Development of an immunotoxin incorporating the Australian jumper ant toxin, pilosulin 1
- Publication Type:
- Issue Date:
Immunotoxins are therapeutic agents that directly target toxins to specific cells and are generally comprised of an antibody or antibody fragment (Fab or scFv) linked to a toxic moiety. The assessment of a number of immunotoxins in recent Phase I and 11 clinical trials has been very promising, particularly for the treatment of haematological malignancies. Since the majority of these incorporate large, potent bacterial or plant toxins, their therapeutic potential is limited by dose-limiting non-specific toxicity, immunogenicity and the need to be endocytosed. An alternative approach is to incorporate cytolytic toxins such as melittin and pilosulin that are smaller, less toxic molecules that act at the cell membrane. Both melittin- and pilosulin-based immunotoxins (mel-IT and pil-IT) have been developed by our research group. These cytolytic immunotoxins, which incorporate a scFv moiety specific for the human kappa myeloma antigen (KMA) expressed on human kappa myeloma cells and the human lymphoblastoid cell line, HMy2, display specific cytotoxic activity at micromolar concentrations against the target cell line. In contrast, immunotoxins in clinical and late stage pre-clinical studies are active at picomolar concentrations and thus it was deemed necessary to enhance the specific activity of mel-IT and pil-IT to ensure they could be effective at relevant clinical doses. The pil-IT displayed greater cytotoxic potential as peptide studies indicated that pilosulin was four times more potent than melittin against white blood cells (WBCs) and additionally, the pil-IT was shown to be twice as toxic as the mel-IT on a molar basis. In order to identify the regions of pilosulin essential for cytolytic activity and thus develop a smaller imrnunotoxin, two recombinant constructs were generated that encoded truncated toxin domains; P₁-₂₂F (incorporating the N-terminal helix of pilosulin, amino acid residues 1 to 22) and P₂₃-₅₆F (incorporating the C-terminal helix, residues 23 to 56). Unexpectedly, both recombinant constructs displayed reduced cytolytic activity compared to the parent construct (pil-IT/ Pt₁-₅₆F), due to reduced specific binding of the immunotoxins to the target cells, presumably as a result of incorrect tertiary folding of the expressed proteins. In a further attempt to increase the specific activity of the pilosulin-based immunotoxin, two additional constructs were generated; P₁₂₁F which had a longer linker arm between the full-length pilosulin peptide and the antigen-specific scFv moiety to enhance steric access of the toxin to the target cell membrane, and P A₃₃K₄₁F which contained the C-terminal helical region of pilosulin in an enhanced helical conformation to aid membrane interaction and penetration. While preliminary studies indicated that both constructs had cytotoxicity comparable the parent, neither exhibited enhanced cytolytic activity. Contaminant proteins were observed to be co-purifying with the immunotoxins raising a question as to whether these contaminants may have had the potential to affect the cytolytic activity of pil-IT. The most significant contaminant was identified as apoA-1, a 27-kDa hydrophobic serum protein that had previously been shown to inhibit the activity of cytolytic peptides and also to stabilise damaged membranes. As the FBS used to supplement the expression culture medium was identified as the source of apoA- 1, the pil-IT immunotoxin was expressed in serum-free medium. The recombinant protein expressed under these conditions was extremely susceptible to proteolysis in the cell culture medium and attempts to block this proteolysis by supplementing the cultures with BSA or a2-macroglubulin were ineffective. However, supplementing the serum-free expression cultures with E-64, a specific cysteine protease inhibitor, blocked the majority of the proteolytic degradation of pil-IT and allowed affinity purification of a very pure immunotoxin preparation. Unexpectedly, pil-IT expressed in this manner displayed significant non-specific toxicity compared to previous immunotoxin batches. It is possible that the presence of E-64 in the culture supernatant affected the tertiary fold of the immunotoxin so that it acted independently of the antigen binding specificity encoded by the scFv moiety, or that pil-IT in a very pure form is very toxic and non-specific (i.e. a true result of an 'apoA-1 free'-immunotoxin preparation). Another issue requiring consideration was that the insect cell line used for expression of the non-specific toxic batches of immunotoxin in serum-free medium (High Five cells) was different to that used to express batches of specifically cytotoxic immunotoxin in serum-containing medium (Sf21 insect cells). To address this question, the pil-IT was expressed in High Five insect cells in the presence of FBS and then affinity purified with a Tween-20 wash step to dissociate the apoA-1 from the immunotoxin. While this produced a pure preparation, non-specific cytolytic activity was again observed, although to a lesser degree than that observed for the batches expressed with E-64 supplementation. Tween-20 was found to contribute to some of this non-specific activity but was not the sole factor, as pil-IT expressed in High Five cells in the presence of FBS and purified without Tween also exhibited nonspecific cytotoxicity. Thus it was likely to be a result of either (i) expressing the immunotoxin in High Five insect cells which, in contrast to Sf21 cells, may generate a tertiary fold in the immunotoxin that allows it to act non-specifically, or (ii) the absence (or low levels) of apoA-1, which when present in the immunotoxin preparation, may have inhibited its non-specific cytolytic activity and/or repaired toxin-induced cell membrane damage, with the strength of this effect varying for different cell membranes.
Please use this identifier to cite or link to this item: