Team:UNITN-Trento/Modeling

B. fruity exploits bacterial ethylene to ripen fruit. Given ethylene dangerous characteristics (flammable, asphyxiating, explosive) it is even more important to maintain ethylene concentration under a certain threshold or, even better, inside a certain interval.

For this purpose, we planned to model our light-regulated system and use the result to determine the required light induction pattern. Given the complexity of the system, the high number of variable and the fact that their values are not present in literature (we can at least deduct an approximation) we hypotesized sensed intra-cellular concentrations of reporter protein and estimated the parameters from these hypotetic data. This might be used as a proof of concept.

Of course a better approach would be to obtain experimentally data on the intracellular concentration of the reporter gene. We are in the process of improving the model.

For simplicity, we did not model the entire circuit (that you can find in the datapage). Instead, the model is based on the BBa_K1065302 part, that consists of YF1 and FixJ under R0010 followed by the amilGFP reporter gene under pFixK2. YF1 is the photo-sensor, FixJ is the response regulator (phosphorylated by YF1) and pFixK2 is the target of FixJ-P.

The main differences with the complete circuit are:

1. the inverter, which adds complexity, is not included in the model;
2. instead of EFE, the model contains the amilGFP reporter gene, which will simplify the retrieving of experimental data for parameter estimation.

Also, a couple of assumptions/approximations have been arranged to simplify the model even more:

• the LOV domain is responsible for YF1 autodimerization. From literature it appears that YF1 has kinase net activity in the dark and phosphatase net activity in the light. It is not clear whether YF1 is able to phosphorylate FixJ only if dimerized, so the model do not include the dimerization reaction.
• in the light, YF1 has phosphatase net activity and so is responsible for FixJ-P dephosphorylation. The model does not contain the details of this reaction since they are not meaningful for our purpose. The following Figure 1 shows our model, which can be found here.

Some kinetic constant have been found in literature and used ascenter of an interval to estimate the variables of our model, other were estimated from experimental data within constraints of our choice:

• Starting DNA concentrations were estimated based on hypotetic data within the interval [1e-6;1e-2].
• FixJ phosphorylation and FixJ-P de-phosphorylation kinetics, k4 and k4b, were estimated based on hypotetic data within the interval [0;10] 1/s.
• Transcription kinetic was taken from BIONUMBERS (k = 0.33 1/s) and estimated for each reaction inside the interval [1e-4;1e-1] 1/s
• RNA degradation kinetic was taken from BIONUMBERS (k = 0.011 1/s) and estimated for each reaction inside the interval [1e-3;1e-1] 1/s
• Protein degradation kinetic was taken from BIONUMBERS (k = 0.00083 1/s) and estimated for each reaction inside the interval [1e-3;1e-1] 1/s
• Protein translation kinetic was taken from BIONUMBERS (k = 22 1/s) and estimated within the interval [1e-2;1] 1/s
• The autophosphorylation kinetics in the light and in the dark, k3 and k3b, were taken from (Möglich A, J MOL BIOL, 2009, 385(5):1433-44)

The result of a time-course simulation of 3e+6 seconds of our model is shown in Figure 3.

We added windows of “dark" and “light" exposure into the model. At first we simply added single windows of “dark" of various length, then we repeated the windows to make the enzyme production fluctuate into the desired interval.

This represent a proof of concept, since we do not have yet enough data on EFE production and on the EFE-ethylene relationship to model them.