A combined synthetic approach to create genome-editing protein delivery tools for the mouse retina

Lipidoids have been developed as key components of lipid-based particles to facilitate membrane penetration, cargo release and endosomal escape via receptor mediated endocytosis, interactions with the lipid bilayer, and phase transitions in the acidifying endosome²⁹. However, they do not offer specific interactions with proteins. To enable the interactions between the lipids and proteins, we incorporated a protein-binding domain into lipidoids, as shown in Fig. 1. Our first-generation CBB compounds began with commercially available CBB G-250 (CBB22), which was functionalized through the coupling of amides or esters of N-methyl-1,3-diaminopropane (K1) or PEG₃ (K3) to CBB, utilizing phosphorus trichloride to form the amino-sulfonamide or sulfonic ester (see Supplementary Fig. 1 and Synthesis of CBB intermediates and lipidoids in Supplementary Information). This intermediate was then coupled to diverse lipid and non-lipid tails of various lengths and degrees of saturation. For example, CBB1 contained a cholesteryl moiety, and CBB3 contained docosahexaenoic acid (DHA). CBB5 and CBB16 were isolated as byproducts during silica gel chromatographic purification of CBB3 and CBB15. CBB1, CBB3, CBB6, and CBB13 were further methylated to produce the lipidoids CBB2, CBB4, CBB7, and CBB14 with the positively charged linker K2. CBB21 was also coupled to the C-terminus of a cell-penetrating peptide (CPP5, sequence KLPVM), to yield CBB18 and CBB19. Upon long-term storage in DMSO, we noted oxidation of CBB11, resulting in CBBox11. Mass spectrometry of separated impurities indicated that CBBox11 comprised two isomers, likely representing hydroxylation events on the polyunsaturated fatty acid or the tertiary amine (Supplementary Fig. 2).

We also developed an alternative pathway to generate protein-binding lipid derivatives of CBB. We reasoned that since our first-generation CBB derivatives eliminated the sulfonic-acid functionality of CBB, retention of this group would enhance protein-CBB interactions, as CBB G-250 binds to proteins through hydrophobic interactions with aromatic residues (Phe, Tyr, Trp) and electrostatic interactions between the anionic sulfonate and basic amino acids (His, Lys, Arg)³⁰. We created a CBB derivative, CBBZ0, with a propargyl ether functionality that can be linked by click chemistry to lipid/lipidoid azides. The construction of the click library began with a family of bifunctional linkers, to which lipidoids were joined via Michael addition reactions with acrylates, as illustrated in Supplementary Fig. 3. The alcohol functionality of the bifunctional linker was then converted to an azide via mesylation and displacement with sodium azide. Finally, the CBBZ compounds were assembled via click reactions, creating the CBB-triazole library consisting of compounds CBBZ1-10.

First, we used a small genome-editing protein, Cre recombinase, to assess the CBB lipidoid-mediated intracellular protein delivery. We used a HEK293-loxP-GFP-RFP color-switching reporter cell line (CS) that constitutively expresses GFP, while an upstream stop codon prevents expression of RFP. Cre recombinase activity in the nucleus triggers the excision of the floxed GFP-stop sequence, and RFP expression replaces GFP (Fig. 2a)¹⁵. Nine compounds (CBB2, 7, 11, 12, 13, 14, 18,19, Z1) were selected for the titration of the [CBB] to [Enzyme] ratio. The CBB lipidoids were dissolved in DMSO (1 volume) and complexed at room temperature with Cre recombinase (19 volumes) at a 2:1, 4:1, 8:1 or 16:1 molar ratio before delivery (Supplementary Fig. 4a). The enzymatic activity of cytosolic Cre was evaluated by fluorescence microscopy according to the extent of GFP to RFP conversion, which indicated that Cre recombinase intracellular delivery was dependent on the CBB-lipidoids concentration, and the dose-response relationship showed that the effect was generally saturated at a ratio 8:1 for most compounds. Accordingly, in the overall screening of the CBB lipidoids, the [CBB]/[Enzyme] ratio was set to 8:1, Lipofectamine 3000 (LF) served as a positive-control vehicle, and free Cre protein was used alongside non-treated cells as a negative control. Compounds were considered positive hits if the ratio of red-to-green (RFP/GFP) quantified by flow cytometry (CBB compounds) or fluorescence microscopy (CBBZ compounds) was two-fold higher than in the cells treated with free Cre (Fig. 2b, panel “C”). Six compounds from the amide coupled library (CBB6, CBB11, CBB14, CBB18, CBB19, and CBB21) demonstrated the ability to enhance delivery of Cre (Fig. 2b, c). The CBBZ approach was tested with C12 and C16 aliphatic chains present in CBB12 and CBB6, respectively, but it did not improve the ability of CBB to deliver Cre compared to amide-linked CBB compounds (Fig. 2d, C12: Z1, Z2, Z6, Z10; C16: Z3, Z7, Z8). Although SM102 enabled efficient delivery of ABE-RNPs in our previous study¹⁵, neither of the CBBZ compounds, in which the branched 9-heptadecyl lipid moiety of SM102 or the entire SM102 moiety were attached to CBB via a triazole ring, displayed protein-delivery ability (Fig. 2d, Z4, Z5, Z9). We noted the formation of a substantial blue precipitate immediately following the complexation of Cre with any of the CBBZ compounds, more pronounced than with amide-linked CBB compounds. This precipitation may have resulted from stronger protein binding of the CBBZ compounds than the CBB compounds, due to conservation of the CBB sulfonic acids in the CBBZ compounds, followed by colloidal aggregation. Despite this outcome, the robust click-based synthetic methods described here may hold significant promise for expanding the chemical space around lipidoids with protein-binding domains in the future. The Cre recombinase and CBB compounds were tolerated well by the CS cells, as shown by flow cytometric exclusion of DAPI (Supplementary Fig. 5a).

The highest GFP-to-RFP conversion, 58.2 ± 2.8%, was achieved for Cre complexed with CBB14, which contained butyrate tails. Methylation of the ionizable tertiary-amine linker K1 to the positively ionized quaternary ammonium K2 was important for the protein trafficking activity of CBB14, as CBB13, a precursor with the same butyrate tails but with the K1 linker, did not consistently deliver Cre activity into the reporter cells. Alkylation of other CBB compounds that had lipid chains containing cholesterol (CBB1), DHA (CBB3), or palmitate (CBB6), to CBB2, CBB4, and CBB7, respectively, did not improve in vitro delivery of Cre. Notably, CBB21, a CBB G-250 amide without lipid moieties, delivered Cre recombinase into reporter cells and triggered a pronounced GFP-to-RFP conversion. Although we previously found that delivery of Cre recombinase fused to CPP5 was negligible in vitro and in vivo¹⁵, here, delivery of a CPP5-CBB conjugate (CBB19) was robust, producing up to 22.4 ± 0.6% GFP-to-RFP conversion. CBB18, a combination of CPP5 and TFA, further increased the color conversion to 49.0 ± 3.6%, suggesting synergistic effects between cell-penetrating peptides and CBB. The leading CBB compounds for Cre recombinase delivery, CBB14, 18, 19 and 21, may have a greater positive charge under neutral aqueous conditions, and have less hydrophobic lipid moieties. CBB11-Cre complex consisted of homogenous particles whose size was approximately 1440 nm, whereas CBB14-Cre consisted of two molecular species whose diameters were approximately 167 and 2130 nm (Supplementary Fig. 6). While both CBB compounds aggregated with Cre into large complexes, the presence of smaller species in CBB14-Cre may explain the enhanced Cre delivery activity. Overall positive charge and lower hydrophobicity may be beneficial for Cre delivery, despite the pronounced aggregation. Lower hydrophobicity may increase long-term colloidal stability of the complexes. In our hands, highly hydrophobic CBB1 and CBB2, that were previously proposed as general protein-delivery vehicles via an endocytosis-independent tagging-penetrating mechanism³¹, did not deliver Cre into the reporter cells.

After in vitro nomination of CBB lead compounds, Cre recombinase was delivered with the in vitro hits CBB6, CBB14, CBB11, and CBBox11, or Cre recombinase alone, into the eyes of mT/mG Cre reporter mice via subretinal injection. Analogous to the reporter cell line described above, the mT/mG mouse is a global fluorescence-reporter strain for monitoring Cre recombinase activity, but with the inverse color change. The Rosa26 locus in mT/mG mice constitutively expresses membrane-localized tdTomato; however, an upstream polyadenylation signal stops the expression of eGFP (Fig. 3a). Upon excision of the floxed tdTomato cassette by Cre recombinase, eGFP expression replaces tdTomato. This red-to-green color switching in mouse tissues, including the retina, can be non-invasively monitored by two-photon microscopy of intact eyes. In control mouse eyes, both the retinal pigment epithelium (RPE) and the photoreceptors exhibited uniform tdTomato expression (Fig. 3b). The injection of 20 µM Cre recombinase (850 ng in 1 µl) in buffer with 7.25% DMSO resulted in minimal eGFP expression in the RPE and photoreceptor layers. Injection of 20 µM Cre complexed with 160 μM (208 ng) CBB14 increased eGFP expression in the RPE, producing up to 90% red-to-green color switching (Fig. 3b and Supplementary Fig. 7). CBB6, CBB11, and CBB14 also mediated delivery of Cre into photoreceptors, with CBB14 achieving up to 20% tdTomato-to-eGFP conversion in photoreceptors (Fig. 3b and Supplementary Fig. 7). Notably, we observed that in vitro, the GFP-to-RFP conversion mediated by CBB11 was moderate, while in vivo CBB11 enhanced Cre-delivery-triggered eGFP expression in both outer and inner segments of the photoreceptors, with up to 30% red-to-green conversion. Although in vitro delivery of Cre recombinase mediated by CBBox11, an oxidation product of CBB11, was marginal, it boosted in vivo eGFP expression exclusively in RPE cells to a level comparable to that mediated by CBB14.

After confirming the ability of CBB to deliver Cre in vitro and in vivo, we next investigated the potential for CBB-lipidoids to transfer CRISPR genome editors. We used the adenine base editor (ABE) ABE8e as a model. To this end, we used our NIH/3T3 fluorescent reporter rd12 cell line³² that contains mCherry and eGFP separated by a 204-bp Rpe65 cDNA fragment containing the nonsense rd12 mutation (c.130 C > T; p.R44X) (Fig. 4a). In this cell line, eGFP expression is suppressed by a stop codon, which can be corrected to an Arg codon by adenine base editing. After the editing, both mCherry and eGFP are expressed, appearing as yellow in merged fluorescent microscopic images.

ABE8e RNP was prepared as previously described³² and applied to the reporter cell lines with or without delivery vehicles. When complexed with Lipofectamine 3000, ABE-RNPs successfully induced eGFP expression, whereas ABE-RNPs incubated with DMSO alone enabled only marginal conversion of the reporter cells (Fig. 4b). We titrated the ABE RNP with selected CBB compounds (CBB2, 7, 11, 12, 13, 14, 18,19, Z1 and found that higher concentrations of CBB compounds may better facilitate RNP delivery (Supplementary Fig. 4b). However, at higher concentrations we also observed pronounced precipitation of CBB compounds and decided to maintain the CBB:RNP ratio of 8:1 to prevent complications such as clogging of flow cytometric fluidics or injection syringes. We then screened our CBB compounds for ABE RNP delivery ability and found that among CBB1-22 and CBBZ1-Z7, consistent eGFP expression in reporter cells was only achieved by CBB3 (3.9 ± 0.4%) and CBB11 (28.6 ± 0.7%) (Fig. 4b–d). The complexing of ABE-RNPs with CBB lipidoids with 10% DMSO also did not affect the deaminase activity (Fig. 4e–g)¹⁵. The common feature of CBB3 and CBB11 are unsaturated lipid domains, DHA (CBB3) and linoleate (CBB11), that contain multiple unsaturated C-C bonds. This may increase the fluidity of the lipid layer and promote fusion of the colloids with the cell membrane. Similarly to Cre, the flow cytometric DAPI exclusion assay showed that the ABE RNP-CBB complexes were tolerated by the rd12 reporter cells (Supplementary Fig. 5b).

We found that the CBB-RNP complexes have poor colloidal stability. Precipitation occurred at high concentrations or upon prolonged incubation, and was associated with a decrease in delivery efficiency. The colloidal instability of CBB11 could have hindered the uptake of ABE-RNPs by the cells or interfered with editing. To improve the distribution of CBB11 in an aqueous environment, we formulated liposomes (Lipo1/3/5) containing, in molar ratio, 5% CBB11, 50% ionizable lipids (either SM102, ALC-0315, or MC3), 10% 1,2-distearoyl-sn-glycero-3-phosphocholine (DSPC), 33.5% cholesterol, and 1.5% 1,2-dimyristoyl-rac-glycero-3-methoxypolyethylene glycol-2000 (DMG-PEG 2000), and a control liposome (Lipo2/4/6) without CBB11 (Fig. 5a). The composition of Lipo2 (50% SM102, 10% DSPC, 38.5% cholesterol, and 1.5% DMG-PEG 2000) resembled the lipid nanoparticles described in our previous work³². We noted an improvement of colloidal properties of CBB11 after incorporating it into the liposomes. Upon buffer exchange, free CBB11 was absorbed by a size-exclusion chromatography resin, whereas liposome-bound CBB11 was recovered in the eluate (Supplementary Fig. 8). Dynamic light-scattering measurements indicated that the two groups of liposomes were highly monodisperse (polydispersity index = 0.16/0.19/0.17, and 0.19/0.20/0.15 for Lipo1/3/5 and Lipo2/4/6, respectively), with an average hydrodynamic diameter of 140-155 nm and 155-161 nm, respectively. In contrast, the particle size of CBB11 in the TAS buffer ranged from 400-500 nm (Fig. 5b and Supplementary Fig. 9a–d). We also did not observe precipitation of ABE-RNP with either Lipo1/3/5 or Lipo2/4/6. The cryo-TEM studies on Lipo1 revealed that it consists of unilamellar vesicles with an aqueous core, and that the ABE-RNP is prone to be recruited to the surface of the liposome (Fig. 5e and Supplementary Fig. 10). Furthermore, the presence of ABE-RNP dramatically changed the zeta potential on the Lipo1 surface (a net decrease of 10.16 mV), but marginally affected the zeta potential on the Lipo2 surface (a net decrease of 0.64 mV) (Fig. 5d). No ABE-RNP was observed to be incorporated into Lipo1. The ABE-RNP complexed with Lipo1 remained accessible to Proteinase K (Supplementary Fig. 11). Thus, the protein binding occurred on the surface of Lipo1, and was intensified by CBB11. As expected, this enhanced liposome-protein interaction boosted intracellular protein delivery. Compared to our positive control Lipofectamine 3000, Lipo1 and Lipo2 increased intracellular Cre delivery to CS reporter cells 2.4- and 2.2-fold, respectively, indicated by increased GFP to RFP conversion visualized with fluorescence microscopy (Fig. 5c). At 40 nM RNP, Lipo1 increased the efficiency of ABE8e-RNP delivery into the rd12-reporter cells, as monitored by eGFP fluorescence, 10.5-, 2.9-, and 1.3-fold when compared to CBB11, Lipo2, and LF, respectively (Fig. 5f–i). The RNP delivery enhancement mediated by CBB11 did not depend on the ionizable lipid, as we observed similar enhancement with liposomes made with ALC-0315 and MC3 (Fig. 5g, h and Supplementary Fig. 9e), and with liposomes formulated without ionizable lipids (Supplementary Fig. 12). With 25 nM RNP, the eGFP fluorescence relative to mCherry in reporter cells treated with Lipo3 was 4.4-fold higher than with Lipo4, and the eGFP expression in cells treated with Lipo5 was 31.9-fold higher than with Lipo6 under the same conditions (Fig. 5g, h). We also investigated the delivery of ABE-RNPs into TIGER HEK cells, where functional tdTomato expression is induced by precise A to G editing of the mutated tdTomato at position 7 of the protospacer (Supplementary Fig. 13a)³³. With 25 nM ABE8e RNP, Lipo1 induced approximately 2-fold higher red fluorescence than our positive control, Lipofectamine 3000. Higher doses of ABE8e led to decreased red fluorescence, originating from bystander editing at of neighboring adenine A9 that leads to the expression of tdTomato with a missense mutation that decreases its fluorescence (Supplementary Fig. 13a, b). The use of a more precise ABE8e N108Q³⁴ led to enhanced rescue of tdTomato fluorescence at a broader range of concentrations (Supplementary Fig. 13c). The incorporation of CBB and the ionizable lipid in the liposomes leads to the creation of a broadly applicable macromolecular delivery system optimized for use with ABE RNP. Separately, CBB and ionizable lipids specifically engage the ABE protein and guide RNA, respectively, as shown by the delivery activity of ionizable-lipid-only and CBB-only liposomes built on a scaffold of phospholipid-cholesterol-PEG liposome. Use of CBB and ionizable lipids leads to synergistic engagement of ABE RNP, as demonstrated by the increased potency of Lipo1/3/5 versus Lipo2/4/6. The dispersion of the CBB-RNP complexes likely is achieved by the embedding of the lipid chains of the CBB lipidoids within the liposomes.

Based on the promising results with Lipo1 and Lipo2 in vitro, we assessed liposome-mediated delivery of Cre recombinase to the retina. Lipo1 and Lipo2 were mixed with Cre recombinase to a final concentration of 3.2 mM total lipids and 20 µM Cre. In the Lipo1-Cre complex, the ratio of CBB11 to RNP was set at 8:1. The resulting mixtures were delivered to the eyes of mT/mG-reporter mice via subretinal injections, and tdTomato to eGFP conversion was monitored by two-photon microscopy (Supplementary Fig. S7)³². Both Lipo1 and Lipo2 selectively delivered Cre into RPE cells, triggering color change with similar efficiency. In contrast to free CBB11, Lipo1 that contained 5% CBB11 did not deliver Cre to photoreceptors, but to RPE (Fig. 3b and Supplementary Fig. 7). A possible explanation of this change in CBB11-delivery specificity is that liposome endocytosis was mediated by receptors such as low-density lipoprotein receptor-related protein 1 (LRP1) whose expression levels in RPE are much higher than in photoreceptors³⁵.

Finally, we investigated the non-viral delivery of ABE-RNPs via our liposome formulations for therapeutic genome editing. To assess the delivery of ABE8e-RNPs, we utilized the rd12 mouse model of Leber congenital amaurosis. These mice harbor a nonsense mutation in Rpe65, which effectively eliminates RPE65 expression within the RPE. Without the enzyme to regenerate the visual chromophore 11-cis-retinal, the visual cycle is non-functional, and the electroretinography (ERG) waveform representing the photoresponse to dim-light stimulus is negligible. Effective editing of the Rpe65 gene restores functional RPE65 expression, the visual cycle, and in turn, induces the ERG response^36–38.

To assess the editing capabilities of ABE8e-RNPs in conjunction with our CBB lipidoids, we measured restoration of functional vision mediated by several delivery vehicles. Two weeks after subretinal injections of 20 µM ABE-RNPs in these vehicles, rd12 mice were dark-adapted overnight, and their scotopic ERG responses were recorded (Fig. 6a). We did not observe an ERG-flash response in untreated rd12 mice, while mice treated with ABE-RNPs in TAS buffer or in TAS with 7.25% DMSO regained recordable ERG responses. The scotopic b-wave amplitudes in response to a 0.5 cd s m^-2 stimulus were 45.8 ± 12.9 μV and 50.6 ± 33.9 μV, respectively. CBB11 and CBB14 complexed to ABE8e performed marginally better than untagged RNPs (60.8 ± 14.9 μV and 58.5 ± 37.4 μV, respectively). These results indicate that our observations in vitro with ABE-RNPs with CBB11, or in vivo with Cre recombinase delivery by CBB14, are not directly translatable to ABE-RNP delivery in vivo. This disconnect could originate from the previously observed propensity of CBB-RNP complexes to aggregate, which, together with the intrinsic propensity of ABE to dimerize^15,39,40 might exacerbate the CBB-induced oligomerization and complicate RNP delivery to a larger extent than observed for Cre. Consistent with in vitro observations for the CBB-RNP complexes, the liposome formulations outperformed the non-liposomal CBB formulations. The restoration of ERG responses was robust, especially when CBB11-containing Lipo1 was used. The b-wave amplitudes were measured at 227 ± 69 μV for Lipo1, and 95.3 ± 62.5 μV for Lipo2. Notably, the best example of Lipo1-mediated editing achieved a b-wave amplitude of 353 μV, estimated to be 76% of the ERG response of wild-type mice. These results support the proposed synergy between nucleic acid engagement by the liposome and protein engagement by the CBB that enables the therapeutic effect of our RNP formulation.

We further analyzed the outcomes of the phenotypic rescue by performing retinoid analysis of eyes treated with ABE-RNP complexed with Lipo1 or Lipo2. After 24 h of dark adaptation, 11-cis-retinal levels reached approximately 40 and 16 pmol/eye for Lipo1 and Lipo2, respectively, estimated to be 10.5 % and 4.2% of the level for wild-type mice (Fig. 6b). We also demonstrated precise editing of the rd12 mutation in the DNA and transcripts isolated from RPE samples obtained from the mice whose ERG measurements are reported in Fig. 6a. Next-generation sequencing revealed up to 9.9% total adenine editing in Rpe65 genomic amplicons (Fig. 6c), which was correlated with up to 28.8% sequence editing of the Rpe65 transcripts (Fig. 6d) in eyes treated with Lipo1. Strikingly, although ABE8e has the propensity to deaminate multiple adenosines within its activity window even upon transient delivery via eVLPs,³⁸ in this case, when delivered in complex with Lipo1, we recorded precise A-to-G corrections by ABE8e up to 5.0% in gDNA sequences (Figs. 6e) and 11.7% in transcripts (Fig. 6f). Precisely edited gDNA sequences in mice injected with ABE-RNPs delivered by Lipo1 were 127-, 20-, 50-, 9.7-, and 6.5-fold higher than those with TAS, DMSO, CBB11, CBB14, and Lipo2, respectively. These findings indicate that the benefits of augmenting liposomes with CBB11 were corroborated by improved visual response and increased editing of the genome.

We also compared the ERG responses and genomic editing of rd12 mice treated with Lipo1-ABE8e-RNPs to the analogous measurements for mice treated with ABE8e-eVLP described previously (Fig. 6a, c, e)⁵. The ABE8e-RNP carried by eVLPs achieved 21.6 ± 6.6% total genomic editing, outperforming lentiviral delivery (11.6 ± 4.5%). Although the genome editing by Lipo1-ABE8e-RNP in this study (5.5 ± 3.3%) is less efficient, both the precise genomic correction rate (2.7 ± 1.5%) and ERG response (227 ± 69 μV) in rd12 mice treated with Lipo1-ABE8e-RNPs are almost 2.2- and 2.3-fold greater, respectively, than those in mice treated with eVLP-ABE8e-RNPs. We attribute this increase of vision-function recovery in rd12 mice to a more transient gene editing in RPE cells by the Lipo1-ABE8e-RNP system, effecting more precise correction. Furthermore, off-target editing analysis by CIRCLE-seq revealed A to G substitutions at background levels in the rd12 mice treated with Lipo1-ABE8e-RNPs, comparable to those observed in untreated animals (Fig. 6g).

To evaluate the safety of the ABE-RNP delivery to the retina mediated by Lipo1, we also administrated ABE-RNP-Lipo1 complex to WT mice (Supplementary Fig. 14) via subretinal injection. Since the oxidation of CBB11 might induce epoxides in the liposome, only freshly formulated Lipo1 was applied to the WT mice. Different batches of Lipo1 showed similar morphology and comparable ABE-RNP delivery efficiency to rd12 reporter cell lines (Supplementary Fig. 14a, b). In vivo retinal imaging by optical coherence tomography (OCT) indicated no substantial alterations to the retinal structure of WT mice administrated the complex of RNP (20 μM) and Lipo1. We noticed the presence of hyperreflective spots visible in the vitreous, indicating possible inflammation⁴¹, which could be ameliorated by lowering the RNP concentration (Supplementary Fig. 14c). The ERG response and the generation of the visual chromophore, 11-cis retinal, of WT mice treated with Lipo1 and RNP (10 μM) were also comparable to those of untreated WT mice (Supplementary Fig. 14d, e). In parallel experiments in which rd12 mice were treated with the same batch of Lipo1 and RNP complex, subretinal injections of 1 μl of Lipo1 with 10 μM RNP achieved optimal ERG response and visual chromophore recovery (Supplementary Fig. 14d, e), comparable to those of rd12 mice treated with Lipo1-ABE-RNP (20 μM) (Fig. 6). These results suggested that optimal vision-function rescue requires a balance between the biological tolerance of the cells and the editing efficiency of the gene editors.

The C57BL/6 J (”WT”, JAX 000664), B6(A)-Rpe65 rd12/J (rd12) (JAX 005379), and B6.129(Cg)-Gt(ROSA)26Sortm4(ACTB-tdTomato,-EGFP)Luo/J (mTmG) (JAX 007676) mice were purchased from Jackson Laboratories (Bar Harbor, ME, USA). mT/mG mice were crossed with BALB/cJ albino mice (JAX 000651) to establish an albino mT/mG line used in this study. The mice of both sexes between 4–13 weeks of age with a body weight of 15–25 grams were used for the experiments (see Source data). The mice were housed in the vivarium at the University of California, Irvine, where they were maintained on a normal mouse chow diet and a 12 h light ( < 10 lux)/12 h dark cycle. The temperature ranged from 75-76 °F, humidity from 30-40%, and animals were given food and water ad libitum. No sex-based stratification or sex-based analysis was performed in this study. Both male and female mice were used. The research applies to both sexes. The functional readout is electroretinography (ERG), and there are no documented differences in ERG responses between male and female mice⁴⁸. We did not find literature evidence of the influence of sex on genome editing in mice. Here, the cohorts are too small to conduct sex-based analysis.

HEK293-loxP-GFP-RFP cells (referred to as “CS cells” GenTarget Inc., San Diego, CA, USA, SC018-Bsd), reporter-rd12 cells (generated previously in Palczewski laboratory)¹⁵ and TIGER HEK cells (generated previously in Palczewski laboratory)³³ were maintained in DMEM/F-12 medium with GlutaMAX supplement (Thermo Fisher, Waltham, MA, USA, 10565018), or in DMEM with glutamine (Thermo Fisher, 11965092), both supplemented with 10% fetal bovine serum (FBS) (Genesee Scientific, San Diego, CA, USA, 25-514H) and optional 100 U ml^-1 penicillin-streptomycin (Thermo Fisher, 15140122) (complete medium) in a humidified incubator at 37 °C, 5% CO₂.

The cells were rapidly thawed from liquid nitrogen storage and passaged at least twice before use in experiments. For protein-delivery experiments, the cells were seeded 24 h prior to the experiment in 24-, 48-, and 96-well plates to reach 50–70% confluence (approximately 100,000, 50,000 and 25,000 cells per well, respectively, for the CS cell line; and 50,000, 24,000 and 10,000 cells per well, respectively, for rd12 reporter cells).

All reactions were performed in oven-dried glassware, under an inert atmosphere with exclusion of moisture from reagents and solvents. Reagents were used as supplied. Liquid chromatography was performed using flash chromatography on silica gel (230-400 mesh), on a CombiFlash NextGen 300+ system, using eluting solvent (reported as a V:V-ratio mixture). Analytical thin layer chromatography (TLC) was performed on 0.25 mm glass-backed Silicycle 60 F254 plates. Visualization of the developed chromatogram was accomplished with UV light (254 nm). Detailed procedures and analytical data are in the Supplementary Information.

The Coomassie compound solution in methanol (Fisher Scientific) was diluted to approximately 5 µM and centrifuged at 17,000 g for 15 min. Five microliters of the supernatant were injected into a Vanquish HPLC system coupled with a Q-Exactive Mass spectrometer (Thermo Fisher Scientific, Waltham, MA). The separation was carried out on a BEH C18 Colum (1.7 µm, 2.1 × 50 mm, Waters, Milford, MA) using a mobile phase consisting of 0.1% formic acid (Thermo Fisher) in water (A) and 0.1% formic acid and 50% 2-propanol (Thermo Fisher) in acetonitrile (B) at a flow rate of 300 µl min⁻¹ with a gradient of 30% B at 0-1 min, 30-100% B at 1–5 min and 100% B at 5–8.5 min. The normalized collision energy (NCE) was set at 40. The data was extracted using Xcalibur 1.4 (Thermo Fisher Scientific, Waltham, MA) and processed with FreeStyle 1.8 P2 (Thermo Fisher Scientific) and GraphPad Prism 10 (GraphPad, San Diego, CA).

¹H and ¹³C-NMR spectra were recorded on a Bruker Ascend Avance III HDTM spectrometer operating at 400 MHz, 500 MHz, or 600 MHz. Chemical shifts reported in δ units (part per million (ppm)) with reference to the residual solvent peak CD₃OD (MilliporeSigma) (δ 4.87, 3.33) for ¹H and (δ 49.15) for ¹³C spectra. Bruker TopSpin 4.4.1 software was used to collect the data, and Mnova 14.3.1 was used to analyze the results. NMR data are presented in the following order: chemical shift, peak multiplicity (s = singlet, bs = broad singlet, d = doublet, t = triplet, q = quartet, dd = doublet of doublets, m = multiplet), coupling constant (in Hz).

Plasmid encoding Cre recombinase for expression in Escherichia coli was a gift from Niels Geijsen (Addgene plasmid # 62730; http://n2t.net/addgene:62730↗; RRID:Addgene_62730). The plasmid sequences were viewed using Snapgene 8.2.1 software. The Cre plasmid was transformed into Escherichia coli BL21star (DE3) bacteria (Thermo Fisher Scientific, C601003) and plated on LB agar (MilliporeSigma) plates, with 50 µg ml⁻¹ kanamycin (Goldbio, St. Louis, MO, USA, K-120-SL25); selected clones were grown in Terrific Broth (Thermo Fisher Scientific, 22711-022) containing 50 µg ml⁻¹ kanamycin, overnight at 37 °C with mixing at 190 rpm. The production cultures were inoculated with overnight cultures and incubated at 37 °C with mixing at 190 rpm. After A_600nm of the culture reached 0.5, protein expression was induced with 0.5 mM IPTG (Goldbio, I2481C25), and incubation was continued at 20 °C for 16 h. The cells were harvested by centrifugation at 7000 × g at 4 °C for 10 min, and the pellets were stored at − 80 °C until used further for protein purification. Cre recombinase was kept on ice or refrigerated at 4–8 °C throughout the purification procedure. The cell pellet from a 1.5-liter culture was thawed in a room-temperature water bath, resuspended in Cre-lysis buffer (50 mM Na phosphate (Fisher Scientific) pH 7.4, 1 M NaCl (MilliporeSigma), 1 complete EDTA-free protease inhibitor cocktail tablet (MilliporeSigma, St. Louis, MO, USA, COEDTAF-RO) per 50 ml buffer), and lysed by sonication (125-W pulses (5 sec on / 5 sec off) for 10 min total) or by French press (3 passes at up to 15,000 psi). The lysate was centrifuged at 48,500 × g for 15 min, and the supernatant was incubated with 1 ml of a suspension of Ni-Sepharose High Performance beads (Cytiva, Uppsala, Sweden, 17526801) in a rotating mixer for 1 h. The resin was centrifuged at 500 × g for 5 min, washed with 40 ml of the Cre-wash buffer (25 mM Na phosphate pH 7.4, 500 mM NaCl), centrifuged again, and then packed in a Tricorn 5/50 (Cytiva) column connected to a Bio-Rad DuoFlow system (Bio-Rad, Hercules, CA, USA) with Bio-Rad BioLogic DuoFlow 5.30 software, which was perfused at 0.5 ml min^-1. The resin was washed with Cre-wash buffer (20 ml, or until a stable absorbance baseline (280 nm) was observed); then, the proteins were eluted with 30 ml of a continuous gradient of the Cre-elution buffer (25 mM Na phosphate, pH 7.4, 500 mM NaCl, and 0–500 mM imidazole (Fisher Scientific)). The fractions containing Cre recombinase that were identified by SDS-PAGE were concentrated and subjected to size-exclusion chromatography on a Superdex 200 Increase 10/300 GL column (Cytiva, 28990944), or a HiLoad 16/600 Superdex 200 pg column (28989335), with P500G buffer (Na phosphate 20 mM pH 7.4, NaCl 500 mM, glycerol (Fisher Scientific) 20% (v/v)) as the mobile phase. Fractions containing pure Cre recombinase were concentrated, snap frozen in liquid nitrogen, and stored at −80 °C.

E. coli BL21star (DE3) bacteria (Thermo Fisher Scientific, C601003) were transformed with the ABE8e SpCas9 NG (referred to as ABE8e) expression plasmid described previously,¹⁵ and grown overnight on Luria broth (LB)-agar plates containing 25 μg ml^-1 kanamycin; single clones were used to inoculate starter cultures in Terrific broth (TB, Thermo Fisher) containing 25 μg ml⁻¹ kanamycin, and grown overnight. One and a half liters of TB containing 25 μg ml⁻¹ kanamycin was inoculated with 10 ml of starter culture, and the culture was grown at 37 °C with mixing at 190 rpm, until it reached an A_600nm of 1.5. Then, the culture was cooled in an ice-water slurry, and protein expression was induced with 0.8% (w/v) rhamnose (Goldbio, R-105-250). The proteins were expressed at 17 °C with mixing of the culture at 190 rpm for 16–24 h. The bacteria were harvested by centrifugation at 5000 × g, 10 min, 4 °C, and the pellets were stored at − 80 °C.

All protein purification steps were conducted in a cold room ( ~ 4-8 °C), or on ice. The thawed bacterial pellet from a 1.5-liter culture was homogenized with a 40 ml Dounce homogenizer in lysis buffer (100 mM Bis-Tris Propane (Millipore Sigma), pH 8.0, 1 M NaCl, 20% glycerol (v/v), 5 mM TCEP (Hampton Research), 1 cOmplete Ultra EDTA-Free protease inhibitor tablet per 40 ml of the buffer), and lysed by sonication with a Qsonica Q125 sonicator (125 W) with a 1/8-inch microtip at 100% amplitude for a total of 20 min (intermittent pulses: 5 sec on; 5 sec off). The lysate was clarified at 4 °C by centrifugation at 48,500 × g for 10 min. ABE protein was captured on a 3 ml TALON metal-affinity resin (Takara Bio, San Jose, CA, USA, 635502). The resin was then washed with 100 ml of the lysis buffer without inhibitors, and with 10 ml of ABE wash buffer (100 mM Bis-Tris Propane, pH 8.0, 500 mM NaCl, 20% glycerol (v/v), 1 mM TCEP). The proteins were eluted with the wash buffer supplemented with 150 mM imidazole.

In a second step, ABE was purified by immunoaffinity chromatography using a 1D4 resin (CNBr-activated Sepharose 4B(Cytiva, 17043001), with immobilized 5–10 mg ml⁻¹ 1D4 antibody purified in house). The 1D4 resin (4 ml) was packed in a DWK Life Sciences Kimble™ Kontes™ FlexColumn™. The column was equilibrated with ABE wash buffer. ABE was loaded by gravity flow. The column was washed with at least 40 ml of the wash buffer at 0.5 ml min⁻¹, and then ABE was eluted with 1 mg ml⁻¹ 1D4 peptide in the wash buffer at approximately 1 ml h⁻¹. Fractions containing ABE were pooled, concentrated, and further purified to remove contaminating nucleic acids and aggregates by size exclusion chromatography on a Superdex 200 Increase 10/300 GL column, or a Superdex 200 16/60 Prep Grade column. The protein was eluted with ABE storage buffer (HEPES 10 mM (Goldbio), pH 7.0, 500 mM NaCl, 20% glycerol). The fractions containing pure ABE were concentrated using Amicon Ultra centrifugal filters with 30-kDa MWCO membranes (Merck, UFC903024), sterilized by passage through 0.22-μm filters, quantified using UV absorption at 280 nm, divided into aliquots, and snap frozen in a chilled metal block for storage at -80 °C. Protein purity was assessed using SDS-PAGE in hand-cast Tris-glycine-SDS discontinuous gels with 4% acrylamide in the stacking gel (pH 6.8) and 10% acrylamide in the resolving gel (pH 8.8), with 2.7% cross-linker (acrylamide:bis-acrylamide ratio of 37.5:1, Bio-Rad 1610158). The samples were mixed with 4x-concentrated Laemmli sample buffer (Bio-Rad 1610747) and supplemented with 50 mM dithiothreitol (DTT, MilliporeSigma D9779), denatured at 70 °C for 10 min and centrifuged before applying the supernatant on to the gel. The protein concentration was quantified using UV absorption spectroscopy by measuring absorbance at 280 nm with a Nanodrop ND-1000 spectrophotometer (software version 3.8.1). Separation of contaminating nucleic acids was followed by monitoring the A_260nm/A_280nm ratio; a ratio < 0.60 was used as an indication of protein free from nucleic acids. A typical A_{260 nm}/A_{280 nm} value for purified ABE was 0.55. Extinction coefficients of the constructs were estimated using the ProtParam tool (https://web.expasy.org/protparam/↗).

The CS cells were plated on 48-well treated tissue-culture plates in complete medium with 10% FBS. For each sample, Cre recombinase was first complexed for 15 min at room temperature with CBB compound in 12.5 μl of OptiMEM medium (Thermo Fisher, 31985070), at final concentrations of 10 μM of Cre enzyme and 20–160 μM of CBB compound. The mixture was then diluted with OptiMEM medium (Thermo Fisher, 31985070) to a final volume of 250 μl and applied to cells. After incubation for 3 h at 37 °C, 5% CO₂, the medium was exchanged with DMEM medium (Thermo Fisher, 10566016) containing 10 % FBS, and the cells were incubated for an additional 24 h. For imaging, the medium was changed to FluoroBrite DMEM (Thermo Fisher, A1896701), and the cells were imaged using a Keyence BZ-X810 microscope with optical filters for GFP and Texas Red. Keyence BZ-X800 Viewer 1.1.1.3 was used to collect the images, and Keyence BZ-X800 Analyzer 1.1.1.2 and XnView 2.51.4 were used to analyze the images. Adobe Illustrator 28.4.1 was used to prepare the figures.

Cells were washed with PBS (Thermo Fisher, 10010023), detached with 0.05% trypsin (Thermo Fisher Scientific, 25300054), transferred to a 96-well round-bottom plate, and centrifuged at 180 × g for 5 min at room temperature. Centrifuged cells were washed with a FACS buffer (PBS with 2% FBS, 100 U ml^-1 penicillin-streptomycin (Fisher Scientific, 15-140-122)), centrifuged again and resuspended in the FACS buffer with 1 µg ml^-1 DAPI (Thermo Fisher, 62248). The cells were analyzed using a Novocyte Quanteon (Agilent) flow cytometer with Pacific Blue (445/45 nm), FITC (530/30 nm), and PE (586/20 nm) optical filters, with NovoExpress 1.6.2 software. Cells were gated on forward and side scatter, and single cells, and analyzed for viability via DAPI exclusion. The gating strategies are outlined in Supplementary Fig. 15.

Synthetic 60-bp-long DNA oligonucleotides (MilliporeSigma, see Supplementary Data 1) labelled with fluorescein on the strand undergoing deamination were annealed at a 1:1 ratio in 10 mM Tris pH 8.0, 50 mM NaCl and 1 mM EDTA (MilliporeSigma) by incubation at 95 °C for 5 min and subsequent slow cooling to 20 °C. End-modified sgRNA (Supplementary Data 1) was ordered from IDT Technologies, with 2′-O-methyl groups on the first three and last three nucleotides, and the first and last three phosphodiester bonds were replaced with phosphorothioate bonds. The guide RNAs were dissolved in nuclease-free water at 37 °C for 15 min at 500 rpm in a thermomixer and folded by incubation at ~ 75 °C for 5 min, followed by slow cooling. Prepared nucleic acids were quantified using UV absorption spectroscopy by measuring absorbance at 260 nm. Accordingly, an A_260nm value of 1.0 corresponded to a DNA concentration of 50 μg ml⁻¹ or an RNA concentration of 40 μg ml⁻¹. DNA was stored at − 20 °C and RNA at − 80 °C. For the ABE enzymatic assay, ABE ribonucleoprotein was assembled by incubation with 1.5-fold molar excess of sgRNA in a reaction buffer (20 mM Bis-Tris propane, pH 7.5, 100 mM KCl (Fisher Scientific), 2.5 mM MgSO₄ (Fisher Scientific), 2 mM DTT, 5% (v/v) glycerol) for 15 min at room temperature. Additional 10% (w/v) sucrose (MilliporeSigma) was used when RNPs were assembled above 4 μM. Then, ABE was diluted to 1 μM with the reaction buffer, preheated at 37 °C and added with 15 ng of DNA substrate (0.02 μM final). The deamination was conducted for 10 - 60 min for ABE8e in the absence/presence of 4- to 16-fold of CBB22, 11, 14. The 20 μl reactions were quenched by addition of 30 μl of water preheated to 95 °C and incubated for 2 min at 95 °C. After cooling, the mixtures were treated with 1 μl of RNase A (20 mg ml⁻¹, MilliporeSigma) and 1 μl of proteinase K (20 mg ml⁻¹, Viagen Biotech 501-PK) for 15 min at room temperature; then DNA products were purified using an Oligo Clean and Concentrator kit (Zymo Research, D4061). Purified DNA was nicked with 5 units of Endonuclease V (EndoV, NEB) for 2-3 h at 37 °C, after which the reaction was quenched by addition of TriTrack DNA-loading dye (1 × final) (Thermo Fisher, R1161) and incubation at 95 °C for 2 min. The cleavage products were analyzed by denaturing polyacrylamide gel electrophoresis with urea (Urea-PAGE) in Bio-Rad MiniProtean continuous hand cast 15% acrylamide gels in Tris-borate-EDTA (TBE; Bio-Rad, 1610770) with 7 M urea and 5% crosslinker (acrylamide:bis-acrylamide ratio of 19:1; Bio-Rad,1610144). The voltage was controlled to maintain at least 42 °C in the electrophoresis chamber. Imaging was done using the ChemiDoc MP system (Bio-Rad) and analyzed with ImageLab 6.1.0 software (Bio-Rad).

ABE-RNP was assembled by incubation of up to 20 µM ABE8e with 1.1-fold molar excess of chemically modified, heat-refolded, synthetic sgRNA (IDT and Genscript, see Supplementary Data 1) in OptiMEM medium with 10% (w/v) sucrose for at least 15 min at room temperature. The sgRNA had stabilizing modifications as described⁴⁹. Next, ABE-RNP was diluted in TAS buffer (20.1 mM Tris, 4.3 mM sodium acetate (Fisher Scientific), pH 7.3, sucrose 10% (v/v)) and added to CBB-lipidoids at 1:2 to 1:16 ratios of enzyme:compound, 5%/7.25% (v/v) final DMSO (MilliporeSigma, D2650-100ML), 10 µM final RNP, and incubated for a further 15 min (Figs. 4 or 5 h) (Fig. 5f, i) at room temperature. The mixtures were applied on the cells in a 48-well plate that had been washed with OptiMEM, at a concentration of 0.1 µM ABE in OptiMEM, without added sucrose. After 3 h, the medium was exchanged with 250 µl of complete medium.

To assemble complexes of ABE-RNP with liposomes (Lipo1-Lipo8), 10 µM ABE-RNP was diluted with TAS buffer and added to concentrated liposomes to achieve 1 µM ABE-RNP with 8-fold excess of CBB-11 (Lipo1/3/5/7) or equivalent amount of lipids (Lipo2/4/6/8) and incubated for 15 min at room temperature. Next, ABE-RNP complexed with liposomes was diluted ten-fold with preheated complete cell culture medium, serially diluted and applied onto the cells in 96-well plate, 100 µl per well.

To assemble complexes of ABE-RNP with Lipofectamine 3000 (LF, Fisher Scientific L3000001), 0.5 µl of LF per well of a 48-well plate, or 0.2 µl of LF per well of a 96-well plate, were diluted to 12.5 µl and 5 µl per well, respectively, with OptiMEM containing 10% (w/v) sucrose; the diluted ABE-RNPs were then added to the diluted lipids and incubated for 15 min at room temperature. A lipoplex of ABE RNP with Lipofectamine in a volume of 25 µl was added to the cells with 225 µl fresh complete medium in a 48-well plate, or 10 µl was added to the cells with 90 µl medium in a 96-well plate.

In all assays, cell culture medium was replaced 24 hours, and the analyses were done 48 h after applying ABE. The cells were imaged using a Keyence BZ-X810 microscope with optical filters for GFP and Texas Red after a medium change to Fluorobrite DMEM with 1% FBS, equilibrated for at least 15 min in CO₂ incubator. Activities of ABE in the rd12 reporter cell line were estimated by comparing ratios of intensities of GFP and mCherry. Activities of ABE in the TIGER HEK reporter cell line were estimated by calculating the intensity of tdTomato in areas covered with cells using ImageJ/Fiji 1.54i.

Cre proteins complexed with CBB lipidoids (as above) were diluted in OptiMEM to 20 µM. First, the mice were anesthetized by an intraperitoneal injection of a cocktail consisting of 20 mg ml⁻¹ ketamine (MWI, 120495) and 1.60 mg ml^–1 xylazine (MWI, 008679) in PBS at a dose of 100 mg kg⁻¹ of ketamine and 8 mg kg⁻¹ of xylazine, and their pupils were dilated by topical administration of 1% tropicamide ophthalmic solution (Akorn, 17478-102-12) and 10% phenylephrine ophthalmic solution (MWI Animal Health, 054243). The corneas were hydrated with GenTeal Severe Lubricant Eye Gel (0.3% Hypromellose, Alcon). Subretinal injections were performed using an ophthalmic surgical microscope (Zeiss). An incision was made through the cornea adjacent to the limbus at the nasal side, using a 27-gauge needle. A 34-gauge blunt-end needle (World Precision Instruments, NF34BL-2), connected to an RPE-KIT (World Precision Instruments, RPE-KIT) with SilFlex tubing (World Precision Instruments, SILFLEX-2), was inserted through the corneal incision while avoiding the lens, and advanced into the subretinal space. Each mouse received a 1-μl injection per eye, and volume and rate were controlled with a UMP3 UltraMicroPump (World Precision Instruments, UMP3-4). After surgery, the mice were placed on a heating pad, and anesthesia was reversed with intraperitoneal atipamezole (2.5 mg kg⁻¹, MWI Animal Health #032800). Triple antibiotic ophthalmic ointment (neomycin, polymyxin, and bacitracin, AmerisourceBergen 044525) was administered to the cornea to promote recovery.

After euthanasia of the mice, their intact enucleated eyes were submerged in room temperature PBS. Pulsing IR light from a Ti:sapphire laser (Coherent, Vision S, Santa Clara, CA; tunable between 690 and 1050 nm) was set to 950 nm and attenuated in a controlled, variable manner with an electro-optic modulator (EOM). To image and spectrally separate GFP and tdTomato signals, two de-scanned spectral detectors were used with their detection bandwidths set to 490–545 nm for GFP, and 590–680 nm for tdTomato. A 1.0 NA 20x Leica objective was used for the imaging^50,51. Maximum projection images were generated from the collected Z-stack function of using LAX 4.7.0 and the Image J analyze particle function was used to quantify transfected cells.

ABE was complexed with 1.1-fold excess of sgRNA in water with 10% (w/v) final sucrose at a protein concentration of 32 µM for 15 min at room temperature, diluted into TAS buffer, and kept at room temperature until injection. Right before injection, ABE RNP in TAS was added to concentrated CBB-lipidoid in DMSO or CBB-liposome in TAS, mixed by rapid pipetting, incubated with the reagents for at least 10 min at room temperature, and injected subretinally into rd12 or WT mice. The remaining reagents were then applied onto rd12 reporter cells in DMEM with 10% FBS in a 96-well plate (Fig. 5f, i).

Mouse eyes were dissected under a light microscope to separate the posterior eyecup (containing RPE, choroid, and sclera) from the retina and anterior segment. Each posterior eyecup was immediately immersed in RLT Plus (Qiagen). RPE, choroid, and scleral cells were detached from the posterior eyecup by gentle pipetting, followed by removal of the remaining posterior eyecup. Cells were then processed for genomic DNA and RNA using the AllPrep DNA/RNA Micro kit according to manufacturer instructions (Qiagen 80284).

Prior to ERG recording, mice were dark adapted overnight. Under a safety light, mice were anesthetized by isoflurane (MWI, 502017) inhalation, and their pupils were dilated with topical administration of 1% tropicamide ophthalmic solution (Akorn; 17478-102-12) and 10% phenylephrine ophthalmic solution (MWI Animal Health #054243), followed by hypromellose (Akorn; 9050-1) for hydration. The mouse was placed on a heated 37 °C Diagnosys Celeris rodent-ERG device (Diagnosys LCC, Lowell, MA, USA). Ocular stimulator electrodes were placed on the corneas, the reference electrode was positioned subdermally between the ears, and a ground electrode was placed in the rear leg. The eyes were stimulated with a green-light stimulus of -0.3 log cd s m^-2 (peak emission 544 nm, bandwidth ∼ 160 nm). The responses for 10 stimuli with an inter-stimulus interval of 10 sec were averaged, and the a- and b-wave amplitudes were acquired from the averaged ERG waveform. Data were analyzed with Espion 6.61.12 software (Diagnosys LLC).

Mice were dark adapted for 24 h before eye enucleation. Eyes were homogenized in 1 ml of a 10 mM Na phosphate buffer (pH 8.0), containing 50% methanol (v/v) (Sigma–Aldrich; 34860-1L-R) and 100 mM hydroxylamine (Sigma–Aldrich; 159417-100 G). After 15 min incubation at room temperature, 2 ml of 3 M NaCl was added. The resulting sample was extracted twice with 3 ml of ethyl acetate (Fisher Scientific; E195-4). Then, the combined organic phase was dried in vacuo and reconstituted in 300 μl hexanes. Extracted retinoids (100 µl) were separated on a normal-phase HPLC column (Zorbax Sil; 5 µm; 4.6 mm × 250 mm; Agilent Technologies) connected to an Agilent Infinity 1260 HPLC system equipped with a diode-array detector. Separation was achieved with a mobile phase of 0.6% ethyl acetate in hexane (Fisher Scientific; H302-4) at a flow rate of 1.4 ml min⁻¹ for 17 min, followed by a step increase to 10% ethyl acetate in hexane for an additional 25 min. Retinoids were detected by monitoring absorbance at 325 nm and 360 nm, using Agilent ChemStation 11 software.

A mixture of CBB11 (0.25 µmol), ionizable lipid (2.5 µmol), DSPC (0.5 µmol, Avanti Polar Lipids), cholesterol (1.675 µmol, MilliporeSigma) and DMG-PEG 2000 (0.075 µmol, BroadPharm, San Diego, CA, USA) in ethanol/chloroform was dried with a gentle flow of nitrogen. The resulting blue film was resuspended in TAS buffer (500 µl), and sonicated for 30 min to form liposomes Lipo1, Lipo3, and Lipo5 that contained ionizable lipids SM-102 (BroadPharm), ALC-0315 (Avanti Polar Lipids), and MC3 (BroadPharm), respectively. Liposomes without CBB11 (Lipo2, Lipo4, Lipo6) contained additional cholesterol in place of CBB11, 1.925 µmol in total. Liposomes without ionizable lipid (Lipo7 with CBB11 and Lipo8 without CBB11) contained additional DSPC (1 µmol total in both Lipo7 and Lipo8) and cholesterol (3.675 µmol total in Lipo7, 3.925 µmol total in Lipo8) instead.

Free CBB11, Lipo1 and Lipo2 were diluted in TAS buffer to an equivalent of 80 µM CBB11 in 100 µl. Then, the samples were applied onto PD-10 desalting columns (Cytiva, 17-0851-01) equilibrated with TAS buffer, and resolved by adding 250-µl aliquots of TAS buffer with manual fractionation. The fractions were transferred to a clear-bottom 96-well plate and scanned for Coomassie fluorescence using a Bio-Rad ChemiDoc MP imager with Bio-Rad ImageLab Touch 2.4 software. The columns were also scanned for Coomassie fluorescence and photographed.

Particle-size distribution was measured using a Malvern Zetasizer Advance Nano instrument (Malvern Panalytical, Malvern, UK) with Malvern ZS Xplorer 3.30 software. Freshly dialyzed liposomes (20 μl) were diluted to 200 μl with TAS buffer and subjected to measurement of particle-size distribution in triplicate. The particle size distributions were processed by the accompanying software to calculate average particle diameter and polydispersity index (PdI).

Cryo-TEM specimen was prepared by plunge-freezing in liquid ethane on a Vitrobot Mark IV (Thermo Fisher Scientific). Five microliters of the sample were applied on a lacey carbon coated grid after glow discharging in a Leica Sputter Coater ACE200 (Leica Microsystems). Excess liquid was removed with blotting of 3.0 seconds under chamber temperature of 8 °C and 100% humidity. The grids were then clipped and transferred into a Glacios TEM (Thermo Fisher Scientific) for examination. The microscope was operated at 200 kV, and images were recorded on a CETA camera (Thermo Fisher Scientific) under low-dose conditions to minimize the radiation damage on the specimen. EPU 2.11.1.11REL (Thermo Fisher Scientific) was used to collect the data.

The genomic context of RPE65 rd12 ABE editing is shown in Supplementary Fig. 16. Complementary DNA (cDNA) was synthesized from RNA with the High-Capacity RNA-to-cDNA kit (Thermo Fisher 4387406), according to manufacturer’s instructions. 0.5 – 1 μl of the isolated genomic DNA or cDNA was used as input for the first of two PCR reactions (PCR1). Genomic loci were amplified in PCR1, using Phusion Plus polymerase (Thermo Fisher Scientific F631S). PCR1 primers are listed in Supplementary Data 1. PCR1 was performed as follows: 98 °C for 30 sec; 30 cycles at 98 °C for 10 sec, 67 °C for 20 sec, and 72 °C for 30 sec; 72 °C for 5 min. PCR1 products were confirmed on a 2% agarose gel. One microliter of PCR1 was used as input for PCR2 to install Illumina barcodes. PCR2 was conducted for nine cycles of amplification using a Phusion HS II kit (Life Technologies). Following PCR2, samples were pooled, and gel-purified in a 1% agarose gel using a Qiaquick Gel Extraction Kit (Qiagen). Library concentration was quantified using the Qubit High-Sensitivity Assay Kit (Thermo Fisher Scientific). Samples were sequenced on an Illumina MiSeq instrument (paired-end read, read 1: 200–280 cycles, read 2: 0 cycles), using an Illumina MiSeq 300 v2 Kit (Illumina) and analyzed using Illumina MiSeq Reporter 2.6.

Sequencing reads were demultiplexed using the MiSeq Reporter 2.6 software (Illumina) and analyzed using CRISPResso2 2.2.7 as previously described^52,53. Batch analysis mode (one batch for each unique amplicon and sgRNA combination analyzed) was used in all cases. The prompt is presented in the Supplementary Information. Reads were filtered according to minimum average quality score (Q > 60) prior to analysis. The following quantification window parameters were used: -w 20 -wc -3. Base editing efficiencies are reported as the percentage of sequencing reads containing a given base conversion at a specific position. Prism 10 (GraphPad) was used to generate the plots.

No sample sizes were pre-determined. Animals were treated by litter cohorts (5-9 animals per litter, 1-2 eyes per animal). Sample sizes were 2-3 animals per group. In vitro cell studies included 2-4 replicates per experiment. Mice with failed subretinal injections were excluded on the basis of no recordable ERG a- or b-wave by a blinded observer. This exclusion criteria were pre-established. In vitro experiments were replicated 2 to 4 times independently. All animal experiments were replicated at least four times independently. In all studies, samples and animals were assigned randomly to treatment groups. Deep sequencing and ERG analysis were conducted blind. Other experiments were not conducted blindly.

No sex-based stratification or sex-based analysis was performed in this study. GraphPad Prism 10 was used to analyze the data. The data were tested for normality, and parametric one-way ANOVA was applied for datasets that passed the normality tests. The data that did not pass the normality test were analyzed with non-parametric ANOVA.

All animal procedures were approved by the Institutional Animal Care and Use Committee of the University of California, Irvine (protocol AUP-24-073), and were conducted in accordance with the NIH guidelines for the care and use of laboratory animals, and with the Association for Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Visual Research.

Further information on research design is available in thelinked to this article. Nature Portfolio Reporting Summary

Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.

The high-throughput sequencing data generated in this study were deposited in the National Center for Biotechnology Information Sequence Read Archive database under accession codes: PRJNA1240374↗ and PRJNA1358750↗. The data underlying the manuscript were deposited in Dryad: doi.org/10.5061/dryad.d51c5b0hd⁵⁴ [https://datadryad.org/dataset/doi:10.5061/dryad.d51c5b0hd↗]. Source data are provided in this paper.