Design and construction of a fluorescent reporter for small molecules

Abstract

This thesis describes a body of work directed to making a versatile and easily customizable binding agent that produces a readily detectable fluorescent signal upon binding a small molecule. Such an agent may find application in reporting the presence of small molecules for example, in microdroplets in microfluidics-based in-vitro evolution systems, or in living cells or tissues. The approach being pursued was to fuse a ligand-binding domain and a fluorescent protein in such a way that a conformation change and concomitant change in fluorescence occurs upon target binding. The project had a number of strands. Firstly, a number of representatives of the PAS family of ligand-binding domains were cloned and screened for ligand binding. Three different PAS domains were studied. TodS PAS domain and largerTodS fragments, derived from Pseudomonas putida, bound only weakly to toluene in contrast to their reported high affinity. CitAP, from Escherichia coli, showed high affinity to its ligand citrate with a dissociation constant of 0.25 {u03E5}M. DcuS, a malate binding PAS domain protein also from E. coli, was expressed and purified using affinity chromatography, but was largely insoluble. Using some of those PAS domains, GFP-PAS fusion proteins were created. Insertion of CitAP and DcuS into EGFP at positions 157, 172 or189 generated various fusion proteins with different fluorescent intensity. A total of 6 EGFP-PAS fusion proteins were constructed. Half of the fusion constructs (GFP172-CitAP, GFP157-DcuS, and GFP172-DcuS) were soluble and fluoresced well. Ligand binding to these constructs was examined by surface plasmon resonance. Dissociation constants for their cognate ligands of GFP172-CitAP, GFP172-DcuS and GFP157-Dcus were estimated to be 0.06 {u03E5}M, 0.6 {u03E5}M and 1.3 {u03E5}M respectively. However, titration of these constructs with the relevant ligands had no effect on fluorescence signal, despite the surface plasmon resonance experiments indicating significant conformational change upon ligand binding. DcuS was then inserted to S65T GFP, a less stable version of EGFP, at positions 157 and 172. S65T GFP172-DcuS was fluorescent, however its fluorescence was independent of the presence of the malate ligand. The successful construction of a series of fluorescent, ligand binding fusion proteins, which showed indications of conformational change upon ligand binding, lead us to attempt to evolve a more conformationally flexible GFP reporter for use in such fusion constructs. The approach was to select from a mutant population of GFP molecules a reversible hyper-responsive version of GFP. Selection using a high throughput method based on fluorescence activated cell sorting of bacterial clones was not reproducible in our hands. Selection of fluorescent colonies on large plates yielded five mutants which were more sensitive to temperature than the parent EGFP. However, none of these mutants showed sufficient reversible fluorescence to be employed in the GFP-PAS fusion constructs. The DcuS PAS domain gene was randomly inserted into the EGFP gene using a transposon based method. This strategy in principle allowed all of the potentially fluorescent fusion proteins to be rapidly selected and screened, rather than relying on the literature to select sites within GFP that were known to accommodate insertions. 700 fluorescent clones were screened for insertion sites, somewhat surprisingly the large majority occurred within the first 120 bp of the 5' end of the EGFP gene. However, two novel insertion sites, at amino acids 6 and 24 were identified. Both fusions, EGFP6-DcuS and EGFP24-DcuS, were reasonably soluble and posses a significant level of fluorescence. However, neither construct responded to the presence of malate by a change in fluorescence. Finally, abandoning GFP as the reporter, an approach based on the fluorescence resonance energy transfer (FRET) was taken. The proposed FRET interaction was between an intrinsically fluorescent tryptophan residue and a fluorescent tag attached site specifically to the CitAP PAS domain. Using site-directed mutagenesis, CitAP was converted to Y11C CitAP allowing the attachment of a fluorescent tag (Alexa Fluor 350) via a disulphide bond. Initial binding data using surface plasmon resonance indicated that Y11C CitAP bound to citrate. However, cruelly, in subsequent expression experiments the Y11C CitAP protein was insoluble, making it impossible to obtain enough protein for labeling experiments. Retransforming fresh cell stocks were not able to regenerate a soluble Y11C CitAP protein and the march of time meant that the experiment remains poised for future success.