Semirational bioengineering of AAV vectors with increased potency and specificity for systemic gene therapy of muscle disorders

Parts used

Show 11 moreparts→

Key findings

• AAVMYO2 (also called AAVS1P1) and AAVMYO3 (also called AAVS10P1) are chimeric AAV capsids assembled in two engineering steps. DNA family shuffling of the cap genes of AAV1, AAV6, AAV8 and AAV9, followed by iterative in vivo cycling of the libraries in quadriceps femoris, diaphragm and heart of intravenously injected C57BL/6J mice, yields the shuffled capsids AAVS1 and AAVS10, into which the P1 peptide of AAVMYO (the 7-mer RGDLGLS) is then transferred by swapping a BsiWI/SpeI cap fragment from a P1-modified AAV9 helper plasmid.
• After three rounds of in vivo cycling of a shuffled library built from AAV1, AAV6, AAV8 and AAV9 cap genes, or two rounds of a library additionally containing AAVpo1, in the musculature of C57BL/6J mice, the C-terminal portion of nearly all enriched chimeric capsid protein sequences derives from AAV9, while no parental serotype is enriched or depleted in the N-terminal half or at the extreme C terminus. AAVpo1 is largely underrepresented among the enriched chimeras, consistent with its low sequence homology to the other four serotypes preventing AAVpo1-containing chimeras from assembling functional particles.
• The shuffled capsids AAVS1 and AAVS10 give an AAV mRNA biodistribution resembling AAVMYO in intravenously injected C57BL/6J mice and are less active than AAV9, AAVMYO and most other chimeras in the off-targets liver, brain, spleen, kidney and lung, but their efficiency is inferior to AAVMYO not only in the off-targets but also in all muscle types. Two further shuffled capsids, AAVH15 and AAVD20, share this AAVMYO-like specificity profile.
• Transferring the P1 peptide into the AAV9-derived C-terminal region of the shuffled capsids AAVS1 and AAVS10 raises their muscle efficiency to within a twofold range of AAVMYO barcode counts in quadriceps, triceps, heart, diaphragm and tongue, in both C57BL/6J and NMRI mice, while AAVMYO2- and AAVMYO3-specific barcodes remain nearly absent in the off-target liver and brain. Combining a shuffled capsid backbone preselected in the musculature with the P1 peptide therefore boosts on-target specificity without enhancing off-target activity.
• Replacing P1 with the peptide P4 (NDVRSAN) at the same capsid loop VIII site of the AAVS10 chimera yields AAVS10P4, which in C57BL/6J mice robustly targets the liver and is much less active in muscle than AAVMYO3 or AAVMYO. The myotropism of AAVMYO3 therefore depends on the P1 sequence itself rather than on the mere presence of an insertion in capsid loop VIII.
• The P1 peptide insertion occupies residues G585 to L594 within variable region VIII of the capsid protein in AAVMYO, AAVMYO2 and AAVMYO3, the most radially protruding motif of the AAV capsid surface, a site whose modification is reported to disrupt the natural laminin receptor binding domain of AAV9 while exposing the foreign peptide to alternative receptors. Despite shuffling and peptide insertion, AAVS1, AAVS10, AAVMYO, AAVMYO2 and AAVMYO3 monomers retain the conserved eight-stranded antiparallel beta barrel that forms the core of the AAV capsid.
• Display of the P1 peptide confers muscle specificity comparable to AAVMYO, AAVMYO2 and AAVMYO3 on capsids based on AAV2, AAV3, AAV5 or AAVpo1 in both C57BL/6J and NMRI mice, albeit at lower efficiency, and it shifts AAV8 tropism toward the musculature and away from the liver, although AAV8P1 remains more liver-efficient than the other P1 variants. AAVrh10P1 performs well in all seven analyzed tissues and surpasses AAV9, whose wide activity across species is established.
• AAVrh74 is inconspicuous in all mouse muscle types but is the top hit in the liver on the AAV mRNA level in a barcoded 17-capsid library delivered intravenously, while ranking higher in skeletal muscle on the vector DNA level than on the mRNA level. Only a limited portion of AAVrh74-delivered vector genomes is therefore transcriptionally active in murine muscle.
• Pairing the synthetic SPc5-12 promoter with the AAVMYO3 capsid produces a two- to three-log increase in Firefly luciferase expression across all skeletal muscle groups of CB17/IcrTac/Prkdcscid mice injected intravenously with 1 x 10^11 vg, relative to an AAV9 vector carrying the same reporter under the human desmin (hDES) promoter. SPc5-12 yields significantly higher luciferase expression than hDES across all skeletal muscle groups, heart and diaphragm, and the low-level leaky liver expression seen from either promoter in AAV9 becomes undetectable when the capsid is switched to AAVMYO2 or AAVMYO3.
• A single intravenous injection of an AAVMYO2 or AAVMYO3 vector expressing human MTM1 from the muscle-specific DES promoter rescues survival of all treated Mtm1-KO mice at each of 1.0 x 10^14, 2.0 x 10^13 and 4.0 x 10^12 vg/kg, whereas the equivalent AAV8-MTM1 vector rescues all mice only at 1.0 x 10^14 vg/kg, 75% at 2.0 x 10^13 vg/kg and none at 4.0 x 10^12 vg/kg. Body weight of Mtm1-KO mice treated with AAVMYO2 or AAVMYO3 at 1.0 x 10^14 vg/kg reaches the level of age-matched wild-type littermates.
• Three months after intravenous delivery of 2.0 x 10^13 vg/kg, AAVMYO2-MTM1 and AAVMYO3-MTM1 restore the global escape-test force of Mtm1-KO mice and the specific tetanic force of their tibialis anterior to wild-type values, normalize muscle fiber size and mitochondria positioning, and strongly reduce the proportion of myofibers with internal nuclei. At the higher dose of 1.0 x 10^14 vg/kg both vectors instead produce abnormally small myofibers in tibialis anterior, consistent with an excessive level of myotubularin expression.
• In Mtm1-KO mice three months after intravenous injection, AAVMYO2-MTM1 and AAVMYO3-MTM1 give higher vector copy numbers in skeletal muscles than AAV8-MTM1 while heart transduction is similar, and they detarget main off-target organs including liver and kidney. At 2.0 x 10^13 vg/kg their MTM1 mRNA levels are 10 to 100 times higher than with AAV8, and myotubularin protein in tibialis anterior, biceps brachii and diaphragm reaches or exceeds the level obtained with a fivefold higher AAV8 dose.
• An AAV9 vector expressing MTM1 at 4.0 x 10^12 vg/kg is less efficacious than AAVMYO2-MTM1 or AAVMYO3-MTM1 at prolonging life span and increasing body weight, muscle mass and strength in Mtm1-KO mice over a 3-month period.
• In dystrophin-deficient mdx mice, AAVMYO3 drives significantly higher micro-dystrophin protein expression in quadriceps femoris than AAV9 at 2 x 10^11 and 1 x 10^12 vg per mouse under the muscle creatine kinase (MCKE) promoter and at 2 x 10^12 vg per mouse under the CMV promoter, with more robust expression also in heart and diaphragm. Four-limb hanging time and four-limb grip strength measured 23 weeks after delivery of the CMV-driven construct trend better with AAVMYO3 than with AAV9 without reaching statistical significance.
• Yields of AAVMYO2 and AAVMYO3 in 250-ml HEK293T suspension cultures are comparable to or higher than those of AAV8 or AAV9 carrying the same vector construct, and in 10-liter bioreactors the two chimeras give two- to fourfold higher titers than AAV8. Both bind 80% to almost 100% of input particles on the Poros AAV9 and Poros AAVX affinity resins, 95 to 100% of particles are recovered in the eluate after 10-liter-scale Poros AAVX purification, and full capsid content after affinity chromatography without any enrichment step is 31% for AAVMYO2 and 48% for AAVMYO3 against 32% for AAV8.
• AAVMYO2 and AAVMYO3 show a prevalence of pre-existing antibodies in human sera comparable to AAV9, measured both by ELISA and by in vitro neutralization assay, so the serologic profile of these P1-displaying chimeric capsids does not differ from that of the AAV9 serotype that contributes their C-terminal portion.

Abstract

Bioengineering of viral vectors for therapeutic gene delivery is a pivotal strategy to reduce doses, facilitate manufacturing, and improve efficacy and patient safety. Here, we engineered myotropic adeno-associated viral (AAV) vectors via a semirational, combinatorial approach that merges AAV capsid and peptide library screens. We first identified shuffled AAVs with increased specificity in the murine skeletal muscle, diaphragm, and heart, concurrent with liver detargeting. Next, we boosted muscle specificity by displaying a myotropic peptide on the capsid surface. In a mouse model of X-linked myotubular myopathy, the best vectors—AAVMYO2 and AAVMYO3—prolonged survival, corrected growth, restored strength, and ameliorated muscle fiber size and centronucleation. In a mouse model of Duchenne muscular dystrophy, our lead capsid induced robust microdystrophin expression and improved muscle function. Our pipeline is compatible with complementary AAV genome bioengineering strategies, as demonstrated here with two promoters, and could benefit many clinical applications beyond muscle gene therapy.