Synthetic biology and metabolic engineering are becoming increasingly vital to sustain as well as revolutionise our current ways of life regarding fossil fuel dependency, deteriorating biodiversity, environmental pollution, human and animal health and space exploration.
These biological engineering disciplines are propelled by the fast and efficient DNA readand-write technologies which are used to, e.g., generate immense libraries of microbial
cell factories (MCFs) to produce all kinds of natural and even new-to-nature molecules of interest in a sustainable manner. Despite these fast design and build tools, the classic
design-build-test-learn cycle for strain and product development is still impeded by the lack
of fast and efficient screening, selecting, control and quantification tools for the metabolites of interest.
Due to its intrinsic metabolic complexity, nature itself provides the necessary tools to overcome this obstacle in the form of transcriptional regulatory circuits. These ubiquitous circuits enable organisms to detect and respond to changes in intra- or extracellular small
molecule or ion concentrations and physical parameters. Because of this compelling ability, such regulatory circuits are being remodelled into transcriptional biosensors which enable the online, fast, efficient, specific, (simultaneous) and in vivo detection of (multiple) metabolites of interest. Consequently, transcriptional biosensors bear the potential to spur
and improve current synthetic biology and metabolic engineering strategies through key applications such as high-throughput screening, adaptive laboratory evolution (ALE) and
dynamic pathway control.
When developing biosensors for specific in vivo applications, the bio-engineer in casu
should have the ability to fully customise the two fundamental characteristics of biosensors, namely the response curve and molecular specificity. The response curve represents the input/output relationship of the biosensor and is determined by the sensitivity and operational and dynamic range. The molecular specificity profile determines how well a biosensor can distinguish between different, more or less related molecules. Biosensor engineering principles and strategies to customise these characteristics, however, are currently not satisfactory for the specific biosensors applications desired by the academic and industrial communities.
Therefore, novel biosensor engineering principles and strategies were developed, evaluated and put to the test in this doctoral research to enable the creation of tailor-made transcriptional biosensors to gratify any researcher’s needs.
As a case study, transcriptional biosensors were developed, customised and applied for monitoring and quantifying specific flavonoids in Escherichia coli. Flavonoids form a large
group of specialised plant metabolites which, due to their broad structural diversity (over 9000 identified compounds), demonstrate a wide variety of biological activities and consequently, convey compelling industrial applications. Therefore, flavonoid-responsive biosensors would be of great interest for the fast and efficient development and optimisation of (E. coli-based) MCFs towards the sustainable production of flavonoids. After a comprehensive review of current biosensor engineering approaches, a synthetic biosensor chassis
(pSynSens, Synthetic Sensor) was constructed based on the natural flavonoid-responsive regulatory circuit from Herbaspirillum seropedicae (FdeR-PfdeAR) and demonstrates compelling response curve characteristics towards naringenin, the central scaffold molecule in the flavonoid biosynthetic pathway. This pSynSens biosensor constitutes a biosensor architecture in which the detector and effector module were decoupled to enable independent and informative engineering efforts at each module. Due to this modularised architecture,
the pSynSens biosensor was used as the biosensor chassis for evaluating novel strategies to customise both the response curve towards naringenin and the specificity profile towards
three closely related flavonoids, naringenin, apigenin and luteolin.
The ability to tune the response curve of any biosensor towards the desired attributes is crucial for proper biosensor application. In the context of this doctoral research, the developed pSynSens biosensor chassis was engineered through a fast, module-based strategy to customise the naringenin-response curve. Through this strategy, the detector and effector module were targeted separately to generate two collections of each ten synthetic biosensor variants demonstrating a wide variety of response curve parameters. To facilitate the extensive characterisation and comparison of biosensor response curves, the available mathematical toolbox was expanded to incorporate a Noise parameter and application-oriented operational range calculation. Each collection demonstrates a unique diversity of response
curve characteristics spanning a 128-fold change in dynamic and 2.5-fold change in operational ranges and 3-fold change in levels of Noise. The established biosensor engineering
strategy as well as the novel biosensor collections contribute to future efforts of biosensor response curve customisation in general, to any desired flavonoid-responsive biosensor
application and as compelling alternatives for the commonly used LacI-, TetR- and AraCbased inducible circuits.
The immense variety of metabolites within a single organism, and even within a single heterologous biosynthetic pathway of interest, demands transcriptional biosensors with the
ability to respond to one specific metabolite of choice without the interference of other, closely related metabolites. In this doctoral research, two distinct chimera-based strategies
were developed and evaluated which target either the detector or effector module of the non-specific flavonoid-responsive pSynSens biosensor chassis to customise its specificity
profile. As a case study, stringent luteolin-specificity was aimed for and acquired by introducing specific parts of an E. coli-incompatible luteolin-specific regulatory circuit from
Sinorhizobium meliloti into the pSynSens biosensor chassis in different configurations at both the DNA and protein level. Through both strategies novel functional chimeric biosensor variants were generated with enhanced luteolin-specificity. While the original pSynSens biosensor chassis demonstrates only 27.5% luteolin-specificity, the chimeric biosensor variants demonstrate luteolin-specificities up to 95.3%. The developed chimera-based strategies provide a fast, modular and efficient route towards molecular specificity customisation in
biosensors. In addition, the biosensor chassis proved to be a robust framework to incorporate parts from otherwise E. coli-incompatible regulatory circuits. Finally, this biosensor
collection comprises both the first reported luteolin-specific biosensor as well as the first chimeric LysR-type biosensor circuits in the current biosensor repertoire.
To demonstrate the speed and performance of biosensor-driven metabolic engineering strategies, an MCF was developed for the de novo biosynthesis of naringenin from glucose by introducing and combinatorially balancing the heterologous naringenin biosynthetic pathway in E. coli. More specifically, this four-enzyme pathway was assembled by combining different enzyme variants with different promoter sequences (with differing promoter strengths) for each of the four enzymatic steps in a single-step assembly reaction. The resulting library of circa 4 12 different pathway expression plasmids was co-transformed in E. coli with one of the developed tailor-made naringenin-responsive synthetic biosensors, selected for its specific set of custom response curve characteristics best suited for this highthroughput screening approach. From this library, MCFs were identified which demonstrate various naringenin titers up to 35.6 mg/L. Moreover, the applied biosensor proved to be a
precise and accurate tool for the prediction of in vivo naringenin production, in addition to being an effective high-throughput screening tool. Next, additional metabolic engineering efforts were investigated towards the biosynthesis of the specialised flavonoid, luteolin.
To this end, specific (membrane-bound) plant enzyme variant collections were constructed and screened separately for the in vivo production of apigenin from naringenin and luteolin from apigenin. Here, two MCFs were identified producing 48.8 mg/L apigenin from naringenin and 1.39 mg/L luteolin from apigenin. Besides the effective demonstration of biosensor-driven metabolic engineering, all three produced flavonoids have compelling industrial applications with the naringenin-producing MCF bearing the potential towards the biosynthesis of a wide variety of specialised flavonoids.
In general, novel engineering principles and strategies were established and applied during this doctoral research which could be used to customise transcriptional biosensors for
a wide variety of molecules of interest and, therefore, assist in accelerating and augmenting the design-build-test-learn cycle in multiple biological engineering disciplines. Moreover, the current biosensor repertoire was expanded with a collection of synthetic flavonoidresponsive biosensors in E. coli demonstrating custom response curves and molecular specificity profiles and are now available to be used in key applications for the development and optimisation of flavonoid-producing MCFs. Finally, these biosensors were put to use for the successful development and optimisation of MCFs producing naringenin, the key metabolite in the large and industrially inciting class of specialised plant metabolites, i.e. flavonoids.