Gene editing has enormous potential for inherited retinal disease, but getting an editor into the right cells – and, crucially, into the nucleus – still remains a major obstacle. This is particularly true in the retina, where post-mitotic cells present a very different delivery challenge from rapidly dividing cells or cells edited ex vivo.
A new Nature Communications study takes an unusual approach to the issue: rather than focusing solely on improving the delivery vehicle, it asks whether the host cell itself can be modified to make genome editing more efficient. The researchers performed a genome-wide CRISPR knockout screen targeting 19,114 human genes, linking individual genetic perturbations directly to the efficiency of nonviral Cas9 editing.
The screen identified 26 candidate genes whose loss increased editing efficiency. Subsequent validation narrowed these to six robust negative regulators of nonviral editing: BET1L, MS4A13, RBM44, SLCO1C1, GJB2 and ZNF584. Their effects were reproduced using both lipofection and lipid nanoparticles (LNPs), and with different editor formats. Importantly, experiments in which Cas9 was introduced by electroporation largely abolished the advantage conferred by gene knockout, pointing towards a bottleneck in cellular uptake or intracellular trafficking rather than simply altered DNA repair.
Two genes emerged as particularly interesting: GJB2, which encodes connexin 26 at the plasma membrane, and BET1L, a Golgi-associated vesicle SNARE. Depleting either substantially increased base-editing efficiency in human cell models. For correction of the pathogenic KCNJ13 W53X mutation associated with Leber congenital amaurosis type 16 (LCA16), GJB2 knockout increased adenine base editing 6.7-fold, reaching 19.3%, while BET1L knockout produced a fivefold increase, to 14.5%. Indel frequencies remained below 0.3%, suggesting that the improvement was not accompanied by a substantial loss of editing precision.
The most compelling findings came in a patient-derived retinal model. The investigators generated retinal pigment epithelium (RPE) from induced pluripotent stem cells carrying the homozygous KCNJ13 W53X mutation. Following LNP delivery of an adenine base editor, GJB2 depletion increased correction from approximately 5.8% to 46.6%, while BET1L depletion increased it to 20.9%. MS4A13 knockdown also produced an approximately eightfold improvement.
The findings suggest a potentially important shift in thinking: that the cell may be part of the delivery system. Instead of continually engineering more sophisticated nanoparticles, transiently suppressing endogenous cellular barriers could create a brief “window” during which a therapeutic editor gains greater intracellular access.
There is, however, an important caveat. Permanent disruption of genes such as GJB2 would not be clinically acceptable; GJB2 has essential roles in hearing and skin development. The study authors therefore envisage transient modulation – potentially using siRNA, CRISPR interference or small-molecule inhibitors – alongside the genome editor.
The concept is definitely intriguing. Efficient, precise delivery remains one of the principal hurdles to bringing genome editing into the retina. If the host cell can be temporarily conditioned to accept and process an editor more effectively, improving the biology of delivery may prove as important as improving the delivery vehicle itself.
