Search

Bioluminescence imaging (BLI) is widely used in pre-clinical research to monitor the location and migration of different cell types, and the growth of cancerous tumours and response to treatments within murine models. However, the quantitative accuracy of the technique is limited. The position of the animal is known to affect the measured bioluminescence, with a change in position causing a change in measurement. Work presented here will address this problem, validating a free space model in a murine model to produce surface bioluminescence measurements which are independent of the position of the animal. The position of the source within the animal and the underlying tissue attenuation also affect the quantitative accuracy of bioluminescence measurements. An extension to bioluminescence imaging, bioluminescence tomography (BLT), aims to overcome these problems by recovering the three-dimensional bioluminescent source distribution within the animal. However, there are limitations to the quantitative accuracy of BLT. Current reconstruction algorithms ignore the bandwidth of band-pass filters used for multi-spectral data collection for BLT. This work develops a model which accounts for filter bandwidth in the BLT reconstruction, improving the quantitative accuracy of the technique. An additional limitation to the quantitative accuracy of BLT is that accurate knowledge of the optical properties of the animal are required but are difficult to acquire. Work to improve the quantitative accuracy by obtaining subject-specific optical properties via a spectral derivative reconstruction method for diffuse optical tomography (DOT) is presented. The initial results are promising for the application of the method in vivo.

E. coli NfsB nitroreductase (NTR) is currently being studied in combination with the prodrug CB1954, as a gene directed enzyme prodrug therapy. NTR reduces CB1954 at either the 2- or 4-nitro groups to produce highly cytotoxic hydroxylamine derivatives, using either NADH or NADPH as cofactor. Initial clinical trials suggest activity, with reduction in PSA and apparent delayed progression in some patients. This thesis is concerned with engineering the NTR enzyme to improve the efficiency of CB1954 activation, since this would be expected to improve the potential clinical efficacy. Site directed mutagenesis has been used to generate four libraries of NTR mutants with three small libraries having a maximum diversity of 6.4 x101 – 8.2 x 103 and a large library of 6.3 x 106 possible nucleotide combinations. A direct positive selection strategy, which utilises bacteriophage lambda and the SOS response, was used to select NTR variants with increase sensitivity for CB1954 from each library. NTR mutants with greatest sensitivity were analysed by bacterial IC50 assays. Despite considerable work to generate and optimise the large library, only one NTR enzyme was improved over WT. However the best mutant (T41G N71S) came from a small library, with an IC50 of 8 µM. The most promising NTR enzyme variants were column purified and their kinetic parameters and activity determined. T41G N71S showed good activity for CB1954 (0.52 µM s-1), being the second best mutant to date behind T41L N71S (0.90 µM s-1). Adenoviral vectors were generated and their ability to sensitise SKOV3 cancer cells tested. T41G N71S was ~2-fold improved relative to WT NTR. Purified enzymes were also tested with SKOV3 cells, with T41G N71S being ~2.6-fold improved relative to WT NTR. My best enzyme should be considered when planning future clinical trials using prodrug activation gene therapy with NTR.