Positive feedback loop switch

We designed an upgraded bistable genetic toggle switch based on orthogonal TAL repressors and activators which is composed of a pair of mutual repressors and a pair of activators,upgradingthe classical toggle switch with positive feedback loop. Simulations of the positive feedback loop switch demonstrated bistability even at low or no cooperativity. Stochastic and deterministic simulations indicate higher robustness in comparison to the mutual repressor switch. We experimentally tested the switch by monitoring production of two fluorescent protein reporters, confirming a clear bimodal distribution of reporter fluorescence and demonstrated adoption of stable states by induction with corresponding inducer molecules. The switch persisted in a stable state after the removal of inducer molecules, which confirmed the epigenetic bistability of our system. (NEW) Cells could switch from one state to the other by the addition of the second inducer, which was confirmed by two methods.

Bistable genetic switch based on non-cooperative elements

Mathematical analysis of genetic switches from the literature indicates that cooperativity, which introduces a nonlinear response, is required for a functional bistable switch, consisting of two mutual repressors (Cherry et al. 2000). Macia et al. (2009) and Widder et al. (2009) proposed that bistability could be introduced by non-cooperative elements, when nonlinearity is introduced for exampleif protein A is able to repress the transcription of protein B and at the same time activate its own transcription and vice versa (Figure 1).

Figure 1. Macia et al. (2009) and Widder et al. (2009) proposed a mathematical solution for creation of a bistable switch that would not require the cooperative binding of transcriptional regulators. According to the proposed model, transcription factor A should activate its own transcription and repress the transcription of gene B and vice versa.

We, as molecular biologists, werenot aware of a transcriptional effector or effector pair that could act simultaneously as a repressor and activator, therefore we were not surprised that, to our knowledge, this type of bistable switch has not yet been experimentally implemented.

The crucial moment in our project was when we realized that the ability to design TAL repressors and activators directed against the same binding site could offer a solution to this problem and provide a unique opportunity to construct orthogonal bistable switches based on noncooperative elements (Figure 2).

Figure 2. A repressor (left) and an activator (right) competing for the same binding site.

Results

Design

We designed an upgraded mutual repressor switch, introducing two additional positive feedback loops (Figure 3), consisting of two TAL activators targeted against the same binding sites as apair of mutual TAL repressors. In other words, rather than having the same protein function as an activator and repressor, we used an activator and repressor pair, competing for the same operator. The same binding sequence for the activator and repressor introduced nonlinearity required for the bistability based oncompetition for the binding site. For the purpose of our project, we designed a bistable switch with a positive feedback loop, capable of switching between the two states through regulation by inducer molecules (Figure 4).

Figure 3. Scheme of the genetic circuit for the bistable toggle switch composed of a pair of mutual repressors and a pair of autoactivators under the control of two operators.

Figure 4. Designed states of the bistable toggle switch with a positive feedback loop. (A) Addition of inducer 1 triggers dissociation of its cognate repressor from its DNA-binding site. As a consequence, repressor A and activator B are transcribed. Activator B further activates transcription of repressor A and itself, which forms the positive feedback loop of the switch state one. Meanwhile repressor A inhibits the other state of the switch. (B) When the inducer is removed, the inducible repressor binds back to its DNA-binding site, but since activator B is still present, autoactivation is achieved, resulting in a stable state. (C) and (D) depict the switch switched to state two and the function of its other positive feedback loop.

Modeling

Before proceeding with experimental verification, we performed a thorough modeling analysis of the designed switch. We incorporated parameters obtained from the repression and activation experimental results into our simulations. We also implemented a new modular, hybrid modeling algorithm that introduced stochasticity into an otherwise deterministic approach and enabled us to explicitly model competitive transcription factor binding and a limited number of binding site repeats. Modeling is described in details in the modeling section. Both deterministic and stochastic simulations demonstrate that the positive feedback loop switch is significantly more stable than the mutual repressor switch. Most importantly, it can exhibit bistability even without cooperativity (Figure 5). This switch is also more robust than the simple mutual repressor switch in regard to leaky expression.

Figure 5. Robustness of the positive feedback loop switch. The positive feedback loop switch exhibited bistability even in the absence of cooperativity and for the minimal promoter's leaky transcription rate of 10% or constitutive promoter leaky transcription rate of 5%. In comparison, the same leaky transcription rate (with equal values of all other parameters) of 5% caused the mutual repressor switch (without the positive feedback loops) to exhibit no bistability unless high cooperativity (2 or more) was introduced into the model.

Construction and experimental testing of the bistability of the switch

For experimental implementation of the positive feedback loop toggle switch we introduced the following components into cells (Figure 6): a pair of TAL:KRAB repressors (TALA and TALB), controlled by the opposite TAL (TALA controls the transcription of TALB and TALB controls the transcription of TALA), exactly as in the mutual repressor switch, a pair of TAL:VP16 activators (TALA and TALB), each activating its own transcription (autoactivator) and transcription of the opposite TAL repressor , two of the constructs were tagged with fluorescent reporter proteins (BFP and mCitrine) via a t2A sequence, which ensured the equimolar production of the fluorescent reporter and TAL regulator, Both TAL repressor and activator pairs (A and B), controlled by inducible repressors, Constitutively expressed inducible repressor constructs, Inducer molecules (pristinamycin and erythromycin) .

Figure 6. Functional components (operons and inducers) of the positive feedback loop switch. .

We analyzed the performance of the switch by using two fluorescent proteins with a sufficientspectral distance (BFP and mCitrine), which enabled easy detection and quantification by confocal microscopy and flow cytometry.

To analyze the bistability we first used flow cytometry, a technique which has a unique ability to determine the number of cells expressing either one or both of the fluorescent reporter proteins. Although cells were transfected with the complete switch device including both reporters, the analysis of cells demonstrated clear bimodal distribution - the majority of the transfected noninduced cells expressed only one of the two fluorescent proteins.The expression of one or the other reporter proteins is most likely a result of stochastic events or possibly a slight imbalance of the amount of transfected plasmids (Figure 7A).Importantly, a very low number of cells expressed both reporters. This bimodal distribution of fluorescence clearly demonstrates the intrinsic bistability of our system in comparison to the mutual repressor switch (classical toggle) topology, where a large fraction of cells expressed both fluorescent reporters. The addition of either one of the inducers switched the reporter production towards the corresponding fluorescent protein (Figures 7B and 7C).

Figure 7. The bistable switch with a positive feedback loop exhibits a bimodal distribution of fluorescence. (A) Non-induced HEK293 cells transfected with plasmids forming the switch (see Figure 6) transcribe either BFP or mCitrine as determined by flow cytometry. (B) HEK293T with the switch plasmids induced with erythromycin (2 mg/L) express mainly mCitrine and (C) cells induced with pristinamycin (2 mg/L) express BFP. Samples were analyzed five days after induction. NOTE: A high fraction of cells in the lower left quadrant, typically between 50-70%,representsnontransfected cells.This is a typical fraction in mammalian cell transfection experiments and will be solved for the therapeutic use by the preparation of stable cell lines, whose selection however typically takes several months.

Confocal microscopy confirmed high expression of the expected and no expression of the opposite fluorescent reporter protein in both induced states (Figure 8). This means the addition of an inducer shifts cells to a corresponding state which is preserved even after the inducer has been removed (Figure 9). The system remained in a stable state several days after the removal of the signal, which further confirms the epigenetic bistability of our positive feedback loop switch.

Figure 8. Positive feedback loop switch exhibiting two different states at induction with pristinamycin (PI) and erythromycin (ER) inducer molecules. HEK293T cells were cotransfected with the following plasmids : PCMV_mCherry transfection control (20 ng), [A]_PMIN_TALB:KRAB, [A]_PMIN_TALA:VP16_t2A_BFP, [B]_PMIN_TALA:KRAB_t2A_mCitrine, [B]_PMIN_TALB:VP16, (all 5 ng), PCMV_[PIR]_TALB:KRAB, PCMV_[PIR]_TALA:VP16, PCMV_[ETR]_TALA:KRAB, PCMV_[ETR]_TALB:VP16, (all 10 ng), PCMV_PIP:KRAB, PCMV_E:KRAB (both 200 ng). Pristinamycin and erythromycin were added to final concentration of 2 µg/ml. Fluorescence was measured 3 days after induction.

Figure 9. Positive feedback loop switch exhibiting stable states at removal of inducer molecules. HEK293T cells were cotransfected with the following plasmids: PCMV_mCherry (20 ng), [A]_PMIN_TALB:KRAB, [A]_PMIN_TALA:VP16_t2A_BFP, [B]_PMIN_TALA:KRAB_t2A_mCitrine, [B]_PMIN_TALB:VP16, (all 5 ng), PCMV_[PIR]_TALB:KRAB, PCMV_[PIR]_TALA:VP16, PCMV_[ETR]_TALA:KRAB, PCMV_[ETR]_TALB:VP16, (all 10 ng), PCMV_PIP:KRAB, PCMV_E:KRAB (both 200 ng). Pristinamycin and erythromycin were added at final concentration of 2 µg/ml. Medium was replaced 3 days after induction and fluorescence was measured 3 days after removal of inducers.

(NEW) Switching between the two states

The ultimate test of the functionality of the bistable switch with a positive feedback loop is to switch the system between the two states by the addition of the second inducer. This is a longer experiment as it requires the system to settle first into the initial selected state by the addition of the first inducer and then switch into the second state after the addition of the second inducer. Nevertheless we managed to demonstrate this functionality of our switch during the time between the regional and world jamboree using a secretory luciferase assay and fluorescence microscopy.

Results

After the induction of state I with pristinamycin we observed expression of reporter I, i.e. blue fluorescent protein (BFP) (Figure 10). Two days after the induction with pristinamycin, inducer was removed and replaced with erythromycin. We observed expression of the reporter mCitrine, an indicator of state II of the switch only one day after inducer exchange. We can still see some BFP emission in the next few days due to a relatively long half-life, but it is clear that mCitrine expression eventually prevails.

Figure 10. Switching from state I to state II determined by confocal microscopy. Expression of mCitrine was induced only after the addition of erythromycin. HEK293T cells were transfected with PCMV_mCherry (20 ng), [A]_PMIN_TALB:KRAB, [B]_PMIN_TALA:KRAB_t2A_mCitrine (10ng), [A]_PMIN_TALA:VP16_t2A_BFP, [B]_PMIN_TALB:VP16, (2 ng), PCMV_[PIR]_TALB:KRAB, PCMV_[ETR]_TALA:KRAB (20 ng), PCMV_[PIR]_TALA:VP16, CMV_[ETR]_TALB:VP16, (5 ng), PCMV_PIP:KRAB, PCMV_E:KRAB (200 ng). Pristinamycin (2 µg/ml) was added to the cells to induce state I. After two days pristinamycin was removed and replaced with erythromycin to switch to state II. Each day images of cells were recorded with a confocal microscope. mCherry was used as a transfection control.

Switching was also demonstrated using a cytosolic firefly luciferase (fLuc) and a secreted alkaline phosphatase ( SEAP) as state I and state II reporters. Secretory Renilla luciferase was used for normalization. Cells were induced with either erythromycin (2 mg/L) to trigger expression of fLuc or with pristinamycin (2 mg/L) to trigger expression of SEAP. Inducer I was removed after one day and replaced with inducer II. Expression of reporters was analyzed 3 and 5 days after inducer exchange. We observed reduced expression of firefly luciferase and elevated expression of SEAP when pristinamycin was replaced by erythromycin and vice versa when the cells were first stimulated with erythromycin and then with pristinamycin (Figure 11).

Figure 11: Switching monitored by the activity of SEAP and firefly luciferase. HEK293T cells were co-transfected with plasmids of the switch and10x[B]_PCMV_fLuc and 10x[A]_CMV_SEAP reporter plasmids. Left, switch from pristinamycin to erythromycin state. Right, switch from erythromycin to pristinamycin induced state.

Our system has considerable inertia as first the transcriptional factors must be degraded as well as the reporter proteins.The degradation could be accelerated by the addition of PEST degradation signals, which we already anticipated and prepared parts with TAL regulators that include pest signals. These regulators exhibit increased fold induction but have not yet been tested within the switch context.

References

Cherry, J., L., Adler, F., R. (2000) How to make a Biological Switch. J. Theor. Biol. 203, 117-133.

Cong, L., Zhou, R., Kuo, Y.C., Cunniff, M., Zhang, F.(2012) Comprehensive interrogation of natural TALE DNA-binding modules and transcriptional repressor domains.Nat Commun.3,968.

Macía, J., Widder, S., Solé, R. (2009) Why are cellular switches Boolean? General conditions for multistable genetic circuits. J. Theor. Biol. 261, 126-135.

Widder, S., Macía, J., and Solé, R. (2009) Monomeric Bistability and the Role of Autoloops in Gene Regulation. PLoS ONE 4, e5399.

---

Next: Controls >>