---

Antithetic integral controller

We take inspiration from Aoki et al (2019) where an antithetic integral controller is implemented in bacteria. An antithetic motif refers to the process in which two controller proteins permanently annihilate each other. The two controller proteins are designated Z₁ and Z₂. One controller protein (Z₁) promotes the expression of the output (X) and consequently the desired protein (or reporter protein in this case). The output (X) in turn promotes the expression of the second controller protein (Z₂). Z₁ and Z₂ annihilate each other. Together an integral controller has been created.

The antithetic circuit motifs implemented has always been an activation motif i.e. Z₁ activates the expression of X and X activates the expression of Z₂ (Figure 1). However, we are also interested in a repression motif i.e. Z₁ repressing the expression of X and X represses the expression of Z₂ (Figure 2). To implement both of the activation and repression circuit, we have decided to use ECF20/AS20 and ECF16/AS16 as Z₁ and Z₂ respectively for activation and sinR/sinI as Z₁ and Z₂ respectively for repression. Through modelling and mathematical analysis, we realise the repression system has region of instability where we need to tune the system by choosing the appropriate values of parameters to ensures stability. Despite this, we are keen to try out the repressor case by tuning the parameters.

The standard antithetic circuit itself is divided into 3 modules. Module A contains the promoter activated/repressed by Z₁, RBS, X and MegaT. Module B contains a constitutive Anderson promoter, RBS, Z₁ and DT3. Module C contains the promoter activated/repressed by X, RBS, Z₂ and DT5. As JUMP assembly requires cloning in 4 modules at a time, we put in a terminator as a placeholder into module D to allow JUMP assembly into pJUMP47-2A (Figure 3).

Through modelling, we discovered that different parameters for the promoter and RBS strengths can lead to different circuit behaviour. To try out the different combinations of parameters, we plan to take a combinatorial approach, changing the RBS strength, the strength of the constitutive Anderson promoter and the amount of inducers added to investigate the dynamics of these different circuits.

Circuit Testing

Positive perturbation

In order to test our genetic circuit and show that it is behaving as an integral controller, a perturbation circuit is used to increase the production rate of X to see whether the circuit can robustly adapt. To incorporate the perturbation circuit into our antithetic integral controller circuit, we make use of the module D site and the downstream site of the Level 2 JUMP plasmid (pJUMP47-2A (sfGFP, pSC101 ori)). In module D, we put in PTet*, RBS B0032m, X and MegaT while for the downstream site, we put in J23106, RBS B0032m, CDS tetR and T_ECK120029600.

To clone into the downstream site, we have to linearly ligate the components of the downstream site, together with a 5’ and 3’ adapter for the recognition site of BbsI through linear golden gate assembly with the enzyme BsaI. After linear golden gate assembly, it can be incorporated into the downstream site of the plasmid pJUMP47-2A through restriction digestion with BbsI. Figure 9 shows the antithetic integral controller motif together with the perturbation circuit. To see the circuit design implemented with the actual antithetic species, see Figure 10, 11 and 12.

How does the perturbation circuit work?

The perturbation circuit produces tetR at a constant rate through the action of J23106 which act as a repressor to PTet* so essentially only the antithetic controller circuit is being expressed. However, by adding aTc, the repression of tetR on Ptet* is lifted, leading to an increased production of X, thus positive perturbation.

Global Perturbation- Temperature:

Temperature always fluctuates. Given that the temperature is still in physiological range, increase in temperature increases reaction rate while decrease in temperature leads to a lower reaction rate. Therefore, we plan to test the circuit at different temperatures to see whether the circuit will adapt to different temperature conditions.

Ammonia

Ammonia is a common by-product of biomanufacturing that can be toxic to cells at high concentrations. Therefore, we aim to test how ammonia concentration can affect the expression level of X and whether the circuit can robustly adapt in the presence of ammonia perturbation.

Negative perturbation- Chloramphenicol

Chlorampheicol is an antibiotic that inhibits the protein production in bacteria through binding to ribosomes, leading to a global negative perturbation of reducing production rates of all the species. By giving a low concentration of chloramphenicol, we can investigate whether the circuit adapts to growing in an overall unfavourable environment that leads to lower rate of protein production. Testing the circuit in response to a perturbation by antibiotics have significant implications on microbial drug delivery where antibiotics may be present in patient’s gut when microbial drug delivery is taking place.

Choices of parts

In Aoki et al (2019), SigW and RSiW is the controller species being used for annihilation. Through reading different literature on the theoretical aspects of the antithetic integral controller circuit, we recognised that for the activation motif, sigma and anti-sigma factor will be the best choice. Sigma factors and anti-sigma factors are essential for bacteria, thus we have to choose a sigma and anti-sigma pair that is orthogonal to other sigma factors. Through reading the paper “Design of orthogonal genetic switches based on a crosstalk map of σs, anti-σs, and promoters” by Rhodius et al (2013), we found that ECF20/AS20, followed by ECF16/AS16 is the most orthogonal to host metabolism, with the highest fold change to the activity of the promoter when bound to it and low toxicity.

For repression, we have decided to use sinI/sinR as the controller species where the paper “Molecular Basis of the Activity of SinR Protein, the Master Regulator of Biofilm Formation in Bacillus subtilis” by Newman et al. (2013) documents the interaction of sinI and sinR and with DNA and shows that sinI and sinR are very strongly bound together.

For the feedback species (X), we have decided to use the species used in the paper “Escherichia coli "Marionette" strains with 12 highly optimized small-molecule sensors” by Meyer et al. (2019) which characterised certain inducible promoters. Looking at the data from the paper, we have chosen the 2 activators (araC, cinR) and 1 repressor (vanR^AM) with the best dynamics i.e. low leakiness and wider dynamic range. Although araC has good dynamics, it is natively produced by wild type E. coli. As it is difficult to get hold of araC knockout strain, we have decided not to araC as X in our final circuit as araC is not orthogonal to the cell’s native system.

In our circuit, we utilise mVenus instead of GFP as our reporter because the maturation time of mVenus is not affected by temperature. As we want to do a global perturbation with temperature, GFP will not mature quick enough to give us a reliable measurement for global perturbation with temperature at 30 degrees.

The terminators we are using for our circuit are double terminators coming from the paper “Precision design of stable genetic circuits carried in highly-insulated E. coli genomic landing pads”. These terminators are ultrastrong double terminators that terminate transcription in both directions. This is important as during the codon optimisation process when ordering new parts, there is a chance that a reverse promoter can be produced somewhere along the genetic circuit. Therefore, it is important to be able to also terminate these reverse transcription if they happen. We choose from the paper, the 3 strongest double terminator (DT3, DT5 and DT54) in both forward and reverse direction to use for our project. DT5 is the strongest in the forward direction amongst all the terminators in the paper while DT3 in second strongest in the forward direction. However, DT3 is stronger in the reverse direction compared to DT5. DT54 is considerably weaker than DT3 and DT5 in the forward direction. Therefore, after consideration, we have decided to use DT3 and DT5 only for our main antithetic circuit and DT54 for module D of our perturbation circuit. In order to avoid the risk of recombination, we use different terminator sequences for each module.

Open loop

To show adaptation strategies, we have designed an open loop circuit to compare with in response to perturbation. To make it comparable with the closed loop, we use the same reporter compared to the closed loop in the open loop system. For the repression case (sinR), the circuit contains psinR, RBS B0032m, CDS VanRAM and MegaT in module A and J23106, RBS B0032m, sinR and DT3 in module B. This is an open loop system because as VanR^AM is produced, it is not feedback into the circuit. Under constitutive expression, sinR will be produced which will act on psinR. To perturb the open loop circuit, we make use of module C and D. To perturb the circuit, Ptet*, RBS B0032m, CDS sinR and DT5 is cloned into module C and J23106, RBS B0032m, tetR and Terminator is cloned into module D. Without aTc, tetR will suppress ptet* so the perturbation circuit will not be active. However, when aTc is added, the repression of tetR on Ptet* will be lifted, leading to an increased amount of sinR which acts on psinR, perturbing the system.
For activation (ECF20 and ECF16), the circuit architecture of the open loop circuit is the same but sinR is replaced by ECF20 or ECF16, psinR with p_20992 or p_16332 and vanR^AM with cinR^AM. The following diagram shows the design of the open loop circuits.

Characterisation of controller species

The use of sequestration species where they act as activators and repressors on promoters is a crucial part of our project. The sequestration species - ECF20/AS20 (activator), ECF16/AS16 and sinR/sinI (repressor) are being characterised with the following plasmid inside a Marionette wild type strain (E. Coli MG1655) in EZRDM medium. The use of MG1655 strain is because the promoters are well characterised and the E. Coli intrinsically have TetR and araC being produced which is necessary for the inducible promotor to work.

For Module A, we are putting in P_z1, RBS (B0032), mVenus and DT5. For Module B, we are putting in P_Bad wt (from CIDAR MoClo), RBS (B0032), Z₁ and DT3. For Module C, we are putting in P_tet*, RBS (B0032), Z₂ and DT54. For Module D, a placeholder terminator is placed as JUMP assembly allows only for 4 module integration.

Characterisation of Z₁

To characterise Z₁ only, (sinR, ECF20 and ECF16), we add arabinose so Z₁ will be produced. The presence of tetR in E. Coli MG1655 suppresses P_tetR*, so Z₂ (sinI, AS20 and AS16) is not produced. If Z₁ is ECF20 or ECF16, the addition of arabinose will increase the expression of ECF20 or ECF16 which will activate p_ECF20 or p_ECF16 respectively and produce increasing amounts of reporter, therefore expecting an increasing sigmoidal curve as arabinose concentration increases. If Z₁ is sinR, the addition of arabinose will increase the expression of sinR which will repress p_sinR and produce reducing amounts of reporter, therefore expecting a decreasing sigmoidal curve as arabinose concentration increases.

Characterising the annihilation of Z₁ and Z₂

To characterise the annihilation of Z₁ and Z₂, we add aTc to the system which will lift the inhibition of TetR on P_tetR*. For sinI/sinR system, as aTc is added, sinI is being produced which will annihilate with sinR produced under the p_BAD promoter. With less free sinR available to repress P_sinR, the reporter level will increase as more aTc is added. For the ECF20/AS20 and ECF16/AS16 system, as aTc is added, AS20 or AS16 is being produced which will annihilate with ECF20 or ECF16 produced under the p_BAD promoter. With less free ECF20 or ECF16 available to activate p_ECF20 or p_ECF16, the reporter level will decrease as more aTc is added.

References

1. Marcos Valenzuela-Ortega, Christopher French, Joint universal modular plasmids (JUMP): a flexible vector platform for synthetic biology, Synthetic Biology, Volume 6, Issue 1, 2021, ysab003, https://doi.org/10.1093/synbio/ysab003


2. Aoki SK, Lillacci G, Gupta A, Baumschlager A, Schweingruber D, Khammash M. A universal biomolecular integral feedback controller for robust perfect adaptation. Nature. 2019 Jun;570(7762):533-537. doi: 10.1038/s41586-019-1321-1. Epub 2019 Jun 19. PMID: 31217585.


3. Rhodius VA, Segall-Shapiro TH, Sharon BD, Ghodasara A, Orlova E, Tabakh H, Burkhardt DH, Clancy K, Peterson TC, Gross CA, Voigt CA. Design of orthogonal genetic switches based on a crosstalk map of σs, anti-σs, and promoters. Mol Syst Biol. 2013 Oct 29;9:702. doi: 10.1038/msb.2013.58. PMID: 24169405; PMCID: PMC3817407.


4. Newman JA, Rodrigues C, Lewis RJ. Molecular basis of the activity of SinR protein, the master regulator of biofilm formation in Bacillus subtilis. J Biol Chem. 2013 Apr 12;288(15):10766-78. doi: 10.1074/jbc.M113.455592. Epub 2013 Feb 21. PMID: 23430750; PMCID: PMC3624457.


5. Meyer AJ, Segall-Shapiro TH, Glassey E, Zhang J, Voigt CA. Escherichia coli "Marionette" strains with 12 highly optimized small-molecule sensors. Nat Chem Biol. 2019 Feb;15(2):196-204. doi: 10.1038/s41589-018-0168-3. Epub 2018 Nov 26. PMID: 30478458.


6. Park Y, Espah Borujeni A, Gorochowski TE, Shin J, Voigt CA. Precision design of stable genetic circuits carried in highly-insulated E. coli genomic landing pads. Mol Syst Biol. 2020 Aug;16(8):e9584. doi: 10.15252/msb.20209584. PMID: 32812710; PMCID: PMC7436927.