Engineered Cas9 extracellular vesicles as a novel gene editing tool

Abstract

Extracellular vesicles (EVs) have shown promise as biological delivery vehicles, but therapeutic applications require efficient cargo loading. Here, we developed new methods for CRISPR/Cas9 loading into EVs through reversible heterodimerization of Cas9-fusions with EV sorting partners. Cas9-loaded EVs were collected from engineered Expi293F cells using standard methodology, characterized using nanoparticle tracking analysis, western blotting, and transmission electron microscopy and analysed for CRISPR/Cas9-mediated functional gene editing in a Cre-reporter cellular assay. Light-induced dimerization using Cryptochrome 2 combined with CD9 or a Myristoylation-Palmitoylation-Palmitoylation lipid modification resulted in efficient loading with approximately 25 Cas9 molecules per EV and high functional delivery with 51% gene editing of the Cre reporter cassette in HEK293 and 25% in HepG2 cells, respectively. This approach was also effective for targeting knock-down of the therapeutically relevant PCSK9 gene with 6% indel efficiency in HEK293. Cas9 transfer was detergent-sensitive and associated with the EV fractions after size exclusion chromatography, indicative of EV-mediated transfer. Considering the advantages of EVs over other delivery vectors we envision that this study will prove useful for a range of therapeutic applications, including CRISPR/Cas9 mediated genome editing.

Keywords: CRISPR/Cas9 delivery; exosomes; extracellular vesicles; gene editing; optogenetics.

FIGURE 1

EV characterization. (A) Transmission Electron Microscopy reveals the presence of vesicles with the typical size and lipid‐enclosed morphology observed in EV preparations (insert scale bar 100 nm, whole field scale bar 400 nm). (B) Schematic diagram of the four inducible dimerization systems used in this study. (C) EV size distribution determined by nanoparticle tracking analysis reveals no significant differences in diameters of EV secreted by wild‐type cells and by cells engineered to express each of the inducible dimerization systems with the MysPalm EV sorting motif and Cas9 cargo (n = 4 per condition). (D) Western blot shows that EV isolates from non‐transfected conditions, or from each of the inducible dimerization systems with the MysPalm EV sorting motif and Cas9 cargo, are enriched compared to their secreting cells in proteins commonly found in EVs, such as Alix, TSG‐101, Flotilin‐1, β‐actin, syntenin‐1 and the most accepted EV markers CD63 and CD81 (6 μg total protein loaded per sample)

FIGURE 2

Cas9 loading into EVs. (A) Cas9 quantification using calibration curve of wild type (WT) recombinant Cas9 protein alongside tested EV isolates (n = 4 per condition, representative blot for CIBN‐CRY2 loading shown). (B) Comparison of Cas9 loading by the different heterodimerization systems using CD9 or MysPalm for recruitment into EV, compared to the control condition without heterodimerization partner (n = 4, representative blot shown), (C) Cas9 molecules/EV calculated from quantitative blots as shown in C and NTA particle quantification for each isolate (Mean ± SD, n = 4 per condition, statistical significance determined with Kruskal‐Wallis test)

FIGURE 3

Impact of fusion proteins on Cas9 functionality. (A) Schematic of Cre‐Lox reporter cassette adapted for Cas9 cleavage detection with a sgRNA design targeting both LoxP sites. Cas9 is targeted by LoxP sgRNA to cleave DNA at both LoxP sites resulting in excision of GFP‐STOP cassette and expression of RFP. (B) Visualization of gene editing in HEK293‐loxP‐GFP‐STOP‐loxP‐RFP cells after transfection with plasmids encoding the indicated Cas9 fusion proteins, with appearance of RFP upon successful editing of loxP‐GFP (n = 5 per condition, representative images). (C) Relative gene editing efficiency of Cas9 fusion proteins compared to Wild Type (WT) Cas9 and Non‐Transfected (NT) Cells (Mean ± SD, n = 5 per condition, statistical significance determined with one‐way ANOVA)

FIGURE 4

In vitro Cas9 delivery by EVs. (A) Cas9 was delivered by EVs loaded with and complemented by sgRNA from plasmid transfection of reporter cells. Gene editing occurs in transfected cells with the wild type Cas9 and sgRNA and with Cas9CRY2 EVs loaded through MysPalm achieving 50.7% while CD9 loaded, or control passively loaded EVs without CIBN resulted in 4.4‐6.3% gene editing. (Mean ± SD, n = 4, statistical significance determined with Kruskal‐Wallis test). (B) Cas9 and LoxP sgRNA was delivered by EVs. Gene editing occurs in transfected cells with the wild type Cas9 and sgRNA and with Cas9CRY2 and LoxP sgRNA EVs loaded through MysPalm achieving 41.6% while while CD9 loaded, or control passively loaded EVs without CIBN resulted in no gene editing (Mean ± SD, n = 4, statistical significance determined with Kruskal‐Wallis test). Cells were stimulated with an equal amount of EVs (5.5 × 10⁴ EVs/cell)

FIGURE 5

Cas9 transfer by EVs is also observed in transwell co‐cultures or other cell backgrounds. (A) Co‐cultures separated by 0.4 μm transwells show statistically significant gene editing with cells secreting Cas9CRY2 loaded EVs with or without LoxP sgRNA through MysPalm achieving 28.2% and 30.6%, respectively, while cells secreting Cas9CRY2 loaded EVs with all other conditions resulted in 8.6–12.8% gene editing (Mean ± SD, n = 4, statistical significance determined with Kruskal‐Wallis test). (B) Cas9 or Cas9 and LoxP sgRNA was delivered by EVs to HepG2 cells transfected to express a loxP‐tdTomato‐STOP‐loxP‐GFP Cre reporter construct. Gene editing efficiency was determined by flow cytometry to be 25.4% for with Cas9CRY2 EVs loaded through MysPalm while EVs loaded with both Cas9Cry2 and LoxP sgRNA resulted in 19.2% gene editing (Mean ± SD, n = 4, statistical significance determined with one‐way ANOVA)

FIGURE 6

In vitro Cas9 delivery by EVs for PCSK9 gene editing. Cas9 or Cas9 and PCSK9 sgRNA was delivered by EVs. In the case Cas9 alone delivery by EVs, PCSK9 sgRNA was delivered by plasmid transfection of Hek293T cells. Gene editing efficiency was determined by next‐generation sequencing to be 5.8% for with Cas9CRY2 EVs loaded through MysPalm while EVs loaded with both Cas9Cry2 and PCSK9 sgRNA resulted in 4.4% gene editing (Mean ± SD, n = 4, statistical significance determined with one‐way ANOVA). Cells were stimulated with an equal amount of EVs (5.5 × 10⁴ EVs/cell)