In vivo gene therapy: AAV and lentiviral vectors

In vivo gene therapy is the delivery of a nucleic acid into the nucleus of a patient’s cell using a virus stripped of its own genes. What the approach can and cannot do follows from the biology of the chosen virus rather than from the therapeutic idea.

AAV: an episome that lasts until the cell divides

Adeno-associated virus is a small non-enveloped parvovirus with a single-stranded DNA genome of about 4.7 kb. In a recombinant vector nothing of the virus remains except the inverted terminal repeats; everything between them is transgene and promoter.

Once in the nucleus, the vector genome does not integrate to any meaningful degree. It resolves into double-stranded circular concatemers and persists as an episome. The consequence is decisive: expression lasts for years in post-mitotic tissue — retina, central nervous system, skeletal muscle — but is diluted at every cell division. An infant liver that is still growing loses the transgene; an adult liver holds it far longer. That, rather than protein stability, is what the durability question is really about.

Tropism is set by the capsid: serotypes engage different cell-surface glycans and then the shared entry receptor AAVR (KIAA0319L). Capsid engineering — directed evolution, peptide insertion, rational design — is an attempt to shift tropism far enough that the required dose comes down.

Three constraints that a higher dose does not solve

Capacity. About 4.7 kb fits, regulatory elements included. The dystrophin coding sequence is over 11 kb, which is why the field works with micro-dystrophins that retain the actin-binding domain and a subset of the spectrin repeats. Dual-vector strategies relying on intracellular recombination exist, but they pay for the extra room in efficiency.

Pre-existing immunity. AAV is a common human virus, and a substantial fraction of adults carry neutralising antibodies against one serotype or another; those patients are excluded. The administered dose itself raises a strong humoral response, so redosing with the same serotype is effectively impossible. “One and done” is a constraint dressed as a feature.

Dose as toxicology. Systemic administration for neuromuscular indications requires on the order of 10^14 vector genomes per kilogram — an enormous load of capsid protein and DNA. The observed consequences are complement activation, thrombotic microangiopathy, acute liver injury, and a CD8 response against capsid peptides that clears transduced hepatocytes some weeks after infusion. Prophylactic immunosuppression blunts this without removing it.

Lentivirus: a different bargain

An HIV-1-derived lentiviral vector carries an RNA genome; reverse transcription produces DNA and integrase inserts it into a chromosome. Capacity is larger — roughly 8 kb — and the transgene is inherited by daughter cells, which is what dividing targets, above all haematopoietic stem cells, require.

The price is insertional mutagenesis. First-generation gammaretroviral vectors in X-linked SCID trials activated LMO2 through the enhancer in their own LTR and caused leukaemias; self-inactivating designs, with the LTR enhancer deleted and an internal promoter driving the transgene, became standard afterwards. Lentiviruses also favour the bodies of actively transcribed genes over promoter regions. The risk is reduced rather than abolished, and clonal tracking of integration sites remains part of long-term follow-up.