Somatic Gene Editing of GUCY2D by AAV-CRISPR/Cas9 Alters Retinal Structure and Function in Mouse and Macaque

Abstract

Mutations in GUCY2D, the gene encoding retinal guanylate cyclase-1 (retGC1), are the leading cause of autosomal dominant cone-rod dystrophy (CORD6). Significant progress toward clinical application of gene replacement therapy for Leber congenital amaurosis (LCA) due to recessive mutations in GUCY2D (LCA1) has been made, but a different approach is needed to treat CORD6 where gain of function mutations cause dysfunction and dystrophy. The CRISPR/Cas9 gene editing system efficiently disrupts genes at desired loci, enabling complete gene knockout or homology directed repair. Here, adeno-associated virus (AAV)-delivered CRISPR/Cas9 was used specifically to edit/disrupt this gene's early coding sequence in mouse and macaque photoreceptors in vivo, thereby knocking out retGC1 expression and demonstrably altering retinal function and structure. Neither preexisting nor induced Cas9-specific T-cell responses resulted in ocular inflammation in macaques, nor did it limit GUCY2D editing. The results show, for the first time, the ability to perform somatic gene editing in primates using AAV-CRISPR/Cas9 and demonstrate the viability this approach for treating inherited retinal diseases in general and CORD6 in particular.

Keywords: AAV; CRISPR/Cas9; GUCY2D; cone rod dystrophy; gene editing; retina.

Figure 1.

Staphylococcus aureus (Sa)Cas9 guide RNAs (gRNAs) targeting macaque (Macaca fascicularis)/human GUCY2D and mouse Gucy2e loci were selected to target early coding sequence within the extracellular domain of the genes. Protospacer and protospacer adjacent motif recognition sequences are shown in an alignment of the corresponding mouse, human, and macaque sequence. Green star represents location of R838S mutation (A). AAV vector plasmids containing Gucy2e gRNA16, GUCY2D gRNA27, and SaCa9 are shown (B).

Figure 2.

AAV-CRISPR/Cas9-based editing of Gucy2e in photoreceptors of GC1^+/–:GC2^−/− mice alters retinal function and structure as a consequence of reduced retGC1 expression. By 20 weeks post injection, photopic (cone-mediated) and scotopic (rod-mediated) function were significantly reduced (p = 0.012 and p = 0.013, respectively) in eyes injected with AAV-Cas9 + AAV-Gucy2e gRNA (n = 10) relative to those injected with AAV-Cas9 + AAV-orthologous gRNA control (n = 10) or vehicle alone (A and B). AAV-Cas9 + AAV-orthologous gRNA control-injected eyes did not differ from vehicle-injected controls (A). By 20 weeks post injection, significant loss of retinal structure was observed in Gucy2e-edited eyes relative to controls (C). Cas9 expression was detectable in all vector-treated eyes (D). retGC1 expression was reduced by ∼70% in Gucy2e-edited eyes relative to eyes injected with AAV-Cas9 + AAV-orthologous gRNA control (D and E). Transcript analysis revealed both Cas9 and gRNA expression in green fluorescent protein (GFP)+ sorted cells from treated eyes (n = 10 pooled for each group) (F). Mice in this cohort received two co-delivered vectors, each at 3 × 10¹² vector genomes (vg)/mL, for a total concentration of 6 × 10¹² vg/mL.

Figure 3.

Representative retinal cross-sections from GC1^+/–:GC2^−/− mice injected with either AAV-Cas9 + AAV-Gucy2e gRNA (A–G) or AAV-Cas9 + AAV-orthologous gRNA control (H). The low magnification image from the edited eye clearly shows a reduction of retGC1 expression and ONL thinning within the injection bleb (A). High magnification images taken outside the injection bleb of this eye (B–D) reveal clear retGC1 expression in cone outer segments, whereas retGC1 is absent from cone outer segments within the bleb where editing occurred (G). Sections were stained with antibodies raised against retGC1 (red) and cone arrestin (purple), and counterstained with DAPI (blue). Confocal images are focused on the edge of the subretinal injection bleb. Red and green schematics above the low magnification images reflect the levels of retGC1 and gRNA/GFP expression across each respective section. ONL, outer nuclear layer; INL, inner nuclear layer; GC1, ganglion cell layer. Scale bars in (A) and (H) = 25 μm; scale bars in (B) and (E) = 10 μm.

Figure 4.

Post-injection fluorescent fundus images and corresponding pre-and post-injection OCT images in macaques. Representative images from OD and OS eyes of GR114QB are shown. Left panels: GFP fluorescence in right and left eyes was detected by 488 nm Spectralis imaging. Horizontal red/blue arrows denote the OCT line scan location, with red denoting the region injected subretinally and blue denoting the uninjected region of the OCT scan. Right panels: OCT images before and 2 months post injection. Vertical red (injected) and blue (uninjected) arrows delimit the region between the outer plexiform layer and the interdigitation zone. Higher power inserts of selected retinal regions show the loss in the right eye post surgically of the ELM and photoreceptors in the injected region. The ELM and photoreceptors are clearly present in the region of the subretinal “control” injection in the left eye. Ch, choroid; ELM, external limiting membrane; EZ, ellipsoid zone of the photoreceptors; IZ, interdigitation zone of cones in RPE; OCT, optical coherence tomography; OPL, outer plexiform layer; RPE, retinal pigment epithelium. Scale bars = 200 μm.

Figure 5.

Representative retinal cross-sections from macaques injected with either AAV-Cas9 + AAV-GUCY2D gRNA (A–D) or AAV-Cas9 + AAV-orthologous gRNA control (E–H). Sections were stained with antibodies raised against retGC1 (red) and rhodopsin (purple), and counterstained with DAPI (blue). Confocal images are focused on the edge of the subretinal injection bleb. Red and green schematics above low magnification images reflect the expression of retGC1 and gRNA/GFP across each respective cross-section. Protein from retinal blocks of GUCY2D-edited and control eyes were probed for retGC1 and cone arrestin (I). retGC1 expression (normalized to cone arrestin) was reduced by ∼80% in GUCY2D-edited eyes (J). Scale bars in (A) and (H) = 25 μm; scale bars in (B) and (E) = 10 μm.

Figure 6.

Enzyme-linked immunospot (ELISPOT) assay of T-cell responses to SaCas9. (A) A representative ELISPOT plate image. Peripheral blood mononuclear cells isolated from JR40D at specified time points were plated at 160,000–370,000 cells per well and mixed with individual SaCas9 peptide pool containing ∼36 peptides of either 9mers overlapping by four amino acids, or 15mers overlapping by 10 amino acids. The total peptides from six pools tile the entire SaCas9 protein sequences. The control NHP from CepheusBio was immunized with viral antigens in Pepmix and Tetanus. No T-cell antigens were added to the wells with cells alone. INF-γ-positive spots normalized to 1E6 peripheral blood mononuclear cells are presented in (B) for CD8 T-cell responses to SaCas9 9mer antigens, and (C) for CD4 T-cell responses to SaCas9 15mer antigens.