INTRODUCTION
The ability to manipulate RNA expression and splicing is one of the most
powerful tools in research and holds great promise for therapeutic
applications. In basic science, the ability to target and knockdown a
specific gene of interest using tools such as siRNA and CRISPRi has
allowed us to unravel the intricacies of biological pathways, leading to
many new discoveries (Cong et al., 2013; Gilbert et al., 2014; Kampmann,
2018; Mullenders et al., 2009). In medicine, targeted manipulation of
expression and splicing through antisense oligonucleotides (ASOs) has
led to several promising therapeutics. For instance, the FDA has
approved ASO-based treatments for both Duchenne muscular dystrophy, a
child-onset disease in which patients progressively lose muscle function
resulting in wheelchair dependency, the need for ventilation assistance
and ultimately premature death (Aartsma-Rus & Krieg, 2017), and spinal
muscular atrophy in infants, a disease caused by mutations in SNM1
leading to musculoskeletal wasting and respiratory failure (Stein &
Castanotto, 2017).
While a variety of technologies exist to knockdown or modify pre-mRNA in
the cell, few options are available that bridge the gap between research
and clinical settings. Tools such as siRNA, lentivirus, and plasmid
expression are frequently employed as screening tools to find
therapeutic targets (Koike-Yusa et al., 2014). However, the end products
of these methods often run into roadblocks as therapeutics themselves.
siRNA therapies are challenged by double-stranded RNA degradation by the
immune system, off-target effects due to improper strand loading, and
toxicity due to oversaturation of the RNAi machinery (Nogrady, 2019).
CRISPR-based technologies require the introduction of the exogenous Cas9
protein, which poses a significant barrier for many applications, and
their off-target effects are still being evaluated for safety (Tycko et
al., 2019). These limitations result in the identification and
characterization of gene targets with technologies that are either
clinically unviable or suboptimal, thus requiring their re-evaluation
with small molecule screens or other therapeutic methods, resulting in
significant delays in translating bench observations to the clinical
setting.
While ASOs have found purchase as therapeutics, they are heavily reliant
on a variety of nucleotide chemistries and backbone modifications to
achieve the desired results (Roberts et al., 2020). ASOs that degrade
their targets are often composed of RNA with central DNA “gap”
sequences that hybridize with homologous regions of RNAs and trigger
RNase H mediated degradation, whereas sterically hindering ASOs, such as
morpholinos, incorporate morpholine moieties instead of ribose backbones
and modified phosphonodiamidite linkages to evade degradation (Ekker,
2006). Reliance on these chemistries and modifications makes it
difficult to perform large ASO based screens to identify genes that can
be targeted with this modality. For example, this methodology is
incompatible with the production of lentiviral libraries, which infect
cells clonally and thereby can greatly parallelize the search for
candidate molecules through a combination of cell culture and
next-generation sequencing.
A technology that could seamlessly transition between construct-based
screening and the synthesis of biologically functional molecules would
greatly benefit both the research and clinical communities. The ideal
technology would require minimal chemical modifications to transition
from screening construct to lead compound, be compatible with modern
molecular biology techniques, such as plasmid or viral expression, and
would not require exogenous proteins to function. To address this need,
we created hnRNPA1 recruiting oligonucleotides (AROs) to serve as both a
standalone technology and paradigm for using endogenous RNA binding
proteins to manipulate pre-mRNA.
hnRNPA1 is a ubiquitously expressed RNA binding protein with multiple
roles that are still being uncovered; however, the protein is best known
for the role it plays in alternative splicing and RNA processing
(Jean-Philippe et al., 2013). Splice sites are defined by short 5’ (GU)
and 3’ (AG) sequences flanking introns. While these dinucleotides are
required for splicing, they do not contain enough information to
correctly specify splice junctions. As a result, regulatory proteins
such as SR proteins and heterogeneous nuclear ribonucleoproteins
(hnRNPs) must bind to exonic splice enhancers (ESEs) and exonic splice
silencers (ESSs) respectively to properly define exon-intron boundaries
(Lee & Rio, 2015) . Binding of hnRNPA1 to an ESS on pre-mRNA blocks the
binding of SR proteins and splicing machinery leading to cooperative
recruitment of additional hnRNPA1 and suppression of the splice site
(Jean-Philippe et al., 2013).
We sought to harness the ability of hnRNPA1 to suppress splice site
selection by recruiting it to targeted pre-mRNA molecules. To do so, we
constructed single-strand RNA molecules herein referred to as hnRNPA1
recruiting oligonucleotides or “AROs” (for A 1 r ecruitingo ligonucleotide). AROs consist of two parts, a short (20 – 25
bp) RNA oligonucleotide targeting domain which is complementary to the
target pre-mRNA, and an hnRNPA1 recruiting loop derived from the HIV ESS
3, which binds to the RNA binding domain RRM1 of hnRNPA1 and recruits
the protein (Jain et al., 2017) (Figure 1A). We hypothesized that, upon
recruitment to the target pre-mRNA, hnRNPA1 would displace local SR
proteins and splicing machinery (Figure 1B) resulting in the suppression
of regional splice sites, leading to frameshifts caused by aberrant exon
skipping or intron inclusion.
Due to the ubiquitous nature of endogenous hnRNPA1 across cell types,
AROs do not require any exogenous proteins to function. Further, as AROs
do not rely on DNA-RNA hybrids to trigger RNAase H degradation, AROs can
be produced entirely as RNAs within the cell. Finally, their
single-stranded nature and simple mechanism of action allow them to be
transcribed in vivo using a standard Pol II promotor, so they are
compatible with standard lentiviral screening methodologies. In the
following experiments, we demonstrate that AROs can suppress target mRNA
transcripts, are biologically functional, and can be expressed using
standard molecular biology constructs or delivered directly as
single-strand RNA oligonucleotides.