Overview

Introduction of complex regulatory circuits in mammalian cells requires the availability of large number of designable orthogonal switches. We provide this tool to the synthetic biology community by creating a new type of bistable toggle switch based on designable TAL regulators.

We created several TAL repressors, activators and their reporter plasmids with their corresponding DNA-binding sites and promoters. The TAL repressor constructs exhibited over 90% repression and TAL activator constructs enhanced expression over 1500 fold.

We tested five induction systems for mammalian cells and adapted three of them for driving TAL regulators in the switch.

We designed and modelled a mutual repressor toggle switch based on two TAL repressors repressing each other.

Experimental results demonstrated low stability of the basic mutual repressor switch, based on the classical toggle switch topology - this was confirmed by modeling analysis, which indicated high sensitivity to leaky gene expression and a requirement for high cooperativity of TAL repressors in order to exhibit bistable behavior.

We designed an improved toggle switch that included additional positive feedback loops based on orthogonal TAL repressors and activators.

Models demonstrated higher robustness in comparison to the mutual repressor switch and exhibited bistability even without cooperative binding to DNA, since the positive feedback loop introduces nonlinearity.

Experimental results demonstrated stable states at induction with corresponding inducers and a clear bimodal distribution of fluorescence. The switch remained in a stable state after removal of inducers, which confirmed the epigenetic bistability of our system.

Our project is based on production of biopharmaceuticals by engineered cells and with tight but versatile regulation and safety in mind, we set out to contribute some fundamental advances to synthetic biology.

We considered two options of regulating production of therapeutic proteins in mammalian cells:

• The first would be a prosthetic network where the cells are engineered to sense endogenous signals such as for example blood glucose levels. This signal would control the production of a therapeutic protein (such as insulin), which in turn affects the level of the endogenous signal (i.e. glucose) and is thus regulated by a kind of a feedback loop. This type of system can replace the function of a defective tissue or organ, hence the name prosthetic. Although a very attractive option, such a system would need to be tailor-made for a specific disease.
• For the second option we regarded a cellular genetic network that produces the therapeutic protein of choice, but can be controlled by an external signal such as small molecular drugs or metabolites which can be administered orally or topically. Instead of requiring a continuous presence of an activating or repressing signal, the system should function as a bistable or multistable switch, requiring only a short signal pulse to change into any selected state. This type of a toggle switch has already been implemented in mammalian cells based on prokaryotic transcription factors fused to the eukaryotic transactivator (eg. VP16, VP64) or transrepressor (eg. KRAB) domains (Carlsson et al., 2012).

We consulted several physicians concerning different potential therapeutic applications of such a delivery system. As examples of medical problems we considered hepatitis C infection, wound healing and cardiovascular ischemia.

Physicians preferred the ability of the device to be controlled by the therapist rather than autonomous switching by sensing internal signals. An additional advantage of a controlled switch is that a very similar device core could be used for different therapeutic applications.

Synthetic biology has already demonstrated the ability of a genetic device to assume many different states. However, while prokaryotic DNA-binding domains appear to be the simplest tool for device construction, their number is limited and currently does not allow scalability to construct very complex logical functions. On the other hand, modular protein DNA-binding domains are known that can be designed to bind to almost any DNA sequence, providing high orthogonality.

Therefore the challenge was to prepare an epigenetic switch based on designable DNA binding domains, such as zinc finger proteins or TAL effectors that can be designed to recognize a selected DNA sequence. This would allow simultaneous introduction of several bistable or more complex switches into mammalian cells for advanced applications. To our knowledge, a designed DNA binding protein-based toggle switch has not been implemented experimentally so far, either in mammalian or in bacterial cells. This endeavor has to be supported by modeling to evaluate which type of genetic switch topology has the best performance, is more robust and which parameters are most critical for the performance.

Epigenetic toggle switches have, however, been implemented using natural bacterial transcription factors (Gardner et al., 2000). An epigenetic toggle switch is a genetic network, designed to memorize its state and maintain expression of a gene of choice even after the inducer has been removed (Cherry & Adler, 2000). Only when a second signal (inducer) is provided to the cells, the system will switch to a different state, shutting off expression of the first protein and/or initiating production of a second one. This type of a toggle switch allows the system to assume two discrete states, e.g. no production or production of an effector at a preset level, or selection between the production of two different protein effectors. A bistable toggle switch or combinations of several switches would permit selection of two or several therapeutic regimens, respectively.

For the switches to function independently of each other or independent of other cell processes, they need to be based on orthologous DNA-binding domains. We therefore decided to construct the genetic core of our device from TAL effectors, whose code of DNA recognition has been deciphered recently (Boch et al., 2009). TAL effectors have modular structure, with each module recognizing a single base pair. TAL effectors, fused to FokI nuclease, were quickly adopted as highly specific and easily customizable genome editing tools (Christian et al, 2010; Bogdanove et al., 2011, Li et al., 2011). For synthetic biologists, TAL effectors represent perfect candidates for engineering artificial transcription factors due to their high specificity, modularity, ability to target virtually any sequence and last but not least, they do not exhibit toxicity when expressed in cells, except in the case that they target an essential endogenous gene. Indeed, in addition to endonucleases, transcription factors employing TALs as DNA binding domains have also been reported recently: transcription activators were constructed by fusing TALs to the mammalian activation domain VP16 or VP64 (Zhang et al., 2011) and, during the initial stages of our iGEM project, transcription repressors were reported, based on fusion of TALs with a KRAB heterochromatin silencing domain (Garg et al., 2012).