Abstract

For the development of improved polyethylenimine (PEI) polyplexes towards ‘artificial viruses’ two key issues are i) to improve the toxicity profile of the applied vectors and ii) to enhance endosomal release, one of the major barriers to efficient gene transfer with PEI polyplexes. Nonviral vectors based on PEI usually contain an excess of PEI that is not complexed to DNA. Since free PEI contributes to cellular and systemic toxicity purification of polyplexes from unbound PEI is highly desirable. In this thesis an easy and efficient method based on size exclusion chromatography (SEC) was developed, which for the first time allowed complete removal of PEI from PEI polyplexes. Moreover, purification of polyplexes enabled to clarify the role of free PEI in gene delivery at the cellular level. Most importantly, the removal of unbound PEI significantly reduced toxicity of the applied polyplexes. However, purified polyplexes without free PEI were less efficient in transfection compared to non-purified polyplexes. Mechanistic studies showed that free PEI most likely enhanced endosomal release of polyplexes and therefore contributed to efficient gene transfer with PEI polyplexes. Nevertheless, the availability of a defined, well-tolerated gene transfer formulation is a vital precondition for the further development of nonviral gene therapeutics, and a purification method like the one developed in this thesis will help to fulfill these requirements. To enhance endosomal release of PEI polyplexes, the membrane active peptide melittin was incorporated into the vector particles. It has been shown previously that PEI bound to the N-terminus of natural all-(L)-melittin (all-(L)-N-mel-PEI) enhanced gene delivery with PEI polyplexes. Here, it was demonstrated that transfection efficiency of N-mel-PEI is independent of the enantiomeric configuration of the bound peptide which allowed the use of non-immunogenic all-(D)-melittin for the generation of optimized melittin-PEI conjugates. To determine the optimal site of melittin-linkage to PEI, the polycation PEI was covalently attached to the N-terminus (N-mel-PEI) or the C-terminus of melittin (C-mel-PEI) in all-(D)-configuration. The site of melittin-linkage had a strong impact on the membrane destabilizing activities of the resulting melittin-PEI conjugates. C-mel-PEI was highly lytic at neutral pH and therefore elevated doses of C-mel-PEI polyplexes induced high toxicity. In contrast, N-mel-PEI was less lytic at neutral pH but retained higher lytic activity than C-mel-PEI at endosomal pH 5. This apparently promoted better endosomal release of N-mel-PEI polyplexes resulting in efficient gene delivery in different cell lines. The high potency of C-mel-PEI to destabilize membranes at neutral pH is presumably due to a reported destabilization mechanism proceeding through membrane insertion of the peptide. In contrast, N-mel-PEI is supposed to induce lysis by insertion-independent pore formation according to the toroidal pore model. Since membrane destabilization by membrane insertion requires lower peptide to lipid ratios than destabilization by pore formation, C-mel-PEI was considered as the more potent template to generate improved endosomolytic melittin-PEI conjugates. The new melittin-PEI conjugates should display pronounced lytic activities at endosomal pH 5. Therefore, PEI was attached to the C-terminus of melittin analogs which were modified with acidic residues. The conjugates with the highest lytic activities at endosomal pH 5 were indeed the most efficient in transfection. This apparent correlation of gene transfer efficiency with lytic activity at pH 5 was in excellent agreement with results obtained with unmodified melittin-PEI conjugates and other membrane-active peptides used in gene transfer. The most efficient melittin-PEI conjugates were incorporated into surface-shielded and receptor-targeted PEI polyplexes, and the resulting particles were further purified by SEC to remove unbound toxic polycations. Endosomolytic melittin-PEI conjugates stably incorporated into such purified polyplexes significantly enhanced transfection efficiency in comparison to polyplexes lacking melittin. Most importantly, these polyplexes exposed an improved toxicity profile providing an artificial virus-like vector that is efficient and safe also for potential in vivo administration.