(594) Correction of a CFTR canonical splice site variant, 3120+1G >A, by adenine base editing

Background: The CFTR variant 3120+1G >A (c.2988+1G >A) results in complete skipping of exon 18, absence of CFTR protein, and no response to modulators. It is the 20^th most common CF-causing variant in the CF population and the most common variant in individuals of African descent. Currently, half of individuals bearing 3120+1G >A are not eligible for modulators. Canonical splice site variants remain challenging to treat at the RNA or protein level. However, these variants are relatively safe targets for genome editing as bystander edits are more likely to alter intronic rather than exonic (coding) sequence.

Methods: Adenine base editing (ABE) allows for targeted A >G conversion with a high on-target efficiency and a reduced likelihood of undesired genome alterations, as compared to other editing approaches. We used the recently evolved NRCH-ABE8e editor in isogenic HEK293Flp cells stably expressing a CFTR expression minigene (EMGi14-i18) with 3120+1G >A to compare ten different sgRNA designs, which placed the target site at different positions within the protospacer (A₃-A₉, relative to the 5’ end). In this cell model, the endogenous CFTR genes remain intact, allowing for analysis of editing at a target with a protospacer mismatch at the disease-causing nucleotide, analogous to a compound heterozygote. To determine the functional consequence of editing, we used our top four designs to edit CF bronchial epithelial (CFBE) isogenic cells stably expressing 3120+1G >A EMG and then performed short circuit current measurements. Lipofectamine-mediated delivery of plasmids was used for editing of both HEK293s and CFBEs. Base editor mRNA and the two most efficient sgRNAs were delivered via electroporation to primary human bronchial epithelial (HBE) cells with genotype 3120+1G >A/F508del.

Results: The most efficient sgRNA design had 10.1±2.6% (mean±SEM) editing of the target site at A₄ in HEK293s, as well as bystander editing of 2.0±0.4%, 15.0±3.8%, and 9.6±2.2% at positions A₂ (c.2987A >G; Q996R), A₆ (c.2988+3A >G) and A₁₀ (c.2988+7A >G), respectively. The endogenous CFTR within the same cells had editing of 18.3±2.6% at A₆ and 1.7±0.7% at A₂. Splice evaluation showed 23.1±0.01% exon 18 skipping in cells expressing c.2988+3A >G EMG and normal splicing when the other two bystander edits were expressed. Expression of CFTR bearing the exonic bystander Q996R in CFBEs showed no alteration in CFTR function. CFTR function increased after editing of CFBEs stably expressing 3120+1G >A EMG from 0.12±0.03μA/cm² (mean±SEM) to 17.2±1.0μA/cm². This function was augmented to 25.6±0.6μA/cm² by modulators, corresponding to ~24%WT. In primary HBEs (3120+1G >A/F508del), we observed higher levels of editing with an allelic conversion of 75% at the +1 nucleotide as well as 87% A >G conversion at the +3 site. Notably, the +3 site editing rate being in excess of 50% indicates that some bystander editing must be occurring at the non-target (F508del) allele. However, this change is predicted to result in only modest loss of full-length transcript (87% conversion x 20% missplicing= 17% reduction in mRNA). Consistent with this finding, editing in HBEs increased CFTR channel function to ~40%WT.

Conclusions: Adenine base editing corrects 3120+1G >A, while also editing nearby intronic adenines at a high rate. Despite the presence of these bystander edits in both the target and non-target alleles, CFTR function in primary cells was restored to a level that should yield significant clinical improvement.

Acknowledgements:

References: