Engineering: The DBTL cycle
We describe a core set of activities that enable each step of the Design-Build-Test-Learn (DBTL) cycle of synthetic biology, focusing particularly on the development of Agrobactory's Module 2-Delivery and Realease (composite part BBa_K3893030).
Module 2 was designed for delivering and realeasing dsRNA molecules of the velvet gene produced by Module 1. However, we replaced Module 1 by a fluorescent reporter to test the behaviour of the genetic device and the effectiveness of cell lysis. Thus, LuxR-AHL transcription factor activates simoultanously expression of GFP and the lysis protein PhiX174E, when sufficient amounts of cells in the population produce AHL.
Before building our device in the lab, we made a mathematical ODE model to check this design was able to produce the desired oscilations. Although the ODE model only considers N=20 cells, we see qualitatively confirm that our device is able to develop the desired behaviour, i.e. the population oscilations based on quorum sensing.ODE Model
As the ODE model of Module 2-Delivery and Release, this new model has 7 states. Particularly N is the number cells growing in the culture, and GFP is the system output. Ae is the number of molecules in the culture medium, and the remaining species are intracellular ones.
From the computational simulation performed with our model (see the detailes in our modeling page), as expected, we can confirm that the designed gene circuit shows an oscillatory behaviour since cells die only when the lysis protein is activated. Therefore, this ODE model used as a proxy of Module 2-Delivery and Release fulfills the requirements for delivering and releasing dsRNA molecules due to cell lysis.
In the lab, we needed to build our device. For this we planned a golden gate assembly with 3 levels using the Golden Braid assembly method from Valencia_UPV. First, we combined together two composite parts that were already in the Part Registry (BBa_K2656122 a Level 1 transcriptional unit expresing GFP under the control of the pux promoter and BBa_K2656114 a Level 1 transcriptional unit constitutively expressing luxR gene) to create part BBa_K3893028 .
Then we created 2 more Level 1 Transcriptional units using basic parts from the Valencia_UPV Part Collection: parts BBa_K3893026 (a Level 1 transcriptional unit constitutively expressing luxI gene) and BBa_K3893027 (a Level 1 transcriptional unit expresing phiX174 lysis protein under the control of the pLux promoter). This way, we combined these two parts and we created a new composite part BBa_K3893029. Finally, combining these two Level 2 parts we obtained a contruction with four transcriptional units implementing our initial design BBa_K3893030.
Level 2 parts
Part K3893028: AHL induced expression of GFP and luxR constitutive expression
Part K3893029: AHL induced expression of phiX174 lysis protein and luxI constitutive expression
Level 3 part
Part K3893030: QS-based lysis protein/population oscilator
We designed a temporal experiment to assess and quantify (if possible) cell lysate. We collected absorbance and fluorescence data from the gene circuit in vivo. One of the assays is shown in the Figure below.
Figures A and B depict how cells death when the lysis protein is activated by the lux promoter K2656003. The number of cells was calibrated using standardized particle units from Engineering Committee (Measurement Committee) and iGEM Interlab study 2018-2019.
The total GFP fluorescence expressed by the population and a single-cell is shown in Figures C and D, respectively. At the beginning of the experiment, the number of cells producing GFP is very low (OD 600=3.05e-6). But after lysis, GFP molecules are released to the medium and some cells are still producing GFP. We used MEFL/Particle (molecules of equivalent fluorescein per particle) as a standardized unit to quantify GFP expression per cell.
- All experimental measurements were taken with Biotek CytationTM 3, using a 96-well plate. The experimental conditions used in this work are in the Table below
- The cells were E. cloni® cells (Lucigen) transformed with the corresponding plasmid.
- For all experiments, we worked with an isolated colony that contains the corresponding plasmid.
- Cultures of 3mL falcon tubes with sLB medium with the corresponding antibiotic were prepared and incubated overnight (18 hours) at 37°C, 200 rpm.
- The overnight cultures were refreshed in 3 mL of sBL with the corresponding antibiotic and incubated at 37°C, 200 rpm for 4 hours.
- Culture tubes at OD600=0.05 were sit in cold water for 30 mins.
- We measured absorbance and fluorescence of four replicas of the sample starting with OD600=3.05e-6 during 14 hours.
- PROTOCOL: Time between measures 5 min, Temperature 37°C, Shaking Double orbital (Continuously), Absorbance wavelength 600 nm, Excitation wavelength 485 nm, Emission wavelength 528 nm, 96-well volume (individual) 200 μl.
In parallel to this DBTL cycle to develop our Module 2, we performed a smaller inned DBTL cycle including design, assembly (build), measurements (test), and model and characterization of the part (learn) using part BBa_K3893028 and contributing to characterize the pLux promoter (Part BBa_K2656003 which is the Golden Braid compatible, together with its Biobrick starndard sister part BBa_R0062) in the context of the transcriptional unit built in part BBa_K2656122.
Using the information we obtained from this smaller DBTL, we could characterize the pLux promoter (in the context of the transcriptional units data was taken, which is the same as the context of our oscilator). From this characterization of the pLux promoter in the mentioned context, we extracted the following parameters and used them to learn:
- Effective dissociation constant: Kd = 15.34 nM
- Hill coefficient: n = 0.958
- Basal expression: 𝛽 = 0.1049
With these parameters from the experimental data and model of pLux promoter, we went back to our model and learned, by including the new values and performing new computational simulations.
Here, we show the results of the new computational simulations. And we can see that the information we learnt and pluged back into the initial model makes our model better. By comparing the in silico vs. in vivo results, we see that our model captures the temporal dynamics of the oscilator, and also recapitulates with the Total dsRNA being similar to the Total GFP in the experimental data.
An interesting point is that we did not fit the oscillator model with data from the experiment from part BBa_K3893030, but we incorporated the characterization made of one of the components of the system, and still, we got good agreement between the model of Module 2 and its corresponding experimental data. This, speaks itself about the power of modeling and the DBTL cycle in Synthetic Biology. Finally, our newly adjusted ODE model can be used to redesign our device, by predicting outcomes for different conditions of components, and also it can be used in an optimization process to improve other aspects of our devices we need to improve. Now we have a starting point for optimization that can be done in silico, instead of building a new set of circuits to see which one performs better than the original.