Neuronal expression and retargeting of VLPs

For the neuronal delivery and transgene expression we followed two approaches:

• the restriction of expression by a neuronal promotor, enabling a broad delivery

• retargeting of VLPs, which would allow the targeted delivery of a strong expression promotor

Similar to the rAAVs and lentiviruses which we tested for the activity of the CaMKII and the hSyn promotors in HeLa cells (see 3.1), we used linear DNA derived from the pAAVs and delivered them via VLPs. We previously observed that VLPs well transduce HeLa cells and lead to a GFP-expression driven by the CAG promotor (see 3.4.2). But none of both constructs with the neuron specific promotors led to a transgene expression.

We conclude that the lack of expression was due to the combination of the rather weak promotors and the delivery via VLPs.

To exclude cell line artifacts, we tested the VLPs, similar to the rAAVs (see 3.1.3), on primary cortical cultures of the Wistar rat. No GFP-expression was detectable and the culture viability decreased dramatically in a dose-dependent manner. Pre-equilibration and re-buffering of the VLP sample pre-application could not circumvent this effect.

Thus we could transduce primary rat osteoblasts and the neuroblastoma cell line SH-SY5Y with VLPs and detect expression of the CAG-GFP construct, it could be explained by the high sensitivity of neuronal cultures. Supported by the previous observation that a culture ofCallithrix jacchusglia cells was susceptible for VLPs delivering a flourescence dye (Cy3) coupled small RNA.

We conclude that the ability of neuronal targeting with VLPs has to be elucidated in another ex vivo model which is more robust. Possibly a denser cell structure, like in cerebral organoid models of the human brain, supports the cell viability after the trans-duction with VLPs (Lancaster et al., 2013). VLPs were found to work efficiently in vivo without harming the individuum (Hoffmann et al., 2016). Anin vivo study using MIDGE vector-loaded VLPs with both, constitutive and neuron-specific promotors and subsequent analysis of organs and cell types for the reporter gene expression would help to unravel the transduction profile of VLPs.

Another option to achieve a cell type restricted protein expression is, as previously de-scribed, the use of promotors being inactive in off-target tissues (White et al., 2009). The

option which we provide for the targeting of Her2/neu-receptor positive cell lines (can-cer cell lines) is the modification of the VLPs‘ tropism via the attachment of retargeting molecules. Having developed a functional retargeting molecule, this can also be used to al-ter the tropism of exo-AAVs and lentiviruses. For these two membrane-enveloped viruses, the retargeting molecule would be recombinantly expressed as a membrane-anchored pro-tein in the virus producer cell line. When the virus is released from the cell, it will gain the membraneuos envelope with the exposed retargeting molecule. Taken together, one could compare the three tools and identify the best suitable one for a specific application.

The humanized strain of the yeastP. pastoris stands out due to its humanized glycosyla-tion profile of recombinant expressed proteins in a sense that only a 5 times branching of mannose is seen (Hamilton and Gerngross, 2007). Thereby it is our preferred expression system for retargeting molecule expression, which we used to express the Her2/neu single chain variable fragments (scFv) that altered the VLPs‘ tropism towards human cancer cells (see 3.4.2). We constructed the expression plasmid for a scFv that should bind to the tropomyosin receptor kinase B (TrkB), the receptor for BDNF which is expressed in the central and peripheral nervous system and other tissues (Gupta et al., 2013). Un-fortunately, no recombinant protein expressing P. pastoris clone could be identified in repeated attempts.

Hence, we switched the expression constructs to the pET28-E.coli expression system and achieved the expression of the retargeting molecules. However, the retargeting molecules were unsoluble. The protein transcription rate and the folding of proteins expressed in E.coli can be influenced by several cultivation parameters. In our case, we gained unsol-uble proteins even when reducing the transcription rate by lowering the concentration of the induction reagent IPTG or by decreasing the temperature and thereby increasing the possibility of proper folding of a soluble protein (Schein and Noteborn, 1988).

Alternative approaches to gain a soluble protein would be the use of bacteria strains which secrete the protein to the periplasm or the medium, or the co-expression of chaperons to gain an improved protein folding (Mergulhao et al., 2005; De Marco, 2009). Additionally, one could try to express the protein in insect cells as we do for the VP1 protein or in human cell lines. These alternative approaches were not undertaken due to the restricted time in the course of this study.

As a promising alternative, we tested a synthetic peptide derived from the rabies virus glycoprotein (RVG) that was shown to specifically deliver siRNAs through the blood-brain barrier (Javed et al., 2016). RVG binds to the n-acetylcholine receptor which is expressed, e.g., in the neuroblastoma cell line SH-SY5Y but not in fibroblasts (Lentz, 1990; Javed et al., 2016; Kovalevich and Langford, 2013). In our experiments, the native VLP

trans-duced both cell lines and SKBR3 cells were used as our positive control for functional VLPs. VLPs with the attached crosslinker alone, as well as with the crosslinker and the retargeting molecule showed drastically reduced transduction of all cell lines. Based on this result we can make no conclusion whether the crosslinking of the retargeting molecule did not work or the molecule was crosslinked but did not bind to the cells.

Further experiments with an altered crosslinking protocol to enhance the crosslinking-success will be elucidated. Because we utilised a rather short crosslinker, steric hindrance might have not allowed the small RVG-proteins (scFvs are several folds larger) to bind to the cell receptors when attached on the VLPs‘ surface. The use of a longer crosslinker than the used one of 32.5 angstroms, might be successful.

In summary, we demonstrated that VLPs are a very useful, non-viral tool to deliver DNA expression cassettes in vitro. We combined this technology with MIDGE vectors, which will increase the safety of DNA delivery in in vivo applications. Even if we could not achieve neuronal retargeted VLPs, our results contribute to the previous finding of specific delivery of therapeutic expression cassettes via Her2/neuscFv-retargeted VLPs into cancers cells.