Análisis del instrumento de evaluación final del proceso

The work presented in this chapter demonstrates the potential for utilising horseradish peroxidase (HRP) and indole-3-acetic acid (lAA) as a novel enzyme/prodrug combination for cancer gene therapy.

In initial studies, a synthetic HRP gene, constructed by using codons commonly found and highly expressed in both E. coli and mammalian genes (Smith et ah, 1990), was used. The synthetic gene was cloned into a plasmid vector fused to EGFP to increase HRP stability and allow direct protein detection in transfected cells (Figure 3.1.B). Unfortunately, when the HRP was expressed in human cells the levels o f enzyme produced, although high enough to induce selective cell kill in the presence o f lAA, could not allow a detailed analysis o f HRP-mediated GDEPT in vitro (Figures 3.3, 3.4). Therefore the plasmid pRK34-HRP (Connolly, et al. 1994; Figure 3.1.A) was utilised to transfect four cell lines o f human origin. This plasmid construct, containing the HRP cDNA fused to the signal sequence from the hGH and the KDEL retention motif, has been previously used to monitor traffic through the Golgi apparatus, and shown to result in high HRP production levels in human cells (Connolly et al., 1994). Also, transient rather than stable transfectants were used, as they are more likely to mimic an in vivo scenario, where only a small fraction o f the population expresses the therapeutic gene.

The response o f the three tumour lines analysed, MCF-7 breast carcinoma, FaDu nasopharyngeal carcinoma and T24 bladder carcinoma cells, did not differ considerably (Figure 3.7; Table 3.1). After 24 h-incubation with lAA, at prodrug levels below 1 mM HRP expression conferred a slightly higher sensitivity to MCF-7 cells, whereas above 1 mM HRP^ T24 cells were markedly more affected by the treatment. The p53 status did not appear to play a major role in the response o f these tumour lines to HRP/IAA, since MCF-7 cells have a wild type p53 gene (Wosikowski et al., 1995), while FaDu and T24 cells are characterised by p53 non-sense and missense mutations at codons 248 and 126, respectively (Reiss et al., 1992; Kawasaki et al., 1996). HRP/IAA may therefore function efficiently in different tumours irrespective o fp53 status.

HRP/IAA GDEPT induced significant inhibition o f proliferation also in H M EC-1 endothelial cells (Figure 3.8.A; Table 3.1). This may represent an advantage if the tumour vasculature was to be targeted. Selective killing o f the endothelial cells forming the lining o f tumour blood vessels may cause malignant cells to starve o f nutrients, producing an amplification o f the cytotoxic effects (Chaplin and Dougherty, 1999). Also, endothelial cells lack drug resistance characteristic o f some neoplastic cells, requiring lower doses o f cytotoxic agents. Additionally, the vicinity to the blood stream would allow direct and simplified agent delivery. For HRP/IAA, low prodrug doses would need to be used, as lAA alone showed some toxicity in HRP endothelial cells (Figure 3.8.A; Table 3.1). This may limit therapeutic efficacy in vivo, although in preliminary studies 250 mg/kg lAA i.p. in mice resulted in tumour peak prodrug levels o f ~1 mM, and plasma levels in excess o f ~3 mM, with no associated toxicity (J. Tupper, Gray Cancer Institute, personal communication).

In order to investigate the potential o f the HRP/IAA combination to kill the hypoxic subpopulation in solid tumours, transfectants were exposed to lAA in an oxygen-ffee atmosphere and significant cell kill was measured (Figure 3.7.B, D). Different lAA-metabolites may be produced in oxic and anoxic cells. In the absence o f oxygen, the peroxyl radical and its decay products cannot be formed, but the skatole radical can (Figure 1.12, 3; Candeias et al., 1994). Skatole-type radicals readily abstract hydrogen from donor molecules and have been shown to react with biomolecules such as DNA (Folkes et al., 1999). In the absence o f oxygen, they could therefore lead to cell damage by the formation o f secondary radicals in key biological targets. Involvement o f relatively short-lived reactive species is also suggested by the lack o f transferability o f the toxic agent under anoxia (Figure 3.9.D). W hat remains to be clarified is how the peroxidase cycle is initiated under anoxia, since lAA did not reduce HRP under strict anaerobic conditions (Gazaryan et ah, 1996). Organic peroxides present as impurities in biological media or produced by the cells may be involved.

In normoxic cells, on the other hand, skatole radicals are more likely to react with oxygen to form peroxyl radicals (Augusto, 1993). Prompt attack by radicals on cellular targets in air as the main cause o f cell death may be ruled out, since conditioned

medium-switch experiments under normoxia and hypoxia (0.1% O2; Figure 3.9.B, C)

and incubation o f mammalian cells with filtered products o f lAA oxidation (Folkes et al., 1998) indicated that the toxic agent is a stable species. O f the stable products, indole-3-carbinol (Figure 1.12, 6) was shown to be non toxic to V79 cells at

experimentally produced concentrations, with or without HRP (Folkes and Wardman, 2001). 3-methylene-2-oxindole (MCI; Figure 1.12, 8) has been reported to be toxic in E.

coli and some plants, to react with glutathione and to bind to sulphhydryl regions o f hi stone DNA or RNA (Folkes and Wardman, 2001). The role o f MOI in the HRP/IAA- induced toxicity is currently under investigation (Folkes et al., manuscript in preparation).

An essential requirement for GDEPT is that the activated drug should induce a bystander effect, whereby conversion o f the prodrug to the active form in the enzyme- modified cells leads to the killing o f adjacent untransfected ones. The killing o f neighbouring cells can be due to the transfer o f toxic metabolic products through gap junctions, via apoptotic vesicles, or through the diffusion o f soluble toxic metabolites (section 1.2.5). Our studies suggest that the HRP/IAA system can produce a strong bystander effect. At neutral pH lAA is hydrophilic (polar and soluble) and can cross cell membranes within a few minutes (Pires de Melo et al., 1997; Folkes et al., 1999). In all experiments presented here, HRP-transfectants were estimated to represent at best a quarter o f the cells exposed to lAA, but this mixed population could be almost completely eradicated (Figure 3.7). In mixing experiments, approximately 70% and 90% cell kill were observed under normoxia with only 5% and 20% o f the cells expressing HRP, respectively (Figure 3.9.A). The effect does not appear to require contact between HRP and HRP^ cells, since incubation o f HRP cells with pre-activated lAA resulted in cell death under both normoxia and hypoxia (0.1% O2; Figure 3.9.B, C).

This compares very favourably with in vitro data on the bystander cytotoxicity o f other enzyme/prodrug systems. For example, 95% cell kill after GCV-treatment required expression o f H SV TK in 50% o f the exposed population (Freeman et al., 1993). Similarly, CD production in 5% o f the cells resulted in 50% cell eradication after 5-FC (Lawrence et al., 1998), and 90% growth inhibition could be achieved when 34-50% o f

the cells exposed to CB1954 produced the enzyme NTR (Bridgewater et ah, 1997; Spooner et ah, 2001). Importantly, an even more pronounced bystander effect was observed under anoxic conditions (Figure 3.9.A), which, as previously discussed, is likely to be due to short-lived reactive species.

Compared to HSV TK/GCV, in T24 cells in vitro HRP/IAA induced faster and more effective cell kill. HRP/IAA was selectively toxic after a brief 2 h-exposure (Figure 3.7.A, B), while HSV TK/GCV required a longer incubation to be effective (Figure 3.10). Moreover, after 24 h GCV-incubation, only a 4-fold increase in cytotoxicity (IC50) was induced in TK^ cells under normoxia, and no selective

sensitisation could be detected under anoxia. In previous studies, 5-14 days growth in the presence o f GCV was required to induce a 3 log increase in cell kill in transfected mammalian cells (e.g. Moolten, 1986; Freeman et ah, 1993). It is important to note that the T24 cell line is known to be resistant to a number o f chemotherapeutic agents (Mizutani et ah, 1997). These cells were also more resistant to lAA when activated by purified HRP (Figure 3.2), compared to the mammalian cells used in previous studies (Folkes et ah, 1998, 1999). For instance, after 2 h-exposure to 0.1 mM lAA activated by 1.2 pg/ml HRP 87% o f T24 cells lost their clonogenic potential, while at the same concentration/time a 2 log cell kill was induced in V79 cells (Folkes et ah, 1998, 1999). Overall, the HRP/IAA combination compared favourably with established systems.

Chapter 4