Promoterless gene targeting without nucleases ameliorates haemophilia B in mice

Abstract

Site-specific gene addition can allow stable transgene expression for gene therapy. When possible, this is preferred over the use of promiscuously integrating vectors, which are sometimes associated with clonal expansion and oncogenesis. Site-specific endonucleases that can induce high rates of targeted genome editing are finding increasing applications in biological discovery and gene therapy. However, two safety concerns persist: endonuclease-associated adverse effects, both on-target and off-target; and oncogene activation caused by promoter integration, even without nucleases. Here we perform recombinant adeno-associated virus (rAAV)-mediated promoterless gene targeting without nucleases and demonstrate amelioration of the bleeding diathesis in haemophilia B mice. In particular, we target a promoterless human coagulation factor IX (F9) gene to the liver-expressed mouse albumin (Alb) locus. F9 is targeted, along with a preceding 2A-peptide coding sequence, to be integrated just upstream to the Alb stop codon. While F9 is fused to Alb at the DNA and RNA levels, two separate proteins are synthesized by way of ribosomal skipping. Thus, F9 expression is linked to robust hepatic albumin expression without disrupting it. We injected an AAV8-F9 vector into neonatal and adult mice and achieved on-target integration into ∼0.5% of the albumin alleles in hepatocytes. We established that F9 was produced only from on-target integration, and ribosomal skipping was highly efficient. Stable F9 plasma levels at 7-20% of normal were obtained, and treated F9-deficient mice had normal coagulation times. In conclusion, transgene integration as a 2A-fusion to a highly expressed endogenous gene may obviate the requirement for nucleases and/or vector-borne promoters. This method may allow for safe and efficacious gene targeting in both infants and adults by greatly diminishing off-target effects while still providing therapeutic levels of expression from integration.

Extended Data Figure 1. hF9 liver immunohistochemistry

From top to bottom, panels show human factor IX staining (red) with DAPI nuclear counterstain (blue) in positive control human liver, negative control untreated mouse liver, and two sets of representative stains from mice treated as neonates or adults with AAV8-P2A-hF9.

Extended Data Figure 2. Scheme of targeting rate assessment

Assessment of on-target integration rate begins using linear amplification (LAM) with biotinylated primer 1 (black), annealing to the genomic locus but not to the vector (step 1). Linear amplicons are then bound to streptavidinylated beads and washed to exclude episomal vectors (Step 2). Subsequent second-strand DNA synthesis with random primers (Step 3) was followed by CviQI restriction digestion (Step 4). A compatible linker is then ligated (Step 5) followed by two rounds of nested PCR amplifications (primers 2–3 in blue- Step 6, and then primers 4–5 in red- Step 7). CviQI cleaves at the same distance from the homology border in both targeted and wild-type alleles, thus allowing for unbiased amplification. The amplicons of the 2^nd nested PCR then serve as a template for qPCR assays with either primers 6–7 (green) or 8–9 (orange) (Step 8).

Extended Data Figure 3. Standard curves for targeting rate assessment by qPCR

qPCR standard curves for the targeted allele (primers 8 and 9, Figure 3) and non-targeted allele (primers 6 and 7, Figure 3). Mass units used are functionally equivalent to molarity because all amplicons used were of equal length.

Extended Data Figure 4. Toxicity assessment by ALT measurement

Alanine transaminase levels (ALT) were evaluated 7 days post-injection in mice injected with AAV8 coding for our experimental vector (1e12) or a negative control coding for a known non-toxic cassette (1e12 of H1 promoter-driven shRNA), or a positive control coding for a known toxic cassette (5e11 of U6 promoter-driven shRNA). Data represent mean of two measurements of four independent mice for each groups. The statistical significance is defined here as having p<0.05 in a one-tailed t test between samples of different variance.

Extended Data Figure 5. Vector copy number

Vector copy number assessed by qPCR using primers 8 and 9 (Figure 3). N = 7 for mice injected as adults and N = 6 for mice injected as neonates and analyzed before or after partial hepatectomy. Error bars represent standard deviation.

Figure 1. Vector design and experimental scheme

a. The rAAV8 vector encodes a codon-optimized hF9 cDNA and preceding 2A-peptide coding sequence flanked by homology arms spanning the Alb stop codon. Length of the 5′ and 3′ arms are 1.3 and 1.4 Kb, respectively. Following integration by homologous recombination, Alb and hF9 are fused at the DNA and RNA levels, but two separate proteins are produced as the result of ribosomal skipping. b. With respect to the Alb homology arms, the AAV inverse control has hF9 inverted along with the 2A-peptide coding sequence, the adjacent Alb exon and the preceding splice junction. Thin white lines: Alb introns; dark gray boxes: Alb exons; white boxes: P2A; white arrows: hF9 transgene; light gray boxes: extragenic DNA; P: proline.

Figure 2. Human factor IX expression and activity in injected mice

a. Plasma hF9 measured by ELISA following IP injections of 2-day-old B6 mice with 2.5e11 vg per mouse of either the hF9 experimental construct (n = 6) or inverse control (n = 3). The limit of detection was 20 ng/mL. PH = partial hepatectomy. Error bars represent standard deviation. Dashed lines denote 5% and 20% of normal F9 levels. b. Plasma hF9 measured by ELISA following tail vain injections of 9-week-old B6 mice with 1e12 vg per mouse of either the AAV hF9 experimental construct (n = 7), or inverse control (n = 3), or a hydrodynamic injection of 30 μg plasmid (3.5e12 copy number) coding for the hF9 construct in the “correct” orientation. The limit of detection was 20 ng/mL. Error bars and dashed lines as in (a). c. Plasma hF9 measured by ELISA following tail vain injections of 9-week-old B6 mice with the designated vector dose of AAV-hF9 experimental construct (n = 4 for each dose group). Error bars represent standard deviation. d. Measurement of coagulation efficiency by activated partial thromboplastin time (aPTT) 2 weeks after tail vain injections of AAV8-hF9 at 1e12 vg per mouse (n = 5). Error bars represent standard deviation. e. Western blot analysis for hF9 in liver samples from mice injected with the AAV8-hF9 construct or inverse control. The expected size of hF9 is 55-Kd.

Figure 3. Rate of Alb targeting at the DNA and RNA levels

a. Assessment of on-target integration rate begins using linear amplification (LAM) with biotinylated primer 1 (black), annealing to the genomic locus but not to the vector. Linear amplicons are then bound to streptavidinylated beads and washed to exclude episomal vectors. Subsequent second-strand DNA synthesis with random primers was followed by CviQI restriction digestion. A compatible linker is then ligated, followed by two rounds of nested PCR (primers 2–3 in blue, and then primers 4–5 in red). CviQI cleaves at the same distance from the homology border in both targeted and wild-type alleles, thus allowing for unbiased amplification. The amplicons of the 2^nd nested PCR then serve as a template for qPCR assays with either primers 6–7 (green) or 8–9 (orange). b. For mRNA quantification, primers 10–11 or 11–12 were used to generate a cDNA for qPCR assays. Shape and fill code as in Fig 1. c. Black bars represent the targeting rate of Alb alleles as the ratio between the abundance of the DNA template amplified by primers 6–7 to the abundance of the DNA template amplified by primers 8–9, corrected by a factor of 0.7 to account for hepatocyte frequency. Gray bars represent the expression rate of targeted Alb alleles as the ratio between the abundance of the cDNA template amplified by primers 10–11 to the abundance of the cDNA template amplified by primers 11–12. N = 3 for each group, error bars represent standard deviation.

Figure 4. Specificity of hF9 expression

a. cDNA, produced from RT with a poly-dT primer, served as a template for either a qPCR assay with primers 13–14 or 14–15. b. Bars represent the rate of Alb_hF9 mRNAs to total hF9-containing mRNAs as the ratio between the abundance of the cDNA template amplified by primers 13–14 to the abundance of the cDNA template amplified by primers 14–15. N = 3 for each group, error bars represent standard deviation. c. Northern blot analysis of liver samples with a probe against P2A. The lower non-specific signal corresponds in size to 18S rRNA. d. Western blot analysis of P2A from liver samples of mice injected with the AAV-P2A-hF9 construct or inverse control. P2A is expected to be fused to Albumin (66.5-Kd).