Functionalizing RNA with Structural Elements

Abstract

The structure of an RNA defines its function. As genetic medicine platforms are gaining prominence in the therapeutic landscape, investigating the RNA structure and its dynamics is becoming imperative. Most genetic medicine platforms already integrate structural considerations into engineering design. At the same time, RNA structural probing methods are starting to uncover functional RNA networks. Within this context, accurate and rapid determination of RNA structure-function relationships and integration of RNA structures with known functions into DNA and RNA modality designs can unlock safe and efficacious therapies. Recombinant adeno-associated viruses (rAAVs) and lipid nanoparticles (LNPs) are the main delivery vectors for genetic medicines. Despite an increasing number of approved products, both vectors must overcome several challenges such as uncontrolled biodistribution, cytotoxicity, and high manufacturing costs to attain widespread adoption. Specifically, in the case of rAAV, overexpression of Factor VIII (FVIII) protein from the transgene for treatment of severe hemophilia A has been repeatedly shown to induce hepatotoxicity. In chapter 2 of this dissertation, I describe a structural RNA element that can act as a rheostat and modulate therapeutic protein expression from delivered nucleic acids when overexpression leads to cellular toxicity. In finding this solution, we took inspiration from an unconventional cytoplasmic splicing mechanism during which inositol-requiring enzyme 1 alpha (IRE1α) recognizes a double stem-loop structure on X-box binding protein 1 (XBP1) mRNA and excises out a 26-nucleotide intron. This splicing event occurs when the protein folding demand in the endoplasmic reticulum (ER) surpasses its capacity, activating ER stress and the unfolded protein response. We created an ER stress responsive circuit by attaching the full length XBP1 mRNA and its functional fragments (XBP1F) at the 5’ ends of transcripts coding for therapeutic proteins with a tendency for misfolding in the ER, including a derivative of FVIII and Leronlimab, a monoclonal antibody targeting the CCR5 receptor. With this transcript architecture, XBP1F splicing leads to a shift in the protein reading frame and halts protein expression under ER stress conditions. Through this process, we detected an up to 6-fold drop in therapy induced ER stress in cell culture studies as measured by the abundance of BiP/GRP78 transcripts, which is a marker of ER stress. In addition, we observed lower serum Leronlimab levels on day 7 in mice that received rAAV packaging an XBP1F incorporated transgene compared to the no-XBP1F control group. Since the amount of Leronlimab in serum was equivalent for all rAAV treated mice on day 28, we inferred that ER stress was elevated at the onset of protein expression. While only in the preliminary phase, this technology can counteract side-effects attributed to overexpression of rAAV and LNP/mRNA products, accelerating clinical translation and ameliorating post-clinical outcomes. In the last couple of years, RNA trans-splicing to correct disease-causing mutations by replacing mutated exons with repaired ones has emerged as a therapeutic modality alternative to protein overexpression and DNA editing. A less common form of natural RNA splicing, RNA trans-splicing takes place when splice sites from two separate pre-mRNAs are recombined due to interactions between their introns. Several groups have harnessed the eukaryotic splicing machinery to drive RNA trans-splicing, whereby recombinant exon(s) are incorporated into the endogenous mRNA sequence. In order to achieve this splicing event, a pre-trans-splicing RNA comprised of an antisense sequence linked to a hemi-intron and one or more exons is delivered to the nucleus. In the nucleus, the antisense guide sequence base pairs with the target mRNA, followed by recruitment of the spliceosome by the intronic elements and fusion of the recombinant exon(s) in trans. Despite its strong potential as a novel RNA editing approach, RNA trans-splicing performed with low efficiency across several targets, which decelerated its preclinical and clinical progress. In chapter 3 of this dissertation, I propose to exploit strategies employed by RNA viruses to combat low RNA trans-splicing efficiency. With this purpose, we built a library of viral RNA segments by tiling the noncoding regions of viral RNA genomes with a sliding window approach. We introduced the library upstream of a pre-trans-splicing RNA targeting an endogenous intron and evaluated trans-splicing efficiency in cells integrated with a split-EGFP reporter. After two library transfections and next-generation sequencing of amplicons containing unique barcodes corresponding to viral elements, we calculated metrics such as RNA/DNA ratio, fold change, and p-value to shortlist candidates. Out of 3848 library members, we individually assessed trans-splicing efficiency for 24, most of which upregulated restored EGFP expression when pre-trans-splicing RNA plasmid was co-transfected with a split-EGFP plasmid encoding a defective EGFP exon. One ongoing study involves splicing efficiency quantification in a CRISPR-RBP knockout cell line to identify host RNA binding proteins that interact with the viral segments to up- or downregulate trans-splicing. To conclude, we demonstrated in two separate projects the power of functionalizing RNA with structural elements to tackle safety and efficacy issues encountered in rAAV and LNP/mRNA delivered genetic medicines. We postulate that in the near future novel machine learning based RNA folding models and structural optimization experiments in relevant animal studies will help bridge the gap between innovative therapies and patients.