TIDES 2026: Summary of Recent Progress in Therapeutic Oligonucleotide Development

By Steve Wolk, Ph.D., founder and chief consultant, Sinawali Biotechnology Solutions

The 2026 TIDES conference was held in Boston, May 11-15.  As always, it was a great opportunity to reconnect with old colleagues, establish new relationships, and of course hear the newest updates across the broad spectrum of oligonucleotide and peptide technologies and the latest clinical advances. 

As a high-level overview, it seems reasonable to organize the highlights into four main buckets:  1) the “shiny things”, meaning new approaches or areas with compelling recent advances and/or those garnering a lot of excitement/attention; 2) areas with continued growth; 3) sectors that are becoming well-established, e.g., boiler plate approaches; and 4) surprises.  Some approaches are on the fence between these buckets, and will be identified as such.

Because of my less extensive expertise with peptides, I will focus on oligonucleotide sectors.  I will make brief comments on peptide technologies where appropriate, but leave the more complete summary to my colleagues with deeper expertise.  Also, for full disclosure, I still own stock in Editas Medicine, and therefore will not discuss them in this review.  I have no financial interests in any of the other companies mentioned.

Figure 1

Overview of Buckets for Highlights from TIDES 2026 Oligonucleotide Coverage

The Shiny Things	Continued Growth	Becoming Boiler Plate	Surprises (underrepresented)
•   high purity oligonucleotides •   AI	•   in vivo genetic medicines •   gene editing technologies •   mRNA/LNP characterization1 •   targeted delivery2	•  AOC Analytics •  bioconjugation1 •  characterization of germline transmission	•   in vivo T-Cell therapeutics •   AI •   alternate in vivo delivery tech

1) hybrid: continued growth/boiler plate

2) hybrid: continued growth/shiny things

The Shiny Things: New and/or Increasingly Hot Tech

High Purity Oligonucleotides - Enzymatic Ligation

One very desirable property for moving oligonucleotide-based therapeutics into the clinic has been the manufacture of high purity molecules, which has always been a challenge for oligonucleotides produced via solid phase phosphoramidite chemistry, which has been the industry standard for decades.   This is especially true for longer oligonucleotides.  For example, in the field of CRISPR therapeutics, the manufacture long of long guide RNAs (gRNAs), can be about 100 nucleotides long, and gRNAs for newer technologies such as prime editing can be 130 nucleotides or more.  For these long guides, phosphoramidite chemistry, struggles to produce high purity products because 1) the large number of coupling steps required during synthesis (n-1 steps for an oligonucleotide of length n); 2) the increasing difficulty in purifying full-length product from chemically similar impurities at longer lengths, and 3) the increasing difficulty of analytically characterizing the impurity profile at longer lengths.

One solution to the purity problem gaining significant traction is synthesis of shorter oligonucleotides via standard solid phase methods, then utilizing enzymatic ligation to create full length sequences.  This approach has a number of advantages, including:

• the ability to manufacture shorter (and therefore higher purity) sequences via solid phase approaches

  • the shorter pieces may not require purification prior to ligation, saving time and money in the manufacturing process

• the most significant impurities in solid phase synthesis (e.g., canonical n-X truncations) are filtered out for the oligonucleotide on the 3’ side of ligation reactions since they cannot ligate

• for gRNAs, key sequence sections such as the spacer sequence in the guide can be chosen as shorter sequences to ensure high purity in these key sections

• for gRNAs, the tracer sequence, which is constant for each type of guide (e.g., asCas9), can be manufactured in bulk and used for every gRNA within the group

• can facilitate generating high purity manufacture of modified oligonucleotides (e.g., HNA, LNA)

  
  

Potential challenges include:

• the need for an enzymatic process at large scale, which often presents scale-up challenges

• the need to remove protein in the purification process

• sequence dependence of the ligases, which would add process optimization time and cost

• the need for removal of bridging sequences for gRNAs

  • it should be noted that double stranded siRNAs are self-templating and do not require bridging sequences

    
    

Credit should be given to Ajinomoto on this subject, since they were giving talks on this approach many years ago for siRNAs, before the approach gained traction.   In addition to Ajinomoto, other contract manufacturers, including Hongene, Avecia, Almac Sciences, have shown significant progress in creating high purity siRNAs and gRNAs.  BaChem presented on scale-up, to gram and kg manufacturing. 

Within biotech/big pharma, Alnylam has also presented on this approach for siRNAs for the last few years.  Key drivers in the siRNA arena include cost, timelines, and quality. Caleb Coly and Sabine Fenner (Lilly) showed new results for optimizing their process for large scale (3 kg) manufacture of siRNAs via ligation, indicating Lily’s belief in the long-term success of the technology at large scale.  Key variables in the Lilly study included temperature, pH, enzyme loading, and fragment ratios.  Purification yield was ~88% via IEX-MCSGP (ion exchange chromatography (IEX) and multicolumn countercurrent solvent gradient purification).  Enzyme removal was assessed via a Bradford assay (in-process) and LC/MS/MS (release).  It is also interesting to note that the peptide world utilizes a similar concept.  In Michael Kopach’s plenary talk, he noted that Lilly is also applying a hybrid approach to peptide synthesis, where shorter peptides are manufactured by a solid phase approach, then linked together via solution methods.

Other Methods for High Purity Oligonucleotides

Novel Phosphoramidite Chemistries.  Another option for generating high quality long guides is the use of alternate chemistries.  For example, Cirena, founded by Marv Caruthers and Doug Dellinger, manufactures long RNA oligonucleotides via solid phase synthesis utilizing 2’-O-thionocarbamate nucleoside phosphoramidites chemistry, which can generate RNA sequences of >200 nucleotides at much higher purity than possible with standard TBDMS phosphoramidite approaches.

Full Enzymatic Synthesis.  Ella Meyer (EnPlusOne Bio) discussed full enzymatic approaches to oligonucleotide synthesis, though this strategy has not seen broad implementation.  The cost, timelines, and quality of this approach will need to be assessed as new innovations are introduced.

Artificial Intelligence (AI)

AI, of course, is currently a shiny thing well beyond the biotech sector and therefore worthy of highlighting.  In fact, some even attribute increased investment in the AI sector as a factor in the current biotech winter.  The impact of DeepMind’s AI-based AlphaFold on predicting protein folding led to sharing the 2024 Nobel Prize in Chemistry.

At TIDES this year, presentations included Kyler Lugo (Nosis Bio), who discussed the use of deep learning to analyze the receptome in order to discover new receptors for the facilitation of extrahepatic delivery of siRNAs.  He noted that currently, 80% of the papers for extrahepatic targeting focus on 2% of the receptors.  The current lead program at Nosis Bio is focused distal alveolar cells in the lung via clathrin-mediated endocytosis, where their in silico approach has led to the development of a low nM binding affinity.  Greg Hoffman (Deep Genomics) discussed the prediction of sites for ADAR (adenosine deaminase acting on RNA) editing, as well as gRNA design.  Similarly, Gerald Platenburg (ProQR) noted that AI reduced candidate design discovery from ~30 months to ~2 months.  Devan Shah (RNAV8) discussed the data quality problem in using AI to guide mRNA sequence design.  Factors impacting data quality included biological nuances such as the effect of sequence modifications (e.g., Y  on ribosome loading, UTR secondary structural variation, cell-type dependence, and the impact of ribosomal binding proteins (RBPs).  Loic Vincent (Korro Bio) shared that machine learning applied to oligonucleotides chemistries is driving meaningful gains in oligonucleotide potency and durability.

On the peptide side, several vendors (XtalPi, Aurigene Pharmaceutical Services) discussed the AI in peptide drug design.

Continued Growth Areas

In Vivo Genetic Medicines

In Vivo CAR-T.  As described by John Zuris (Zuris Consulting), ex vivo CAR-T treatments have shown great success, showing response rates of 40-81% in B-cell malignancies.  However, similar to other ex vivo gene editing treatments such as those for sickle cell anemia, the lack of scalability, the extensive infrastructure (apheresis, cryopreservation, GMP manufacturing, infusion, hospital stays for immunocompromised patients), long processing times (3-4 weeks), and high cost of goods ($300-500K) can be prohibitive.  In addition, the necessity of lymphodepletion (significantly reducing existing the unmodified T-cell population) and T-cell exhaustion during ex vivo expansion can all limit effective implementation and/or create additional health risks.

These limitations are driving the development of in vivo delivery therapeutics.  Among the ~60 companies working in this space, the delivery approaches are currently divided between engineered viral particles (lentiviral, VLPs, and fusosomes) and targeted LNPs (tLNPs), where receptor choices include CD3e, CD4, CD5, CD7, CD8, and CD3+CD28 bispecific.  Each approach has specific advantages and challenges. For example, targeting CD8 is good for autoimmune applications since it targets cytotoxic T cells with good specificity, but is not as potent as CD3e, CD5, or CD7.

Jane Wang (Tessera Therapeutics) discussed their strategy to optimize tLNPs for extrahepatic targeting, which included iLipid screening, LNP formulation optimization, target moiety discovery, and tLNP engineering.  tLNPs optimized by this process contain mRNAs for Tessera’s gene writer and the template a +CD20 CAR.  The data showed that T cells maintained their normal phenotype, and complete B-cell clearance was observed in the spleen at day 32.  Similarly, CD20 CAR-T generated in NHPs drove B-cell clearance in blood and lymph nodes (confirmed by lymph node histology) with a single low dose (0.05 mg/kg).

Ali Nahvi (Nava Therapeutics) discussed tLNPs that are specific to cytotoxic T cells; two systemic administrations achieved full B-cell depletion across tissues, transient CAR expression, and a clinical chemistry profile consistent with a tolerable LNP.

Neoantigen Therapies.  Kristine McKinney gave an overview of Moderna’s neoantigen program, where whole exome sequencing (WES) is done on patient tumors to identify neoantigens (tumor specific antigens) which can then be programmed into a patient’s T cells.  The process involves identifying tumor specific mutations, evaluating them for tumor response, encoding the strong responders in mRNA (up to 34 per therapy), then introducing the mRNA via LNP intramuscularly.  The neoantigens are then produced and presented by antigen presenting cells (APCs), which trains and activates neoantigen-specific T cells.  The treatment, in combination with a PD-1 inhibitor (pembrolizumab), showed significant improvements in the risks of recurrence, distant metastasis, and death compared to pembrolizumab alone in various cancers.  Future off-the-shelf (OTS concepts) are under development, which can be organized into three buckets – immune cell targeted, tumor targeted antigens shared across patients, and cell therapy enhancing approaches to improve T-cell persistence.

Platform Approaches.  Several CDMOs presented on platform approaches for the manufacture of  in vivo CAR-T therapeutics, demonstrating their belief in the strong future of this therapeutic approach.  Telmo Graca (Lonza) discussed optimizing mRNA and tLNP through refined lipid compositions, optimized formulation parameters, and enhanced methods for characterization, which all support more effective targeted delivery and controlled biodistribution.  Martin Rable (Cytiva) and Shimobi Onuoha (Chimeris) discussed their novel ionizable lipids and LNP/tLNP platform designed to efficiently reach lymphoid tissues and T cells for enhanced CAR expression, which is combined with scalable manufacturing and a strong analytical toolbox.

Gene Editing Technologies

“Next Gen” CRISPR Gene Editing.  Base Editing, Prime Editing, Gene Writing continue to make strides with regards to enzyme engineering.  This year, Ben Kleinstiver (MGH, Harvard Med) discussed the continued development of CRISPR base editors, which represent a powerful approach to treat single nucleotide variants (SNVs), which are the most frequently occurring cause of human genetic diseases. Base editors contain a Cas9 linked to a deaminase.  Recent optimizations to the Cas9 focused on creating  a “PAM-less” variant, which would potentially allow the Cas9 to bind at more mutation sites instead of being limited to sites adjacent to the PAM.  Engineering of the deaminase utilized machine learning models to develop the edit type (e.g., which base conversion), the edit window, and the reduction of bystander edits (where nearby bases identical to the target base are also deaminated).  Many successful variants were created using and a 24 or 96 well format of sequence variants and a biochemical assay to test activity.  The next step will be to reduce the new set of variants to a small toolbox that can address most known mutations.

Large Knock-ins.  The ability to introduce large gene knock-ins (e.g., the addition of an exon or even an entire gene) is one of the current holy grails of the gene editing field.  For example, diseases like cystic fibrosis or retinitis pigmentosa have hundreds or even thousands of pathogenic variants, so the feasibility of having an approved therapeutic for each variant of a disease quickly becomes untenable.  The ability to replace an entire exon, or even an entire gene, to treat some or all of the variants with a single therapeutic, becomes a much more pragmatic solution.  Current technologies to achieve exon or whole gene knock-ins include viral approaches (gene therapy), prime editing variants, site-specific recombinases (SSRs), combining prime editing with integrases (PASTE and PASSIGE), transposons, and retrotransposons.  A full analysis of these approaches will be the subject of a future article.  Some of these approaches require delivery of a DNA template, which represents an unsolved technical challenge, and is discussed in the next section.  Other solutions, such as Prime editing and retrotransposons, utilize an RNA template to circumvent this difficulty.

Ari Friedland (Typewriter Therapeutics) discussed the use of retrotransposon technology for development of in vivo CAR-T therapies and liver-directed genetic medicines.  Their construct is a two RNA system (an R2 mRNA and template RNA) that are engineered and optimized for transgene cassette insertion in the 28S ribosomal DNA via the target primed reverse transcription (TPRT) mechanism.  The resulting constructs showed strong expression in vitro in T cells and primary human hepatocytes, and yielded 3-4% expression in vivo in rats when delivered in an LNP.

Gregory Cost (Addition Therapeutics) showed a very similar two RNA platform (dubbed PRINT, developed in Kathleen Collin’s lab at UC Berkeley), demonstrating delivery of GFP to rat, NHP, and human cells, with durable expression in vivo (rats and NHPs) for >1 year with a single dose.  Using this technology, Addition has generated preclinic data for obesity, Fabry disease, HIV protection, and wet AMD.

DNA Donor Technology.  Nonviral delivery of DNA templates remains an unsolved technical problem for gene editing technologies from the perspective of efficiency of delivery and immune response caused by double stranded DNA immune sensors (e.g., the cGAS and STING inflammatory pathways).  Potential workaround technologies include terminal modifications, circular templates, circular single-stranded templates, and RNA templates.

Ben Kleinstiver (MGH, Harvard Med) discussed a potential solution for use with recombinase approaches for insertion of large DNA donors.  The template is delivered as a circular single-stranded DNA (cssDNA), which largely avoids the cGAS and STING inflammatory pathways.  In the INSTALL-2e platform, a short chemically modified oligonucleotide is annealed to the cssDNA, which creates a double-stranded DNA region large enough for the recombinase to recognize.  In initial studies with INSTALL 2e, no toxicity was observed below doses of 3 mg/kg, and 1% integration was observed.

Personalized Gene Editing.  A major milestone was achieved last year with the successful treatment of a patient (baby KJ) with a personalized CRISPR gene editing therapy within a remarkably short development time (~8 months).  This achievement resulted from a collaboration between researchers at Children’s Hospital of Philadelphia (CHOP), University of Pennsylvania (U Penn), and various industrial collaborators.  This year, Kiran Musunuru (U Penn) discussed progress on the regulatory side for future custom gene editing treatments, where similar base editors would be needed for a potentially large number of individual mutations.  Dr. Musunuru discussed ongoing discussion with agencies regarding the possibility of developing umbrella clinical trials (e.g., for phenylketonuria (PKU) and urea cycle disorder (UCD)) and permitting amendments to INDs with in vitro data for each variant specific guide RNA.

New Advances in Characterization of mRNA Therapeutics (biology, production, and analytics)

Biology. Andre Dziembowski (Intl. Ins. of MCB) discussed the characterization of poly(A) tails, the shortening of which is a known regulatory mechanism for endogenous mRNAs as well as mRNA vaccines and therapeutics.  The Oxford nanopore technology (ONT) was utilized for the characterization, since modifications are sometimes involved, e.g., SpikeVax from Moderna ends with mYmYAG.  In model cell lines (e.g., HEK293 cells), the poly(A) tail of SpikeVax was degraded within 48 hours, as expected.  In vivo studies with mice, however, the poly(A) tail was observed to be extended up to 200 nucleotides after removal of the mYmYAG.  TENT5A poly(A) polymerase was shown to be involved in this process and therefore the stabilization of the vaccine.

Production – elimination of dsRNA.  Double stranded RNA (dsRNA) as a known impurity in mRNA produced by in vitro transcription (IVT), and is known to cause immune activation that can reduce efficacy and cause toxicity.  Dong-Eun Kim (Konkuk Univ.) proposed that the dominant mechanism for formation of dsRNA is 3’-end annealing and transcription from the anti-sense strand.  Reduction in the dsRNA level can be addressed via upstream approaches (e.g., preventing its formation) and downstream approaches (removal during purification).  Athul Sanjeev (Verve/Eli Lilly)) discussed tethering a dsDNA binding domain to the T7 RNA polymerase to increase selectivity for dsDNA template over the mRNA product as the template.  The development process involved surveying linker lengths and attachment sites, and the optimization yielded constructs with reduced dsRNA production and also improved salt tolerance of the polymerase.  Dong-Eun Kim (Konkuk Univ.) developed a nicked template which reduced antisense transcription.

Advances in mRNA Analytics.  Standard release assays to measure critical quality attributes (CQAs) are becoming well established for mRNA analytics, as demonstrated by recent white papers from USP and Biophorum.  However, recent innovations are being developed that will increase the depth of characterization and will likely lead to deeper understanding of the CQAs.

mRNA Analytics - Higher order structure (HOS).   HOS includes intramolecular folding as well as intermolecular associations.  The intermolecular species include distinct multimer structures, which are likely reversible, as well as amorphous aggregates, which are likely irreversible.  With the exception of the measurement of aggregate levels by SEC in the USP document, there is no other specific reference to HOS in either the USP or Biophorum white paper.

Eivor Ornskov (AZ Gothenberg) showed that HOS will impact standard purity analyses, such as the observation of additional peaks in CGE chromatograms resulting from incomplete denaturation at 8M urea, and the appearance of additional peaks and changes in retention times in IP-RP-UPLC.  Both of these artifacts can result in overestimation of shortmers and therefore underestimation of purity.   For characterization of HOS elements, additional assays include SHAPE analysis (Selective 2’-Hydroxyl Acylation analyzed by Primer Extension), which identifies flexible and unpaired nucleotides via chemical modification followed by detection via using primer extension reactions, and mass photometry, which can detect dimers and aggregates via light scattering from particles a glass surface.  Circular dichroism can also to be used, though changes may be difficult to interpret directly in terms of structural changes.  In this case, comparison to standards is a pragmatic solution.

Similar to the SHAPE approach, Wayne Doyle (Eclipse Bio) showed that treatment of the mRNA with chemical reagents followed by sequencing analysis (though the exact method was not specified) can reveal unpaired nucleotides.

Mark Dickman (Univ. Sheffield) showed a novel “non-denaturing” LC method (20% acetonitrile in mobile phase B) that allowed detection of mRNA dimers and multimers.  The multimer peaks were collected as fractions and then analyzed using mass photometry.  He used this method to show that multimerization was concentration dependent, that the multimers could be removed by heat treatment (e.g., 75 °C for 5 minutes), and also showed how modifications such as psuedouridine and N1-methylpseudouridine impacted dimer formation.

mRNA Analytics – Identity and Sequence.  Mark Dickman (Univ. Sheffield) discussed the limitations of traditional indirect sequencing methods for mRNA identity (Sanger, NGS, and PCR), which included inability to detect modifications, sample prep and data analysis time, amplification bias, and inaccurate poly(A) tail analysis.  Similarly, direct methods such as ONT can struggle with modifications, accuracy of base calls, and validation issues, and inaccurate poly(A) tail analysis.  He then showed an automated IP-RP-UPLC/MS/MS method that included an online partial digestion step, followed by UPLC/MS/MS, which yielded 90% sequence coverage in less than 60 minutes, and could detect modifications.

mRNA Analytics – Integrity.  Historically, integrity has been measured either CGE or IP-RP-UPLC, where the main peak intensity is compared to the total intensity in the electropherogram or chromatogram, respectively.  The methods show the presence of impurities, but can only provide identity of the impurities if coupled to mass spectrometry, as described by Mark Dickman (Univ. Sheffield) above.  More granular information can also be obtained via sequencing data.  Wayne Doyle (Eclipse Bio) showed that sequencing of the impurities can reveal locations in the sequence where breaks are occurring, and that these locations are most often loops and other unpaired regions that are susceptible to enzymatic (e.g., nucleases) and chemical (e.g., hydrolysis, sometimes metal mediated) attacks.

As discussed above, recent work presented at TIDES by Eivor Ornskov (AZ Gothenberg) show that additional peaks resulting from HOS moieties can be observed in the CGE or IP-RP-UPLC chromatograms that distort measured purity values.

Analytics - dsRNA.  A remaining unsolved technical challenge regarding dsRNA impurities is that little is known about the dominant mechanisms that produce the impurities, which forms are immunogenic, which forms are detected by current assays such as dot blot or K2/J2 ELISAs, and whether the standards used generate accurate quantitation.  Wayne Doyle (Eclipse Bio) showed that sequencing methods, including Illumina-based capture methods followed by sequencing analysis, provided more granular insight into the mechanisms that generated the impurity (e.g., 3’-end annealing and transcription from the anti-sense strand).

Analytics – 5’-Cap and 3’-Poly(A) Tail Characterization.  Mark Dickman (Univ Sheffield) showed that a complete digestion followed by LC/MS/MS could identify of the 5’-cap structure, and partial digestion methods could yield quantitation of capping efficiency, as well as the length and distribution of the poly(A) tail.

Both Andre Dziembowski (Intl Ins of MCB) and Telmo Graca (Lonza) discussed characterization of the poly(A) by ONT, which included length, and identity (e.g., the presence of modifications).

Analytical Control Strategies.  Mike Webb, (Mike Webb Pharma) spoke about quality by design (QbD), as defined in ICH Q8, which develops predefined objectives that emphasize product and process understanding and control.  Critical process parameters and analytical testing are central to the control strategy.  For example, CVs in analytical methods could be due to true process variation, sample handling, and/or variability in the analytical method is this must be understood.  Under QbD, the target product profile (TPP) must be defined, then critical process parameters (CPPs) and critical quality attributes (CQAs) must be defined to achieve the TPP (including parameter ranges), and then DOE experiments combined with a failure mode analysis (e.g., FMEA) are implemented to better understand and test the process.  This set of steps, along with literature and previous experience, allows development of parameter ranges test during process validation.

New Advances in LNP Characterization (hybrid: continued growth/boiler plate)

Telmo Graca (Lonza) discussed the analysis of tLNPs with a flow NanoAnalyzer, which is a miniaturized flow cytometer that measures the interaction of single particle’s targeting moiety with a fluorescently labeled antigen/protein.  Readouts include the fraction of moiety-positive LNPs, the ligand number and density per LNP, as well as the tLNP size and particle number.  He also mentioned the efficiencies gained from defining and evaluating analytical target profiles and CQAs, digital workflows, and cryoEM. Though he didn’t discuss the details for cryoEM, this technique allows direct observation of particle size distribution without the averaging artifacts observed with scattering methods.  It also shows deviations from spherical shape and blebbing.

Targeted delivery (hybrid: continued growth/shiny things)

The majority of the reported progress for targeted delivery utilized LNP technology.  As summarized by Jane Wang of Tessera Therapeutics, LNPs naturally target the liver, therefore extrahepatic targeted delivery involves de-targeting the liver through the choice of the ionizable lipid and LNP formulation, and active targeting is achieved through both the choice of targeting moieties and the engineering of the LNPs to add the moieties to the surface.  On the road to in vivo delivery to treat sickle cell disease and beta thalassemia, Tessera has achieved 60% editing of long-term hematopoietic stem cells (LT-HSCs) in the bone marrow in mice and nonhuman primates, well above the predicted threshold of 30% needed for therapeutic impact.

For therapies treating diseases of the brain/central nervous system, efficiently crossing the blood brain barrier remains one of the biggest hurdles.   Using the transferrin (TfR) receptor for receptor mediated transcytosis is currently one of the dominant approaches.  Padma Akkapeddi (Denali Therapeutics), discussed this approach in the context AOCs.  Many of the antibodies currently in use target the apical domain of the receptor, though some target protease domain, and it is important to not interfere with transferrin binding.  While effective at crossing the BBB, accumulation in other high TfR expressers (bone marrow and spleen) was observed.  Intermediate receptor affinities have been found to be ideal, e.g. pM binders don’t release to complete transcytosis, and the receptors degrade.

Vadim Dudkin (Souffle Tx) discussed his company’s vision of “finding a GalNAc for every cell in the body” via their platform of ligands (mAb, nanobody, peptide, etc.), conjugation chemistries (Click, maleimide, and chem-enzymatic approaches (sortase, transglutamase)), and potent payloads.  He discussed targeting muscle via an undisclosed receptor C633, which was found to be more potent than TfR.

Liqun Wang (Wyss Inst, Harvard) discussed POC for brain targeting of AOCs via anti-TfR shuttles (BTSTfR).  Data included , 55% KD of Malat1 transcript after IV administration of an ASO conjugate in mice, and 90% knockdown of SNCA transcript in whole brain lysates using an anti-TfR-siSNCA conjugate.

Becoming Boiler Plate 

Antibody Oligonucleotide Conjugate (AOC) Analytics

Jacob Guo (Genscript) gave an overview of their current standardized menu of analytics for antibody oligonucleotide conjugates, mentioning that the first AOC approvals are expected in 2026 or 2027.  Companies in the space include Avidity Biosciences, PepGen, Dyne Therapeutics, Entrada Therapeutics, Vertex, Denail Therapeutics, Tallac Therapetics and Aro Biotherapeutics.

CQAs include drug-antibody-ratio (DAR), purity, heterogeneity, aggregation, and efficacy.  Key measurements and associated methods include conc. (BCA), purity (SEC, IEX), residual siRNA (IEX), residual antibody (HIC), pI (CIEF), and modification site (digestion + mass spectrometry (MS)), and DAR distribution (MS, SDS-PAGE, IEX, HIC and/or CE-SDS, depending on Ab construct).

Bioconjugation (hybrid: continued growth/boiler plate)

Bioconjugation refers to modification of a biomolecule to change its functionality.  In the context of oligonucleotide therapeutics, common examples include attaching a small molecule or antibody variant to an siRNA to achieve targeting (e.g., GalNAc-siRNAs or AOCs, respectively), or modification of the surface of an LNP with a targeting moiety.

As pointed out by Vadim Dudkin (Souffle), widely used approaches include chemical (e.g, maleimide, Click) and enzymatic (e.g., sortase, transglutamase) approaches.  Padma Akkapeddi (Denali Therapeutics) covered the most advanced clinical programs for AOCs that use the transferrin receptor to cross the blood brain barrier (BBB) via transcytosis.  Dyne Therapeutics, Vect-Horus and Denali Therapeutics utilize cysteine-maleimide conjugates, and Roche’s “brain shuttle” technology, which uses transglutamase mediated conjugation.  She also pointed out that in addition to the choice of chemistry, the conjugation site of matters (e.g., it can affect clearance), as does the drug-antibody ratio (e.g., 1:1 is optimum for Ab-ASOs and Ab-siRNAs, but is >4:1 for Ab-PMOs).

Other speakers focused on the chemistry for creating modifications of the oligonucleotides.  Michael Booth (Univ, College, London) noted that small molecules can be conjugated to the backbone modifications through reaction of phosphorothioates with alkyl halides, terminal sugars can modified via Click chemistry (e.g., terminal linker-N₃ structures are reacted with alkynes linked to small molecules, nanoparticles, or nanogels), and exocyclic amines on nucleobases can be reacted peptide/proteins modified with NHS-esters.

Hassan Fakih (U. Mass. Chan) focused on the current challenges in extrahepatic delivery of siRNAs, which include stability and immune response, PK/biodistribution, and productive tissue extravasation/cellular uptake.  He pointed out the well-established ideas that the backbone modifications solve stability to nucleases, improve potency, lower immune activation, and that both the type of modifications and the pattern of the modifications impact these performance parameters.

Novel chemistries were discussed by Derek O’Flaherty (U Guelph), including novel disulfide linkages of pyrimidine nucleobases) and Kurt Vesterager (Arhaus Univ.) covered new azide modification at phosphorous, and reviewed previous work with sulfonyl azides and guanidinium azides.

Gene Editing Characterization - Germline Transmission

Testing for vertical germline transmission verifies whether intentionally edited or unintentionally affected genetic modifications enter the reproductive cells (sperm, egg, or embryos) and whether they are passed down to subsequent generations.

Laura Sepp-Lorenzino (GNMmeds LLC) pointed out the EMA now has a draft ATMP guideline which was written for viral vectors, but is adaptable to CRISPR gene editing.  The FDA currently has no specific guidance, but references the ICH considerations document.

In preclinical testing of Intellia candidates NTLA-2001 and NTLA-2002 in NHPs, editing was observed in the ovaries of the females, but no editing was observed in the sperm of the males.  Subsequent vertical germline transmission (per agency guidelines) of female mice that were treated with a high dose resulted in about 1% editing detected in the ovaries.  These animals were then mated with untreated males, and genes of the progeny (323 pups) were analyzed.  If germline transmission occurred, 50% editing would be observed in the offspring, and none was observed.  Subsequent testing with LNPs containing luciferase showed that delivery was happening in theca cells, but not the oocytes, consistent with the lack of germline transmission observed.

Surprises – Underrepresented Sectors

In Vivo T-Cell Therapeutics

As described by John Zuris (Zuris Consulting), ex vivo CAR-T treatments have shown significant continued growth, with ~60 companies working in the space.   Considering the amount of recent progress, it was a little surprising that there were only a few presentations during the conference.  This may have been driven by the fact that ASGCT was being held at the same time across town, and many companies chose to present their work at this more clinically focused conference.

AI

Considering the current magnitude of the field, the depth and breadth of the coverage of AI-based work was less than expected this year.  I suspect that investment in AI approaches in the private sector is increasing, but the current results are viewed as proprietary and/or not quite ready for prime time, and/or presented at other conferences.  It is also possible that the data quality problem is hindering progress at this time.

Alternate In Vivo Delivery Tech

Alternatives to LNP/tLNP delivery continue to be explored including viral approaches (AAVs, virus-like particles (VLPs), and bacteriophages) and nonviral delivery (NVD) approaches (liposomes/ liposomal LNPs, extracellular vesicles (EVs), polymer nanoparticles, inorganic nanoparticles, protein-based nanoparticles, and unencapsulated variants.  Progress in this sector may also have chosen to presented at ASGCT, or the progress has been slow and/or is proprietary.

Conclusions

The progress in the oligonucleotide and peptide therapeutic space continues to be compelling, as reflected by the content in three main buckets detailed in Table I (shiny things, continued progress, and becoming boiler plate).  Continued progress is being made in traditionally challenging areas like safety, delivery, stability, potency, characterization, and manufacture of high purity materials.  There is continued growth in oligonucleotide-based therapeutics going into the clinic and gaining regulatory approvals each year, which of course continues to create comfort within the regulatory agencies regarding the safety and efficacy of these therapeutics.  Another marker is the consistent growth of the TIDES Conference year after year, and the corresponding continued growth in the depth and the breadth of the science.  The next few years should be really exciting to watch.

About the Author

Steve Wolk, Ph.D., is founder and chief consultant for Sinawali Biotechnology Solutions. He was formerly vice president of chemistry and Boulder site head for Editas Medicine, where he led the advanced technology, analytical sciences & structural biology, and process chemistry teams. In addition to an extensive history directing analytical groups for characterizing oligonucleotides, proteins, small molecules, and polymers, Wolk has led efforts to develop new technologies for delivery of CRISPR therapeutics and worked on an aptamer-based proteomic technology. He received his bachelor’s degree in chemistry from U.C. San Diego, where he received the Harold Urey Award. Wolk completed his Ph.D. work in biophysical chemistry at U.C. Berkeley under Ignacio Tinoco, Ph.D., where he used 2D-NMR and other spectroscopic techniques to characterize nonstandard DNA structures. Wolk also has broad experience in leadership, including management, developing future leaders, and creating/revising governance programs.