Development of a Subretinally Delivered CEP290-Specific CRISPR Medicine for the Treatment of Leber Congenital Amaurosis 10 (LCA10)

11 Hurley St Michael Stefanidakis, Morgan Maeder, George Bounoutas, Minerva Sanchez, Eugenio Marco, Cambridge, MA 02141 Georgia Giannoukos, Dawn Ciulla, Clifford Yudkoff, Steven Samuelsson, Hoson Chao, Maxwell Skor, Gregory Gotta, Sebastian Gloskowski, Christopher Wilson, Charlie Albright, Haiyan Jiang

Introduction Results

For ocular diseases with well-defined genetic defects, such as Leber Congenital Amaurosis Efficient Editing of Mouse Retina by Subretinal Delivery of EDIT-101 (LCA), CRISPR-based genome editing represents a novel therapeutic modality for previously unaddressable disease targets. LCA10 is an early-onset retinal degeneration A. B. C. caused by mutations in the CEP290 . LCA10 is not amenable to AAV-mediated gene GFP 60 replacement therapy because the large size of the CEP290 cDNA (~7.5kb) exceeds the g n i

packaging capacity of AAV. AAV t i d 40 E

• In normal photoreceptor cells, CEP290 is located in the connecting , where it e v i t

is required for protein trafficking that supporting outer segments development and vision. c u d o • In CEP290-associated LCA10 patients, despite severe vision loss and rapid loss of rods, r 20 P

central nonfunctional foveal cones and intracranial visual pathways remain structural % intact. 0

WT Photoreceptor LCA10 Photoreceptor Whole Retina GFP+ Retinal Cells

Degenerated discs HuCEP290 IVS26 KI mice were treated with AAV5-GRK1-GFP either alone or with EDIT- Discs Protein t

n trafficking 101 by subretinal injection. Six weeks later, GFP+ area was quantified in flat-mounted e m t g n e e s

retinas which showed that approximately 30% of the neuroretina were positive for GFP m r g e t CEP290 e u s

r O

e expression (A), which is consistent with the ISH of AAV vector genome (B). Productive t

u No protein Connecting cilium O trafficking HuCEP290 editing rates were determined in genomic DNAs isolated from either total neuroretinal cells, or sorted GFP+ photoreceptor cells (C). t t n n e e m m g g e e s s

r r e e

n Rapid Onset and Stable CEP290 Gene Editing by EDIT-101 in Mice n n n I I

A. * B. 5000 n

) a l i

Boye et al. 2014 20 l d e e c * /

M 4000 s

e l - 16

• The majority of LCA10 patients are homozygous or compound heterozygous for a u ) c % e l ( 3000 common intron 26 (IVS26) mutation that creates an aberrant splice site, leading to the * o

g 12 m n i ( t

misincorporation of a cryptic exon of 128 nucleotides, and consequently a mutant, non- i A d 2000 E N

8 R functional CEP290 protein. e v i m t

c 9 1000 u 4 s * d • Our therapeutic strategy is to use a SaCas9/gRNA pair to specifically remove the intronic a o C r sequences flanking the mutation, thus restore normal CEP290 RNA splicing and protein P 0 0 0.42 1 2 4 8 16 26 0.42 1 2 4 8 16 26 expression. Time Post Dosing (Week) Time Post Dosing (Week)

CEP290 with IVS26 mutation CEP290 corrected with EDIT-101 C. D. 20000 Inversion

IVS26 IVS26 ) l

l *

Exon 26 Exon 27 Exon26 Exon27 e DNA * * c 15000 / s e l u

Deletion c * e l 10000 o SaCas9 m (

Transcription A

N @ ONL 5000 R

g * mRNA Exon 26 X Exon 27 Exon 26 Exon 27 0 0.42 1 2 4 8 16 26 Time Post Dosing (Weeks)

Translation In EDIT-101-treated HuCEP290 KI mice, the productive edits were detectable as early as on Day 3, and increased significantly by Week 1 (p<0.0001 vs Day 3). The editing rates p.Cys998X Full length, Protein were maintained through 6 months (A). The levels of SaCas9 mRNA/gRNA increased Prematurely truncated functional significantly by 2 weeks post dosing and then reduced significantly by 6 months (p<0.05) (B and non-functional CEP290 & C), without impacting the editing rate. The Cas9 protein was detected exclusively in photoreceptor cells (D). Previously, we demonstrated that CEP290 gRNAs induce targeted deletion leading to normal CEP290 mRNA and protein expression in LCA10 patient-derived fibroblasts. Dose Response in CEP290 Gene Editing and CRISPR Expression Productive edits include sequence inversions and deletions that correct the LCA10- a a n associated splicing defect of CEP290. GUIDE-Seq followed by next generation targeted n i i t A. t 80 B. 100 e e R sequencing did not detect any off-target cutting by CEP290 gRNAs in a number of human R

d d e e c cells. Finally, the surrogate NHP vector achieved targeted CEP290 gene editing in macular c 80 u u 60 d d s s n photoreceptor cells following a single subretinal injection in cynomolgus macaques. n a a 60 r r T T / /

(Maeder et al 2017)

40 t t i i d d E

E 40

e In this study, we determined the kinetics and the dose response of EDIT-101 in the e v v i i t t 20 c c u transgene Cas9/gRNA expression and on-target CEP290 editing in human CEP290 IVS26 u Minimal 20 d d o o Therapeutic r knock-in (KI) mice to support First-in-Human trials. r P P Target

0 0 % % 1010 1011 1012 1013 1014 10 100 1000 10000 100000 Cas9 mRNA (Mol/Cell) Methods EDIT-101 Dose Concentration (vg/mL) Given the stable gene editing efficiency over time, we pooled the productive gene editing EDIT-101, which is an rAAV5 vector expressing SaCas9 and human CEP290-specific data across different time points and normalized to the transduction efficiency of the total gRNAs, was produced by transient plasmid transfection of HEK293 cells, and processed retina. (A) The dose range of EDIT-101 to achieve target therapeutic threshold of 10% of with final sterilize filtration. productive CEP290 edits in photoreceptor cells (Geller and Sieving 1993; Geller, Sieving Human CEP290 IVS26 KI mouse contains the human CEP290 exon 26, intron 26 with the and Green 1992). (B) The correlation of productive CEP290 editing efficiency with the LCA mutation c.2991+1655A>G and exon 27 in the murine CEP290 gene through expression levels of SaCas9 mRNA. homologous recombination (Garanto, Duijkers, and Collin 2015), All mice used in these studies were heterozygous for the HuCEP290 IVS26 mutation, thus the % of CEP290 gene Conclusions editing is the same as the % of cells edited. Mixed gender of HuCEP290 IVS26 KI mice, at 6 – 8 weeks of age, were treated in both Subretinal delivery of EDIT-101 has demonstrated efficient transduction of mouse eyes with a single subretinal injection of either vehicle or escalating doses of EDIT-101. neuroretina and achieved predictive therapeutic levels of targeted CEP290 gene editing in Animals were sacrificed at specified time points at Day 3 – 6 months. Fresh neuroretinal HuCEP290 IVS26 KI mice. The onset of CEP290 gene editing is rapid, correlates with the samples were collected for genomic DNA and RNA extraction. On-target CEP290 gene levels of Cas9/gRNA and lasting, while the expression of CRISPR reduces over time. editing was determined by UDiTaS™ deep sequencing method. The expression levels of The results provide strong support for clinical development of EDIT-101 for treatment of Cas9 mRNA and gRNA were measured by RT-QPCR. Mouse eye cups were also fixed for LCA10. If successful, this in vivo CRISPR approach may have broad application to other immunohistochemistry of Cas9 protein and in-situ hybridization (ISH) of AAV vectors. inherited retinal degenerations with significant unmet medical needs.

References: (1) Boye, S. E. et al. 2014. PLoS One, 9: e92928. (2) Maeder, M. et al. 2017. Mol Ther, 25(5S1): 353. (3) Garanto, A. et al, 2013. PLoS One, 8: e79369. (4) Geller, A. M., and P. A. Sieving. 1993. Vision Res, 33: 1509-24. (5) Geller, A. M., P. A. Sieving, and D. G. Green. 1992. J Opt Soc Am A, 9: 472-7.