Team:OUC-China/Model
This year, OUC-China is devoted to creating a whole-cell biosensor for antibiotic detection, which is different from traditional whole-cell biosensors. We hoped our biosensor have more advantages, such as higher sensitivity, lower leakage, better signal-noise ratio, and larger dynamic range. To achieve these goals, we chose RNA aptamer as our reporter, which saves time and helps our system become more sensitive. We introduced T7 promoter system to amplify the signal and improve the sensitivity; we also introduced KB2(a kind of RNA complementary to some sequence of 3WJdB) to 3WJdB combined with the leaky expression without antibiotics to reduce the leakage and improve the signal-noise ratio; We also introduced CRISPRi system to inhibit the expression of KB2 in the presence of antibiotics and make 3WJdB better expression.
There are many RNA aptamers to choose from, which is the most suitable for us? Through structural analysis, we selected the most stable aptamer: 3WJdB. However, when we construct plasmid, we found that 3WJdB in vivo has three bases more GGA which is absent in another literature whose experiments are performed in vitro. Why are they different? Through NUPACK, we analyzed the structure of the two groups of sequences which guided us to explore the specific role of these three bases. Finally, we added the GGA sequence at the 5 'end of our 3WJdB sequence. Due to the lack of data provided in the literature and the lack of time, we can not verify the feasibility of the KB2 module through experiments. Through modelling, we also analyzed the structure of this module. We found that KB2 can effectively prevent 3WJdB from combining with dye to emit fluorescence, so as to reduce leakage and improve the signal-noise ratio.
Figure 1. The structure of 3WJdB
In this figure, we can know that 3WJdB is relatively stable, and the base ‘GGA’ is not a crucial part of 3WJdB
In our system, the constitutive promoters before allosteric transcription factors and KB2 play roles in the realization of our circuit advantage. However, in the initial design of plasmids, we were not sure what strength should be selected for the two constitutive promoters. Therefore, for the promoter before the allosteric transcription factor, we first selected a weak promoter J23114, placed it in the basic circuit, characterized it by modelling, explained the operation mechanism of this circuit by mathematical method, predicted the impact of different promoters on the circuit by adjusting the parameters representing constitutive promoters, and selected the appropriate promoter. For the constitutive promoter before KB2, we also designed experiments to explore the operation mechanism of the module containing this promoter. Similarly, we predicted the circuit by adjusting the promoter parameter and selecting the appropriate promoter.
Figure 2. promoter Pc1, we can infer that the strongest promoter is our best choice
Figure 3. promoter Pc2, we can infer that a weaker promoter such as J23113 and J23103 are our best choices
To sum up, the modelling work this year has assisted the design and development of the whole project from different aspects. Click the following section to learn more about our work this year!
RNA-based systems have a smaller genetic footprint and they provide better time-sensitivity because signal amplification and attenuation are controlled by the rate of transcription, degradation, and assembly of RNA. So we choose RNA aptamer as our reporter to improve the response speed this year. However, we have a lot of choices. Which one is the most suitable? We compared with the secondary structure of four aptamers: 3WJdB(Three-way Junction dimeric Broccoli)[1], Mango[2], Spinach[3], and Broccoli with NUPACK(Figure 1). Results are as follows:
Figure 1. (A)3WJdB;(B)Mango;(C)Spinach;(D)Broccoli
By comparison, we find that the free energy of the secondary structure of 3WJdB is the lowest. It has the advantage of being expressed in vivo, so we choose it for our experiment. However, when we construct plasmids, we found that 3WJdB in vivo has three bases more GGA which is absent in another literature whose experiments are performed in vitro.[1] We wonder if there are any differences between in vitro and in vivo. Therefore, we first used NUPACK to analyze the structure of 3WJdB(Figure 2). From the structure, we infer that ‘GGA’ is not a crucial part of 3WJdB.
Figure 2. (A)the structure in vitro;(B)the structure in vivo
Through contacting the authors and reviewing the literature, we learned that the function of bases ’GGA’ is mainly to assist T7 polymerase to better bind to DNA. So, we ended up adding an extra three bases ’GGA’ to our sequence.
Because of that, the paper can’t give us enough data and lack of time, we can not direct that if our idea: KB2 binding to 3WJdB reduce leakage and improve signal-noise ratio[5] is feasibility by experiment. Therefore, we tested the effect of the combination of KB2 and 3WJdB by using NUPACK. By comparing the structure after KB2 and 3WJdB are complementary(Figure 3) and the figure of the combination of 3WJdB and DFHBI-1T in the paper(Figure 4), we found that the part binding DFHBI-1T was disappeared because of the changing of 3WJdB’s secondary structure! Our idea works!
Figure 3. The secondary structure that KB2 binding to 3WJdB
Figure 4. DFHBI-1T binding to 3WJdB[6]（Jaeyoung K. Jung et al. 2020）
Reference
[1]Alam K K , Tawiah K D , Lichte M F, et al. A Fluorescent Split Aptamer for Visualizing RNA-RNA Assembly in Vivo[J]. Acs Synthetic Biology, 2017: acssynbio.7b00059.
[2]Dolgosheina E V , Jeng S, Panchapakesan S, et al. RNA mango aptamer-fluorophore: a bright, high-affinity complex for RNA labeling and tracking.[J]. Acs Chemical Biology, 2014, 9(10):2412-2420.
[3]Pothoulakis G , Ceroni F, Reeve B , et al. The Spinach RNA Aptamer as a Characterization Tool for Synthetic Biology[J]. Acs Synthetic Biology, 2014, 3(3).
[4]Filonov G S , Moon J D, Svensen N , et al. Broccoli: Rapid Selection of an RNA Mimic of Green Fluorescent Protein by Fluorescence-Based Selection and Directed Evolution[J]. Journal of the American Chemical Society, 2014, 136(46):16299-308.
[5]Jonathan, Lloyd, Claire, et al. Dynamic Control of Aptamer–Ligand Activity Using Strand Displacement Reactions[J]. ACS Synthetic Biology, 2017, 7(1).
[6]Jung J K , Alam K K , Verosloff M S, et al. Cell-free biosensors for rapid detection of water contaminants[J]. Nature Biotechnology, 2020.
Constitutive promoters that control the expression of allosteric transcription factors have a certain effect on the performance of our circuit. Different strengths of promoters will express different concentrations of allosteric transcription factors, which will bind to operons. If the promoter strength is too low, it may not be able to compete with other promoters. If the promoter strength is too high and the concentration of repressor protein is high, it needs a high concentration of antibiotics to relieve the inhibition and express fluorescence. In this way, the sensitivity of our device will be insufficient, so we need a constitutive promoter with appropriate strength.
1.Modelling method and data source
1.1 Modelling method
Considering that our laboratory can mainly provide the relationship between fluorescence intensity and time, which meets the requirement that the state variable changes with time, we choose ordinary differential equation to characterize the system. By setting parameters, the parameter characterization system is solved, and then the change of system output (i.e. fluorescence intensity) when the promoter strength changes is predicted by adjusting the value of the parameter characterizing the promoter strength.
1.2 data source
We characterized the constitutive promoter PC2 before allosteric transcription factor in the basic circuit (as shown in the figure below). In the experiment, we first selected the weaker promoter J23114 of Anderson series for testing. Within four hours, we measured the fluorescence value under the concentration gradient of a series of antibiotics. Using the experimental data measured in the laboratory, we fitted the parameters. For the expression of different promoter strength, we know the relative strength of different promoters from the part library, and characterize different promoters by relative strength.
Figure1: basic circuit
Table1: the relative strength of Anderson promoters
2.The assumptions and simplification of the model
(1)The ratio of fluorescence intensity to bacterial density was used to characterize the concentration of reporter molecules. Because the amount of fluorescent ligand remains the same in each experiment, we also ignore the binding process of 3wjdb and fluorescent dye in the modelling.
(2)We need to characterize the inducible promoter in our system. We use ‘a0’ to represent the leakage, ‘a’ to represent the maximum transcription rate of the part after the promoter, and characterize the transcription rate of the inducible promoter by the occupancy of the operator.
(3)The binding process between different allosteric transcription factors and operators is different at the molecular level. For example, TetR will dimerize and bind to tetO, but this process has little impact on our system, so we ignore this process.
3.Reaction and species
Figure2:TetR-basic circuit.Pc2 is a constitutive promoter that controls the expression of TetR. PT7 is followed by tetO. In the absence of antibiotics, TetR will bind to tetO. At this time, there is a small amount of leakage expression of 3WJdB. With the increase of antibiotic concentration, more and more TetR binds to tetracycline, releases tetO, and the expression of 3WJdB is also increasing.
Take the circuit for detecting tetracycline as an example, we can describe the process of our system as follow:
(1) null -> mRNA_TetR
(2) mRNA_TetR -> mRNA_TetR + TetR
(3) TetR + tetracycline <-> TetR-tetracycline
(4) null -> 3WJdB
(5) mRNA_TetR -> null
(6) TetR -> null
(7) TetR-tetracycline -> null
(8) tetracycline -> null
(9) 3WJdB -> null
Table2:the species in our system
4.ordinary differential equations(ODEs)
For readability, the complex symbol is simplified as:
5.Fitting results and parameters
5.1 Fitting results
The following figure shows our fitting results. The circle is the data measured in our experiment. Different colors represent different antibiotic concentrations, the horizontal axis represents time, and the vertical axis represents the fluorescence intensity after our standardization. The curve we fitted can approximate the experimental data, and the standard deviation is also in the appropriate range. It can be seen that our fitting result is accurate.
Figure3:Fitting results of experimental data.It can be seen that the fitting curve is very close to the observed value, indicating that our model is accurate
5.2 Parameter
6.Results of adjusting constitutive promoter parameters
Figure4:Results of adjusting constitutive promoter parameters
As we can see in the figure, the fluorescence output of the system decreases with the increase of promoter intensity. Based on the existing promoters, we can infer that J23113 and J23103 are the best choices for the promoters of allosteric transcription factors.
We used the same method to fit the parameters and predict the circuit performance of erythromycin and macrolide antibiotic detection circuits. Similarly, our result is that the weakest promoter is the most appropriate in these two detection circuits.
The constitutive promoter before KB2 also has an impact on the implementation of our circuit advantage. In our project, the role of KB2 is to bind to the 3WJdB expressed by leakage in the absence of antibiotics, thus further reducing leakage and improving the signal-noise ratio. If the constitutive promoter controlling KB2 is too weak, the effect of reducing leakage cannot be maximized. If the promoter strength is too high, too much KB2 may affect the expression of 3WJdB in the presence of antibiotics, making our device not sensitive enough. Therefore, selecting the constitutive promoter before KB2 is also crucial to our circuit. Therefore, we constructed a circuit containing the promoter in front of KB2 to characterize the strengh of the promoter in front of KB2 and then predicted the influence of different promoters on the expression of the circuit by adjusting the strength of the promoter.
1.Modelling method and data source
1.1 Modelling methods
Considering that our laboratory can mainly provide the relationship between fluorescence intensity and time, which meets the requirement that the state variable changes with time, we choose ordinary differential equation to characterize the system. By setting parameters, the parameter characterization system is solved, and then the change of system output (i.e. fluorescence intensity) when the promoter strength changes is predicted by adjusting the value of the parameter characterizing the promoter strength.
1.2 Data sources
In the CRISPRi circuit, we included the promoter in front of KB2 , and in this test, we replaced KB2 with sfGFP. By measuring the expression of the circuit, we characterize the system and get the parameters of the corresponding module. As for the strength of different promoters, we also referenced the relative strength data provided by the Part library.
Figure1: the genetic circuit we tested. Our inducer is AHL. Different inducer concentrations will induce the expression of different concentrations of dCas9. The complexes of dCas9 and sgRNA will bind to the binding site after PC1 and inhibit the expression of sfGFP. That is, different concentrations of inducers express different concentrations of sfGFP.
2. Model assumptions and simplification
(1)The ratio of fluorescence intensity to bacterial density was used to characterize the concentration of reporter molecules.
(2) In the characterization of promoter Pc1, we considered its own strength, and because its expression can be inhibited after being combined by dCas9 and sgRNA complex, which is similar to the action mechanism of inhibitory inducible promoter, we introduced the method of characterizing inhibitory inducible promoter to characterize this promoter.
3. reaction and species
We can describe the process of our system as follow:
null->RNA_LuxR
RNA_LuxR->LuxR+RNA_LuxR
LuxR+AHL<->LuxR - AHL
null->sgRNA
null->RNA_dCas9
RNA_dCas9->dCas9+RNA_dCas9
dCas9+sgRNA<->dCas9 - sgRNA
null->RNA_sfGFP
RNA_sfGFP->sfGFP+RNA_sfGFP
RNA_LuxR->null
LuxR->null
AHL->null
LuxR - AHL->null
sgRNA->null
RNA_dCas9->null
dCas9->null
dCas9 - sgRNA->null
RNA_sfGFP->null
sfGFP->null
sfGFP->null
4. reaction and species
4.1 ordinary differential equations(ODEs)
For readability, the complex symbol is simplified as:
5.Fitting results and parameters
5.1 Fitting result
The following figure shows our fitting results. The circle is the data measured in our experiment. Different colors represent different AHL concentrations, the horizontal axis represents time, and the vertical axis represents the fluorescence intensity after our standardization. The curve we fitted can approximate the experimental data, and the standard deviation is also in the appropriate range. It can be seen that our fitting result is accurate.
5.2 parameters
6. Results of adjusting constitutive promoter parameters
As we can see in the figure, the fluorescence output of the system increases with the increase of promoter strength. Based on the existing promoters, we can infer that J23100 is the best choice for the promoter of KB2.
We used the same method to fit the parameters and predict the circuit performance of erythromycin and macrolide antibiotic detection circuits. Similarly, our result is that the strongest promoter is the most appropriate in these two detection circuits.
Due to the lack of experimental time, we can not experimentally verify the complete circuit we designed. In our experiments, we have obtained the characterization parameters of each part. We constructed a complete circuit in simbiology and replaced the two promoters we predicted to predict the performance of the whole circuit. Scanning the change of fluorescence when the concentration of antibiotics changes. As shown in the figure below, the antibiotic concentration increases from bottom to top, and the vertical axis is FI / OD. As expected, the curve is sparse at low concentration; The curve is dense at high concentration, indicating that our system is also sensitive enough at low antibiotic concentration. The lowest antibiotic concentration in the figure is 3 μ g/L, lower than the existing detection limit of 5.17 μ g/L. Our system can realize our ideal!
Figure 1. Results of scanning antibiotic concentration changes
Our original intention is to design a broad and sensitive platform for the detection of antibiotics. In our characterization and prediction, our line can better achieve the effect we want. Although we can not completely turn our ideas into application tools in a limited time, our work provides a platform and direction for future generations. By changing the ATF, we can detect different antibiotics. By changing the promoter and other modules, we can adjust the detection limit, signal-to-noise ratio, and dynamic range. Our project well embodies the advantages of modularity and platform of synthetic biology, But this is only a small part of synthetic biology. It still has broader fields to explore!