Is Gene Editing Stuck at the Cellular Gate? Genome-Wide Screen Identifies Six Barriers to Nonviral Delivery

Even an accurate gene editor must first cross the cell membrane, escape intracellular vesicles, and reach the nucleus before it has a chance to rewrite DNA. This journey particularly limits the ability of nonviral carriers such as lipid nanoparticles to work in difficult-to-treat tissues including the retina and nervous system. A study published in *Nature Communications* took a different approach: instead of continuing to modify the carriers themselves, the researchers searched for factors within cells that block the editors.

The research team established a genome-wide CRISPR knockout screen in human HEK293 cells, covering 19,114 genes and more than 200 million cells. The design enabled guide RNAs and editing outcomes to be identified from the same sequencing read, allowing the researchers to determine, within a vast mixed-cell population, which disrupted genes made it easier for nonvirally delivered Cas9 to complete editing. The initial analysis yielded 26 candidate genes. After validation across different targets, payloads, and delivery methods, the list converged on six high-confidence restriction factors: BET1L, MS4A13, RBM44, SLCO1C1, GJB2, and ZNF584.

Experiments showed that knocking out these genes substantially improved editing outcomes when Cas9 was delivered into cells using lipid formulations or lipid nanoparticles. However, when electroporation was used to bypass the cell membrane and endocytic pathways, most of the gains disappeared. Imaging analysis also found greater accumulation of Cas9 signals in the nuclei of some knockout cells. These results support the conclusion that the six factors primarily affect cellular uptake, intracellular trafficking, or release, rather than simply altering DNA repair. BET1L, however, may also be involved in processes within the nucleus, and the precise mechanism has not yet been fully clarified.

The researchers further tested GJB2 and BET1L. In a cell model carrying a pathogenic KCNJ13 mutation, after an adenine base editor was delivered by lipid nanoparticles, GJB2 knockout increased the target base conversion by 6.7-fold, reaching approximately 19.3% at the highest level. BET1L knockout produced a 5-fold increase, reaching approximately 14.5%, while insertion or deletion mutations remained below 0.3%. The effects of both factors were also observed in another cytosine base-editing model, indicating that the gains may not be limited to a single editor or locus.

Validation in a setting closer to the disease context used retinal pigment epithelial cells differentiated from patient-derived induced pluripotent stem cells carrying the KCNJ13 W53X mutation. After GJB2 or BET1L was reduced, base-editing efficiency increased by more than 3.5-fold, and some corrected cells regained Kir7.1 potassium ion channel function. This advances the screening signals reported in an early preprint and conference abstract to peer-reviewed evidence in patient-derived cells with a functional readout. However, it remains an in vitro study and has not yet demonstrated safe delivery to the retina in vivo or improvement in vision.

The real translational challenge is that these “barriers” also have normal physiological functions. GJB2 is important for hearing and skin development, and permanently or systemically shutting it down could cause harm. The research team therefore does not envision permanently editing these genes again, but instead temporarily suppressing their activity with siRNA, CRISPR interference, or small molecules to open a limited time window for the therapeutic editor. Whether this dual-delivery approach can control the site and duration of action, immune responses, and off-target editing must still be addressed in animal models and subsequent toxicology studies.