Development of a transcriptional regulator-based bioreporter : towards a generic selection method for novel enzymes

Enzymes are proteins that catalyse chemical reactions. They accelerate the reaction by lowering the activation energy, thereby allowing the equilibrium to be reached more quickly. The molecules upon which enzymes may act are called substrates and the enzyme converts the substrates into molecules known as products. Nearly all metabolic processes in a living cell need enzyme catalysis in order to proceed at rates fast enough to sustain life. In Chapter 1, the relevance of enzymes for society is described. This dates back to more than 8000 years ago, when people unknowingly already made use of enzymatic conversions via fermentation by whole-cell microorganisms to make early forms of bread and beer. In the last two centuries, our knowledge on how enzymes are functioning increased tremendously. They have remarkable features that make them interesting for industrial applications, such as a high specificity and selectivity, and they can make processes ‘greener’ by replacing often polluting or toxic chemical reactions. This resulted in the large scale industrial production and application of enzymes in diverse areas ranging from food industry and detergents to pharma and DNA technology. Currently, there is both the need and the room to increase the number of enzyme applications. A more extensive implementation of enzymes in industry is very important because, sooner or later, we have to make the shift from a fossil fuel based economy to a biobased economy. At this moment, only 5% of all chemical products are produced biologically. For the latter, only 150-170 of the 3000 different types of known enzymes are being applied. It is estimated that only 1% of enzymes is known and, therefore, there is much room to extend the number of industrial enzyme applications. But how to find novel enzymes for these applications? Enzymes can be obtained from nature (natural evolution, screening of metagenomics libraries) or by enzyme engineering via laboratory evolution (screening of random or semi-random enzyme variant libraries) or computational design (in silico generation and screening of enzyme variant libraries followed by experimental verification screens). Together these approaches comprise a multitude of methods to find or generate an enormous amount of genetic diversity. To obtain the desired variants from these large libraries it is essential to have an efficient screening method, in which phenotype and genotype are linked. However, since screening methods are often time-consuming, complicated and/or require expensive equipment, this screening step is the main bottleneck in obtaining novel enzymes. Reporter-based in vivo screening and selection is a new development in order to deal with the large numbers in screening for novel enzymes. This approach is extensively discussed in Chapter 2, including a comparison of other in vivo screening and selection strategies and the various reporter-based mechanisms. In reporter-based approaches, it is not the enzymatic conversion or its product that result in a measurable property, but rather a genetically encoded reporter that gives a discriminating phenotype. Since the enzyme activity is measured indirectly, this approach is independent of the reaction and thus general. A sensor part couples the enzyme activity to the reporter and, hence the reporter choice determines the detection output, e.g. bioluminescence or fluorescence. The cell containing these sensor and reporter components, and functions as a reporter, is called a whole-cell bioreporter or simply bioreporter. As sensor part, various biomolecules are possible, either protein- or RNA-based (riboswitches), but the most popular sensor is a transcriptional regulator. This transcriptional regulator very specifically binds the enzymatic product, which results in a conformational change that modifies its DNA binding capacity and switches on expression of a reporter gene. Consequently, the specificity of the sensor has to be modified for each product. Although the development of this and other reporter-based strategies takes time and effort, the many advantages make this a very powerful screening method. These advantages include: wide applicability, screening for enantioselectivity, signal enhancement, no need for artificial substrates and high-throughput screening. The most commonly used reporter-based screening method combines the reporter GFP with screening by fluorescence activated cell sorting (FACS). Though this is a high-throughput method that has proven very successful in obtaining new and/or improved enzymes, it does entail expensive equipment and experienced people operating this equipment. The aim of this thesis is to make this technology easier by developing a generic and high-throughput in vivo reporter-based system that involves selection instead of screening. The advantage of selection over screening is that only positive cells, containing the active enzyme, stay in the library pool, which allows for a quick reduction of the initially large library size. Although this thesis is not describing the first reporter-based selection system, other systems often are not applicable for a wide range of enzymes. In this thesis, the modular set-up of the system should make it more generic. To show that the developed system can be used in finding novel enzymes, a proof of principle is required. This consist of three aspects: (1) detect a product of an enzymatic activity, (2) apply the system at library scale, and (3) change the specificity of the system to make it applicable for a wide range of enzymatic products and thus different enzymes. In Chapter 3, the first two aspects are demonstrated, while Chapters 4 and 5 each focus on a different approach for the third aspect. Our bioreporter is based on the most common reporter-based strategy, namely the transcriptional regulatory-based strategy, and couples enzymatic activity to growth of the bacterium Escherichia coli. Chapter 3 covers the development of this in vivo transcriptional regulator-based selection system. A high false positive rate is a returning problem for growth-based selection and, therefore, our system was designed with dual reporters, both a selection and a screening reporter. The sensor part of the bioreporter is based on the transcriptional regulator AraC, which is involved in L-arabinose metabolism in E. coli, because AraC has been well studied. Furthermore, protein structures of AraC with and without ligand are available and it has been a topic of several engineering studies. In our system, the AraC sensor binds the product of the enzymatic reaction and switches on transcription of both a selection reporter (LeuB or KmR; enabling growth), for rapid reduction of the initially large library size, and a screening reporter (LuxCDABE; causing bioluminescence), for exclusion of false positives and quantification of positive variants. The characteristics of four different systems, differing in the selection reporter (LeuB or KmR) and in the plasmid origin of replication (low or medium copy number), are compared. The medium copy number system with KmR as selection reporter was found to be the best performing system based on leakiness, maximal signal, dynamic range and sensitivity in both selection and screening. Most importantly, a proof of principle of this system was provided by selecting cells expressing an L-arabinose isomerase derived from mesophilic E. coli or thermophilic Geobacillus thermodenitrificans. A more than a millionfold enrichment of cells with L-arabinose isomerase activity was established by selection and exclusion of false positives by screening. This shows the value of the dual selection and screening system for the detection of both mesophilic and thermophilic enzymes at library scale. However, in order to demonstrate that our bioreporter is generic and can be applied for a wide range of enzymes, its specificity needs to be adaptable towards the product of any enzyme. In Chapters 4 and 5, two different approaches to change the specificity of the bioreporter are presented. In Chapter 4, the replacement of the transcriptional regulator AraC by LacI (the regulator of lactose metabolism in E. coli), is described. The characteristics of four different systems, all having LacI as transcriptional regulator, but varying in the selection reporter (LeuB or KmR) and in the plasmid origin of replication (low or medium copy number), were compared. The low copy system with LeuB as selection reporter was selected as best performing system and using this system, it was demonstrated that previously described weak inducers or anti-inducers can be detected. The newly developed LacI-based system was compared with the original AraC-based system. The LacI-based system has a better sensitivity and a higher fold change of maximal signal over leakiness, but its dynamic range for selection is lower than that of the AraC-based system. Although some optimization is required, the replacement of the transcriptional regulator is rather straightforward due to the modularity of the system. It is a good approach to alter the specificity of the dual selection and screening system and thereby to broaden its range of potential target molecules. A second approach to change the system’s specificity is described in Chapter 5. This approach is based on engineering the ligand specificity of AraC, from L-arabinose to D-xylose, by targeting residues in the ligand binding pocket with combinatorial site-saturation mutagenesis. Others have already successfully modified the specificity of AraC using a GFP- and FACS-based screening of transcriptional-regulator variants. The aim here was to offer a simpler and alternative method by using growth-based selection instead. To this end, the dual reporter system itself was applied for selection and screening of transcriptional-regulator variants. The complete process is described, starting from library design and construction up to kanamycin resistance-based selection and bioluminescence-based screening of these libraries in the presence of D-xylose. Some of the developed AraC variants showed an altered, albeit small, response to D-xylose and several other tested monosaccharides. The selected variants yet have to be investigated in more depth to verify whether their ligand specificity is truly modified. Nonetheless, these variants will be interesting starting points for further engineering and indicate that the right positions in the protein were targeted. However, to obtain variants that give a better response, the selection and screening set-up needs to be optimized. After this optimization step, the same set-up could be used to select not only AraC variants with a better response to D-xylose, but also variants specific to other target molecules. Inhibitory and stimulatory effects of L-arabinose on growth of E. coli were observed during the experimental work with the AraC-based dual reporter system (Chapters 3 and 5). In Chapter 6, these observations are supplemented with follow-up experiments to understand the underlying regulatory mechanisms of these effects. The growth effects caused by L-arabinose are described for the system strain, its parent strain E. coli BW25113 and various single gene knockout strains derived from BW25113. In LB medium, L-arabinose negatively affects growth of wildtype strain BW25113 (lower final OD600), but not of ΔaraC or Δcrp strains. In addition, the effect is stronger in strains with ΔxylA, encoding the first enzyme in D-xylose catabolism. Growth of strains in which wildtype araC is replaced by sacB or by araC variants that encode L-arabinose unresponsive AraC mutants, is still inhibited by L-arabinose. Other related monosaccharides show to various extent also inhibition. In M9 minimal medium, L-arabinose stimulates growth of both BW25113 and ΔaraC strains in the early phase of growth, but ultimately reduces the final OD600 of only BW25113. Based on the different genotypes and phenotypes of the various tested strains, hypothetical regulatory mechanisms that may explain the effects of L-arabinose on growth of E. coli are discussed. To critically question the work presented in this thesis, a general discussion on the developed bioreporter is provided below. The bioreporter is compared to other screening and selection methods and suggestions for further improvements are outlined.