HBC Since its inception ADEPT approach has
Since its inception, ADEPT approach has been widely reported by many groups using a variety of enzymes and prodrugs (see reviews) , . Both mammalian and non-mammalian enzymes have been utilised. Of the mammalian variety, human β-glucuronidase in combination with many prodrugs ,  and as part of either a fusion protein with humanised antibody fragment  or fully human single chain Fv  has been one of the most commonly reported enzymes. In addition, non-mammalian enzymes with a mammalian homologue such as nitroreductase  and enzymes with no mammalian homologue including carboxypeptidase G2 and beta lactamase ,  have been widely reported. These studies are interesting and encouraging but none of these have reached clinical stage for reasons unknown.
The only system to have been studied in clinical application is the original ADEPT using carboxypeptidase G2 (CPG2). CPG2, a bacterial enzyme isolated from a Pseudomonas sp.  has no known human analogue and catalyses cleavage of reduced and non- reduced folates. Preclinical studies of CPG2 conjugated to non-internalising HBC directed at secreted tumour associated antigens such as human chorionic gonadotrophin (hCG) and carcinoembryonic antigen (CEA) have been carried out in nude mice bearing either CC3 or LS174T (respectively) xenografts in combination with a CMDA prodrug. These showed complete regression of the CC3 model  and growth delay in the LS174T model  As the preclinical development has been described in many studies, this review will focus on the clinical studies of ADEPT. These were all carried out using anti-CEA antibody-enzyme constructs.
Immunogenicity Although the generated drug produces inter-strand cross-links in tumour cell DNA within 1h, evidence showed these to be repaired by the cell within 24h . Therefore repeated ADEPT is necessary for sustained therapeutic benefit. Repeated ADEPT has only been possible with CPG2 if patients were given cyclosporine to suppress immune response . Thus immunogenicity of CPG2 has been a key limitation of this system. In all ADEPT clinical studies using antibody-enzyme conjugates, antibodies to murine monoclonal antibodies and the bacterial enzyme (CPG2) were detectable in serum of patients within 10days after a single ADEPT cycle. Two or three ADEPT cycles were possible when patients were given cyclosporine A. However, cyclosporine added to the toxicity. Attempts to reduce CPG2 immunogenicity included identification and modification of B cell epitopes. This resulted in identification of a discontinuous immunodominant B cell epitope at the C terminus of CPG2 . A hexa-histidine (His-tag) was engineered at this site for the MFECP1 fusion protein to potentially mask this epitope. This was confirmed as there was no evidence of immune response to the C terminus in both mice and human given MFECP1 . There was no immune response to the His-tag in patients even though human anti-MFE and human anti-CPG2 antibodies were detectable . Modification of T cell epitopes provides another strategy ,  and although identification of T cell epitopes was reported  for CPG2, modification of these epitopes to eliminate immunogenicity has not been reported. However, T cell modification approach has been successfully applied to the enzyme β-lactamase from Enterobacter cloacae to result in a modified enzyme that retained enzyme activity but with a significant reduction in immunogenicity as shown by in vitro T cell activation assays. It has to be noted that this modified enzyme has not been applied in the ADEPT situation in man so it remains to be seen if it would be non-immunogenic. Other approaches to make enzymes non-immunogenic include directed evolution , application of algorithms to optimise for enzyme function and low immunogenicity , , , or selection from combinatorial libraries . Of course, human enzymes can also be used in ADEPT but to prevent toxicity by premature endogenous activation of prodrug, human enzymes need to be mutated such that the prodrug is a substrate for the modified enzyme only and that the endogenous counterpart will not be able to activate the prodrug. This has been achieved with human carboxypeptidase A1 which has been modified to activate a methotrexate prodrug as well as with a reverse polarity human pancreatic ribonuclease which catalysed a bis-chloro phenol prodrug . Human beta glucuronidase is an attractive enzyme in the ADEPT context which has been widely reported , . It is surprising that no clinical study has been reported.