Engineering self-deliverable ribonucleoproteins for genome editing in the brain

Abstract

The delivery of CRISPR ribonucleoproteins (RNPs) for genome editing in vitro and in vivo has important advantages over other delivery methods, including reduced off-target and immunogenic effects. However, effective delivery of RNPs remains challenging in certain cell types due to low efficiency and cell toxicity. To address these issues, we engineer self-deliverable RNPs that can promote efficient cellular uptake and carry out robust genome editing without the need for helper materials or biomolecules. Screening of cell-penetrating peptides (CPPs) fused to CRISPR-Cas9 protein identifies potent constructs capable of efficient genome editing of neural progenitor cells. Further engineering of these fusion proteins establishes a C-terminal Cas9 fusion with three copies of A22p, a peptide derived from human semaphorin-3a, that exhibits substantially improved editing efficacy compared to other constructs. We find that self-deliverable Cas9 RNPs generate robust genome edits in clinically relevant genes when injected directly into the mouse striatum. Overall, self-deliverable Cas9 proteins provide a facile and effective platform for genome editing in vitro and in vivo.

Fig. 1. Comparison between SpyCas9 and LbCas12a for cell-penetrating SV40 NLS-assisted delivery and gene editing in neural cells in vitro and in vivo.

a Graphic illustrations of two RNP constructs. b Schematic of Cas9 or Cas12a RNP-mediated editing of Ai9 tdTomato NPCs to turn on fluorescent signals. c Quantification of tdTom⁺ NPCs based on the direct delivery of gene editors. n = 4 for each group, data are presented as mean values with individual data points. d Comparison of the gene editing activities of cell-permeable RNPs based on SpyCas9 and iCas12a in Ai9 mouse brain. e Editing volumes in the striatal tissue based on the injection of different RNP dosages. n = 6 for each group. f Co-expression of tdTomato and NeuN quantified per regions of interest (ROIs), e.g., edited area per hemisphere. n = 4 for the group of iCas12a (250 pmole), n = 5 for the other groups. Statistical analyses (unpaired t test) were performed for (c, e, f) and p values (**p < 0.01, ****p < 0.0001, and n.s. - not significant) were indicated with each set of quantification. Data are presented in box plots for (e, f) where the lower bound of the lower whisker shows the minimum, the lower bound of the box shows the lower quartile, the center of the box shows the median, the upper bound of the box shows the upper quartile and the upper bound of the upper whisker shows the maximum. Images in (a, b, d) were created with biorender.com.

Fig. 2. Screening of different cell-penetrating peptides to evaluate their ability in cellular delivery of SpyCas9 RNPs.

a Graphic illustrations of Cas9 RNP constructs fused with CPPs. b Sequence information of representative CPPs used in the screening. c Schematic of the expression and purification systems for Cas9-CPP fusion proteins in the screening. d Screening of Cas9-CPP constructs for self-delivery and gene editing activities using Ai9 tdTom NPCs. Fold change in tdTom⁺ NPCs% based on (upper) direct delivery and (lower) nucleofection of RNPs with 2x-SpyCas9-2x as the standard reference. 100 pmol RNP used for each NPC editing experiment. n = 4 for each group, data are presented as mean values with individual data points. Statistical analyses (unpaired t test) were performed, and p values (*p < 0.1, **p < 0.01, ***p < 0.001, ****p < 0.0001, and n.s. - not significant) were indicated with each set of quantification. Images in (a, c) were created with biorender.com.

Fig. 3. Further optimization of SpyCas9-CPP fusions for improved delivery efficacy and mechanistic profiles of CPP-assisted delivery.

The bar graphs in this figure present the quantification of tdTom⁺ NPCs based on the direct delivery of corresponding RNPs. 100 pmol RNP used for each NPC editing experiment, unless specified. a Table of Cas9-Bac7/A22p constructs (* indicates removal of S3). b Locational effects of the peptides at Cas9 termini in combination with different NLS peptides. c Terminal exposure effect of the peptides. d Insertion of Bac7 peptide in Cas9 backbone. e CPP-promoted RNP delivery in trans. f Systematic engineering of Cas9-A22p constructs. n = 4 for each group, data are presented as mean values with individual data points for (b–f). Statistical analyses (unpaired t test) were performed for (b–f), and p values (*p < 0.1, **p < 0.01, ***p < 0.001, ****p < 0.0001, and n.s. - not significant) were indicated with each set of quantification. Images in (b, d) were created with biorender.com.

Fig. 4. Comparison of SV40 NLS-assisted and A22p peptide-assisted delivery of SpyCas9 for gene editing in neural cells in vitro and in vivo.

a Quantification of tdTom⁺ NPCs based on the direct delivery of RNPs to NPCs in vitro. n = 4 for each group, data are presented as mean values with individual data points. Statistical analyses (unpaired t test) were performed, and p-values (****p < 0.0001) were indicated with each set of quantification. b Comparison of the gene editing activities of cell-permeable Cas9 RNPs in Ai9 mouse brain based on the injection of different RNP dosages. c Editing volumes in the striatal tissue. n = 6 for each group. d Co-expression of tdTomato and NeuN quantified per regions of interest (ROIs), e.g., edited area per hemisphere. n = 4 for the group of 2x-A22p3*, n = 5 for the other groups. Statistical analyses (unpaired t test) were performed for (c, d) and p values were indicated with each set of quantification. Data are presented in box plots for (c, d) where the lower bound of the lower whisker shows the minimum, the lower bound of the box shows the lower quartile, the center of the box shows the median, the upper bound of the box shows the upper quartile and the upper bound of the upper whisker shows the maximum. Images in (b) were created with biorender.com.

Fig. 5. A22p peptide-assisted delivery of SpyCas9 for gene editing at disease-relevant genomic sites.

a Schematic of the target sequences of Cas9 RNPs and the protospacer adjacent motif (PAM) for TH and mGluR5 knockout. b In vitro knock-out of the TH and mGluR5 genes based on direct delivery of the Cas9 RNPs. Indels quantified by NGS. n = 4 for each group, data are presented as mean values with individual data points. c In vivo knock-out of the TH and mGluR5 genes based on intraparenchymal injections of the Cas9 RNPs into mouse brains. RNPs assembled with a 1.5:1 mole ratio of sgRNA to Cas9 protein. Editing efficiency at the DNA level (indel throughout the whole striatum) quantified by NGS. n = 4 for the non-targeting control group and n = 6 for the experimental group. Data are presented in box plots where the lower bound of the lower whisker shows the minimum, the lower bound of the box shows the lower quartile, the center of the box shows the median, the upper bound of the box shows the upper quartile and the upper bound of the upper whisker shows the maximum. d Editing efficiency at the mRNA level (throughout the whole striatum) quantified by qPCR. n = 4 for the non-targeting control group and n = 5 for the experimental group, data are presented as median values ±s.e.m. Statistical analyses (unpaired t test) were performed for (c, d), and p values were indicated with each set of quantification. Images in (b, c) were created with biorender.com.