Abstract
Lux-type quorum sensing systems enable communication in bacteria with only two protein components: a signal synthase and an inducible transcription activator. The simplicity of these systems makes them a popular choice for engineering collaborative behaviors in synthetic bacterial consortia, such as photographic edge detection and synchronized oscillation. To add to this body of work, we propose a pulsatile communication circuit that enables dynamic patterning and long-distance communication analogous to action potentials traveling through nerve tissue. We employed a model-driven design paradigm involving incremental characterization of in vivo design candidates with increasing circuit complexity. Beginning with a simple inducible reporter system, we screened a small number of circuits varying in their promoter and ribosomal binding site (RBS) strengths. From this candidate pool, we selected a candidate to be the seed network for the subsequent round of more complex circuit variants, likewise variable in promoter and RBS strengths. The selection criteria at each level of complexity is tailored to optimize a different desirable performance characteristic. By this approach we individually optimized reporter signal-to-background ratio, pulsatile response to induction, and quiescent basal transcription, avoiding large library screens while ensuring robust performance of the composite circuit.
1 Introduction
Natural quorum sensing circuits implement positive feedback to establish bistable control over quorum sensing components. The feedback is mediated by the quorum sensing transcription factor, which activates transcription from quorum sensing promoters when bound to its cognate acyl-homoserine lactone (AHL) signal molecule. Activated transcription factor promotes synthase expression and, in turn, AHL accumulation. This feedback, along with rapid diffusion of AHL, ensures two stable states: one with only basal expression of quorum sensing-related expression and the other with high, saturated expression of quorum sensing components and maximal synthesis of AHL [11, 10]. In this paradigm, a quorum sensing circuit essentially contributes one bit of information: high or low density.
Augmenting the natural quorum sensing circuit with a negative feedforward arm forms an incoherent feedforward loop (IFFL), a well-studied circuit family capable of a variety of behaviors such as bistability, oscillation, and pulse generation [9, 3, 2]. Groundbreaking work in synthetic synchronized oscillators has established that, in accordance with the Turing model of reaction-diffusion systems, extending the spatial domain relative to the diffusion coefficient results in qualitatively different behaviors [13, 6]. Namely, a circuit that produces synchronized oscillations over small spatial domains will behave as a wave generator in larger arenas. In demonstrations involving extended arenas, it was shown that a single cluster spontaneously became the source of signaling activity for the entire community. This cluster emits, at regular intervals, signal packets traveling away from their source in all directions. Taking inspiration from these results, we aim to apply a similar IFFL gene circuit in a pulsatile communication circuit. Our goal is to tune the transcriptional balances such that there exists stable a low-expression state of the circuit, yet it will emit a single pulse in response to increased signal chemical concentration. This system would enable cells to initiate and transmit communications without permanently altering their internal transcriptional state or spontaneously generating signals. In this construction, a single quorum sensing system emulates a serial port, where a single channel may transmit more than one bit of information.
We hope that work on this system will open the door to more distributed logic and pattern formation applications in synthetic biology. Previous work has shown that dynamic transcriptional programs and heterogeneous populations may allow for more informative chemical event detectors, embedded control of cell growth, and photographic edge detection [5, 1, 12]. One of our goals for this research is to build an E. coli strain capable of detecting the temporal coincidence of spatially separated chemical events. As mentioned above, when used as a serial port, a signaling channel may communicate more than one bit of information. Specifically, a signal pulse generated in response to a single event carries information on the time of event in addition to the occurrence of the event itself. In a scenario with two such circuits, based on orthogonal quorum sensing components, the collision of pulses indicates that they were both emitted within a window of time that is dependent on the size of the consortia and the transmission velocity of the pulses.
2 Circuit design and modeling
The proposed circuit is an incoherent feedforward loop composed of a Lux-type quorum sensing system augmented with a negative feedforward arm. Lux-type quorum sensing systems consist a transcription factor, a signal molecule, and synthase enzyme, and a transcriptional repressor (Fig. 1). In this system, the quorum sensing transcription factor is expressed constitutively and, when bound to its cognate quorum sensing molecule, activates expression of the repressor and signal synthase. One regulatory loop is formed by the quorum sensing components, a positive-feedback loop wherein the quorum sensing signal molecule promotes its own production via transcriptional activation of its associated synthase protein. The transcriptional repressor acts on the synthase's promoter region, forming negative feedforward arm which is delayed relative to the positive feedback arm.
Numerical simulations of the circuit show that it is capable of producing synchronized oscillations, bistability, or pulse transmission, among other behaviors (Fig. 2). Indeed, similar circuit architectures been applied successfully in synchronized oscillators [13, 3]. However, these behaviors are sensitive to variations in parameter values and accurately predicting transcriptional and enzymatic parameters from literature is haphazard [7]. To quickly search through design space, we build and characterize the circuit incrementally.
3 Model-guided design and experimental characterization
In our approach, we translate the reaction-diffusion model to a well-mixed solution model. Then, we perform liquid-culture in vivo experiments of many candidate gene circuits and fit the data to the well-mixed model. At each step, we define an optimality measure used to select an optimal candidate from the pool of candidate circuits to seed the following round of screening. In the following round, the seed circuit is augmented with variants of an additional component and the process is repeated.
Different optimality measures are required at each step. In the first round, we characterize variants of the constitutive quorum sensing transcription factor source and promoter variants of a fluorescent promoter and select for the highest inducible range of reporter expression. We add the repressor during the second round and select for the candidate whose maximal expression rate is largest with respect to its fully-repressed expression rate. In the final round, we add the synthase and select for the candidate that is quiescent when left alone and has a pulsatile response to externally provided AHL.
4 Experimental Results and Discussion
To date, we have experimentally characterized variations on the simplest circuit, an AHL-inducible fluorescent reporter. We compose these circuit candidates using two plasmids. One bears an inducible fluorescent protein and the other constitutively expresses a quorum sensing transcription factor. The reporter promoters are combinatorial, having activating operator sites for CinR binding and repressing operator sites for binding by either TetR or LacI. Experiments were performed using DH5a-Z1 E. coli as the host strain, which expresses both TetR and LacI from its genome. This way, cells began to express GFP at the beginning of the experiment, when aTc or IPTG were added.
The model predicts increasing intracellular GFP concentration at a constant rate during the first few hours of cell growth following induction by AHL along with IPTG or aTc. For each candidate, we determined its OD-normalized GFP expression rate in both 1μM AHL and AHL-absent media [3]. The difference between these rates is the inducible range of expression rate, which is the optimality measure at this stage. The results (Fig.3) show that J23106 B0032 CinR yields high inducible range when paired with either pCinTetO B0032 GFP or pCinLacO_m B0034 GFP reporters (Table 1). Two reporters genotypes were selected so that we may test both TetR and LacI as repressors in the second round of screening.
For the second round of screening, a transcriptional repressor is added to the circuit architecture and variants are screened for pulsatile response to induction. In our first attempt, the repressor gene was inserted immediately downstream of the constitutively-expressed CinR. Preliminary screening data from shows almost no reporter repression resulting from the addition of a transcriptional repressor. We know that the promoters are repressible, having observed that repressors expressed from the DH5a-Z1 genome effectively suppress reporter transcription. In future work, we will investigate whether compositional context negatively impacts repressor expression, and can be ameliorated by reorganizing the gene allocation.
In addition, we will investigate how employing plasmids with different origins of replication can improve the performance and repeatability of the circuit. During the first stage of screening, plasmids pairs had different resistance genes but the same origin of replication, pMB1. Transforming two plasmids with related replication origins may result in heterogeneous plasmid copy number. Future work will include testing plasmid pairs with compatible pairs of replication origins.
5 Acknowledgments
James M. Parkin is supported by the Institute for Collaborative Biotechnologies through grant W911NF-09-0001 from the U.S. Army Research Office. The content of the information does not necessarily reflect the position or the policy of the Government, and no official endorsement should be inferred. Plasmid vectors and non-coding regions were provided as a generous gift of Douglas Densmore at the Cross-disciplinary Integration of Design Automation Research lab (Addgene Kit # 1000000059). Quorum sensing promoters and coding sequences were provided as a generous gift from Matthew Bennet (Addgene Plasmid # 65954, 65952).