Systematic identification and characterization of high efficiency Cas9 guide RNAs for therapeutic targeting of ADAR

Abstract

Therapeutic targeting of the adenosine deaminase ADAR has great potential in cancer and other indications; however, it remains unclear what approach can enable effective and selective therapeutic inhibition. Herein, we conduct multi-staged guide RNA screening and identify high efficiency Cas9 guide RNAs to enable a CRISPR/Cas-based approach for ADAR knockout. Through characterization in human primary immune cell systems we observe similar activity with two-part guide RNA and single guide RNA, dose responsive activity, similar guide activity rank order across different cell types, and favorable computational off-target profiles of candidate guide RNAs. We determine that knockout of ADAR using these guide RNAs induces pharmacodynamic responses primarily consisting of immunological responses such as a type I interferon response, consistent with the known function of ADAR as a key regulator of dsRNA sensing. We observe similar biological effects with targeting only the p150 isoform or both p110 and p150 isoforms of ADAR, indicating that at least in the contexts evaluated, loss of p150 ADAR mediates the primary response. These findings provide a resource of well-characterized, high efficiency ADAR-targeting Cas9 guide RNAs suitable for genomic medicines utilizing different delivery modalities and addressing different therapeutic areas.

Fig 1. Screening for highly efficient ADAR-targeting gRNAs in human T cells.

A) Diagram of the workflow used for designing, screening, and validating ADAR gRNAs. B) Schematic diagram of the ADAR gene, corresponding protein domains, and the target space used for gRNA design. C-D) Results of the ADAR gRNA primary screen in human T cells, with gRNAs arranged in rank-order by average frameshift (C) or by location in the gene body (D). Points represent the average of 3 biological replicates. Codon location corresponds to the cut site relative to the coding sequence of the p150 isoform. E) Editing efficiency observed with ADAR gRNA secondary gRNA screen in primary human T cells. Guide RNAs are sorted by rank order of the 4 µM dose. Data points represent the average of 3 biological replicates ±  S.D. F) Comparison of frameshift mutation rates observed in the primary vs. secondary screens for the 22 gRNAs chosen for follow-up. Data are plotted as mean ±  S.D.

Fig 2. Guide RNA validation in human macrophages.

A) Frameshift mutation rate for the 22 ADAR-targeting gRNAs chosen for follow-up in primary human macrophages. B) Comparison of frameshift mutation rates observed in primary human T cells vs. macrophages. Values for T cells are the same as shown in Fig 1F. Guide RNAs that were advanced for further validation are shown in orange. Data points represent the average of 3 biological replicates ±  S.D.

Fig 3. Loss of ADAR protein after editing with ADAR-targeting gRNAs.

A) Loss of ADAR protein as measured by intracellular flow cytometry. Samples for flow cytometry were taken in parallel with gDNA collection in samples shown in Fig 1E. Data are plotted as mean ±  S.D. B) Comparison of the average frameshift mutation rate vs. protein loss observed in the secondary screen in human T cells. The average frameshift mutation rate and average protein loss are shown for the 4 µM RNP dose taken from Fig 1E (frameshift) and Fig 3B (protein loss) (Fig 1E) vs. average protein loss at the 4 µM RNP dose. C) Results of a kinetic experiment tracking the appearance of frameshift mutations and ADAR protein loss over time after T cells were treated with an ADAR-targeting RNP. For each time point after treatment, samples were taken for gDNA isolation and quantification of frameshift mutation by NGS, or for analysis of ADAR protein loss by intracellular flow cytometry. The average value for frameshift mutation and protein loss at each time point is plotted in blue and red, respectively. Data points represent the mean of 3 biological replicates.

Fig 4. Immunological responses induced by ADAR targeting revealed by RNA-Seq analysis

A) Comparison of p150 isoform specific targeting and p110/p150 dual isoform ADAR-targeting gRNAs as assessed by principal component analysis of RNASeq data. Points are shown in the first two principal components. B) The top 20 genes weighted in principal component 1 of the RNAseq principal component analysis. C) Principal component 1 values of RNAseq analysis displayed as a function of editing activity of gRNAs are displayed on a per sample basis. Guide RNAs located in the most 5’ region of ADAR are denoted (cg007, cg019, cg020). D) Upregulated gene ontology terms and statistical significance measured by -log₁₀(adjusted-p-value) are displayed for representative gRNAs cg096 (p150 specific targeting) and cg223 (p110/p150 dual targeting). E) The fold-change of CXCL10 mRNA measured by RNAseq is plotted as a function of frameshift editing rate.