Engineering self-deliverable ribonucleoproteins for genome editing in the brain

CRISPR-Cas technology, derived from bacterial immune systems, revolutionizes genome engineering. Its RNA-guided Cas protein creates double-stranded DNA breaks, inducing site-specific DNA repair. While effective *in vitro*, efficient and safe *in vivo* delivery of CRISPR-Cas complexes, particularly to the central nervous system (CNS), remains challenging. Viral vectors, like adeno-associated virus (AAV), are used but have immunogenic and insertional mutagenesis drawbacks. Direct delivery of CRISPR-Cas RNPs offers advantages, avoiding viral limitations. However, efficient non-viral delivery to brain cells remains a hurdle. Various nanoparticle strategies exist, but they require complex optimization and manufacturing. Cell-permeable Cas9 RNPs, enabled by cell-penetrating peptides (CPPs), offer a simpler and broader solution. Previous studies have shown some success with NLS-fused Cas9 and endosomolytic peptides, but optimization for *in vivo* brain delivery is crucial. This research aimed to engineer self-deliverable Cas9 RNPs by fusing them with CPPs, focusing on *in vitro* and *in vivo* genome editing in neural cells.

Existing literature highlights the transformative potential of CRISPR-Cas technology for treating genetic disorders but underscores the significant challenges in delivering CRISPR components efficiently and safely *in vivo*, especially to the brain. While viral delivery methods, such as AAV, have been employed, their immunogenicity and potential for insertional mutagenesis necessitate the exploration of safer alternatives. Direct delivery of RNPs presents a promising approach, minimizing off-target effects and immune responses associated with viral vectors. The literature demonstrates varying success with nanoparticle-based RNP delivery systems, each with inherent complexities in optimization and scalability. Prior research has explored the use of NLS-fused Cas9 and synthetic endosomolytic peptides to enhance RNP delivery, yet achieving robust and consistent results, particularly *in vivo*, requires further improvement. This background establishes the need for a simpler, more broadly applicable strategy, focusing on the development of self-deliverable RNPs for CNS applications.

The study began by comparing the *in vitro* and *in vivo* genome editing efficiency of SpyCas9 and LbCas12a, fused with varying numbers of NLS peptides, in mouse-derived neural progenitor cells (NPCs). A tdTomato reporter system was used to quantify editing efficiency. *In vivo* studies involved intraparenchymal injection of RNPs into the striatum of Ai9 mice. Next, a screen of 34 CPPs, fused to the C-terminus of Cas9 (with N-terminal NLS), was conducted to identify peptides enhancing RNP delivery to NPCs. The CPPs were categorized into different classes: nuclear localization signals, viral-derived peptides, cell-permeable miniature proteins, peptides from signaling proteins, peptides from functional proteins, antimicrobial peptides, and membrane-active peptides. A bacterial expression system with CL7 and His tags enabled efficient purification. The most promising CPPs were further optimized by varying peptide location, NLS combinations, and CPP copy numbers. The mechanism of CPP-assisted delivery was investigated using endocytosis inhibitors. Finally, the optimized SpyCas9-A22p3 construct was tested for *in vivo* editing efficacy at endogenous genomic sites (tyrosine hydroxylase (TH) and metabotropic glutamate receptor 5 (mGluR5)) in the mouse striatum. Editing efficiency was quantified using next-generation sequencing (NGS) and qPCR. Detailed methods on plasmid construction, nucleic acid/peptide preparation, protein expression and purification (including endotoxin removal for *in vivo* studies), RNP assembly and characterization, cell culture, gene editing (nucleofection and direct delivery), flow cytometry, NGS, stereotaxic infusion, tissue collection and immunostaining, fluorescent imaging and quantification, DNA/RNA/protein extraction, cDNA synthesis, qPCR, and statistical analyses are provided.

SpyCas9 demonstrated superior *in vivo* editing efficiency compared to iCas12a in the mouse striatum. Screening of CPPs revealed Bac7, HBP, CA-Tat, and A22p as highly effective for enhancing RNP delivery to NPCs. Optimization studies showed that A22p, particularly three copies fused to the Cas9 C-terminus (A22p3), significantly improved editing efficiency *in vitro* (up to 72% in the tdTomato assay) while maintaining low cytotoxicity. The A22p3 construct demonstrated robust *in vivo* genome editing in Ai9 mice at doses as low as 25 pmol. *In vivo* editing of endogenous genes (TH and mGluR5) using A22p3-conjugated RNPs resulted in 1.5–5% editing at the DNA level and a 15–20% reduction in mRNA expression. Mechanistic studies suggested that endocytosis plays a key role in CPP-assisted RNP delivery. The improved delivery efficiency of the engineered Cas9 RNPs compared to previously tested NLS-enriched constructs were observed in both *in vitro* and *in vivo* experiments.

This study successfully demonstrated the development and optimization of self-deliverable CRISPR RNPs for efficient genome editing in the brain. The use of CPPs, particularly A22p, drastically improved the delivery efficiency of Cas9 RNPs compared to NLS-based approaches, resulting in high levels of genome editing both *in vitro* and *in vivo* with minimal cytotoxicity. The findings overcome limitations of previous RNP delivery strategies by offering a facile and potentially cost-effective platform for CNS gene therapy. The *in vivo* editing of medically relevant genes suggests the therapeutic potential of this approach for treating neurological disorders. The study also provides insights into the mechanism of CPP-mediated RNP delivery, highlighting the role of endocytosis. The success with A22p, a peptide with potential receptor interactions, opens avenues for targeting other cell types by employing similar ligand-peptide strategies.

Self-deliverable CRISPR RNPs, engineered by fusing cell-penetrating peptides like A22p to Cas9, offer a significant advancement in genome editing technology. This approach achieved high editing efficiency *in vitro* and *in vivo* in the CNS, showing promise for treating neurological disorders. The relative simplicity and scalability of this method compared to nanoparticle-based delivery make it a compelling tool for therapeutic applications. Future research could explore additional CPPs, optimize targeting to specific brain regions, and investigate the long-term effects of this technology.

The study primarily focused on a mouse model. The translational potential to humans needs further investigation. The achieved *in vivo* editing efficiency of endogenous genes was modest (1.5–5% at the DNA level), highlighting the need for further optimization for therapeutic efficacy. While the mechanism of CPP-mediated delivery is partially elucidated, further studies are needed to fully understand it. The long-term effects and potential off-target effects in *in vivo* settings require comprehensive investigation.