PackGene’s Latest Breakthrough: Spatially Controlled AAV Gene Therapy Decouples Efficacy from Toxicity

High-dose intravenous (IV) administration of recombinant adeno-associated virus (rAAV) often results in significant off-target liver accumulation, leading to severe clinical complications such as acute liver failure. To overcome this critical hurdle in AAV therapy, PackGene’s research team developed a precision strategy to decouple therapeutic efficacy from hepatotoxicity by engineering spatial control over transgene expression. By integrating advanced AAV capsid engineering (the liver-detargeted, muscle-tropic AAV.eM variant) with microRNA-mediated regulation and intramuscular (IM) delivery, the study achieved a 10-fold dose reduction while maintaining with full anti-tumor efficacy in a lymphoma model. Recently published as a bioRxiv preprint titled ‘Decoupling Efficacy from Toxicity: Engineering Spatial Control in AAV-Mediated Gene Therapy,‘ this work provides a precision spatial control strategy with an engineered AAV variant, effectively breaking the dose–toxicity paradox that limits the safety of high-dose gene therapeutics.

Hepatotoxicity in systemic AAV gene therapy

High-dose systemic AAV gene therapy is powerful for achieving widespread transduction of tissues such as muscle and the central nervous system, but it faces a recurring clinical bottleneck: disproportionate liver uptake and off-target transgene expression, leading to dose-limiting hepatotoxicity. The liver’s natural tropism for AAV serotypes like AAV8 and AAV9 acts as a double-edged sword—beneficial for treating liver-specific diseases but a major safety hazard for therapies targeting extrahepatic tissues. This “hepatotropic” phenotype arises from the liver’s role in clearing circulating exogenous substances via the reticuloendothelial system, its high blood flow, microvascular structure providing extensive and surface area for uptake. As a result, a substantial fraction of intravenously administered AAV accumulates in the liver, triggering multifactorial toxicity through mechanisms including immune-mediated T-cell destruction of transduced hepatocytes, direct injury from capsid components or transgene overexpression, vector overload causing endoplasmic reticulum stress and apoptosis, and exacerbation by pre-existing conditions.

Clinical evidence underscores the severity of this issue, with sobering cases of fatal acute liver failure in high-dose AAV programs: two deaths following Zolgensma (AAV9-based for SMA) in 2022, four deaths in the ASPIRO trial of an AAV8-based therapy for X-linked myotubular myopathy, and, by late 2025, two fatal acute liver failure cases in Elevidys (AAVrh74-based for Duchenne muscular dystrophy) that triggered regulatory actions including labeling updates, a boxed warning, indication narrowing, and suspension of use in certain patient subgroups.

These observations make clear that simply delivering more AAV is not a sustainable solution: the field urgently needs strategies that reduce hepatic sequestration and off-target expression while preserving efficient delivery to extrahepatic targets like muscle and brain. This raises a fundamental design question: can we engineer AAV systems that maximize therapeutic protein production at the disease site, yet minimize or eliminate dangerous off-target expression and vector burden in the liver?

PackGene’s innovation and solution

PackGene proposes a stackable spatial-control toolkit to separate efficacy from toxicity, combining three complementary levers:

1. AAV capsid tropism control: At the capsid level, the novel myotropic variant AAV.eM generated from the π-Icosa AAV serotype screening platform was engineered to prioritize muscle transduction while significantly reducing off-target hepatic accumulation compared to wild-type AAV9.
2. Transcriptional and post-transcriptional control: By utilizing the muscle-restricted MHCK7 promoter, the system shifts transgene expression away from the liver, establishing the skeletal muscle as the primary production hub for the therapeutic protein. To achieve even more high-resolution spatial control, the study utilizes microRNA-mediated ’tissue filtering.’ By incorporating tissue-specific microRNA binding sites into the transgene’s 3′ UTR, expression patterns are fine-tuned even between related tissues. This regulatory layer provides a critical fail-safe, successfully silencing the transgene in cardiac muscle while maintaining robust therapeutic activity in skeletal muscle.
3. Strategic localized delivery: By shifting from systemic intravenous (IV) administration to intramuscular (IM) delivery, the study minimizes the total systemic viral burden. This approach effectively transforms the localized injection site into a high-efficiency ‘bio-factory,’ secreting therapeutic proteins into the bloodstream to maintain systemic efficacy while shielding the liver from direct viral exposure.

Together, this work is presented as a framework to reduce liver/off-target expression burden while preserving systemic therapeutic exposure—the essence of “decoupling efficacy from toxicity.

Methods and Experimental Design

To rigorously validate the effectiveness of spatial control, the study utilized a high-stakes therapeutic model involving the in vivo secretion of a cancer immunotherapy protein. Specifically, a bispecific T-cell engager (BiTE) against CD19 and CD3 (modeled on blinatumomab) was selected as the therapeutic transgene, which is a payload that requires high circulating blood levels to effectively recruit human T cells to destroy CD19+ tumor cells. This was tested in humanized B-NDG mice engrafted with Raji-Luc lymphoma, where efficacy was measured through real-time bioluminescence tumor tracking and overall survival outcomes.

The experimental design followed a systematic matrix of variables—capsid (AAV9 vs. the engineered AAV.eM), promoter (ubiquitous vs. muscle-restricted MHCK7), regulatory circuits (including miR-208a binding sites for cardiac-specific silencing), and delivery route (IV vs. IM)—to determine the optimal configuration for decoupling efficacy from toxicity. By benchmarking these combinations across 10-fold dose-reduction cohorts and utilizing diverse readouts—including tumor burden and survival measurement, serum protein exposure, tissue biodistribution, transcript levels, and protein expression—the study established a logical and granular connection between engineered spatial control and therapeutic potency.

Key Results

1. AAV.eM + MHCK7 outperforms AAV9 + MHCK7 after IV delivery (Fig. 1–2)

The experimental design first compares a single IV injection of various AAV configurations—AAV9 with a ubiquitous CAG promoter (AAV9-CAG-αCD19αCD3 or AAV9-CAG-BiTE), AAV9 with a muscle-specific MHCK7 promoter (AAV9-MHCK7-BiTE), and the engineered AAV.eM with the MHCK7 promoter (AAV.eM-MHCK7-BiTE)—to mice with established lymphomas. While the AAV9-CAG-BiTE cohort exhibited rapid tumor clearance and high serum BiTE levels, this systemic potency was coupled with severe toxicity, including rapid weight loss and significant mortality. In stark contrast, the AAV.eM-MHCK7 group achieved robust tumor regression and superior survival, outperforming the AAV9-MHCK7 group, which failed to effectively control tumor progression (Figure 1).

Biodistribution analysis confirmed that AAV9 sequestered heavily in the liver, whereas the engineered AAV.eM capsid showed markedly reduced hepatic transduction and high accumulation in skeletal muscle (Figure 2). These results demonstrate that while high-level, non-specific expression from wild-type AAV9 is therapeutically potent, it carries a lethal risk due to off-target liver involvement. By successfully redirecting the therapeutic payload from the liver to the muscle, the AAV.eM capsid provides a significantly wider therapeutic window, improving the safety and efficacy profile over standard AAV9 for systemic applications.

Fig 1. Evaluation of IV administered AAV9- and AAV.eM- BiTE therapy in a humanized lymphoma model

Fig 2. PK, safety, and biodistribution profile of AAV9- and AAV.eM- BiTE treated lymphoma model

2. MicroRNA elements add fine spatial control without losing efficacy (Fig. 3–4)

To refine the spatial precision of the therapy, the team addressed the fact that myotropic capsids like AAV.eM naturally target both cardiac and skeletal muscle. By inserting binding sites for miR-208a—a microRNA highly and specifically expressed in the heart—into the 3′ UTR of the transgene, we engineered a post-transcriptional “fail-safe” to selectively silence expression in cardiac tissue. Experimental observations using eGFP reporters demonstrated that even a single miR-208a binding site was sufficient to nearly eliminate heart expression while leaving skeletal muscle unaffected (Figure 3). When integrated into the therapeutic BiTE vector, this “spatial control” circuit maintained full tumor clearance efficacy, with quantitative PCR and protein analysis confirming that BiTE production was robustly suppressed in the heart while remaining high in systemic circulation (Figure 4).

By leveraging microRNA-mediated silencing, we can proactively prevent off-target cardiotoxicity, adding a sophisticated and essential layer of safety to custom AAV therapies. This approach allows for the safe targeting of large skeletal muscle groups as “bio-factories” without risking the functional integrity of the heart, thereby further widening the therapeutic window of the treatment.

Fig 3. microRNA binding sites provide a tunable “expression filter” on top of tropism and promoters

Fig 4. Cardiac silencing reduces heart expression without comprising anti-lymphoma activity

3. Intramuscular AAV.eM supports dose reduction while maintaining efficacy (Fig. 5–6)

To further minimize systemic toxicity and investigate the potential of localized delivery, the researchers evaluated IM administration of AAV.eM-MHCK7 and AAV9-MHCK7 across high and low-dose cohorts. While both vectors effectively cleared tumors at the high dose (5E12 vg/kg) (Figure S4), a stark difference emerged at the 10-fold lower dose (5E11 vg/kg). At this reduced concentration, the therapeutic effect of the standard AAV9 was significantly diminished, whereas AAV.eM-MHCK7 maintained potent tumor growth restriction and complete eradication (Figure 5). Biodistribution analysis confirmed this disparity, showing that AAV.eM produced markedly higher levels of BiTE mRNA in the injected muscle compared to AAV9 at the same low dose (Figure 6).

These findings indicate that the superior transduction efficiency of the engineered AAV.eM capsid allows for significant “dose-sparing.” By achieving the same therapeutic outcome with a 10-fold lower dose, this strategy can drastically reduce AAV manufacturing costs and minimize the risks associated with high-dose-related immune responses. Consequently, localized muscle delivery using engineered capsids serves as a highly attractive, safer alternative to traditional systemic IV injection, effectively turning the injection site into a high-performance bio-factory for systemic cancer therapy.

Fig 5. IM delivery of AAV.eM-MHCK7-BiTE enables efficacy at muscle-focused, lower-burden delivery

Fig 6. Confirmation of muscle-biased presence and expression with IM delivery of AAV.eM-MHCK7-BiTE

Conclusion and impact

PackGene’s “spatial control” framework provide a new strategy in AAV therapy, offering a sophisticated solution to the industry’s pervasive dose–toxicity paradox. By integrating liver-detargeted capsids, muscle-specific promoters, microRNA-based expression circuits, and localized delivery, the study provides a robust template for minimizing liver exposure without compromising systemic efficacy.

This work carries several transformative implications for the future of genetic medicine:

• Maximizing Muscle-Targeted Efficiency: Our engineered AAV.eM variant consistently outperforms AAV9 in muscle transduction, significantly widening the therapeutic window for muscle-directed therapies.
• Precision Spatial Control: Incorporating microRNA binding sites in the 3′ UTR effectively decouples physical biodistribution from functional expression. This vector-based engineering enables high-resolution tissue-sparing profiles that are easily adaptable for a broad spectrum of rare neuromuscular diseases.
• Reduced Manufacturing & Regulatory Burden: Achieving full efficacy with lower doses allows viral vector manufacturing to be more efficient, treating more patients per batch while reducing dose-dependent immunogenicity and safety concerns.
• AAV as an In Vivo Biologics Platform: By demonstrating sustained production of an αCD19αCD3 BiTE from muscle, the study establishes AAV as a powerful platform for antibody and T-cell engager expression in oncology, potentially replacing expensive, repeated biologic infusions.

In conclusion, PackGene’s strategy systematically dismantles the barrier between potency and safety. It marks a transition from high-risk systemic administration into a new era of spatially precise, safe, and commercially viable genetic medicine.

This research serves as a premier example of our rigorous approach to innovation, showcasing how PackGene’s superior AAV services and development capabilities provide the high-quality vectors and advanced engineering necessary to overcome the most complex clinical challenges. By choosing PackGene, partners gain access to this industry-leading expertise, ensuring their therapies are built on a foundation of safety, precision, and manufacturing excellence.

Reference:

1.Fan, Ying, et al. “Decoupling Efficacy from Toxicity: Engineering Spatial Control in AAV-Mediated Gene Therapy.” bioRxiv, 2025, https://doi.org/10.64898/2025.12.26.696588.

2.Pan, Yue, et al. “Engineering Novel AAV Capsids by Global De-targeting and Subsequent Muscle-Specific Tropism in Mice and NHPs.” bioRxiv, 2025, https://doi.org/10.1101/2025.05.19.654800.