In the world of gene therapy, the Adeno-associated virus (AAV) is the industry-standard delivery vehicle. It is safe, efficient, and has a natural knack for finding specific tissues. However, AAV has a "trunk space" problem: it can only carry about 4.7 to 4.9 kilobases (kb) of genetic material.
This physical constraint creates a massive bottleneck. Many of the most important human genes—like dystrophin for muscular dystrophy or Factor VIII for hemophilia—are far larger than 5 kb. Furthermore, modern "molecular machines" like CRISPR-Cas9, along with the regulatory sequences needed to control them, often exceed this limit. Researchers are now employing a variety of strategies to bypass this bottleneck, ranging from shrinking the tools themselves to redesigning the delivery vehicles from scratch.
Shrinking the Molecular Machinery
The most direct way to fit a payload into a small vector is to make the payload smaller. Standard genome-editing tools like SpCas9 are often too bulky when paired with the necessary guide RNAs and promoters.
To solve this, researchers are hunting for "compact" alternatives. Wang‐Yu Tong and colleagues discovered SviCas3, a single-protein editor that functions without the bulky multi-protein complexes usually required by Type I CRISPR systems Tong et al. (2023). Similarly, Shaoting Weng demonstrated that using the smaller Staphylococcus aureus Cas9 (SaCas9) combined with space-saving tRNA promoters allows an entire editing system to fit within a single AAV Weng et al. (2020).
This "miniaturization" extends to the regulatory elements that turn genes on and off. Large, naturally occurring human promoters often take up too much room. A. Gomes and Oriol Fornés developed Ple389, a "MiniPromoter" that is small enough to fit in an AAV while maintaining highly selective expression in specific brain neurons Gomes et al. (2024).
Truncating Large Therapeutic Proteins
When the therapeutic gene itself is too large, researchers create "mini-versions" by identifying and keeping only the most essential parts. This is a primary strategy for Duchenne Muscular Dystrophy (DMD), where the dystrophin gene is a staggering 2.4 million base pairs long.
Jackie L. McCourt and James M. Ervasti have evaluated truncated "mini-dystrophins" that fit into viral vectors, though they note that the specific way a protein is "cut" significantly impacts its structural stability McCourt et al. (2015). Taking this a step further, Ethan Baron has proposed SCISOR, a generative AI model that uses discrete diffusion to "shrink" proteins by identifying non-essential amino acid segments, effectively automating the design of these condensed therapeutic variants Baron (2026).
Architectural Efficiency and Multi-Vector Systems
If a payload cannot be shrunk, it can be split or organized more efficiently. Bernd Zetsche and colleagues showed that the Cas12a (Cpf1) enzyme can process its own guide RNA arrays. This eliminates the need for multiple bulky promoters, allowing researchers to target many genes at once within a single, compact construct Zetsche et al. (2016).
For truly massive payloads, researchers use "multi-vector" systems. In this approach, the genetic components are divided among two or more separate AAVs that are co-administered. Once inside the cell, these components interact—for example, via protein splicing or DNA recombination—to form the complete therapeutic machine Mahajan & Tsang (2018).
Expanding the Container
Finally, some researchers are looking to replace or modify the delivery vehicle itself. Michael A. Thomas and Marjorie Robert-Guroff successfully expanded the cargo capacity of adenovirus vectors by systematically deleting non-essential viral genes, creating more room for complex vaccine antigens Thomas et al. (2013).
Others are moving away from viruses entirely. Wolfgang Wurst and his team used AI to design Synthetic Transfer Vehicles (STVs)—non-natural protein assemblies that bypass the rigid volumetric constraints of biological evolution, potentially allowing for the delivery of much larger mRNA strands or gene-editing complexes Wurst et al. (2024). Similarly, Carlos Panizo has explored using transposon-based technologies like Sleeping Beauty to deliver large chimeric receptors for cancer therapy without the cost or capacity limits of viral vectors Panizo (2023).
By attacking the packaging problem from both ends—shrinking the cargo and expanding the container—the field is moving toward a future where the size of a gene is no longer a barrier to its cure.