Serotype-dependent packaging of large genes in adeno-associated viral vectors results in effective gene delivery in mice

Abstract

Vectors derived from adeno-associated virus (AAV) are promising for human gene therapy, including treatment for retinal blindness. One major limitation of AAVs as vectors is that AAV cargo capacity has been considered to be restricted to 4.7 kb. Here we demonstrate that vectors with an AAV5 capsid (i.e., rAAV2/5) incorporated up to 8.9 kb of genome more efficiently than 6 other serotypes tested, independent of the efficiency of the rAAV2/5 production process. Efficient packaging of the large murine Abca4 and human MYO7A and CEP290 genes, which are mutated in common blinding diseases, was obtained, suggesting that this packaging efficiency is independent of the specific sequence packaged. Expression of proteins of the appropriate size and function was observed following transduction with rAAV2/5 carrying large genes. Intraocular administration of rAAV2/5 encoding ABCA4 resulted in protein localization to rod outer segments and significant and stable morphological and functional improvement of the retina in Abca4(-/-) mice. This use of rAAV2/5 may be a promising therapeutic strategy for recessive Stargardt disease, the most common form of inherited macular degeneration. The possibility of packaging large genes in AAV greatly expands the therapeutic potential of this vector system.

Figure 2. Genome integrity of rAAV2/5-CMV-Abca4, -MYO7A, -CEP290, and -EGFP-8.9.

(A) Southern blot analysis of vector DNA isolated directly from rAAV large preparations (2.5 × 10¹⁰ GC/lane) and separated on alkaline agarose gels. Lane 1 contains a marker DNA fragment obtained by restriction digestion from the pAAV2.1-CMV-Abca4 plasmid; lane 2 contains the same DNA fragment as in lane 1 digested with DNaseI, as control of DNaseI activity; lanes 3 and 4: genomes isolated from rAAV2/5-CMV-Abca4; lanes 5 and 6: genomes isolated from rAAV2/5-CMV-MYO7A; lanes 7 and 8: genomes isolated from rAAV2/5-CMV-CEP290; lanes 9 and 10: genomes isolated from rAAV2/5-CMV-EGFP-8.9 (lanes 1–10 belong to the same gel but were noncontiguous); lanes 11 and 12: genomes isolated from rAAV2/5-CMV-EGFP (2.6 kb). Samples in lanes 3, 5, 7, 9, and 11 were treated with DNaseI. (B) Assessment of rAAV2/5-CMV-Abca4 genome length following in vivo delivery. Top panel: Schematic representation of the rAAV2/5-CMV-Abca4 genome with the 2 probes used for the Southern blot analysis. Bottom panel, upper blots: Southern blot analysis of genomic DNA from uninjected muscles (lanes 1 and 3) and an equivalent amount of genomic DNA from murine muscle injected with rAAV2/5-CMV-Abca4 (lanes 2 and 4) digested with NcoI and NotI (lanes 1 and 2) or NcoI alone (lanes 3 and 4). Lanes belong to the same gel but were noncontiguous. The arrows indicate the bands of the expected size. Lower blots: Southern blot analysis with a probe specific for the PDE6B gene used as loading control. Molecular weights are indicated on the left. (C) Western blot analysis with anti-ABCA4 (upper blot) or anti–α-tubulin (lower blot) antibodies of lysates from Cos cells transduced with rAAV2/5. Lane 1: retina from wild-type mouse; lane 2: samples transduced with rAAV2/5-CMV-Abca4; lane 3: samples transduced with rAAV2/5-CMV-EGFP. Anti–α-tubulin was used as loading control. The amounts (μg) of protein loaded are indicated under the respective lanes.

Figure 1. Packaging capacity of rAAV serotypes.

(A) Average titers of rAAV2/1, 2, 3, 4, 5, 7, 8, and 9 containing the Abca4 cDNA. The rAAV genome is composed of AAV2 ITRs, CMV promoter, and Abca4 cDNA sequence (rAAV genome size 8.9 kb). Data are shown as average ± SE. *P ≤ 0.05, compared with rAAV2/5; n, number of rAAV large preparations. The numbers above the SE bars represent the average titers. (B) Large genome to small genome titer ratio for each of the rAAV serotypes tested (average ± SE). The large genome was represented by Abca4 (rAAV genome size 8.9 kb). The small genomes (containing various transgenes) were 4.7 kb or less. n, number of rAAV preparations. (C) Average titers ± SE of rAAV2/5-CMV-Abca4 vectors with genome size of 8.9, 10, and 11.4 kb. The 10- and 11.4-kb genomes were obtained by adding 1.1 and 2.5 kb of stuffer sequence, respectively, to the rAAV2/5-CMV-Abca4 genome. The numbers above the SE bars represent the average titers. (D) Average titers (±SE) of rAAV2/5- (gray bars) and rAAV2/2-CMV-EGFP (white bars) vectors with genome sizes of 4.1, 5.1, 6.1, 7.1, and 8.9 kb. The different-size genomes were obtained by adding stuffer sequences of different lengths after the poly(A) sequence in the pAAV2.1-CMV-EGFP plasmid. The results shown are the average of titers obtained by RT-PCR from 3 small-scale vector preparations.

Figure 3. EGFP, ABCA4, and MYO7A expression following rAAV2/5 delivery.

(A) Transduction efficiency (assessed as green forming units [GFU]) of rAAV2/5-CMV-EGFP with genome size of 4.1 (black bars), 5.1 (gray bars), and 8.9 kb (white bars) at various MOI (GC/cells) on Cos cells 72 hours after the infection. Results are shown as the average of 3 independent experiments ± SE. (B) Western blot analysis with anti-ABCA4 (top panel) and anti–α-tubulin (middle panel) antibodies and 8-azido-[α-³²P]-ATP labeling of ABCA4 (bottom panel) of lysates from Abca4^–/– retinas transduced with rAAV2/5. Lane 1: retina from wild-type mouse; lane 2: samples transduced with rAAV2/5-CMV-Abca4; lane 3: samples transduced with rAAV2/5-CMV-EGFP. Anti–α-tubulin was used as loading control. The amounts of protein (μg) loaded are indicated under the respective lanes. (C) Western blot analysis with anti-MYO7A (top panel) and anti–retinal G protein–coupled receptor (anti-RGR; bottom panel) antibodies of lysates from primary RPE cell cultures. The cells were from Myo7a-null mice and were transduced with rAAV2/5-CMV-MYO7A. Lane 1 and 2: 7 days after infection with rAAV2/5-CMV-MYO7A; lanes 3 and 4: 17 days after infection with rAAV2/5-CMV-MYO7A; lane 5: 17 days after infection with rAAV2/5-CMV-EGFP; lane 6: uninfected RPE lysates from wild-type C57BL/6 mice 7 days after plating. Anti-RGR antibody was used as loading control. Molecular weights are indicated on the left. The amounts of protein (μg) loaded are indicated below each lane.

Figure 4. Morphological analysis of Abca4^–/– retinas following rAAV-mediated gene transfer.

(A) Immunohistochemical analysis with anti-ABCA4 (Rim 3F4) antibody of retinal sections from 4-month-old Abca4^+/+ mice and Abca4^–/– pigmented mice injected subretinally at 1 month of age with rAAV2/5-CMV-EGFP and the contralateral eye with rAAV2/5-CMV-Abca4. OS, outer segment (photoreceptors); ONL, outer nuclear layer; INL, inner nuclear layer; GCL, ganglion cell layer. Original magnification, ×20. (B) Electron microscopic analysis of RPE from pigmented 5-month-old Abca4^–/– mice. RPE from 1 eye injected subretinally at 1 month of age with rAAV2/5-CMV-EGFP (left) and the contralateral eye with rAAV2/5-CMV-Abca4 (right). Ch, choroid; BrM, Bruch’s membrane. White arrows indicate the irregularly shaped lipofuscin pigment granules to be distinguished from the larger oval melanosomes. Micrographs were obtained at the same magnification (×6,000). (C) Number of lipofuscin granules (left) and RPE thickness (right) in the RPE of Abca4^+/+ or Abca4^–/– mice injected subretinally with rAAV2/5-CMV-EGFP or rAAV2/5-CMV-Abca4 (n = 2 eyes/group).

Figure 5. Reduction of lipofuscin levels and improved recovery from photoreceptor desensitization in Abca4^–/– mice injected with rAAV2/5-CMV-Abca4.

(A) Effect of rAAV2/5-mediated Abca4 gene transfer on lipofuscin accumulation in the retina of Abca4^–/– mice. A2E (combined A2E and iso-A2E), atRALdi-E, and atRALdi-PE levels in eyecups of 4- and 6-month-old albino and pigmented Abca4^–/– mice, respectively, injected at postnatal day 30 in 1 eye with rAAV2/5-CMV-Abca4 (black bars) and in the contralateral eye with rAAV2/5-CMV-EGFP (white bars). Age-matched albino BALB/c and pigmented Abca4^+/+ mice are represented in gray bars. Values are the average of 2 independent samples containing 4 eye cups each. (B) Rescue from delayed recovery from photoreceptor desensitization in Abca4^–/– mice treated with rAAV2/5-CMV-Abca4. Progressive recovery after bleaching of the b-wave amplitude in 4-month-old Abca4^–/– mice injected subretinally with either rAAV2/5-CMV-Abca4 (red triangles; n = 4 eyes) or rAAV2/5-CMV-EGFP (green squares; n = 4 eyes) and in age-matched wild-type BALB/c mice (black circles; n = 10 eyes). Data are shown as average ± SE. *P ≤ 0.05.