mRNA trans-splicing dual AAV vectors for (epi)genome editing and gene therapy

Abstract

Large genes including several CRISPR-Cas modules like gene activators (CRISPRa) require dual adeno-associated viral (AAV) vectors for an efficient in vivo delivery and expression. Current dual AAV vector approaches have important limitations, e.g., low reconstitution efficiency, production of alien proteins, or low flexibility in split site selection. Here, we present a dual AAV vector technology based on reconstitution via mRNA trans-splicing (REVeRT). REVeRT is flexible in split site selection and can efficiently reconstitute different split genes in numerous in vitro models, in human organoids, and in vivo. Furthermore, REVeRT can functionally reconstitute a CRISPRa module targeting genes in various mouse tissues and organs in single or multiplexed approaches upon different routes of administration. Finally, REVeRT enabled the reconstitution of full-length ABCA4 after intravitreal injection in a mouse model of Stargardt disease. Due to its flexibility and efficiency REVeRT harbors great potential for basic research and clinical applications.

Fig. 1. Development and optimization of mRNA trans-splicing in vitro.

a Illustration of split fluorophore assay to test reconstitution via mRNA trans-splicing. A cis-splicing vector served as positive control (cis-ctrl). v1, vector 1; v2, vector 2; BD, binding domain; SDS, splice donor site; SAS, splice acceptor site; pA, polyadenylation signal. b Confocal images of HEK293 cells transfected with v1 and v2 containing different BDs. Scale bar: 50 µm. c RT-PCR of reconstituted full-length cerulean. GAPDH served as input reference. Sequencing result of the splicing junction shown below. bp, base pairs. d Western blots of transfected cells shown in (b). α-cer, anti-cerulean antibody; α-tub, anti-β-tubulin antibody; kDa, kilodalton. e Quantification of reconstitution efficiency relative to cis-ctrl using western blots (WB, minimum of n = 3) or flow cytometry (FC, n = 3). f Western blot of cells transfected with v1 and v2 containing different human (RHO intron 2-derived, BD 5) or bacterial (lacZ-derived, BD 9–12) binding domains. g Ratiometric quantification of reconstitution efficiency of the transfections shown in (f) relative to cis-ctrl. n = 3. h Western blots of cells transfected with v1 and v2 containing different SDS or SAS. i Ratiometric quantification of reconstitution efficiencies relative to cis-ctrl. n = 3. Welch’s ANOVA with Dunnett T3 (SAS) or Kruskal-Wallis test with Dunnett T3 (SDS) was used. j Western blots of cells transfected with versions of v1 containing (+pA) or lacking (−pA) a polyadenylation signal and with versions of v2 containing (+prom) or lacking (−prom) a promoter. k Ratiometric quantification of reconstitution efficiencies relative to cis-ctrl. n = 5 (pA), n = 3 (prom). Two-tailed Mann–Whitney test was used. Scatter plots show mean ± SEM. For all co-transfections equimolar amounts have been used (1 µmol v1 + 1 µmol v2, corresponding to 1 µmol cis-ctrl). All ‘n’ in Fig. 1 represent independent transfections. All source data are provided as a Source Data file.

Fig. 2. Reconstitution of luciferase via dual REVeRT AAVs in vivo and in human retinal organoids.

a Dual REVeRT AAVs expressing a split luciferase gene. Lower panel, Time scale for experiments in WT mice (b, c). b Luminescence obtained from WT mice co-injected with titer-matched dual REVeRT AAV9 vectors. Untreated mice injected with luciferin served as negative control. c Relative quantification of luciferase expression in different organs three weeks post-injection. n = 3 mice. RLU, relative light units. d Time scale for experiments in human retinal organoids transduced with dual REVeRT AAV9 vectors at differentiation day 215 (DD215). dpt, days post-transduction. e Luminescence obtained from transduced organoids. Regions of interest indicating the organoid position were used for quantification and are shown in red. A AAV vector expressing full-length luciferase served as positive control. f Quantification of the results shown in (e). n = 3 retinal organoids. Scatter plots show mean ± SEM. All source data in this figure are provided as a Source Data file.

Fig. 3. Reconstitution of dCas9-VPR in various tissues in vivo upon local and systemic application.

a Upper panel, dCas9-VPR size and the split site used for REVeRT. Lower panel, REVeRT vector cassettes expressing split dCas9-VPR and three sgRNAs (g1–g3) targeting the promoter region of Myo7b. Titer-matched dual REVeRT AAV8(Y733F) vectors were injected via different routes of administration shown in (c, f, h). b dCas9-VPR targeting the Myo7b promoter. c Experimental design for (d, e). OD, Oculus dexter; OS, Oculus sinister. Contralateral eyes injected with AAV formulation buffer served as reference. d qRT-PCR from injected retinas. Shown are the relative expression levels of reconstituted dCas9-VPR (left) and the transactivated Myo7b (right). n = 5 retinas. e Western blot from the retinas of two mice (#1 and #2) injected with REVeRT AAVs or buffer. f Experimental design for (g). g qRT-PCR from injected hippocampi. Shown are the relative expression levels of reconstituted dCas9-VPR (left) and transactivated Myo7b (right). Untreated (untrd) hippocampi served as reference. n = 4 hippocampi (REVeRT), n = 6 hippocampi (untrd). h Experimental design for (i). i qRT-PCR from organs or tissues harvested from intraperitoneally injected mice. Shown are the relative expression levels of reconstituted dCas9-VPR (left) and transactivated Myo7b (right). Untreated organs served as reference. n = 3 organs/tissues. A two-tailed unpaired t test with Welch’s correction was used in (d, g). Scatter plots show mean ± SEM. All source data in this figure are provided as a Source Data file.

Fig. 4. Simultaneous gene knockout and transactivation using REVeRT.

a Upper panel, Dual REVeRT AAV8(Y733F) vector cassettes expressing split Cas9-VPR and three sgRNAs targeting either exon 1 of Rho (g1) or the promoter region of Opn1mw (g2–g4). A lacZ-targeting sgRNA (gL) is used as a non-targeting control. Lower panel, The corresponding dual split intein AAV8(Y733F) vector cassettes used for side-by-side comparison to REVeRT shown in (c). b Experimental design and timeline of experiments shown in c-e. sgRNAs with different spacer lengths are used for a simultaneous knockout (20 nt, g1) of Rho and transcriptional activation (15 nt, g2-g4) of Opn1mw. g4 was added in (d) as indicated. c Relative expression of Rho and Opn1mw in retinas of subretinally injected WT mice. Saline-injected retinas served as a reference. n = 9 retinas (saline, REVeRT), n = 10 retinas (intein). d Relative expression of Rho and Opn1mw upon subretinal injection of Rho^P23H/+ mice with dual REVeRT vectors expressing g1 and either two (REVeRT(3xg)) or three (REVeRT(4xg)) Opn1mw transactivating sgRNAs. Retinas injected with dual REVeRT vectors expressing a lacZ-targeting sgRNA served as a reference. The three retinas used for RNA-Seq analysis (Fig. S5h) are shown in dark pink. n = 15 retinas (saline), n = 6 retinas (REVeRT(3xg)), n = 9 retinas (REVeRT(4xg)). Welch’s ANOVA with Dunnett T3 was used in c and d. e Immunostainings of retinal cryosections of Rho^P23H/+ mice subretinally injected with REVeRT(3xg) (left). A control site distal from the injection site is shown as a reference (right). Magnification of areas marked by white rectangles are shown. Arrowheads indicate rod outer segments expressing M-opsin while lacking rhodopsin signal. Peanut agglutinin lectin (PNA) served as a cone outer segment marker. In the control site, M-opsin is only expressed in cones. Scale bar: 30 µm. Scatter plots show mean ± SEM. All source data are provided as a Source Data file.

Fig. 5. ABCA4 reconstitution and full-length protein expression in a Stargardt mouse model using REVeRT.

a Upper panel, ABCA4 size and the split site (arrowhead). Lower panel, Dual REVeRT AAV.GL vectors expressing split ABCA4. b Experimental design for (c–e) and Fig. S7. c Representative immunostainings on retinal sections of mouse #1 (left) and mouse (#4) showing a strong or weak improvement in OCT and ERG measurements, respectively. Images of the right eye injected with dual REVeRT AAVs and images of the contralateral control eye injected with buffer solution are shown. Scale bar: 30 µm. Cyclic nucleotide-gated channel (Cngb1) antibody was used as a rod outer segment marker to evaluate ABCA4 localization. Some unspecific ABCA4 staining was detected across all sections above the outer segments. d Higher magnification of the area marked by a gray rectangle in (c). Scale bar: 10 µm. e Western blot from retinas of mouse #2 and mouse #3.