Antisense Oligonucleotide and siRNA Therapeutics

Antisense oligonucleotides (ASOs) and small interfering RNAs (siRNAs) are programmable synthetic nucleic acid therapeutics that modulate gene expression at the post-transcriptional RNA level via sequence-specific Watson–Crick base pairing. Whereas conventional small-molecule drugs and monoclonal antibodies require a folded three-dimensional protein pocket or extracellular epitope—leaving more than 80% of the human proteome pharmacologically "undruggable"—oligonucleotide therapeutics require only the primary nucleotide sequence of a target messenger RNA (mRNA) or pre-mRNA to suppress toxic gain-of-function proteins or restore functional alternative splicing.

The conceptual foundation of antisense pharmacology was introduced in 1978 by Paul Zamecnik and Mary Stephenson, who used a synthetic 13-mer oligodeoxynucleotide complementary to Rous sarcoma virus RNA to inhibit viral replication in chick embryo fibroblasts. Two decades later, in 1998, Andrew Fire and Craig Mello (awarded the 2006 Nobel Prize in Physiology or Medicine) discovered that double-stranded RNA triggers potent, catalytic gene silencing via the endogenous RNA interference (RNAi) pathway.

Unmodified natural phosphodiester DNA or RNA oligonucleotides are unsuitable as systemic drugs for three biophysical reasons: (1) serum exonucleases and endonucleases hydrolyze phosphodiester bonds within minutes in human blood; (2) polyanionic oligonucleotides (molecular weight ~5 to 15 kDa, carrying 15 to 45 negative charges) cannot passively diffuse across hydrophobic lipid bilayers and undergo rapid renal glomerular filtration into urine (t_1/2 < 30 minutes); and (3) unmodified foreign RNA/DNA motifs activate innate immune pattern-recognition receptors, including endosomal Toll-like receptors (TLR3, TLR7, TLR8, TLR9) and cytoplasmic RIG-I / MDA5.

Three generations of chemical modifications solved these pharmacokinetic and immunogenic barriers:

1. Phosphorothioate (PS) Backbone Modification: Developed by Fritz Eckstein, substituting one non-bridging oxygen atom of the internucleotide phosphodiester linkage with a sulfur atom (P=S) increases resistance to nuclease cleavage and imparts amphiphilic character. In addition, a phosphorothioate backbone binds reversibly with low micromolar affinity (K_d ≈ 10–50 μM) to 85% to 95% of circulating plasma proteins (principally human serum albumin), which prevents rapid renal glomerular filtration, extends circulation time, and promotes productive scavenger-receptor-mediated cellular uptake.
2. 2′-Ribose Sugar Modifications: Modifying the 2′-hydroxyl (2′-OH) position of the ribose ring locks the sugar pucker into an RNA-like C3′-endo (North) A-form conformation, which shields the backbone from nucleases, abolishes innate immune TLR7/8 activation, and increases hybridization affinity (measured by the duplex melting temperature, T_m) toward complementary target RNA:

• 2′-O-Methyl (2′-O-Me) and 2′-Fluoro (2′-F): Increase T_m by approximately +0.5 to +1.8 °C per modified nucleotide; alternating 2′-O-Me and 2′-F patterns form the standard stabilization scaffold of clinical siRNAs.
• 2′-O-(2-Methoxyethyl) (2′-O-MOE): Pioneered by Frank Bennett and Stanley Crooke at Ionis Pharmaceuticals, the bulky hydrophilic methoxyethyl substituent traps a structured water network around the phosphate backbone, increasing T_m by +2.0 °C per base and extending tissue elimination half-lives to 2 to 4 weeks.
• Bridged / Locked Nucleic Acids (LNA and 2′,4′-Constrained Ethyl [cEt]): Introduce a covalent methylene or ethyl bridge tethering the 2′-oxygen to the 4′-carbon of the ribose ring, pre-organizing the C3′-endo conformation to reduce entropic binding penalty and boosting T_m by +4 to +8 °C per base.

1. Phosphorodiamidate Morpholino Oligomers (PMOs): Replace the ribose sugar with a six-membered morpholine ring and the anionic phosphodiester linkage with an uncharged phosphorodiamidate group (used in Duchenne muscular dystrophy exon-skipping therapies eteplirsen [exon 51], golodirsen [exon 53], viltolarsen [exon 53], and casimersen [exon 45]).

Single-stranded ASOs (15 to 25 nucleotides) enter cells unassisted ("gymnotic uptake") via adsorption of their phosphorothioate backbone to cell-surface proteins and traffic between the cytoplasm and the cell nucleus, enabling them to target both nuclear pre-mRNAs and mature cytoplasmic mRNAs via two distinct mechanisms:

RNase H1-Mediated Degradation via Chimeric Gapmers

Mammalian RNase H1 is a ubiquitous endogenous endonuclease present in both the nucleus and cytoplasm that selectively recognizes a DNA:RNA heteroduplex and hydrolyzes the phosphodiester bond of the RNA strand while leaving the DNA strand intact to bind another transcript. However, fully 2′-modified oligonucleotides (2′-O-Me, 2′-O-MOE, LNA) adopt an A-form RNA:RNA geometry that cannot be recognized by RNase H1.

To combine the high affinity and nuclease stability of 2′-modifications with RNase H1 cleavage, medicinal chemists designed chimeric "gapmer" ASOs (typically a 5–10–5 architecture): a central "gap" of 8 to 10 deoxynucleotides (PS-DNA) flanked on both the 5′ and 3′ ends by "wings" of 5 nucleotides carrying 2′-O-MOE or cEt/LNA modifications. The wings provide nuclease protection and picomolar binding affinity, while the central DNA gap forms a bona fide DNA:RNA substrate cleaved by RNase H1. Approved gapmers include inotersen and GalNAc-conjugated eplontersen (degrading hepatic mutant and wild-type TTR mRNA in hereditary transthyretin amyloidosis) and intrathecal tofersen (degrading mutant SOD1 mRNA in SOD1-amyotrophic lateral sclerosis).

Steric-Block Splice-Switching ASOs: Nusinersen in SMA

When an ASO is modified at every 2′-ribose position along its entire length (or uses a PMO backbone), it binds its target pre-mRNA with high affinity without recruiting RNase H1. Positioning such a steric-block ASO across an exonic/intronic splicing enhancer or silencer masks the site from trans-acting spliceosomal factors, reprogramming alternative splicing.

The landmark clinical validation of splice modulation is nusinersen (Spinraza), developed by Adrian Krainer and Frank Bennett for Spinal Muscular Atrophy (SMA). SMA is a lethal infantile autosomal recessive motor neuron disease caused by homozygous deletion or mutation of the SMN1 (Survival of Motor Neuron 1) gene. Humans carry a nearly identical centromeric paralog, SMN2, which differs by a silent C-to-T transition at position 6 of exon 7; this single nucleotide change disrupts an exonic splicing enhancer and causes the spliceosome to skip exon 7 in ~85%–90% of SMN2 transcripts, yielding an unstable, rapidly degraded SMNΔ7 protein. Nusinersen is an intrathecally administered 18-mer fully 2′-O-MOE phosphorothioate ASO that hybridizes to Intronic Splicing Silencer N1 (ISS-N1) located just downstream of exon 7 in intron 7 of SMN2 pre-mRNA. By sterically displacing the inhibitory splicing repressors hnRNP A1/A2, nusinersen forces ~90% inclusion of exon 7, restoring full-length functional SMN protein and enabling infants with Type 1 SMA to survive and achieve independent motor milestones. In 2019, Timothy Yu and colleagues extended this platform to create milasen, an individualized N = 1 22-mer 2′-O-MOE splice-correcting ASO designed and administered within 12 months for a single child with a unique retrotransposon insertion in MFSD8 (Batten disease).

The Argonaute-2 (Ago2) Catalytic RISC Cycle

Whereas ASOs are single-stranded and amphiphilic, small interfering RNAs (siRNAs) are hydrophilic, double-stranded 21-to-23-base-pair RNA duplexes (~14 kDa) comprising a sense (passenger) strand and an antisense (guide) strand. Once delivered into the cytoplasm, the siRNA duplex loads into the multi-protein RNA-Induced Silencing Complex (RISC) containing the endonuclease Argonaute-2 (Ago2). Ago2 cleaves or unwinds and ejects the passenger strand, retaining the guide strand anchored via its 5′-phosphate in the MID domain of Ago2.

Guided by nucleotides 2–8 (the seed region) of the antisense strand, activated RISC scans cytoplasmic mRNAs; upon full complementary hybridization, the PIWI endonuclease domain of Ago2 cleaves the target mRNA phosphodiester backbone precisely between nucleotides 10 and 11 measured from the 5′ end of the guide strand. The cleaved mRNA fragments are degraded by cellular exonucleases (XRN1 and the exosome), while the Ago2–guide-strand complex remains intact and recycles catalytically to cleave hundreds of additional mRNA copies.

LNP Delivery vs. Trivalent GalNAc–ASGPR Conjugation

Because large, hydrophilic, polyanionic siRNA duplexes cannot cross cell membranes unassisted, two clinical delivery platforms were engineered (pioneered by Alnylam Pharmaceuticals):

1. Ionizable Lipid Nanoparticles (LNPs): The first FDA-approved siRNA, patisiran (2018) for hereditary transthyretin amyloidosis, encapsulates the siRNA duplex inside ~60-nm lipid nanoparticles containing the ionizable cationic lipid DLin-MC3-DMA (pK_a ≈ 6.4). In blood plasma, the LNP adsorbs endogenous Apolipoprotein E (ApoE), which binds Low-Density Lipoprotein Receptors (LDLRs) on hepatocytes; inside the acidifying endosome (pH < 6.4), DLin-MC3-DMA becomes protonated and interacts with anionic endosomal phospholipids to hexagonalize the membrane and release siRNA into the cytosol.
2. Trivalent N-Acetylgalactosamine (GalNAc) Conjugation: To eliminate intravenous LNP infusions and corticosteroid premedication, chemists covalently tethered a triantennary cluster of three N-acetylgalactosamine (GalNAc) sugars to the 3′ end of the siRNA sense strand. Trivalent GalNAc binds with nanomolar avidity (K_d ≈ 1–5 nM) to the Asialoglycoprotein Receptor (ASGPR), a C-type lectin expressed specifically on the sinusoidal surface of hepatocytes at 500,000 to 1,000,000 copies per cell that internalizes via clathrin-coated pits and recycles to the membrane every ~15 minutes. Following a simple 1.5-mL subcutaneous injection, GalNAc–siRNAs (such as inclisiran targeting hepatic PCSK9 to lower LDL cholesterol, vutrisiran targeting TTR, givosiran targeting ALAS1 in acute hepatic porphyria, and lumasiran targeting HAO1 in primary hyperoxaluria type 1) accumulate inside slow-release acidic hepatocyte endolysosomal depots, continuously feeding Ago2–RISC to sustain >75%–85% target knockdown with once-every-3-month or twice-yearly dosing.

Why can single-stranded ASOs modify nuclear pre-mRNA splicing whereas siRNAs generally act only in the cytoplasm?

Single-stranded phosphorothioate ASOs bind intracellular shuttling proteins (such as nucleolin and paraspeckle proteins) and diffuse freely through nuclear pore complexes to accumulate in the cell nucleus, allowing them to bind nascent pre-mRNA introns and splice sites co-transcriptionally. In contrast, clinical siRNA action relies on cytoplasmic loading into mature Ago2–RISC, which predominantly patrols translated mRNAs in the cytoplasm.

How does the 5′-vinylphosphonate modification improve GalNAc–siRNA potency?

For the guide strand of an siRNA to load stably into the MID domain of Argonaute-2 (Ago2), its 5′ terminus must carry a negatively charged 5′-monophosphate group. Inside acidic hepatocyte endolysosomes, cellular phosphatases rapidly strip natural 5′-phosphates. Replacing the bridging oxygen of the 5′-phosphate with a carbon double bond (5′-(E)-vinylphosphonate [5′-VP]) creates a metabolically stable phosphate mimic that resists lysosomal phosphatases, increasing Ago2–RISC loading and tissue duration up to 10-fold.

Why does inclisiran require only two subcutaneous injections per year to lower LDL cholesterol?

Although free inclisiran clears from blood plasma into the liver within 24 to 48 hours via ASGPR uptake, the chemically stabilized 2′-O-Me / 2′-F / phosphorothioate duplex resists lysosomal nucleases inside hepatocytes for months. An intracellular endolysosomal depot slowly trickles intact siRNA duplexes into the cytoplasm (<1% endosomal escape is sufficient), where Ago2–RISC protects the guide strand and catalytically degrades PCSK9 mRNA over a 6-month dosing interval.

1. Fire, A., Xu, S., Montgomery, M. K., et al. (1998). "Potent and specific genetic interference by double-stranded RNA in Caenorhabditis elegans." Nature, 391(6669), 806–811. https://doi.org/10.1038/35888
2. Crooke, S. T., Witztum, J. L., Bennett, C. F., & Baker, B. F. (2018). "RNA-targeted therapeutics." Cell Metabolism, 27(4), 714–739. https://doi.org/10.1016/j.cmet.2018.03.004
3. Finkel, R. S., Mercuri, E., Darras, B. T., et al. (2017). "Nusinersen versus sham control in infantile-onset spinal muscular atrophy." New England Journal of Medicine, 377(18), 1723–1732. https://doi.org/10.1056/NEJMoa1702752
4. Roberts, T. C., Langer, R., & Wood, M. J. A. (2020). "Advances in oligonucleotide drug delivery." Nature Reviews Drug Discovery, 19(10), 673–694. https://doi.org/10.1038/s41573-020-0075-7