A Synthetic Biology Approach to Large Vector Assembly and Functional Genetic Element Screening

Abstract

Synthetic multi-input gene circuits offer great promise for therapeutic applications (Jaekel et al., 2021). Cell classifiers (Xie et al., 2011) in particular are a class of synthetic gene circuits that are well-suited for therapeutic applications, as they can increase the specificity of a therapy by leveraging AND logic (Nissim et al., 2017), broaden the target range by making use of OR logic (Doshi et al., 2020; Zah et al., 2020) and combine both to implement more complex Boolean expressions (Angelici et al., 2021). One interesting application of cell classifiers would be to incorporate a cancer cell classifier circuit into the genome of an oncolytic virus and use viral replication as circuit output. In doing so, one can create a conditionally replicating adenovirus (CrAd) (Heise & Kirn, 2000) that may be used as an oncolytic agent for cancer immunotherapy (Huang et al., 2019). One of the challenges faced by the aforementioned approach is that it requires the scaffold virus to be amenable to genome engineering. Human adenoviruses, in particular serotype Ad5, are good candidates for genome engineering as they have a compact and well-studied dsDNA genome of 36 kbp. Previous work has established recombineering as a method for the engineering of adenoviral genomes (Stanton, 2008a; Stanton et al., 2008). In this work, we explore recombineering as a way to generate oncolytic adenovirus candidates at high-throughput. Due to limitations associated with inserting the als-cassette into the region of interest, we found that recombineering does not allow for the high-throughput generation of oncolytic adenovirus candidates. To overcome this challenge, a yeast assembly assay toolbox was developed that divides the adenovirus genome into 8 fragments that can be separately modified. For this purpose, each of these 8 fragments has been cloned on a separate plasmid. This strategy prevents the accumulation of mutations during PCR. By mixing and shuffling fragments with different properties, this adenovirus yeast assembly toolbox promises the quick parallel generation of virus variants. Adenovirus genes E1A, E1B and E2 have been shown to be important targets for the regulation of Adenoviral replication (Kneidinger et al., 2012). These genes may be leveraged to actuate circuit activity for future virus variants. To this end, we show that targeting the 3’-UTR in the transcripts of these genes using siRNA is a promising 3 strategy to control viral replication. However, the effectiveness of this approach is likely to depend on Ad5 copy number and the abundance of siRNAs. Therefore, the yeast assembly toolbox also contains a circuit insertion template that allows for the separate integration of miRNA targets in the 3’ UTRs of E1A, E1B and E2 and the cell classifier circuit into the adenovirus genome. Finding efficient input combinations for cell classifier gene circuits remains a challenging task. In the second part of this thesis, we demonstrate, at the proof-ofconcept level, a screening method for functional genomic elements in mammalian cells, which we call mammalian uASPIRE (ultradeep Acquisition of Sequence-Phenotype Interrelations) after a previously published concept (Höllerer et al., 2020), on the example of a set of three promoter fragments expected to demonstrate different strengths (Qin et al., 2010).In this approach, a DNA modifying protein modifies a specific DNA sequence close to the functional genomic element of interest (in this case a promoter), such that both the region of interest and the state (modified or not modified) can be read through NGS sequencing. The portion of modified states gives an approximation of the promoter strength in the cells of interest. In order to avoid crosstalk between different promoters, the muASPIre construct is transferred and integrated into the target cells with a VSVG-lentivirus construct at low MOI.