Taking base editing beyond proof of concept
The Pin-point™ base editing platform is built to support the path from early discovery toward therapeutic development, bringing together innovative base editing technology, scientific expertise, and access to the licensing rights needed to move programs forward.
New data being presented at ESGCT extend the platform into in vivo LNP delivery, alongside an expanding portfolio of CBE and new ABE reagents featuring AI-engineered adenine deaminases developed by Profluent.
Through our Access Program, researchers can work directly with the scientists who developed the Pin-point platform to explore ambitious editing challenges and generate data around new applications. Or, it's easy to get started independently with ready-to-use reagents.
Wherever you are in your translational journey, Discovery and Commercial License options provide flexible paths to take the Pin-point technology forward into therapeutic development.
Tailor GMP reagents to your specific goals
Cell therapy research and development can be challenging and time-consuming - but getting support shouldn’t be.
Leverage our expertise to establish new, validated procedures for your custom GMP project.
Our GMP bioprocessing team is ready to partner with you to transition your research from bench to bedside.
Navigating the future of gene therapies
-
How do you identify and validate the right targets?
Build confidence in your biology before moving forward.
-
How do you engineer the right payload and optimize its delivery?
Combine precise editing with delivery technologies designed to improve performance across relevant cell types.
-
How do you build confidence in therapeutic performance?
Connect cellular, functional and in vivo evidence to understand how your candidate performs.
-
How do you scale a complex process with confidence?
Build reproducible, controlled workflows designed to support the transition from development to manufacturing.
-
How do you maintain confidence in every batch?
Bring together sample preparation and orthogonal analytical technologies to characterize critical quality attributes.
Scientific highlights
Check our posters:
Modular lipid nanoparticle delivery of base editor components enables versatile platform strategies for therapeutic genome editing
Poster Number: P0951
Poster session: 1
Time: Wednesday 28 October from 13:30 to 15:3
Base editors offer a broad range of opportunities for precise and efficient genome editing without requiring the introduction of genotoxic double strand breaks, making them especially suitable for engineering therapeutic products. Despite progress in translating experimental genome editing technologies to the clinic, major obstacles exist in establishing the appropriate manufacturing and regulatory processes required to scale their adoption. One proposed strategy to address these challenges is the development of platform-based therapeutic approaches that leverage prior regulatory approvals of individual editor components. As both the precise correction of pathogenic variants and the engineering of novel therapeutic variants require assembling target-specific configurations of Cas and deaminase components, technologies that enable modular reconfiguration of base editors may be particularly well suited to support such platforms.
Due to the complexity and cost associated with ex vivo genome editing, broadening access to therapeutic base editing will ultimately require development of effective in vivo delivery strategies. By combining established LNP delivery technologies with modular base editing platforms, it may be possible to re-use individually encapsulated editor components across therapeutic programs with greater flexibility than Cas-deaminase fusion editor architectures, thereby addressing key manufacturing, regulatory, and scalability constraints in therapeutic genome editing.
We developed an RNA encoded, aptamer assembled modular base editing platform that supports high efficiency, precision editing in HSPCs and T cells. Editing outcomes at target sequences can be modulated both by the intrinsic activity of the deaminase module and by adjusting the stoichiometry of individual editor components. By modifying the editing outcome at the HBG promoter using either approach, we show that a single A>G conversion is sufficient to efficiently upregulate fetal hemoglobin expression in HSPC-derived erythroid progenitors. Dissociating the DNA binding and DNA modifying activities of modular base editors enabled independent delivery of each component, either by ex vivo electroporation or LNP transfection. Comparable high editing efficiencies (>80%) were achieved in both HSPCs and T cells following electroporation or LNP delivery, whether editing components were delivered individually or co-encapsulated.
These findings demonstrate that the individual mRNA and sgRNA components of modular base editors can be efficiently delivered to primary human HSPCs and T cells by LNP transfection ex vivo, highlighting the potential extensibility of this approach to in vivo applications. The ability to independently or co-encapsulate multiple RNA components underscores the flexibility of the modular base editor architecture and suggests a path toward component-level manufacturing and regulatory strategies for therapeutic genome editing. We envision that the potential for re-use of individual component LNP formulations may create opportunities to iteratively de-risk deaminase and Cas variants with distinct sequence targeting specificities across successive therapeutic programs. Together, these findings indicate that modular delivery architectures may enable scalable, platform-based development of base editing therapies with tunable precision and functionality.
Decoupling deaminase recruitment and transgene targeting resolves tradeoffs in multiplexed base editing workflows
Poster Number: P0948
Poster session: 2
Time: Thursday 29 October from 14:00 to 16:00
The development of advanced cell therapies relies heavily on complex, precise genome engineering processes that include knockout of multiple targets and site-specific integration of exogenous transgenes. Base editors offer a broad range of opportunities for precise and efficient multiplexed genome editing without requiring the introduction of genotoxic double strand breaks and associated risks to genome stability. The nickase component of base editors can also be re-purposed to safely target transgene integration in parallel with base editing; however, parallel deployment is limited by unintended base editing at the transgene integration site. A number of strategies have been proposed to overcome this challenge, including the delivery of additional Cas components with orthogonal sgRNA compatibilities or design of sgRNAs at the integration site lacking base editor target nucleotides. However, these approaches either increase the complexity of the editing machinery or constrain guide RNA design for targeting transgene integration, limiting scalability and flexibility for multiplexed cell engineering workflows.
To develop a flexible, simplified approach to targeted transgene integration in combination with base editing we developed an RNA encoded, modular base editing platform that assembles via an aptamer encoded in the sgRNA, enabling selective recruitment of editing components to distinct genomic sites. By de-targeting the deaminase component of the modular base editor from transgene integration sites it was possible to markedly reduce disruption of the DNA nicking activity required for efficient transgene integration, increasing efficiencies of precise integration from double stranded DNA templates by ~2x in iPSCs. Base editing efficiencies remained high (>70%) when combined with targeted transgene integration in both iPSCs and T cells, enabling the engineering of HLA knockout hypoimmune iPSC clones expressing both constitutive HLA-E and inducible CD47 transgenes, and CAR expressing T cells knockout for HLA-1, PDCD1, CD52 and TRAC. Integration of large transgenes (~7kb) using dsDNA donors was achieved at >40% efficiency in iPSCs, and transgene integration precision could be increased ~2x without impacting base editing efficiency by optimising the relative quantities of the independent base editor modules. Optical genome mapping demonstrated that transgene integration occurred specifically at the target site in simultaneously base edited iPSC clones, and interchromosomal translocations and untargeted AAV donor integrations were undetectable in bulk analysis of multiplex base edited CAR T cells.
Aptamer-recruited modular base editors provide a generalizable strategy to achieve safe, high efficiency multiplexed gene knockout in combination with precise targeted transgene integration in primary human T cells and iPSCs. Aptamer-dependent recruitment of the deaminase component to base editing target sites allows multi-purposing of a single Cas nickase component without unintended base editing of transgene integration sites, improving the efficiency of both precise transgene targeting and target gene disruption. The modularity of the platform allows independent tuning of each editing component to co-optimise the efficiencies of base editing and precise transgene targeting depending on the size of transgene insertions. Together, these results demonstrate that modular, aptamer-dependent assembly enables complex, customisable genome engineering workflows to be executed in a single step, addressing a key scalability challenge in the development of advanced cell therapies.
Decoupling deaminase recruitment reveals a Cas association-dependent origin of off-target base editing
Poster Number: P0949
Poster session: 1
Time: Wednesday 28 October from 13:30 to 15:30
Base editors offer a broad range of opportunities for precise and efficient DNA editing without requiring the introduction of genotoxic double strand breaks. Despite rapid progress toward clinical application, multiple challenges associated with non-specific activity of the DNA deaminase component must be addressed to fully realize the potential of precision genome engineering: imprecise 'bystander' editing of additional nucleotides proximal to the target nucleotide, and off-target editing at sequences in the genome distal to the intended target.
Whereas non-specific editing at sequences with complete or partial homology to the target are attributed to the DNA binding Cas component of base editors, editing of non-homologous sequences is generally considered to occur independently of the Cas component. However, as Cas proteins are known to scan the genome for target sequences, we hypothesised that transient deaminase recruitment by the Cas component during the search process may materially influence non-specific editing outcomes, rather than arising solely from intrinsic deaminase promiscuity.
To investigate whether non-specific editing can be reduced by decoupling the DNA binding and DNA modifying components of base editors we developed a modular system that assembles via an aptamer encoded in the sgRNA, enabling independent and orthogonal control of component expression stoichiometry and association kinetics. Modular adenine and cytosine base editors were transiently expressed in human induced pluripotent stem cells (iPSCs) by electroporation of mRNA and synthetic aptamer-encoding sgRNAs. Each editor component was titrated from saturating to limiting quantities and non-specific editing was analysed using targeted sequencing, immunofluorescence analysis of DNA repair, and whole exome sequencing.
Transient expression of sub-saturating doses of the deaminase component by mRNA electroporation resulted in an unexpected reduction in bystander editing by modular adenine base editors at target sequences while maintaining efficient on-target single-base editing, enabling non-specific editing to be reduced below levels observed for equivalent Cas-deaminase fusion editor architectures. Contrary to prevailing assumptions, editing at Cas9:guide-RNA stabilized R-loops was strongly dependent upon deaminase recruitment at sub-saturating quantities of both adenine and cytosine deaminases. Moreover, sequence-independent off-target editing was undetectable in the absence of deaminase tethering to the Cas module, indicating that the major fraction of non-specific editing arises from an association-dependent process. Similarly, low-level off-target DNA damage associated with cytosine base editors depended upon recruitment of the cytosine deaminase component to Cas9 R-loops and could be reduced to background levels at sub-saturating quantities of the aptamer-encoding sgRNA.
These findings identify deaminase-Cas association, rather than deaminase activity alone, as a primary determinant of both on-target precision and off-target editing of base editors. Our observation that transient expression of limiting quantities of the deaminase component of modular base editors can be leveraged to bias on-target editing profiles to individual bases highlights deaminase recruitment kinetics as a key determinant of editing outcomes. Similarly, we conclude that off-target editing can be minimised by tuning the expression dynamics or recruitment kinetics of the deaminase component of base editors. Taken together, these observations highlight control of the assembly of modular base editors as a promising general strategy to minimize non-specific editing activity that is compatible with mRNA-based delivery formats commonly used in translational genome editing workflows.
Sequence-context–aware design of precision base editors using modular architectures enables scalable therapeutic gene correction
Poster Number: P0952
Poster session: 2
Time: Thursday 29 October from 14:00 to 16:01
Base editors have emerged as attractive solutions for gene correction applications due to their relative simplicity and ability to precisely and efficiently modify single nucleotides in post-mitotic cells. The current paradigm for designing base editors for precise gene correction applications relies on empirical, target-by-target exploration of the combined activities of an ever-expanding repertoire of their DNA targeting Cas and nucleotide modifying deaminase components. The limiting factor for gene correction is therefore no longer editing efficiency or safety, but rather effectively predicting single nucleotide editing outcomes of diverse base editors across therapeutic target sequence contexts. As a result, therapeutic base editor selection remains largely ad hoc, limiting scalability across the thousands of clinically relevant SNVs.
To address the challenge of predicting base editor editing outcomes at scale, we developed a modular base editing system and a suite of compatible screening platforms. Arrayed screening enabled scalable low-resolution charaterisation of the editing outcomes of modular SpCas9-based cytidine and adenine base editors with distinct and complementary activities. Deep analysis of two modular SpCas9-based cytidine base editors by pooled screening of ~7000 pathogenic variant target sequences enabled the development of machine learning models able to predict the distinct position-specific editing efficiencies for each configuration with performances within the range of comparable base editing outcome prediction models (ρ=0.66-0.7). Models were trained on positional preference and sequence context features and performance was tested by evaluation of held-out sequences from the library. Predictive capabilities were further validated at endogenous and exogenously introduced variant target sequences in orthogonal cellular assays.
Promising therapeutic candidates edited with high efficiency and precision by either editor configuration were empirically identified by mining the pooled screening data and subsequently validated through targeted screening in cellular models. Additionally, comparison of the editing outcomes of SpCas9- and OpenCRISPR1-based modular cytidine base editors by targeted screening indicated unexpected differences in the peak of editing activity of each configuration, with editing by OpenCRISPR1-based editors shifted towards the PAM-distal end of the target sequence. We conclude that modular base editor assembly enables scalable screening and prediction of editing outcomes, supporting a shift from empirical discovery to rational design of precision base editors for therapeutic gene correction. Notably, our preliminary analysis indicates that rather than acting in isolation, interactions between the deaminase and Cas components of base editors determine the nucleotide positions within target sequences that can be edited with high efficiency and precision. These observations highlight the need for scalable screening and modeling approaches to building predictive capabilities to underpin the rational design of the myriad bespoke therapeutic base editor configurations required for gene correction applications. We anticipate that the complementary modular base editor screening platforms we have developed will support iterative model refinement, enabling the continuous development of predictive capabilities and accelerating therapeutic candidate selection. This framework provides a foundation for treating therapeutic gene correction as a design problem rather than an empirical optimization exercise.
Establishing a Scalable Suspension-Based Manufacturing Workflow for VSV-G Retargeted Lentiviral Vectors
Poster Number: P0563
Poster session: 1
Time: Wednesday 28 October from 13:30 to 15:30
Targeted lentiviral vectors (LVs) are a promising strategy for in vivo gene delivery, but modifying the viral envelope can reduce manufacturability, potency, or downstream process compatibility. VSV‑G remains the most common pseudotype for LV production, yet its broad tropism and Immunogenicity limits in vivo applications. We evaluated whether our patented approach (EP 3 092 306 B1) of modular VSV‑G fusion constructs containing scFv could be co‑expressed with wild‑type VSV‑G in a scalable suspension system to define the production parameters required for future targeting optimization.
Suspension HEK293 cells cultured in serum free medium were transiently transfected with plasmids encoding wild‑type VSV‑G and modified VSV‑G fusion constructs at 1:1 or 3:1 ratios. Fusion constructs contained an anti-CD3 scFV linked to VSV‑G via a linker, and all vectors carried an eGFP reporter cargo. Vector production was quantified by RT‑qPCR genomic copies (GC/mL). Functional titers (TU/mL) were assessed on target cells, expressing CD3. Purification feasibility was evaluated using anion‑exchange chromatography. LV particles were successfully produced across all VSV‑G:modified ratios tested, indicating compatibility with our scalable suspension‑based production platform. The platform currently uses RUO‑grade reagents, all of which have cGMP‑grade equivalents to support future clinical translation. Higher wild‑type VSV‑G content correlated with increased GC/mL, while TU/mL remained relatively consistent across ratios. Anion‑exchange purification yielded overall recoveries similar to wt, suggesting that the presence of modified VSV‑G constructs does not substantially impact downstream processing under the conditions tested. Early western blot results confirm incorporation of the CD3‑scFv–modified VSV‑G into particles, and further work is underway to understand why GC titers decrease at higher modified:VSV‑G ratios. In vivo translatability will be assessed by incorporating mutated VSV-G variants into our suspension-based production platform.
We have established a scalable suspension‑based platform for co‑expressing wild‑type and VSV‑G fusion constructs during LV production. Although functional targeting is ongoing, these results identify VSV‑G ratios in suspension HEK293 cells that support particle formation and purification. The ratios are consistent with those described in our patented adherent‑cell process, suggesting that the same design principles extend to suspension‑based production. These results provide the foundation for generating stable, retargeted LV particles with enhanced target-specific functionality for ex vivo cell therapy manufacturing programs and in vivo applications.
Development of CNS-specific AAV capsid libraries through peptide insertion and iterative in vivo selection in non-human primates
Poster Number: P0491
Poster session: 1
Time: Wednesday 28 October from 13:30 to 15:30
Central nervous system (CNS) gene therapy faces significant challenges due to limited tissue specificity and poor blood-brain barrier penetration of current adeno-associated virus (AAV) vectors. To address these limitations, we developed a systematic approach for evolving CNS-specific AAV capsids through peptide insertion libraries and iterative selection in non-human primates (NHPs).
We constructed peptide insertion libraries based on the rh74 capsid backbone, incorporating randomized peptide sequences at strategic positions IV and VIII within the capsid structure. In parallel, a library was generated by DNA family shuffling. Libraries were administered intravenously to NHPs, followed by comprehensive biodistribution analysis across multiple brain regions (cerebellum, hippocampus, motor cortex, prefrontal cortex, striatum caudate, thalamus, striatum putamen, and substantia nigra) and off-target tissues (heart, liver, lung, and quadriceps muscle). Capsid genes were recovered and analyzed through next-generation sequencing (NGS) to identify enriched variants.
After two complete selection rounds, the peptide insertion library demonstrated remarkable CNS enrichment with reduced sequence diversity in brain tissues. The peptide libraries outperformed the shuffled library in magnitude of CNS enrichment. Cross-validation between NHP subjects revealed important selection dynamics: while round 1 showed more diverse enrichment patterns between subjects, round 2 demonstrated highly comparable performance between the two subjects, indicating convergent selection toward optimal CNS-targeting variants. The top 10 peptide variants maintained consistent rankings across animals in the second round, confirming.
VP3-based nanobody insertion: Optimization of manufacturing strategy for retargeted AAV vectors
Poster Number: P0061
Poster session: 1
Time: Wednesday 28 October from 13:30 to 15:30
Gene therapy applications have increasingly relied on adeno-associated virus (AAV) vectors for treating a wide range of disease conditions. Following extensive development of naturally occurring AAV serotypes for therapeutic purposes, researchers have focused on engineering enhanced AAV variants with improved targeting capabilities for specific cell types and tissues. The structural composition of AAV capsids includes three viral proteins - VP1, VP2, and VP3 - that accommodate insertion of peptide sequences or larger protein domains within their surface-accessible variable regions. This structural flexibility enables rational capsid engineering to achieve targeted delivery properties. A promising strategy involves incorporating single-domain antibodies from camelids (VHH or nanobodies) as fusion constructs with capsid proteins, allowing their integration into assembled viral particles.
We conducted a quantitative comparison between VP3-only nanobody insertion and multi-protein constructs where nanobodies are simultaneously incorporated across VP1, VP2 and VP3 capsid proteins. Side-by-side manufacturing was performed using suspension cell culture systems with comparative purification approaches: one-step purification (iodixanol gradient centrifugation) versus two-step purification (Protein A affinity chromatography followed by iodixanol gradient).
Results demonstrate equivalent manufacturing performance between VP3-only and multi-protein nanobody constructs. Purification strategy comparison revealed a trade-off between yield and purity: one-step iodixanol purification provided higher yields, while the two-step Protein A/iodixanol approach delivered superior purity profiles with reduced overall yield. The multi-protein VP1/VP2/VP3 construct showed modestly enhanced insertion rates compared to VP3-only approaches. Importantly, transduction efficiencies remained comparable across all conditions tested, indicating functional equivalence.
This comparative study provides practical insights into AAV-Nanobody (AAV-Nb) engineering and purification strategies, supporting the development of scalable manufacturing processes for targeted gene therapy vectors with optimized yield-purity.
DoE‑Driven optimization of empty/full separation with AEX for engineered retinal AAV capsid
Poster Number: P0490
Poster session: 2
Time: Thursday 29 October from 14:00 to 16:00
rAAV vectors remain central to retinal gene therapy, but there is still a need for capsids that combine strong biological activity with scalable manufacturability. We previously identified capsids optimized for the retina through in vivo selection in non‑human primates, where they showed enhanced tropism and efficiency. Although our standard anion exchange (AEX) workflow performed well for some variants, reaching full‑capsid levels as high as 78%, one lead capsid achieved only ~50% full capsids under generic conditions, prompting further optimization.
The engineered capsid variant carrying a reporter transgene was produced in HEK293 suspension cells using triple transfection at the 700 mL scale in a 2 L Biostat B UniVessel. All process reagents used, including transfection materials and lysis buffer, were research‑use‑only, with GMP‑grade equivalents available. The first purification step was affinity capture and captured material was split into equal aliquots and diluted 20‑fold into low‑salt buffer for AEX chromatography using a strong anion exchanger. Because our AEX buffers include both NaCl and MgCl₂, identifying optimal combinations is not straightforward. To streamline this, we implemented a 16‑condition DoE across two anion exchange columns to identify conditions that improve empty/full enrichment. Vector genomes (vg) were quantified and the percentage of full capsids was determined after formulation.
The capsid variant had a crude titer of 2×10¹¹ vg/mL, and recovery after affinity capture was 96%. Through DoE a model was created which helped to identify a condition that improved empty/full enrichment by anion-exchange, yielding full‑capsid levels comparable to iodixanol ultracentrifugation.
Using a DoE‑based approach, we identified an AEX buffer condition that improves empty/full enrichment for our retinal capsid and is compatible with scalable workflows used in clinical manufacturing. This optimization strategy could be applied to other capsid–cargo combinations that are difficult to enrich for full capsids by anion exchange.
Smooth transition from plasmid to cell-free DNA: proof-of-concept testing of dbDNATM technology for rAAV manufacturing
Poster Number: P0865
Poster session: 1
Time: Wednesday 28 October from 13:30 to 15:30
Although plasmid DNA remains widely used across rAAV development, including GMP manufacturing, transitioning to alternative DNA formats such as enzymatic, cell-free DNA becomes a critical consideration as programs approach GMP manufacturing. Uncertainties around process compatibility can arise, which creates a need for early-stage testing to confirm performance. Here, a proof-of-concept evaluation of enzymatically produced linear doggybone™ DNA (dbDNA™) was performed to demonstrate its suitability as an alternative to conventional plasmid DNA within a standard triple transfection workflow.
Methods : To compare the two DNA formats, a design of experiments (DoE) study was performed centered on conditions previously optimised for plasmid-based transfection. Experiments were conducted at 10 mL scale using Falcon tubes with vented caps, enabling parallel evaluation of 30 conditions while minimizing material consumption. Following identification of an optimal condition for dbDNA™, processes were scaled to shake flask level to produce AAV2 and AAV8 vectors carrying a Luciferase-eGFP cargo. Vectors were purified by affinity capture. Process performance and vector quality were assessed by: (1) qPCR for vg/mL titers, (2) percentage of full capsids using UV–Vis absorbance, and (3) residual plasmid DNA using ori-specific PCR. Additionally, dbDNA™ production was evaluated using the NEB EnClose™ cell-free synthesis kit.
Results : One DoE condition yielded the highest titer for AAV8 Luciferase eGFP produced using dbDNA™. At shake flask scale, dbDNA™ and plasmid-based production generated comparable titers for AAV2 (1.3×10¹¹ vs. 1.8×10¹¹ vg/mL) and AAV8 (5.6×10¹¹ vs. 3.5×10¹¹ vg/mL). For AAV8, full capsid levels were slightly higher with dbDNA™ than with plasmid DNA (18% vs. 15%). Residual plasmid DNA in plasmid-derived vector preparations was quantified at 0.1% using ori-specific PCR. Using the NEB EnClose™ cell-free synthesis workflow, an average yield of 33 µg dbDNA™ per reaction was obtained, demonstrating that sufficient dbDNA™ for flask-scale production can be rapidly produced.
Conclusion : dbDNA™ supports rAAV production with titers and full-capsid percentages comparable or greater to plasmid DNA after transfection process optimisation. It also enables workflows that significantly reduce plasmid backbone–derived residual DNA. This work therefore shows that cell-free DNA formats such as dbDNA™ can be used as an alternative to plasmid based workflows, and demonstrates the value of early evaluation and optimisation experiments in preclinical stages.
Automated plasmid assembly with FlexDropTM platform for DoE‑driven rAAV upstream optimization
Poster Number: P0863
Poster session: 1
Time: Wednesday 28 October from 13:30 to 15:30
Optimizing upstream parameters in rAAV manufacturing is often limited by how many samples can be prepared at once. Even with design of experiments (DoE), which reduces the number of combinations that need to be tested, the triple transfection method involves assembling multiple plasmids in different amounts, a process that is highly manual and error prone. To address this, we evaluated the FlexDrop™ platform, an automated, tip less system for high throughput liquid dispensing down to 8 nL. It is compatible with 24 and 96 well plates in standard or deep well formats. We used the FlexDrop™ platform to automate plasmid assembly as part of a DoE driven upstream optimization experiment for rAAV production.
With AAV8 as the capsid serotype and eGFP as the cargo, the FlexDrop™ platform was used to assemble plasmid mixes across 24 conditions for a DoE matrix that varied in plasmid ratios and total DNA amount. Plasmid stocks were placed into a FlexDrop™ platform source plate, which contains porous wells that allow pressure driven dispensing of each plasmid at the required amounts. The instrument then automatically dispensed the plasmid mixtures into a receiver plate according to values specified in the picklist. Complexation was performed by adding a low salt buffer containing the transfection reagent and enhancer to each well, and a complex stabilizer was included to ensure consistent particle formation across all conditions. Once DNA–reagent complexes had formed, cells were added to initiate transfection. Seventy two hours later, the cells were lysed using a detergent–nuclease solution, and vector genomes were quantified directly from crude lysates using qPCR with ITR specific primers.
Droplet based dispensing produced all 24 transfection ready plasmid mixes in under a minute, whereas manual pipetting required at least 30 minutes. qPCR titers showed clear, condition dependent differences driven by changes in plasmid ratios and total DNA.
By coupling 1 mL scale rAAV production with high throughput plasmid assembly on the FlexDrop™ platform, this automated approach accelerates upstream parameter exploration and enables faster iteration with more confident early stage decisions.
Higher ITR Fidelity Does Not Translate into Higher AAV Yield: Evaluation of E. coli JW0387 strain
Poster Number: P0819
Poster session: 1
Time: Wednesday 28 October from 13:30 to 15:30
Background. rAAV manufacturing critically depends on transfer plasmid quality, which encompasses multiple attributes, including topology, yield, and sequence integrity. ITR stability represents one factor contributing to sequence integrity and is particularly susceptible to recombination during E. coli propagation. While specialized strains such as JW0387 have been reported to improve ITR fidelity, their suitability for scalable, standardized plasmid amplification workflows remains unclear.
Methods. We evaluated JW0387 (Dharmacon, Revvity) alongside DH5α and Stbl3 using a challenging AAV transfer plasmid from a discovery gene therapy program containing AAV2-derived 145 bp ITRs with unstable 3′ ITR behavior in standard strains. Amplification was performed at 37°C and 42°C at two culture volumes (150 mL and 500 mL) to reflect scalable plasmid preparation needs. Plasmid yield was compared across strains and temperatures, and plasmid topology (e.g., supercoiled and open circular forms) was measured using the automated LabChipTM GXII TouchTM system (Revviyty) with the LabChip TM plasmid DNA assay (Revvity). ITR sequences were analyzed by long-read sequencing supported by an R-based tool with a web interface to identify ITRs and classify variants within defined ITR subregions. To assess functional impact, plasmids were used in triple-plasmid transfection to produce rAAV9 in a suspension HEK293 producing cell line widely used as an industry benchmark. Upstream titers were quantified by ITR-qPCR, and purified material was analyzed for full capsid content using DLS after affinity capture and formulation.
Results. Increasing culture temperature from 37°C to 42°C improved plasmid yields across strains. Under the tested conditions, JW0387 did not provide a plasmid productivity advantage over DH5α or Stbl3. Plasmid topology analysis indicated that JW0387 produced the lowest supercoiled plasmid fraction compared with the other strains. In contrast, long-read sequencing analysis showed improved ITR integrity in JW0387-amplified plasmids, with the 3′ ITR displaying greater overall susceptibility. Despite improved sequence fidelity, no differences were observed in upstream vector genome titers or downstream full capsid levels across strains and conditions, indicating that fidelity gains alone were insufficient to impact functional output under these conditions.
Conclusion. JW0387 improves ITR sequence integrity but does not enhance plasmid productivity, rAAV yield, or full capsid levels under the tested conditions, indicating that fidelity gains alone are insufficient to impact plasmid functional output. Importantly, a comprehensive assessment of plasmid amplification process parameters, including bacterial growth kinetics and harvest timing, was not yet performed and will be required to determine whether JW0387 can generate plasmids with a combination of quality attributes that translate into improved manufacturing metrics, such as AAV yield and full capsid levels.
Evaluation of an AI‑designed AAV capsid in a HEK293 suspension workflow
Poster Number: P0864
Poster session: 2
Time: Thursday 29 October from 14:00 to 16:00
Dyno Therapeutics uses AI‑guided design to develop next‑generation rAAV capsids with improved biological performance. As part of the Dyno Frontiers Network, we evaluated the manufacturability of the AAV9‑based neuromuscular capsid Dyno-bn8 using our suspension production platform.
We produced Dyno-bn8 with two distinct reporter payloads (GFP-Luciferase and HA-tagged histone H3) in duplicate at the 120 mL scale in HEK293 suspension cells grown in serum‑free medium using plasmid DNA triple transfection. Cells were lysed directly in the growth medium 72 h post‑transfection using detergent and salt‑active nuclease for DNA clearance. All reagents used in this study were research use only, but equivalents suitable for cGMP manufacturing are available. Clarified lysates were purified by affinity chromatography using the POROS™ CaptureSelect™ AAV9 resin (Thermo Fisher Scientific), followed by iodixanol gradient ultracentrifugation. The final material was formulated in DPBS with 0,001 % Poloxamer 188 plus salt. Vector genomes (VG) were quantified by qPCR targeting the AAV2 ITR. The percentage of full capsids and polydispersity index (PDI) were measured by UV/Vis absorbance and dynamic light scattering on the Stunner™ platform.
Without any process optimization, upstream titers for Dyno-bn8-GFP-Luciferase averaged 2,1×10¹¹ vg/mL (13,5 % CV). While roughly 50 % lower than the benchmark serotype, this value is in line with what we often see for engineered capsids. Downstream purification yielded an overall recovery of 18,5 % with a final concentration of 2,0×10¹³ vg/mL. The material contained 79,1 % full capsids, with a PDI of 0,01, indicating a low level of aggregates. Endotoxin levels were <0,1 EU/mL, and only VP1, VP2, and VP3 were detectable by silver‑stained SDS‑PAGE, demonstrating high purity.
Dyno‑bn8 with a harder-to-package payload - H3F3-HA - showed upstream titers and final product quality comparable to eGFP‑Luciferase, but the overall recovery was lower. Process optimization efforts are underway to improve the yields.
This evaluation shows that Dyno‑bn8 performs well on our manufacturing platform, generating consistent upstream titers and high‑quality material across different test cargos. As with all projects, the capsid now enters our manufacturing optimization workflow, where our process development team will focus on improving yield and recovery using approaches compatible with future cGMP production. These efforts will improve our readiness to package sponsor‑provided payloads into Dyno capsids for work within the Dyno Frontiers Network.
Development of a chimeric antigen receptor that specifically and selectively recognizes acute myeloid leukemia (AML) blasts
Poster Number: P0305
Poster session: 1
Time: Wednesday 28 October from 13:30 to 15:30
Acute myeloid leukemia (AML) continues to have a poor prognosis, particularly for patients who are refractory to current treatments or who experience disease relapse. While allogeneic hematopoietic stem cell transplantation remains the primary curative approach, the long-term survival rate post-transplant is below 20%. Chimeric antigen receptor (CAR) T cell therapy has revolutionized the treatment of B-cell malignancies, yet its application in AML is hindered by a lack of target antigens that can safely distinguish leukemic blasts from healthy hematopoietic cells. Building on previous work identifying CD86 as a robust target to achieve safer and more efficacious cell therapy for AML, this project aims to develop an optimized anti-CD86 CAR-T cell manufacturing pipeline. To overcome current developmental bottlenecks, we performed in silico design of fully humanized, second- and third-generation anti-CD86 CAR constructs. Following lentiviral particle generation, we systematically optimized the primary human T cell transduction protocol by evaluating the combinatorial effects of chemical transduction enhancers, spinoculation, and recombinant fibronectin fragments. To ensure preclinical developments are readily translatable, we implemented GMP-quality reagents, enabling successful cell expansion under GMP-like conditions. A critical objective of this manufacturing phase is to compare T cell phenotypic profiles and expansion kinetics between GMP-like cultures and standard research-grade conditions. The resulting anti-CD86 CAR-T cells are currently undergoing rigorous comparative evaluations. In vitro functionality and cytotoxicity kinetics across a panel of AML cell lines are being assessed utilizing real-time live-cell imaging and complementary luminescence-based co-culture assays. These functional readouts are supported by comprehensive cytokine profiling via ELISA and multi-parameter T cell phenotyping. Concurrently, in vivo efficacy and survival advantages are being evaluated in NSG mouse models, comparing the anti-CD86 constructs against an irrelevant CAR control. Ultimately, integrating targeted in silico CAR design with a highly optimized, GMP-like manufacturing workflow and robust multi-modal functional validation establishes a strong foundation for translating this novel anti-CD86 therapy into clinical applications.
For research use only. Not for use in diagnostic procedures.
Pin-point™ base editing reagents are available for research use only and are not for diagnostic use or direct administration into humans or animals. The Pin-point™ base editing platform technology is available for clinical or diagnostic study and commercialization under a commercial license from Revvity.
AAVs Developed with Revvity Technology: A commercial license is required for clinical or commercial applications using AAVs developed with Revvity technologies.
