A myocardium tropic adeno-associated virus (AAV) evolved by DNA shuffling and in vivo selection

Abstract

To engineer gene vectors that target striated muscles after systemic delivery, we constructed a random library of adeno-associated virus (AAV) by shuffling the capsid genes of AAV serotypes 1 to 9, and screened for muscle-targeting capsids by direct in vivo panning after tail vein injection in mice. After 2 rounds of in vivo selection, a capsid gene named M41 was retrieved mainly based on its high frequency in the muscle and low frequency in the liver. Structural analyses revealed that the AAVM41 capsid is a recombinant of AAV1, 6, 7, and 8 with a mosaic capsid surface and a conserved capsid interior. AAVM41 was then subjected to a side-by-side comparison to AAV9, the most robust AAV for systemic heart and muscle gene delivery; to AAV6, a parental AAV with strong muscle tropism. After i.v. delivery of reporter genes, AAVM41 was found more efficient than AAV6 in the heart and muscle, and was similar to AAV9 in the heart but weaker in the muscle. In fact, the myocardium showed the highest gene expression among all tissues tested in mice and hamsters after systemic AAVM41 delivery. However, gene transfer in non-muscle tissues, mainly the liver, was dramatically reduced. AAVM41 was further tested in a genetic cardiomyopathy hamster model and achieved efficient long-term delta-sarcoglycan gene expression and rescue of cardiac functions. Thus, direct in vivo panning of capsid libraries is a simple tool for the de-targeting and retargeting of viral vector tissue tropisms facilitated by acquisition of desirable sequences and properties.

Fig. 1.

Sequence and structure analysis of the M41 capsid. (A) The primary structure of M41 capsid by alignments of its VP1 amino acid sequence with those of the parental AAV serotypes. A D to Y substitution on residue 418 is shown by a triangle. (B and C) The structural model of the 9-mer AAVM41 VP3 subunits reconstructed from the known crystal and homologous model structures of the parental viruses with the exterior surface (B), and interior surface (C), respectively. Sequences derived from AAV1 are colored in purple, AAV6 in hot pink and light pink (2 segments), AAV7 in cyan, AAV8 in green. The axes of symmetry are shown by white pentagon (5-fold), triangle (3-fold) and oval (2-fold) on the structural models. The structural regions around them are highlighted by frames a, b, and c.

Fig. 2.

Luciferase activities and vector genome copy numbers in various mouse tissues after systemic administration of AAV vectors. Comparison of AAV9- and AAVM41-CMV-Luc vectors in luciferase activities (A), and AAV vector genome copy numbers (B), at 2-weeks after i.v. injection of 3 × 10¹¹ vector genomes in mice. Similar comparison of AAV6- or M41-CMV-Luc vectors in luciferase activities (C), and AAV genome copy numbers (D). Data are mean values ± SD. Ab, abdomen muscle; Di, diaphragm; Ta, tibialis anterior; Ga, gastrocnemius; Qd, quadriceps; Fl, forelimb; Ma, masseter; Ht, heart; Lv, liver; Sp, spleen; K_d, kidney; Lu, lung; Pa, pancreas; Te, testis; Br, brain; Tg, tongue. Heart vs. liver ratio in transduction efficiency by 3 rAAV vectors on luciferase activities (E), and vector genome copy numbers (F).

Fig. 3.

Systemic delivery of LacZ transgene by AAVM41 into striated muscles. (A) X-gal staining of cross-sections of hearts after systemic administration of AAV vectors. 3 × 10¹¹ vector genomes of AAV9- or M41-CB-LacZ were injected via tail vein into adult mice; and 1 × 10¹² vector genomes of AAV9- or M41-CMV-LacZ were injected via jugular vein into adult hamsters. Hearts from mice or hamsters were collected at 2-weeks or 3-weeks post injection for X-gal and eosin staining. Two magnifications (4× and 40×) were used for photography. (B) Lacz transgene expression in the tibialis anterior muscles of the same animals as described in (A). [Scale bars: A, 200 μm (4×) and 50 μm (40×); B, 100 μm.]

Fig. 4.

Comparison of gene transfer efficiency in primary cardiomyocytes or skeletal muscles. (A) Representative X-gal staining of cardiomyocytes or skeletal muscle after transduction by AAV9-, AAVM41-, or AAV6-CMV-LacZ vectors. The AAV vectors were inoculated on the primary neonatal rat cardiomyocytes (5 × 10⁵ cells per well) at an infection multiplicity of 3,000. Cells were fixed for X-gal staining 96 h later (Upper). The AAV vectors (5 × 10⁹ v.g.) were also injected into the gastrocnemius muscle of adult C57BJ/6L mice and tissues were sectioned and stained with X-gal and eosin 14 days post-treatment (Lower). (Scale bars, 100 μm.) (B and C) Quantitative β-gal activities and AAV vector genome copies in primary cardiomyocytes; and (D) β-gal activities in mouse skeletal muscles. Data are shown as mean values ± SD.

Fig. 5.

Systemic delivery of δ-sarcoglycan into cardiomyopathic hamster for treatment of heart failure. Totals of 1 × 10¹² vector genomes of M41-Syn-δSG vector were injected into 7-week-old male TO-2 hamsters via the jugular vein (n = 5). (A) Immunofluorescent staining of δ-sarcoglycan on thin sections of heart and skeletal muscle tissues 4 months after vector administration. Two magnifications were used for clear view. (Scale bars, 50 μm.) (B) Western analysis of δ-sarcoglycan in muscle and non-muscle tissues from untreated TO-2 (-) and rM41 vector-treated TO-2 (+) hamsters. Twenty micrograms of total proteins were loaded in each lane. (C) Amelioration of cardiomyopathy in TO-2 hamsters after δ-sarcoglycan gene transfer. Hamster hearts of wild-type F1B, untreated TO-2 and AAVM41-δSG-treated TO-2 were cryosectioned and stained with different methods for histology and pathology. Arrows and arrowheads indicate fibrosis and calcification, respectively. (Scale bars, 100 μm.)