. ComQ1 = ComQ1.kf*Cell02.PreComX $$
$$3. ComQ2 = ComQ2.kf*Cell03.PreComX $$
$$4. ComQ5 = ComQ5.kf*Cell04.PreComX $$
$$5. ComQ6 = ComQ6.kf*cell05.PreComX $$
$$6. ComP = ComP.kf*Cell01.ComA*ComX $$
$$7. ComP1 = ComP1.kf*Cell02.ComA*ComX $$
$$8. ComP2 = ComP2.kf*Cell03.ComA*ComX $$
$$9. ComP3 = ComP3.kf*Cell04.ComA*ComX $$
$$10. ComP4 = ComP4.kf*Cell05.ComA*ComX $$

B) Ordinary differential equations

$$1. \frac {d(ComX)} {dt}= \frac {1} {extracellular*(ComQ + ComQ1 + ComQ2 + ComQ5 + ComQ6 - ComP - ComP1 - ComP2 - ComP3 - ComP4)} $$
$$2. \frac {d(Cell01.PreComX)}{dt} = -ComQ $$
$$3. \frac {d(Cell01.ComA)}{dt} = \frac {1}{Cell01*(-ComP)} $$
$$4. \frac {d(Cell01.[ComA+P])}{dt} = \frac {1}{Cell01*(ComP)} $$
$$5. \frac {d(Cell02.PreCom_X)}{dt} = -ComQ1 $$
$$6. \frac {d(Cell02.ComA)}{dt} =\frac {1}{Cell02*(-ComP1)} $$
$$7. \frac {d(Cell02.[ComA+P])}{dt} = \frac {1}{Cell02*(ComP1)}$$
$$8. \frac {d(Cell03.PreCom_X)}{dt} = -ComQ2 $$
$$9. \frac {d(Cell03.ComA)}{dt} =\frac {1}{Cell03*(-ComP2)} $$
$$10. \frac {d(Cell03.[ComA+P])}{dt} = \frac {1}{Cell03*(ComP2)} $$
$$11. \frac {d(Cell04.PreCom_X)}{dt} = -ComQ5 $$
$$12. \frac {d(Cell04.ComA)}{dt} =\frac {1}{Cell04*(-ComP3)} $$
$$13. \frac {d(Cell04.[ComA+P])}{dt} = \frac {1}{Cell04*(ComP3)} $$
$$14. \frac {d(cell05.PreCom_X)}{dt} = -ComQ6 $$
$$15. \frac {d(cell05.ComA)}{dt} =\frac {1}{cell05*(-ComP4)} $$
$$16. \frac {d(cell05.[ComA+P])}{dt} =\frac {1}{cell05*(ComP4)} $$

Model Two

Question

1. Where do the educts for pre_ComX and ComA come from?
2. Which rate of change values do we need for the Model?

Assumptions

1. The pathways of the quorum-sensing system can be described as rate of changes.
2. The model is built around mass-action equations.
3. One cell stands for an undefined cluster of cells.
4. The educts for the pre_comX and ComA somehow are connected to the promoter we use.

Model description

To address the problems we encountered in the first model, we built a second one. The most fundamental change was that we did not simulate individual cells but rather one cell or more precise an undefined cluster of cells. Like the first model, pre_ComX is converted into ComX and then released into the extracellular space. Then it is transported back into the cell and acts as a cosubstrate for ComP. ComA is phosphorylated and activates the promoter in a new subunit. The new subunit describes the promoter and therefore the step after the phosphorylation.
Another question arose as to where the educts for the pre_ComX and ComA come from. Therefore, we linked the promoter subunit with unknown transition units, which synthesize the educts.

Goal

The goal of the model was to determine the missing values for the model and to prove that it is working, as well as to get values to compare with real lab data.

Results and future development

Although the results of the first simulations looked promising, another problem arised as we were not able to determine the values just by looking into the literature.
If we would set up the model on simple mass-action equations, we would only need the so-called "rate of change" values, which are called kf values and describe the rate with which educts are converted to products. The model may work well with the right values and would have helped us to fine-tune our kill switch by defining the value of proteins we would have needed. Unfortunately, our literature research did not come up with rate of change values we could use in our model, so we considered making the model a concept and help future iGEM teams move their project forward. Some starting points for fine-tuning the model are:
• Integrate an enhancer upstream of the promoter
• Simulate additional interference factors within the cell
• Add a second security layer, e.g. another kills switch-like system
But this is for future iGEM teams to discuss.

**Figure 2:** Here you can see a scheme of the second model. The green ellipses represent the species while the boxes represent the enzymes which catalyzes the reactions. Along the black lines the reactions take place.

Equations Second Model

A) Fluxes

$$1. [Y] = [Y].kf*[Y] $$ $$2. ComQ = ComQ.kf*precomX $$ $$3. ComP = ComP.kf*ComA*ComX $$ $$4. [X] = [X].kf*[X] $$ $$5. reaction3 = reaction3.kf*[ComA+P] $$ $$6. reaction1 = reaction1.kf*Promotor $$ $$7. reaction2 = reaction2.kf*Promotor $$

A) Ordinary differential equations

$$1. \frac {d(ComX)}{dt} = \frac {1}{[Extracellular Space]*(ComQ-ComP)} $$ $$2. \frac {d(precomX)}{dt} = \frac {1}{Cell01*([Y]-ComQ)} $$ $$3. \frac {d([Y])}{dt} = \frac {1}{Cell01*(-[X] + reaction2)} $$ $$4. \frac {d(ComA)}{dt} = \frac {1}{Cell01*(-ComP + [X])} $$ $$5. \frac {d([ComA+P])}{dt} = \frac {1}{Cell01*(ComP-reaction3)} $$ $$6. \frac {d([X])}{dt} = \frac {1}{Cell01*(-[X] + reaction1)} $$

Model Three

Question

1. Is it possible to describe the necessary ComX Threshold in the extracellular space, by using the SimBiology Event functions?

Assumptions

1. The pathways of the quorum-sensing system can be described as rate of changes.
2. The model is built around mass-action equations.
3. One cell stands for an undefined cluster of cells.
4. The extracellular ComX concentration need to be above a certain threshold to be used by comP to phosphorylate ComA.
5. The educts for pre_ComX and ComA are produced continuously.

Model description

The third model built on the second one and refined some aspects of it. On the one hand, a transcription and translation subunit was introduced, which simulate the RNA/DNA synthesis and the associated degradation. On the other hand, event rules were introduced, which simulate a dynamic concentration limit of ComX. That means, only if the ComX concentration rises above a specific threshold, the cells can absorb it back.
translation and transcription unit:
The added transcription and translation subunit simulate the degradation of DNA, RNA, and proteins. For example, not all mRNA reaches the ribosomes and a certain part of mRNA is removed on the way, this must be considered.
ComX Concentration limit:
The cells need a certain concentration of ComX to activate ComP. If the concentration in the extracellular space rises above a concentration of 0.4 nM, the kf value of ComP increases a hundredfold. If the value of ComX is below 0.4 nM, the kf value is set to 0.01 and the reaction does not take place. The 0.4 nM value is an example to demonstrate the function of the event rules and does not relate to real-world data

Goal:

The goal was to conceptualize the model so that future research into the characterization of the kill switch could be used to build a working model.

Results and future development:

The model works in terms of describing the basic function of the kill switch. At this point, the model needs characterization in the lab to be validated. In the future it could be refined by increasing the complexity or using the results from the lab, thereby making it more plausible.

Figure 3: Here you see a scheme of our kill switch model three. The green ellipses represent the species while the boxes represent the enzymes which catalyzes the reaction. Along the black lines the reactions take place. Also here you see a transcription and translation subunit and a tunnelprotein which transports ComX into the extracellular room.

Equations Model Three

A) Fluxes

$$1. ComQ = ComQ.kf*precomX$$ $$2. [Tunnel Protein] = [Tunnel Protein].kf*ComX1$$ $$3. ComP = kf_1*ComA*ComX$$ $$4. [Binding/Unbinding] = [Binding/Unbinding].kf*[ComA+P]$$ $$5. Transkription = Transkription.kf*DNA$$ $$6. Translation = Translation.kf*mRNA$$ $$7. [mRNA Degradation] = [mRNA Degradation].kf*mRNA$$ $$8. [Protein Degradation] = [Protein Degradation].kf*Protein$$

B) Ordinary differential equations

$$1. \frac {d(ComX)}{dt} = [Tunnel Protein] - ComP$$ $$2. \frac {d(PrecomX)}{dt} = -ComQ$$ $$3. \frac {d(ComX1)}{dt} = ComQ - [Tunnel Protein]$$ $$4. \frac {d(ComA)}{dt} = -ComP$$ $$5. \frac {d([ComA+P])}{dt} = ComP - [Binding/Unbinding]$$ $$6. \frac {d(DNA)}{dt} = [Binding/Unbinding]$$ $$7. \frac {d(mRNA)}{dt} = Transkription - [mRNA Degradation]$$ $$8. \frac {d(Protein)}{dt} = Translation - [Protein Degradation]$$

C) Event rules

$$1. ComX < 0.4 kf_1 = 100$$ $$2. ComX > 0.1 kf_1 = 0.1$$

Model download

We from iGEM Technical University Darmstadt want to share the model files with you. We used the Matlab SimBiology version 2020a for our project, but it may work with older versions of Matlab. If you want to download the model you can click here to download it.

References

[1]MATLAB and SimBiology Toolbox 2020a, The MathWorks, Inc., Natick, Massachusetts, United States. [2]ŠPACAPAN,LINKS BETWEEN EXTRACELLULAR PROTEASES, MATRIX COMPONENTS PRODUCTION AND CELL-CELL COMMUNICATION IN A Bacillus subtilis BIOFILM ,DOCTORAL DISSERTATION,2019 [3]Vipul Singhal, Zoltan A. Tuza, Zachary Z. Sun, Richard M. Murray bioRxiv 2020.08.05.237990; doi: https://doi.org/10.1101/2020.08.05.237990

@@ Line 63: / Line 63: @@
               <h4>
-                 Questions:
+                 Questions
              </h4>
-.Is it <b>possible</b> to <b>simulate different cells</b> in the model? <br>
+. Is it <b>possible</b> to <b>simulate different cells</b> in the model? <br>
-.Is it <b>realistic</b> to model the <b>quorum sensing</b> part of our kill switch? <br>
+. Is it <b>realistic</b> to model the <b>quorum sensing</b> part of our kill switch? <br>
-.Which <b>values</b> do we need to <b>look for?</b> <br>
+. What <b>rate of change values</b> do we need to <b>look for?</b> <br>
     <h4>
-                 Assumptions:
+                 Assumptions
              </h4>
-.The <b>pathways</b> of the <b>quorum-sensing system</b> can be described as <b>rate of changes</b> <br>
+. The <b>pathways</b> of the <b>quorum-sensing system</b> can be described as <b>rate of changes</b> <br>
-.The <b>reactions</b> are simple <b>Mass-Action</b> reactions <br>
+. The <b>reactions</b> are simple <b>mass-action</b> reactions <br>
     <h4>
-                 Model description:
+                 Model description
              </h4>
@@ Line 84: / Line 84: @@
     <h4>
-                 Goal:
+                 Goal
              </h4>
-The <b>goal</b> of the <b>first model</b> was to experiment with <b>MATLAB SimBiology</b> and produce the <b>first</b> searchable <b>values</b>. It should be a <b>starting point</b> to <b>refine</b> the <b>model</b>. Also, we wanted to <b>show</b> that it is <b>possible</b> to <b>simulate</b> the <b>quorum sensing</b> part of our model. <br>
+The <b>goal</b> of the <b>first model</b> was to experiment with <b>MATLAB SimBiology</b> and produce the <b>first</b> searchable <b>rate of change values</b>. It should be a <b>starting point</b> to <b>refine</b> the <b>model</b>. Also, we wanted to <b>show</b> that it is <b>possible</b> to <b>simulate</b> the <b>quorum sensing</b> part of our model. <br>
   <h4>
-Results and future development: <br>
+Results and future development <br>
   </h4>
-This <b>initial model</b> helped us to <b>confine first values</b> and to gain an impression of how the software and the <b>modelling works</b>. Also we extrapolated one <b>critical problem</b>:
+This <b>initial model</b> helped us to <b>confine first values</b> and to gain an impression of how the software and the <b>modeling works</b>. Also we extrapolated one <b>critical problem</b>:
 The problem with this model is that the <b>ComX is distributed in a uniform manner</b>, which is <b>unrealistic</b> because the cells are neither equally distributed nor in an equal development stage. Due to the number of cells, there will be some that are more efficient than others. So, we decided to <b>break it up</b> and only <b>simulate one cell</b>. With lab access, it should be possible to <b>determine</b> the needed <b>values</b> by <b>measuring</b> the <b>extracellular ComX concentration</b> of <b>different colonies</b> or we could try to find the <b>minimal number</b> of <b>cells</b> so that the <b>colony survives</b> and the kill switch is not triggered. <br>
 For the further development of this model, it is <b>essential to know</b> how high the <b>intracellular concentration of ComX</b> must be in the cell <b>to phosphorylate ComA</b>. This task needs to be performed in the future, for instance by another iGEM team.
@@ Line 102: / Line 102: @@
 <figure>
                  <img src="https://static.igem.org/mediawiki/2020/b/ba/T--TU_Darmstadt--Killswitch_model_1.svg" alt="GraphicalAbstract" style="width: 100%; height: auto; padding: 10px 0 0 0;">
-                 <figcaption id="Figure1">Figure 1: Here you can see a schematic of the first model. The green ellipses represent the species while the boxes represent the enzymes which catalyzes the reaction. Along the black lines the reactions take place.  </figcaption>
+                 <figcaption id="Figure1"><b>Figure 1:</b> Here you can see a scheme of the first model. The green ellipses represent the species while the boxes represent the enzymes which catalyzes the reaction. Along the black lines the reactions take place.  </figcaption>
              </figure>
@@ Line 170: / Line 170: @@
               <h4>
-                 Question:
+                 Question
              </h4>
 . <b>Where</b> do the </b>educts</b> for <b>pre_ComX and ComA</b> come from? <br>
@@ Line 177: / Line 177: @@
     <h4>
-                 Assumptions:
+                 Assumptions
              </h4>
@@ Line 187: / Line 187: @@
               <h4>
-                 Model description:
+                 Model description
              </h4>
-To address the <b>problems</b> we encountered in the <b>first model</b>, we built a second one. The most <b>fundamental change</b> was, that we did <b>not simulate individual cells</b> but rather one cell or more precise an <b>undefined cluster of cells</b>.
+To address the <b>problems</b> we encountered in the <b>first model</b>, we built a second one. The most <b>fundamental change</b> was that we did <b>not simulate individual cells</b> but rather one cell or more precise an <b>undefined cluster of cells</b>.
 Like the first model, <b>pre_ComX</b>  is converted into <b>ComX</b> and then released into the extracellular space. Then it is transported back into the cell and acts as a <b>cosubstrate for ComP</b>. <b>ComA</b> is phosphorylated and <b>activates</b> the <b>promoter</b> in a new <b>subunit</b>. The new subunit describes the promoter and therefore the step after the phosphorylation. <br>
-The <b>question</b> also arose as to <b>where the educts</b> for the pre_ComX and ComA <b>come from</b>. Therefore, we linked the promoter subunit with unknown transition units, which synthesise the educts.<br>
+Another <b>question</b> arose as to <b>where the educts</b> for the pre_ComX and ComA <b>come from</b>. Therefore, we linked the promoter subunit with unknown transition units, which synthesize the educts.<br>
     <br>
     <h4>
-                 Goal:
+                 Goal
              </h4>
-The <b>goal</b> of the model was to<b> determine</b> the <b>missing values</b> for the model and to <b>proof</b> that it is <b>working</b>. Another goal was to <b>prove</b> that the <b>simulation</b> would <b>work</b> and that we would get <b>values to compare with real lab data</b>.
+The <b>goal</b> of the model was to<b> determine</b> the <b>missing values</b> for the model and to <b>prove</b> that it is <b>working</b>, as well as to get <b>values to compare with real lab data</b>.
              <h4>
-                 Results and future development:
+                 Results and future development
              </h4> <br>
-Although the <b>results</b> of the first simulations looked <b>promising</b>, another <b>problem</b> raised as we were <b>not able</b> to <b>determine the values</b> just by looking into the <b>literature</b>. <br>
+Although the <b>results</b> of the first simulations looked <b>promising</b>, another <b>problem</b> arised as we were <b>not able</b> to <b>determine the values</b> just by looking into the <b>literature</b>. <br>
-If we would set up the model on <b>simple mass action equations</b>, we would only <b>need</b> the so-called <b>"rate of change"</b> values, which are called <b>kf values</b> and describe the rate with which <b>educts are converted to products</b>.
+If we would set up the model on <b>simple mass-action equations</b>, we would only <b>need</b> the so-called <b>"rate of change"</b> values, which are called <b>kf values</b> and describe the rate with which <b>educts are converted to products</b>.
-The model may work well with the right values and would have helped us to <b>fine-tune</b> our <b>kill switch</b> by defining the value of proteins we would have needed. Unfortunately, our <b>literature research did not</b> come up with rate of change <b>values</b> we could use in our model, so we considered making the model a <b>concept</b> and <b>help future iGEM teams</b> moving their project forward.
+The model may work well with the right values and would have helped us to <b>fine-tune</b> our <b>kill switch</b> by defining the value of proteins we would have needed. Unfortunately, our <b>literature research did not</b> come up with rate of change <b>values</b> we could use in our model, so we considered making the model a <b>concept</b> and <b>help future iGEM teams</b> move their project forward.
-Some <b>examples for fine-tuning</b> the model are: <br>
+Some <b>starting points for fine-tuning</b> the model are: <br>
-•	<b>Build</b> an <b>enhancer</b> upstream of the promoter<br>
+•	<b>Integrate</b> an <b>enhancer</b> upstream of the promoter<br>
-•	<b>Simulate</b> additional <b>interferenc</b>e factors <b>within</b> the cell <br>
+•	<b>Simulate</b> additional <b>interference</b> factors <b>within</b> the cell <br>
 •	Add a </b>second security layer</b>, e.g. another kills switch-like system <br>
   But this is for future iGEM teams to discuss. <br>
@@ Line 221: / Line 221: @@
              <figure>
                  <img src="https://static.igem.org/mediawiki/2020/4/4f/T--TU_Darmstadt--Killswitch_model_2.svg" style="width:50%; height:auto; align=center; margin-left:25%;">
-                 <figcaption id="Figure2"><b>Figure 2:</b> Here you can see a schematic of the second model. The green ellipses represent the species while the boxes represent the enzymes which catalyzes the reactions. Along the black lines the reactions take place.  </figcaption>
+                 <figcaption id="Figure2"><b>Figure 2:</b> Here you can see a scheme of the second model. The green ellipses represent the species while the boxes represent the enzymes which catalyzes the reactions. Along the black lines the reactions take place.  </figcaption>
              </figure>
@@ Line 269: / Line 269: @@
               <h4>
-                 Question:
+                 Question
              </h4>
 . Is it <b>possible</b> to <b>describe</b> the necessary <b>ComX Threshold</b> in the extracellular space, by using the <b>SimBiology Event functions</b>?
              <h4>
-                 Assumptions:
+                 Assumptions
              </h4>
 . The <b>pathways</b> of the quorum-sensing system can be described as <b>rate of changes</b>. <br>
 . The <b>model</b> is built around <b>mass-action equations</b>. <br>
-. One <b>Cell</b> stands for an <b>undefined cluster of cells</b>. <br>
+. One <b>cell</b> stands for an <b>undefined cluster of cells</b>. <br>
-. The extracellular <b>ComX concentration</b> need to be above a certain <b>threshold</b> to be used by comP to <b>phosphorylate ComA</b> <br>
+. The extracellular <b>ComX concentration</b> need to be above a certain <b>threshold</b> to be used by comP to <b>phosphorylate ComA.</b> <br>
-. The <b>educts</b> for pre_ComX and ComA are <b>produced continuously</b> <br>
+. The <b>educts</b> for pre_ComX and ComA are <b>produced continuously.</b> <br>
     <h4>
-                 Model description:
+                 Model description
              </h4>
-The <b>third model</b> built on the second one and <b>refined</b> some aspects of it. On the one hand, a  <b>transcription and translation subunit</b> was introduced, which <b>simulate the RNA/DNA synthesis</b> and associated <b>degradation</b>. On the other hand, <b>event rules</b> were introduced, which <b>simulate</b> a dynamic <b>concentration limit</b> of ComX. That means, only if the <b>ComX concentration</b> rises above a specific <b>threshold</b> the cells can <b>absorb</b> it back. <br>
+The <b>third model</b> built on the second one and <b>refined</b> some aspects of it. On the one hand, a  <b>transcription and translation subunit</b> was introduced, which <b>simulate the RNA/DNA synthesis</b> and the associated <b>degradation</b>. On the other hand, <b>event rules</b> were introduced, which <b>simulate</b> a dynamic <b>concentration limit</b> of ComX. That means, only if the <b>ComX concentration</b> rises above a specific <b>threshold</b>, the cells can <b>absorb</b> it back. <br>
 translation and transcription unit:  <br>
 The added <b>transcription and translation subunit</b> simulate the <b>degradation</b> of DNA, RNA, and proteins. For example, not all <b>mRNA</b> reaches the ribosomes and a certain part of mRNA is <b>removed</b> on the way, this must be considered.  <br>
@@ Line 311: / Line 311: @@
              <figure>
                  <img src="https://static.igem.org/mediawiki/2020/a/af/T--TU_Darmstadt--Killswitch_model_3.svg" alt="figure" style="width:50%; height:auto; align=center; margin-left:25%;">
-                 <figcaption id="Figure#">Figure 3: Here you see a schematic of our kill switch model three. The green ellipses represent the species while the boxes represent the enzymes which catalyzes the reaction. Along the black lines the reactions take place. Also here you see a transcription and translation subunit and a tunnelprotein which transports  ComX into the extracellular room.   </figcaption>
+                 <figcaption id="Figure#">Figure 3: Here you see a scheme of our kill switch model three. The green ellipses represent the species while the boxes represent the enzymes which catalyzes the reaction. Along the black lines the reactions take place. Also here you see a transcription and translation subunit and a tunnelprotein which transports  ComX into the extracellular room.   </figcaption>
              </figure>
           </div>