Amphiphilic shuttle peptide delivers base editor ribonucleoprotein to correct the CFTR R553X mutation in well-differentiated airway epithelial cells

Abstract

Base editing could correct nonsense mutations that cause cystic fibrosis (CF), but clinical development is limited by the lack of delivery methods that efficiently breach the barriers presented by airway epithelia. Here, we present a novel amphiphilic shuttle peptide based on the previously reported S10 peptide that substantially improved base editor ribonucleoprotein (RNP) delivery. Studies of the S10 secondary structure revealed that the alpha-helix formed by the endosomal leakage domain (ELD), but not the cell penetrating peptide (CPP), was functionally important for delivery. By isolating and extending the ELD, we created a novel shuttle peptide, termed S237. While S237 achieved lower delivery of green fluorescent protein, it outperformed S10 at Cas9 RNP delivery to cultured human airway epithelial cells and to pig airway epithelia in vivo, possibly due to its lower net charge. In well-differentiated primary human airway epithelial cell cultures, S237 achieved a 4.6-fold increase in base editor RNP delivery, correcting up to 9.4% of the cystic fibrosis transmembrane conductance regulator (CFTR) R553X allele and restoring CFTR channel function close to non-CF levels. These findings deepen the understanding of peptide-mediated delivery and offer a translational approach for base editor RNP delivery for CF airway disease.

Graphical Abstract

Figure 1.

Peptide S10 sequence and structure. (A) Amino acid sequence and schematic structure of S10 peptide. ELD-derived, linker and CPP-derived domains are shown. Cationic residues are highlighted in blue, highly hydrophobic residues in gray, polar residues in orange, weekly hydrophobic residues in yellow. Predicted secondary structure by software PyMol is illustrated. (B) Circular dichroism spectroscopy analysis of S10 in PBS solution containing 0%, 25% or 50% TFE. Alpha-helix structures are characterized by a maximal wavelength at 190 nm, and a minimal at 208 nm as observed for S10 in solutions containing 25 or 50% TFE. Disordered structures are characterized by a minimal wavelength of 195 nm and maximal at 212 nm as observed for S10 in PBS solution (no TFE). (C) Percentage of alpha-helix structure formed by S10 peptide in PBS solution containing 0%, 25% or 50% TFE. Column heights represent means and error bars represent SD. n = 2 per group.

Figure 2.

N-Methyl amino acids incorporated in S10 N-terminal sequence decrease alpha-helix level and delivery activity. (A) Sequence of S10 peptide and N-methyl peptides S10-methyl-N2/C2. N-terminal ELD-derived domain is colored in green, linker in black, and C-terminal CPP-derived domain in blue. (B) Circular dichroism spectroscopy analysis of S10-methyl-N2 peptide in the presence of 0, 25, and 50% TFE. (C) Percentage of alpha-helix structure of S10-methyl-N2 peptide in the presence of 0, 25 and 50% TFE. (D) Circular dichroism spectroscopy analysis of S10-methyl-C2 peptide in the presence of 0, 25 and 50% TFE. (E) Percentage of alpha-helix structure of S10-methyl-C2 peptide in the presence of 0, 25 and 50% TFE. (F) Epifluorescence microscopy images of HeLa cells after delivery of 10 μM GFP-NLS using 7.5μM of respective shuttle peptides. (G) Percentage of GFP positive cells after delivery of 10μM GFP-NLS with 7.5 μM of peptides to HeLa cells, as measured by flow cytometry. Column heights represent means and error bars represent SEM, n = 4 replicates per group.

Figure 3.

S237 containing the N-terminal domain of S10 with glycine-serine extensions preserves GFP-NLS delivery activity, resists the inhibition by Cas9 RNP, and improves Cas9 RNP delivery to primary airway epithelia. (A) Sequences of S10 peptide and its derivative peptides. ELD-derived N-terminal domain is colored in green, linker in black, and C-terminal CPP in blue. (B) GFP delivery activity of S10 peptide and the derived peptides in HeLa cells as measured by flow cytometry. Column heights represent means and error bars represent SEM, n = 4 replicates per group. (C) GFP delivery activity of S10 and S237 to CFF-16HBEge cells in the absence (light gray bar) or presence (dark gray bar) of Cas9 RNP measured by flow cytometry. Column heights represent means and error bars represent SEM. n = 4 replicates per group. Statistical analysis by two-tail t-test, NS: not significant, ***P< 0.005. (D) Editing of the CFTR locus in primary HBE ALI cultures from six different donors after delivery of Cas9-RNP with S10 or S237 as determined by NGS (% indels). Column heights represent means and error bars represent SEM. n = 6 replicates per group. Statistical analysis by two-tail t-test, **P< 0.01.

Figure 4.

Delivery of DRI-NLS-Cy5 peptide and Cas9 RNP to pig airways. (A) Epifluorescent microscopy image of pig lung tissue sections. The lung tissue was collected ∼1hr post-administration of DRI-NLS-Cy5 and S237 peptide. Nucleus (DAPI, blue) and Cy5 (red). Scale bar is 200 μm. (B, C) Confocal microscopy images documenting localization of DRI-NLS-Cy5 fluorescence (white) in ciliated cells (pseudo-colored red in (B), AcTub-stained), and mucus producing cells (pseudo-colored red in (C), Muc5ac-stained). Scale bar is 20 μm. (D) NGS results of % GFP gene editing in airway epithelia isolated from segmental airways of the pig porpoise lobe post two administrations of indicated shuttle peptide with Cas9 RNP. Each dot represents data from one pig at the indicated sampling position (upper or lower porpoise lobe main segmental airway).

Figure 5.

S10- and S237-mediated delivery of ABE8e-Cas9-NG RNP targeting CFTR R553X restores CFTR function in primary CF airway epithelia. (A) S10 and S237 delivery of GFP in CFF-16HBEge cells in the absence or presence of ABE8e-Cas9 RNP as measured by flow cytometry. Column heights represent means and error bars represent SEM, n = 4 replicates per group. Statistical analysis by two-tail t-test, NS: not significant, **P< 0.01. (B) Editing of the B2M locus in primary HBE-ALI culture after delivery of ABE8e-Cas9 RNP as determined by the frequency of A•T base pairs converted to G•C base pairs within the editing window measured by NGS. Column heights represent means and error bars represent SEM. n = 3 replicates per group. Statistical analysis by two-tail t-test, *P< 0.05. (C) Top panel shows the targeted DNA strands and PAM sites (blue text). Yellow highlight denotes the predicted 4–8 nt ABE editing window (numbered). Mutation is highlighted in red text. ABE8e-Cas9-NG RNP targeting R553X locus was delivered to primary human airway epithelial cells (R553X/L671X) cultured at the air liquid interface using shuttle peptides S10 and S237. One week following the first application of ABE8e-Cas9-NG RNP with shuttle peptides, Ussing chamber analysis was conducted, followed by gDNA analyses by NGS. (D) % Editing graphically represented as an average frequency of desired product (A•T converted to G•C) at the target site. As the cells are compound heterozygous (R553X/L671X), the percent allelic editing efficiencies were adjusted from the next generation sequencing data using the formula (% base edited-50)/50 ×100. Statistics by un-paired two-way t-test, *P< 0.05. (E) Summary of changes in short circuit current (ΔI_sc) in response to forskolin and IBMX (F&I) (circle symbols) and CFTR inhibitor GlyH (square symbols) in groups represented in (D) and non-CF control donor cells. In both (D) and (E), each symbol, except in non-CF group, represents a technical replicate culture from the same R553X/L671X donor, n = 3 technical replicates per group. For non-CF group, each symbol represents a biological replicate from three different non-CF donors. Column heights represent means and error bars represent SEM. (F) Representative short circuit current tracings presented in (E), comparing non-CF, mock, ABE8e-Cas9 RNP alone, ABE8e-Cas9 RNP + S237, and ABE8e-Cas9 RNP + S10 treated cells.