What it takes to move an AOC from design to a reproducible, regulatorily defensible drug product
Antibody-oligonucleotide conjugates (AOCs) solve an issue that has limited oligonucleotide therapeutics: delivering a sequence-specific gene modulator to the right cell type. An oligonucleotide can silence, splice, or modulate any RNA target, including targets that small molecules can’t even touch, but on its own it has no way to find the tissue that needs it. Attach it to an antibody, and the antibody supplies the targeting mechanism the oligo lacks.
It’s an idea that has demonstrated some real clinical momentum. Avidity’s AOC 1001, the first AOC to enter clinical trials, reached Phase III for myotonic dystrophy in 2024 and received FDA Breakthrough Therapy designation the same year. Dyne Therapeutics has advanced its lead Duchenne candidate, z-rostudirsen (DYNE-251), through the Phase 1/2 DELIVER trial and, in 2026, both initiated the confirmatory Phase 3 FORZETTO trial and submitted a BLA to the FDA seeking accelerated approval in exon-51 DMD. And Novartis moved fast enough on the idea that it announced a $12 billion acquisition of Avidity, a company built almost entirely around AOCs, in October 2025, completing the deal in February 2026.
None of that changes a harder truth for anyone running a CMC program. Moving from an elegant biological concept to a reproducible, regulatory-acceptable drug product is far from simple, and AOC development can fail or miss a key milestone at several specific points. It’s worth walking through them in the order a program is likely to tackle them.
It all starts with the oligonucleotide
The reason oligonucleotides are so therapeutically appealing (their ability to bind with sequence-specific precision to virtually any RNA target, including one’s small molecules can’t) is inseparable from the reason they’re so challenging to manufacture.
Oligos are built by solid-phase phosphoramidite chemistry, adding nucleotides one at a time to a growing chain anchored to a solid support, with each addition following a four-step cycle of detritylation, coupling, oxidation or sulfurization, and capping. This has been the industry standard for nearly forty years and it works, but at each coupling step, a small fraction of chains fail to extend, and multiplied across twenty or more cycles, that produces a population of truncated sequences (n-1, n-2, missed incorporations) chemically similar enough to the target product to resist easy purification. Incomplete deprotection of intermediate groups adds a further impurity category, one that can affect target binding, immunogenicity, and toxicity in ways that matter to regulators.
The phosphorothioate backbone modifications carried by many therapeutic oligos (replacing a non-bridging phosphate oxygen with sulfur to improve nuclease resistance) add another wrinkle. Each PS linkage creates a chiral center at phosphorus, yielding Rp and Sp diastereomers with measurably different pharmacology, including nuclease sensitivity, stability, and interaction with the enzymes that run the antisense and RNAi pathways. A 20-mer with a fully PS backbone has the theoretical capacity to exist as 2¹⁹ diastereomeric species. You don’t need to separate them all, but you do need to characterize the distribution and be able to defend it.
Then there’s secondary structure. Oligos aren’t passive strings of nucleotides waiting to be conjugated; under physiological buffer conditions they can fold into hairpins, G-quadruplexes, and self-complementary duplexes. An unintended fold can change target engagement, hybridization kinetics, conjugation efficiency, or potency. These structures aren’t always detectable by standard analytical methods, and they don’t always show up at the conjugation step; instead, they surface later, when the potency data looks off, and nobody can immediately understand why.
That’s the impurity and structural picture before the oligo has even been attached to anything.
Conjugation must keep two very different molecules functional at once
The antibody and the oligo aren’t natural companions. The antibody is a large, relatively hydrophobic protein evolved to sit comfortably in plasma; the oligonucleotide is a polyanionic, hydrophilic molecule whose entire therapeutic rationale depends on getting inside a cell. Joining them stably and reproducibly, while preserving the function of both, is the central technical challenge of AOC manufacturing, and it can fail in more than one place at once.
The most common approach is maleimide-thiol chemistry, where the maleimide group on the linker reacts with free cysteine residues on the antibody. It’s reactive, reasonably site-selective, and well understood from years of ADC development, but it’s also susceptible to hydrolysis in an aqueous environment, which is a problem for a molecule that will spend its working life in one. Hydrolysis of the maleimide linker before it couples to cysteine reduces conjugation efficiency and introduces unreacted linker-oligo intermediates as product-related impurities. Conjugated maleimides are also susceptible to post-conjugation retro-Michael/thiol-exchange reactions and deconjugation in plasma, where payload can transfer to albumin and glutathione in circulation and cause off-site toxicity.
At Abzena, we sidestep the maleimide problem with ThioBridge®, a site-specific conjugation chemistry that selectively reduces the antibody’s native interchain disulfides and re-bridges them with a bis-sulfone reagent across a stable three-carbon bridge, restoring disulfide connectivity instead of leaving an unstable thiosuccinimide behind. The result is a stable, homogeneous conjugate with low, defined, site-specific loading, at exactly the point where the standard maleimide approach introduces instability.
Linker design carries its own set of trade-offs that shape the rest of development. Cleavable linkers (releasing the oligo in response to the endosomal pH drop, lysosomal enzymes, or the reductive cytoplasmic environment) are essential when the payload needs intracellular release to function, but a linker that cleaves efficiently in a late endosome needs to be stable enough in plasma to survive getting there without shedding oligo into circulation. Non-cleavable linkers solve the stability problem but rely on lysosomal antibody degradation to release payload, introducing a different timing question and making endosomal escape even harder. Linker length matters too, since a linker that’s too short creates steric hindrance from the bulky antibody that blocks the oligo from engaging its target, while one that’s too long creates new aggregation risk. There’s no default setting, and the right trade-offs for a given antibody-oligo combination come from empirical data across a structured panel of conditions.
The oligonucleotide-to-antibody ratio, or OAR, the AOC equivalent of DAR, adds another layer, and controlling it is harder than controlling DAR. OAR directly governs stability, efficacy, safety, and pharmacokinetics, and the optimal value is payload dependent. Charged payloads such as siRNA and ASOs tend toward low, precisely defined loading, since each additional copy adds anionic charge and hydrophilicity that can accelerate clearance, while neutral chemistries such as the phosphorodiamidate morpholino oligomers used in exon-skipping programs can tolerate higher loading. That’s why site-specific chemistry delivering a low, homogeneous, well-defined OAR is so valuable, since it gives you a single, characterizable species to develop around rather than a distribution to defend.
It’s also why standard DAR analytics don’t transfer cleanly. Hydrophobic interaction chromatography works for DAR because small-molecule payloads create measurable hydrophobicity differences between DAR1, DAR2, and DAR4 species. Oligos, being large and negatively charged, contribute essentially nothing to hydrophobicity shifts between OAR species, which makes HIC far less useful here. The validated alternatives (anion exchange chromatography, capillary zone electrophoresis-mass spectrometry, reduced capillary gel electrophoresis, and size-exclusion chromatography-mass spectrometry) come with their own caveat, since mass spectrometry in positive ion mode can underestimate OAR because the oligo’s negative charge suppresses the signal of high-OAR species. In practice, that means relying on multiple orthogonal methods for every batch.
Analytical characterization needs more than the ADC toolkit
AOCs straddle two fully developed analytical frameworks, antibody characterization and oligo characterization, that were each built independently and are, in several respects, mutually incompatible when applied to the conjugate.
UV absorbance at 280 nm, the standard method for antibody quantification, is confounded by oligonucleotide absorbance at 260 nm. SEC retention times shift due to column-sample interactions specific to the conjugate that don’t occur with either component alone. Mass spectrometry for conjugation site mapping, well-established for ADCs, is complicated by the oligo moiety, which requires nuclease pre-digestion before the antibody can be analyzed, a step that introduces its own impurity risk if digestion is incomplete. Platform methods developed for monoclonal antibodies, in several cases, fail to accommodate RNA’s electrochemical properties at all.
The practical consequence is that method development for a new AOC isn’t a matter of selecting from an established menu. It’s genuine development work requiring dedicated time, material, and expertise, which translates directly into timeline cost and into the quality of data available to support early manufacturing decisions. At Abzena, our analytical suite was built specifically for this dual-biology problem, combining SEC-MS, IEX, ICP-MS, UV correction for oligo absorbance, and hybridization-based bioassays for OAR and activity, all run as orthogonal verification rather than sequential steps, because no single method tells you the full picture, and acting on incomplete analytical data in AOC development tends to be expensive. It’s also worth noting AOCs trend toward subcutaneous administration and high-concentration formulations, a key capability difference from ADCs that CMC teams need to plan for.
Manufacturing can be flawless and the payload still won’t work
Even a perfectly manufactured AOC, with an ideal OAR, stable linker, and clean impurity profile, still has to get its payload into the right intracellular compartment to do anything useful, and this is where development runs into a problem that’s partly biological and partly unsolved.
After binding its target antigen, the AOC is internalized, typically via clathrin-mediated endocytosis, entering an early endosome that acidifies as it matures toward the late endosome and eventually the lysosome. For the oligo to reach the RNA-induced silencing complex (for siRNA), the nucleus (for splice-switching ASOs), or any other cytoplasmic target, it has to escape the endosome before lysosomal hydrolases degrade it. This step is the rate-limiting barrier to oligonucleotide delivery. Quantitative NanoSIMS microscopy has shown that only 1 to 2% of GalNAc-conjugated ASOs escape hepatocyte endosomes in vivo, and for AOCs the escape rates are likely comparable, since the endosomal membrane’s lipid bilayer sequesters roughly 99% of internalized RNA therapeutics.
Linker design is the primary manufacturing lever here. Cleavable linkers responsive to endosomal acidification or enzymatic activity can enable earlier payload release, improving the odds of cytoplasmic escape before lysosomal degradation, which is why assessing endosomal trafficking and escape at the lead characterization stage matters. It tells you whether the linker strategy is working in the biological context that counts, not just in a plasma stability assay.
The honest caveat is that endosomal escape isn’t fully solved at the field level. It’s a constraint that shapes which AOC targets work well (TfR1, which undergoes rapid recycling and efficient intracellular trafficking, is the dominant clinical target for exactly this reason) and which targets remain difficult. No manufacturing capability addresses this entirely, because it’s a biological reality of the endosomal pathway that developers are forced to engineer toward and measure carefully. Acknowledging that limit is the kind of clarity that keeps programs from failing for the wrong reasons.
Regulatory strategy has to be built, not borrowed
Regulators have yet to issue AOC-specific CMC guidance. The FDA has issued oligo-focused guidance for clinical pharmacology and nonclinical development, and EMA published a draft guideline in 2024 on the development and manufacture of oligonucleotides, covering quality expectations for manufacturing, characterization, specifications, and analytical control. But AOCs still sit across oligonucleotide, biologic, and bioconjugate precedent, which means CMC packages have to be built from first principles, drawing simultaneously on ADC precedent, oligonucleotide precedent, and regulatory science the field is still generating. Diastereomeric purity limits, co-eluting impurity thresholds, immunogenicity assessment for novel chemical modifications, and stability acceptance criteria all lack settled regulatory standards for AOC conjugates. Every program is, to some degree, negotiating these standards as it goes.
The practical requirement is regulatory foresight built into process development from the start. If a conjugation chemistry produces a diastereomeric distribution that can’t be defended, that needs to be known at process development stage, not at the pre-IND meeting. If analytical methods don’t resolve OAR species to the resolution a reviewer will expect, that gap needs to be closed before it becomes a deficiency letter. At Abzena, our global CMC and regulatory teams are embedded in program development from early process design, because for AOCs, the regulatory strategy and the manufacturing strategy are the same strategy.
A sustainability problem waiting at the edge of the conversation
There’s one more issue worth raising even though it sits outside the immediate CMC critical path. Oligo manufacturing has a sustainability problem that becomes harder to ignore as production scales.
Process mass intensity, the total mass of raw materials consumed per kilogram of API produced, sits between 3,000 and 7,000 for therapeutic oligos, against a median of 170 to 300 for small-molecule APIs. Oligo synthesis is, by this measure, roughly ten to forty times more material-intensive than small-molecule manufacture, driven primarily by the wash solvents required at each synthesis cycle and by the preparative chromatography needed for purification, which alone accounts for roughly half of all process materials used. The solvents involved, including toluene, acetonitrile, dichloroacetic acid, pyridine, and triethylamine, carry REACH classification concerns in several cases and require specialist waste management.
These inefficiencies are inherent to the phosphoramidite synthesis cycle, which relies on stoichiometric rather than catalytic reagents, extensive protecting group usage, and cumulative yield losses that scale with oligo length. Enzymatic synthesis alternatives would offer genuinely greener process chemistry but don’t yet scale to the chemically modified nucleotides therapeutic oligos require. And lyophilization, the standard isolation method, consumes more than 2 kWh per kilogram and represents a multi-day production bottleneck per batch.
It isn’t a regulatory hurdle for AOCs today. But as clinical programs scale and oligo API demand increases, solvent use, PMI, waste handling, and lyophilization bottlenecks will get harder to ignore. It’s worth asking oligo suppliers and CDMOs about PMI and waste strategy early, since sustainability is also a cost and supply-chain question.
Why it’s worth solving
None of this changes the biological upside. AOCs can silence disease-driving genes in cell types, from skeletal muscle to cardiac tissue, that other conjugates simply don’t reach. They address targets small molecules can’t touch and complement the ADC toolkit with a fundamentally different mechanism of action.
Discover AOC Development & Manufacturing Expertise
