Safe and effective liver-directed AAV-mediated homology-independent targeted integration in mouse models of inherited diseases

Abstract

Liver-directed adeno-associated viral (AAV) vector-mediated homology-independent targeted integration (AAV-HITI) by CRISPR-Cas9 at the highly transcribed albumin locus is under investigation to provide sustained transgene expression following neonatal treatment. We show that targeting the 3' end of the albumin locus results in productive integration in about 15% of mouse hepatocytes achieving therapeutic levels of systemic proteins in two mouse models of inherited diseases. We demonstrate that full-length HITI donor DNA is preferentially integrated upon nuclease cleavage and that, despite partial AAV genome integrations in the target locus, no gross chromosomal rearrangements or insertions/deletions at off-target sites are found. In line with this, no evidence of hepatocellular carcinoma is observed within the 1-year follow-up. Finally, AAV-HITI is effective at vector doses considered safe if directly translated to humans providing therapeutic efficacy in the adult liver in addition to newborn. Overall, our data support the development of this liver-directed AAV-based knockin strategy.

Keywords: AAV; CAST-Seq; CRISPR-Cas9; HITI; genome editing; hemophilia A; homology-independent targeted integration; inherited diseases; in vivo; liver; mucopolysaccharidosis type VI; persistent transgene expression.

Graphical abstract

Figure 1

Schematic representation of homology-independent targeted integration (HITI) (A) The on-target mouse albumin locus and the gRNA sequence designed within the intron 13 are depicted. EX1-13, mouse albumin exons; EX14, last mouse albumin exon;PAM, protospacer adjacent motif. (B) Schematic representation of the two AAVs. One AAV carries the nuclease SpCas9 under control of the small hybrid liver-specific promoter (HLP). The other AAV contains the HITI donor DNA with the desired promoter-less transgene coding sequence (CDS). SAS, splicing acceptor signal; EX14, mouse albumin exon 14; T2A, Thosea asigna virus 2A skipping peptide; pA, polyadenylation signal (synthetic or bovine growth hormone); U6 gRNA, expression cassette for the gRNA (depicted) or the scRNA sequence; EX1-13, mouse albumin exons 1–13; gRNA and PAM sequences are represented with black (inverted gRNA sequences at the extremities of the donor DNA) and gray (within the mouse albumin locus) boxes and white triangle, respectively. (C) The non-homologous end-joining (NHEJ) repair pathway of the cell leads to the integration of the cleaved donor DNA at the on-target site. (D) Upon transgene integration a single fusion transcript is produced, and this results in the expression of both a modified albumin (Alb-2A) and a therapeutic transgene product containing a proline (P) residue at its N terminus.

Figure 2

AAV-HITI-mediated transgene expression from newborn liver (A) Representative fluorescence microscopy images of OCT liver cryo-sections from wild-type mice injected with AAV-HITI using the dsRed transgene (gRNA, N = 5 or scRNA, N = 5) at the total dose of 1.2 × 10¹⁴ total GCs/kg. HITI efficiency is reported below the images. Scale bar, 50 μm. (B) Serum ARSB activity was analyzed at different time points after AAV-HITI treatment. AFgRNA, MPS VI mice treated with AAV-HITI-gRNA (N = 8); AFscRNA (N = 14 and N = 8 survived up to P360), MPS VI mice treated with AAV-HITI-scRNA. Dotted line corresponds to normal (NR) serum ARSB activity (NR = 11,825 ± 334 pg/mL; Alliegro et al., 2016⁴⁴). ARSB measurements in all AFscRNA animals are equal to zero while all treated mice show ARSB activity levels higher than zero; therefore all the comparisons between the two groups are significant. No statistically significant differences were observed in AFgRNA-treated mice among the different time points. (B–C) Each dot corresponds to a single animal within each group at different time points. (C) FVIII activity levels evaluated in HemA mice by chromogenic assay at different time points after AAV-HITI treatment. Statistical differences were assessed by ordinary one-way ANOVA test. ∗∗p = 0.0025, between normal (NR, N = 5) and AFgRNA (N = 6) at 14 days; ∗p = 0.0190, between NR and AFgRNA (N = 5, one sample was excluded) at 30 days; ∗p = 0.0389, between NR and AFgRNA (N = 6) at 42 days. No statistically significant differences were observed between NR and AFgRNA at the following time points. (D) FVIII antigen levels measured in AFgRNA (N = 4) HemA mice at 90 days of age. ∗p = 0.0262 between NR and AFgRNA at 90 days. All data are represented as mean ± standard deviation. See also Figures S1, S2, and Table S1.

Figure 3

Liver-directed AAV-HITI therapeutic efficacy in newborn mice (A) Urinary levels of glycosaminoglycans (GAGs) reported as a percentage of GAG levels in affected MPS VI mice (% of AF). Statistical differences were assessed by Kruskal-Wallis test and Dunn’s multiple comparisons test. p values: ∗p = 0.0379, ∗∗p = 0.0022, and ∗∗∗∗p < 0.0001. AFgRNA (N = 8), MPS VI mice treated with AAV-HITI-gRNA; AFscRNA (N = 14 and only N = 8 survived up to P360), MPS VI mice treated with AAV-HITI-scRNA. Each dot corresponds to a single animal within each group at different time points. (B) Quantification of GAGs in the liver, kidney, and spleen. Statistical differences were assessed by Brown-Forsythe and Welch ANOVA tests. For the liver: ∗∗p = 0.0016 between NR (N = 4) and AFscRNA (N = 6); ∗∗p = 0.0016 between AFscRNA and AFgRNA (N = 8); p = 0.9996 between NR and AFgRNA. For the kidney: ∗∗∗∗p = 0.0001 between NR (N = 4) and AFscRNA (N = 7); ∗∗∗p = 0.0002 between AFscRNA and AFgRNA (N = 8); p = 0.3784 between NR and AFgRNA. For the spleen the Kruskal-Wallis test was used: ∗∗p = 0.0029 between NR (N = 4) and AFscRNA (N = 8); ∗p = 0.0113 between AFscRNA and AFgRNA (N = 8); p > 0.9999 between NR and AFgRNA. (A and B) Each dot corresponds to a single animal within each group at different time points. The differences in the number of analyzed samples within the same group of treatment at different time points were due to sample availability. (C) Representative histological images of GAG storage in mitral heart valve and myocardium. Scale bar, 100 μm. Alcian blue quantification is reported inside the images as Alcian blue-positive area/total area. NR, N = 4; AFscRNA, N = 8; AFgRNA, N = 8. (D–F) Measurement of skull length/width ratio, femur, and tibia lengths; data are reported as the percentage of normal length (% of NR). Males and females were kept separate in the analysis. AFgRNA, N = 8; AFscRNA, N = 8. Statistical differences were assessed by ordinary one-way ANOVA and Tukey’s multiple comparisons test. (D) p value ∗ = 0.0336 between NR and AFscRNA males; p = 0.4847 between NR and AFgRNA males; ∗p = 0.0191 between NR and AFscRNA females; p = 0.6887 between NR and AFgRNA females. (E) ∗∗∗p = 0.0011 between NR and AFscRNA males; p = 0.4491 between NR and AFgRNA males; ∗∗p = 0.0099 between AFscRNA and AFgRNA males; ∗∗p = 0.0122 between NR and AFscRNA females; ∗∗p = 0.0055 between NR and AFgRNA females. (F) ∗∗p = 0.0019 between NR and AFscRNA males; ∗∗p = 0.0089 between NR and AFgRNA males. ∗p = 0.0238 between NR and AFscRNA females; ∗p = 0.0398 between NR and AFgRNA females. (G and H) Tail-clip assay performed at 63 days of age in HemA AAV-HITI-treated male mice (AFgRNA, N = 6; AFscRNA, N = 6) and unaffected (NR; N = 4) controls. (G) ∗∗p = 0.0019; (H) ∗p = 0.0129. (I) Activated partial thromboplastin time (aPTT) measured at 90 days of age in HemA AAV-HITI (AFgRNA, N = 4 and AFscRNA, N = 5) mice and unaffected (NR, N = 6) controls. p values: ∗∗∗∗p < 0.0001 between NR and AFscRNA; ∗∗∗∗p < 0.0001 between NR and AFgRNA; ∗∗∗∗p < 0.0001 between AFscRNA and AFgRNA. All data are represented as mean ± standard deviation. See also Figure S3.

Figure 4

AAV-HITI molecular characterization at the on-target site (A) Pie chart showing the modified reads observed in AAV-HITI (HITIgRNA or HITIscRNA)-treated mice analyzed by next-generation sequencing (NGS). The expected amplicon size is 285 bp, which was covered by 2x250 bp reads in NGS. The percentage of the modified reads is used as an indication of the gRNA efficiency at the on-target site (HITIgRNA∼30% of indel). (B) Sequences from different portions of both AAV vectors (AAV-HITI donor DNA and AAV-SpCas9) are captured at the induced double-strand breaks, mostly ITR sequences (ITR). (C) Schematic representation of the strategy with the expected PCR amplicons and the position of the primers (up) and representative images of the 5′ junction PCR run on a 2% agarose gel (down). Four different biological samples for each group are showed in the gel. The white arrow indicates proper HITI-mediated donor DNA integration band (∼190 bp); the black arrow indicates the ITR-mediated donor DNA integration band (∼260 bp). See also Table S1.

Figure 5

Long-reads analysis shows that full-length HITI donor DNA is integrated predominantly after ITR cleavage (A) Schematic of the different long PCR junction products. White arrows indicate the primers designed to amplify the 5′ junction: the forward primer (white left) was designed in the endogenous locus right before the cleavage site; the reverse primer (white right) was designed at the end of the donor DNA on the polyadenylation signal (BGHpA). Black arrows indicate the primers designed to amplify the 3′junction: the forward primer (black left) was designed at the beginning of the donor DNA on the splicing acceptor signal (SAS); the reverse primer (black right) was designed in the endogenous locus right after the cleavage site. (B) Representative images of the 5′ and 3′ junction PCR products run on an agarose gel for the detection of the full-length HITI donor DNA (∼2 kb). The 5′ junction is showed in Gel 1 and the 3′ junction in Gel 2. (C) Pie chart showing the percentage (%) of long reads in which donor DNA integration was HITI mediated or ITRs mediated (ITRs) at the junction sites. See also Figures S4, S6, and Tables S2 and S3.

Figure 6

Safety following neonatal delivery of high doses of AAV-HITI (A) The circos plot summarize the CAST-Seq analysis performed on genomic DNA extracted from liver samples of AAV-HITI-treated MPS VI mice. The on-target is reported in green. No chromosomal aberrations were found. (B) Off-target activity (indel frequency) measured at the top predicted off-target loci by next-generation sequencing (NGS) analysis on genomic DNA extracted from liver samples of AAV-HITI-treated MPS VI. The off-target 8 (OFF-8) was excluded due to technical issues. (C) Serum albumin levels measured 1-year post-treatment in sera samples from normal controls (NR, N = 3) or AAV-HITI-treated MPS VI mice (AFgRNA, N = 8; AFscRNA, N = 3); data are represented as mean ± standard deviation. Statistical differences were assessed by ordinary one-way ANOVA test. (D) Representative images from histopathological analysis performed on paraffin sections from different liver lobes from AAV-HITI-treated MPS VI mice (AFgRNA, N = 8; AFscRNA, N = 8) and untreated normal controls (NR, N = 4) 1-year post-treatment. Scale bar, 50 μm. See also Figures S7, S8 and Tables S5–S7.

Figure 7

Liver-directed AAV-HITI in adult mice (A) Serum ARSB activity analyzed at different time points in MPS VI mice left untreated (AF, N = 14) or treated with AAV-HITI-gRNA (AFgRNA, N = 3) at the total dose of 1.2 × 10¹³ total GCs/kg. (B) Urinary GAG levels at different time points in MPS VI mice left untreated (AF, N = 14) or treated with AAV-HITI-gRNA (AFgRNA, N = 3) reported as a percentage of GAG levels in untreated MPS VI mice. NR, sera samples from normal animals. Statistical differences were assessed at P90 by Brown-Forsythe and Welch ANOVA tests: ∗∗∗∗p < 0.0001 between NR and AF; ∗∗∗∗p < 0.0001 between AF and AFgRNA; by Kruskal-Wallis test at P180: ∗∗∗∗p < 0.0001 between NR and AF; ∗∗∗∗p < 0.0001 between AF and AFgRNA; by ordinary one-way ANOVA test at P270: ∗∗∗∗p < 0.0001 between NR and AF; ∗∗p < 0.0007 between AF and AFgRNA; at P360: ∗∗p = 0.0012 between NR and AF; p = 0.9999 between NR and AFgRNA; ∗∗p = 0.0078 between AF and AFgRNA. The differences in the number of analyzed samples within the same group of treatment were due to sample availability. Each dot corresponds to a single animal within each group at different time points. (A and B) The differences in the number of analyzed samples within the same group of treatment at different time points were due to sample availability. (C) FVIII antigen levels detected in AAV-HITI (AFscRNA, N = 4; AFgRNA, N = 4)-treated adult HemA mice and normal controls (NR, N = 3) 1-month post AAV-HITI treatment. ∗∗∗p = 0.0001 between NR and AFgRNA. (D) FVIII activity levels evaluated in HemA mice (AFgRNA, N = 4; AFscRNA, N = 4) and normal controls (NR, N = 3) by chromogenic assay 1-month post AAV-HITI treatment. Statistical differences were assessed by ordinary one-way ANOVA test. p value between NR and AFgRNA = 0.0783. (E) Activated partial thromboplastin time (aPTT) measured 1 month post AAV-HITI treatment in HemA AAV-HITI (AFgRNA, N = 4 and AFscRNA, N = 4) mice and normal controls (NR; N = 4). ∗∗∗∗p < 0.0001 between NR and AFscRNA; ∗∗p = 0.0028 between NR and AFgRNA; ∗∗p = 0.0014 between AFscRNA and AFgRNA. All data are represented as mean ± standard deviation. See also Figure S11.