Antisense oligonucleotides redirect pre-mRNA splicing to restore exon inclusion — the nusinersen (Spinraza) paradigm for spinal muscular atrophy
Many genetic diseases are not caused by a missing gene, but by a mis-read one. Pre-mRNA splicing removes introns and joins exons together, and this process is governed by short regulatory sequences that can be disrupted by a single point mutation or by a naturally weak splice site. When a critical exon is skipped, the resulting mRNA is out of frame or missing an essential domain, and the protein made from it is truncated, unstable, or completely absent.
Spinal muscular atrophy is caused by loss of the SMN1 gene, which normally produces full-length Survival Motor Neuron (SMN) protein essential for motor neuron survival. Humans carry a near-identical backup copy, SMN2, but a single C-to-T change in SMN2 exon 7 weakens an exonic splicing enhancer and creates a splicing silencer context.
The result: the spliceosome skips exon 7 in roughly 90% of SMN2 transcripts. The truncated protein (SMNΔ7) is rapidly degraded, so SMN2 alone produces only a small fraction of functional, full-length SMN protein — not enough to prevent motor neuron loss.
This is a general principle in splicing-defect diseases (SMA, Duchenne muscular dystrophy, some forms of retinitis pigmentosa): the DNA sequence for a working protein is often still present, disguised inside an mis-spliced transcript.
Because the coding information for full-length SMN protein is still intact in SMN2, the disease is, in principle, correctable at the RNA level — without touching the genome at all. This insight motivated splice-modulating ASOs as a therapeutic strategy rather than gene replacement.
Splicing decisions are made by a combinatorial code of short cis-regulatory elements read by the spliceosome and its accessory factors:
• Exonic/intronic splicing enhancers (ESE/ISE): recruit SR proteins that promote exon recognition • Exonic/intronic splicing silencers (ESS/ISS): recruit heterogeneous nuclear ribonucleoproteins (hnRNPs, e.g. hnRNP A1) that repress exon recognition • 5’/3’ splice sites: recognized by U1 and U2 snRNPs; strength varies by sequence match to consensus
In SMN2, an intronic splicing silencer element (ISS-N1), located just downstream of exon 7 in intron 7, is a strong binding site for hnRNP A1/A2, which loops the RNA and blocks U1 snRNP recruitment to the weak 5’ splice site — biasing the outcome toward skipping.
An antisense oligonucleotide is a short, chemically stabilized strand of nucleic acid (typically 14–25 nucleotides) synthesized to be exactly complementary to one target sequence in the pre-mRNA. Unlike small-molecule drugs that fit into a protein pocket, an ASO finds its target purely through Watson-Crick base pairing — giving it exquisite, programmable sequence specificity.
A therapeutic splice-switching ASO is designed against a defined target window in the pre-mRNA — usually an intronic silencer, a cryptic splice site, or a splice site itself. Sequence uniqueness is checked genome-wide (BLAST-style off-target screening) so the 18–20 nucleotide sequence occurs, for practical purposes, at only one locus in the transcriptome.
Chemical modifications are essential for a drug that must survive in the body and act catalytically-free (no enzyme required):
• 2’-O-methoxyethyl (2’-MOE) or 2’-O-methyl sugar modifications: block RNase H cleavage of the RNA:ASO duplex (a "steric-block" mechanism, not gene knockdown) and resist nuclease degradation • Phosphorothioate (PS) backbone: one non-bridging oxygen replaced by sulfur at each linkage, improving nuclease resistance and protein binding for tissue distribution • No 2’-OH on the ASO strand: prevents RNase H recognition, which is essential — this class of ASO must NOT degrade the target RNA, only redirect its processing
This is a "steric-block" ASO, mechanistically distinct from gene-silencing ASOs (like mipomersen) that recruit RNase H to destroy the target transcript. Splice-switching ASOs are designed specifically to leave the pre-mRNA intact and instead change how the spliceosome reads it.
For CNS-acting splice-modulating ASOs such as nusinersen, systemic delivery is inefficient because oligonucleotides do not readily cross the blood-brain barrier. Instead, the drug is delivered directly into the cerebrospinal fluid by intrathecal (lumbar puncture) injection, distributing to spinal cord and brain motor neurons where SMN protein deficiency is most consequential.
Once inside the nucleus of the target cell, the single-stranded ASO anneals to its complementary pre-mRNA sequence while the transcript is still undergoing co-transcriptional splicing — the intervention has to occur in this narrow window, before splice-site selection is finalized.
Binding alone does not treat disease — it is where the ASO binds that matters. By hybridizing directly over the intronic splicing silencer (ISS-N1), the ASO forms a stable RNA:ASO duplex that occupies the exact footprint hnRNP A1/A2 needs to dock. The repressor can no longer engage its binding site, so its inhibitory signal to the spliceosome is silenced before it can be read.
The ASO:RNA duplex is thermodynamically far more stable than the transient, sequence-degenerate protein:RNA interaction of hnRNP A1/A2 with ISS-N1. Once the ASO is annealed, the silencer element is physically unavailable — a competitive occlusion mechanism, analogous to a bandage placed exactly over a "stop" sign so it can no longer be read.
Critically, the pre-mRNA backbone itself is untouched: no phosphodiester bond is cleaved, no nucleotide is chemically altered. The oligonucleotide is a reversible, non-covalent occupant of one short stretch of intron 7 for as long as it persists in the nucleus.
With hnRNP A1/A2 excluded, the local RNA secondary structure relaxes and U1 snRNP — the spliceosomal component that recognizes the 5’ splice site — can now access and stably bind the exon 7 / intron 7 boundary despite its intrinsically weak consensus match.
This single molecular event cascades into a global change in splicing outcome: instead of a minority of transcripts including exon 7, the majority now do — without adding a single enhancer sequence or correcting the causal point mutation. The therapy works entirely by removing an inhibitory signal, not by adding a positive one.
Nusinersen exemplifies "splice-switching" therapy: a single 18-nucleotide drug redirects a probabilistic, genome-wide splicing decision made independently in every motor neuron nucleus, every time SMN2 pre-mRNA is transcribed — for as long as sufficient drug concentration is maintained.
With the silencing signal masked, the spliceosome assembles normally across the newly-recognized 5’ splice site, excises both flanking introns as lariats, and ligates exon 7 permanently into the mature mRNA. The corrected transcript is exported to the cytoplasm and translated by ribosomes into full-length, properly folded, functional protein — restoring what the genetic defect had been silently subtracting.
Spliceosome assembly proceeds through the canonical E → A → B → C complex pathway: U1 snRNP marks the 5’ splice site, U2 snRNP recognizes the branch point, and the U4/U6.U5 tri-snRNP triggers catalysis — two sequential transesterification reactions that excise the intron as a lariat and ligate the flanking exons.
Because exon 7 is now retained in essentially every processed transcript, the mature mRNA restores the full, in-frame open reading frame. Ribosomes translate this corrected message into full-length SMN protein, which — unlike the truncated SMNΔ7 isoform — is stable, correctly folds, and retains its self-oligomerization domain required to form functional SMN complexes that support motor neuron RNA metabolism and axonal transport.
Clinical trials of nusinersen in infantile-onset SMA (ENDEAR) and later-onset SMA (CHERISH) demonstrated statistically significant improvement in motor milestones and event-free survival compared to sham-controlled arms — the first disease-modifying outcome ever achieved by redirecting splicing rather than replacing or editing a gene.
It is important to note what has NOT changed: the underlying SMN1 mutation and the SMN2 C840T splice-weakening variant remain in every cell’s genome, untouched. The ASO has not edited DNA. What has changed is the probability distribution of a stochastic RNA-processing decision, shifted pharmacologically in favor of the therapeutic outcome for as long as adequate ASO concentration is present in the nucleus.
This distinguishes splice-modulating ASO therapy fundamentally from gene therapy (SMN1 gene replacement, e.g. onasemnogene abeparvovec) or gene editing — those interventions alter the genome once; ASO splice modulation must be pharmacologically sustained.
Because the ASO acts as a reversible occupant of one RNA sequence rather than a permanent edit to the genome, its concentration in the target tissue inevitably declines through normal nuclease turnover and cellular clearance. As drug levels fall, the splicing silencer becomes accessible to hnRNP repressors again, and exon skipping gradually creeps back toward the untreated baseline — so lifelong maintenance dosing at defined intervals is required to sustain the clinical benefit.
Unlike a corrected gene, which is copied into every daughter cell indefinitely, ASO molecules are diluted by cell turnover, degraded by residual nuclease activity despite chemical stabilization, and gradually cleared from the CSF and nuclear compartment. Splicing correction is therefore not a single event but a dynamic equilibrium that must be continuously re-established.
Clinically, this translates into a defined dosing schedule: an initial loading phase with closely spaced doses to rapidly build nuclear ASO concentration to a therapeutic threshold, followed by lower-frequency maintenance doses to counteract gradual clearance and keep exon-inclusion levels within the effective range.
The reversibility of ASO splice modulation is a double-edged property. On one hand, it obligates patients to a lifelong treatment schedule and intrathecal procedures. On the other, it means the intervention is inherently tunable and stoppable — dose, interval, and even the target sequence can be adjusted for an individual patient without any permanent genomic consequence, and if a safety signal emerged, the effect would fade rather than persist irreversibly.
This stands in contrast to permanent gene-editing approaches (e.g. CRISPR-based correction), which aim for a single durable intervention but cannot easily be reversed or re-titrated once delivered. Splice-modulating ASO therapy trades permanence for controllability and reversibility.
Missing a maintenance dose does not cause an acute crisis, but it does allow the underlying splicing defect to gradually reassert itself over weeks to months — reinforcing why adherence to the maintenance interval, not a one-time treatment, is what sustains clinical benefit over a patient’s lifetime.