Synthetic biology and computer science are two intertwined disciplines. This is be- cause bioengineered cells can display the same characteristics as micro-/nanocomputers. Specific algorithms are associated with different cellular mechanisms, and genetic pro- gramming becomes feasible by changing modular subunits within molecular devices[218,
185,123, 6]. To exemplify this approach, we use the analogy between a simple electri- cal circuit and a light production device in bacteria (Figure 1.2). On the electric board, if the switch is turned on, the light bulb should receive current generated by the bat- tery and start emitting light. If no light is observed, then it is easy to either change the light bulb or the battery, which are the most likely causes of fault. In synthetic biology, we build modular devices that can be adapted to reproduce similar systems. There should thus be known components and methodological controls that allow the troubleshooting of any uprising issue (no light emission protein observed for instance). Since individual cells are the tiny batteries powering genetic devices, it is important to grow them in optimal conditions in order to provide these circuits reliable environmen- tal conditions. Genetic modification of specific circuit transcription or translation units can then be compared to replacing a light bulb, but at the molecular level. Therefore, there are key parameters that allow for the construction of stable biological devices, and optimising one element is analogous to one or multiple cycles through the devel- opment of synthetic circuits presented in Figure 1.1. By the end of this optimisation process, biological devices should be thoroughly tested, well-characterised and show a good response to changes in the environment.
1.1.1
Cellular stochasticity
When studying genetic circuits in a cellular context, there are a lot of processes oper- ating at the same time as the specific device functions. Therefore, there are constant perturbations within cells that restrict our ability to obtain precise parameters over synthetic circuit behaviour[167]. Without the use of external methods of analysis, bi- ological research would be like looking for a needle in a haystack. Fortunately, the development of characterisation methods for single cell or population scale measure- ments facilitated our understanding of biological processes. As displayed in Figure 1.2B, one can use intrinsic biological system properties to activate light emission in bacteria. Actually, this is a common approach in synthetic biology, where a reporter
1.1. Design in synthetic biology 5
FIGURE1.2: (A) A simple classic electrical system lightling up a bulb upon activation of a switch if wired through a power supply (battery). (B) A generic biological circuit, or plasmid, replicating itself with a replicon and selected for with a specific marker. This plasmid drives the expression of a protein of interest (fluorescent reporter F) through transcription (from the promoter to the terminator) and translation (via ribosomes recognising a
ribosome binding site, or rbs).
provides variable illumination given different cellular states. Since bacteria are in- dividual microscopic organisms, cellular signals often need amplification to be per- ceived, and the use of fluorescent reporters allows an easy detection of physiological changes via a range of qualitative and quantitative methods[112]. External tools thus enable us to analyse and to interpret changes invisible to the naked eye, but resulting from significant biological activity performed at the microscale. These physiological changes later need to be compared to the proposed theoretical model. Nevertheless, the background noise behind experimental data may sometimes impair with the use of automated analysis methods, and it makes the matching of experimental data to computational models a trickier process.
Bacteria represent a stochastic environment: cellular responses may vary from ex- periment to experiment, and sometimes lead to false-positive and false-negative re- sults. This highlights the importance of performing biological replicates in order to obtain meaningful data. However, the complementation of biological studies with in silico resources usually facilitates the identification of true-positives. For instance, fluo- rescence measurements derived from in vivo experiments can be analysed, modelled by mathematical models and compared to theoretical results in order to improve genetic devices[176,128]. Generally, this is achieved with a low modelling level of abstraction, better to model biological processes, that is based on simulating the main features of the central dogma of biology for a specific function (cf. Chapter 2). In practice, this
allows us to overcome the noise imposed by millions of surrounding molecules, and to truly characterise a device function. Yet, the modelling process of synthetic constructs is sometimes context specific, and may not provide an exact prediction of the observed physiological changes in different conditions. Thus, the documentation of in silico re- sources linked to biological devices should account for specific features, which allow the easier characterisation of novel circuits.
1.1.2
Host/circuit compatibility
In the microscopic world, any individual that does not fit the environment is quickly eliminated by natural selection. In all biological assays, cells can sustain a specific metabolic load, which is directly connected to their growth profile. Therefore, in syn- thetic biology, any defect in bacterial growth is usually synonym of an increased metabolic load, resulting from the functionality of a genetic circuit creating crosstalk between host and specifically encoded elements[79, 185]. Back to the analogy between genetic and eletrical circuits presented in Figure 1.2, leaving the light on at all times would likely reduce the life time of a light bulb, besides leaving a salty bill. To avoid this sit- uation at the molecular level, we thus try to limit the impact that genetic devices may display on cellular metabolism, since it is the same metabolism that is also responsible for the functionality of a given construct. Hence, in the development of any biological circuit, it is of upmost importance to be aware of its potential metabolic load, and to control the impact it may create on other specific cellular processes.
So far, we have provided an overview of biological circuit design in synthetic bi- ology, and explained how cells may be programmed to perform different kinds of actions. Differential behaviour is obtained by the setup of an ordered nucleotide se- quence that contains regulatory regions, required for a specific biochemical phenomenon to take place in vitro/in vivo. However, many genome and DNA sequences are of un- known function, and some may be deleted without altering the bacterial fitness. In contrast, attempting to remove other unknown regions may also be detrimental to the overall metabolism. Although, per se, the succession of individual nucleotides may not provide direct information about its function, it still follows generic patterns that are recognised and used by the cellular machinery. Therefore, a possibility to minimise the potential for interactions between synthetic circuits and host metabolism is to design bioorthogonal circuits[129]. This consists of engineering a nucleic acid sequence that displays minimal homology with the elements of nature, making it biologically inert. In this area of synthetic biology called DNA programming, nanoengineering methods must account for context specificity (where a circuit evolves), and circuit functionality
1.2. Thesis problem statement and objectives 7