modality · intermediate · partly written
Gene therapy
A medicine whose active ingredient is a gene, usually carried into cells by an emptied-out virus, so that the cell itself makes the protein.
Also called: viral-vector gene therapy, gene replacement therapy
This describes published medicine for readers who want to understand how it works and what happened. It is not medical advice and says nothing about what any person should take.
What it is, in brief
A gene therapy supplies the instruction for a protein, not the protein. The instruction travels in a hollowed-out virus, goes to chosen cells and is made there, for as long as the DNA stays. Three choices decide the outcome: the capsid and route (which cells), the vector type (how long), and the promoter and enhancer (how much and where). 1
Reported, not yet replicated
How the gene gets in: capsid, tropism and route more
Tropism is which cell types a capsid can enter, set by its surface protein. AAV9 crosses the blood-brain barrier; AAV8 and AAV9 reach several muscle types; most natural capsids reach liver after systemic dosing; AAV2 is used in the retina. Route can matter more than capsid. Luxturna goes under the retina at 1.5 x 10^11 vg per eye in 0.3 mL; Zolgensma goes into a vein at 1.1 x 10^14 vg/kg.
The same vector idea gives opposite outcomes depending on dose, route and cell type. 123
Reported, not yet replicated
Persistence: episome or integration more
AAV DNA sits in the nucleus as a loose circle (an episome). In cells that no longer divide, such as neurons, it keeps working; in dividing tissue the circles are diluted and lost. The Zolgensma label says its DNA persists largely in episomal form, that random integration has been reported, and that tumours have been reported after approval with no causal link established.
Integrating vectors are copied with the cell and so last, but they can switch on neighbouring genes, as in the SCID-X1 trial. That is why integrating vectors are used on stem cells and T cells outside the body, and AAV in the body in non-dividing tissue. 1345
Reported, not yet replicated
Promoters, enhancers and expression control more
The promoter is the DNA stretch where the cell starts reading the gene; an enhancer boosts it. Both approved examples use the same kind of strong, always-on switch: Luxturna uses a cytomegalovirus enhancer with a chicken beta-actin promoter driving RPE65, and Zolgensma a hybrid promoter of the same type driving SMN. In both, selectivity comes from delivery, not from the switch.
The tools for adding control, from a review: tissue-specific promoters; miRNA target sites placed in the transgene to cut expression in unwanted cells (miR-208a sites stopped calpain3 being made in heart while sparing muscle; miRNA sites are also used for MeCP2 in Rett syndrome to avoid overexpression toxicity, in trial NCT05898620); and drug-inducible systems such as the Tet system switched by doxycycline. Co-expression of a safety gene exists in engineered T cells (a rimiducid-inducible caspase 9). The AAV review states that too much protein may be toxic. 6341
Reported, not yet replicated
The immune system is part of the drug more
The shell is itself an immune stimulus. In 1999 an adenovirus vector infused into the liver artery at the top dose of 6 x 10^11 particles/kg was followed within about 18 hours by confusion and jaundice in an 18-year-old participant, then systemic inflammation, clotting failure and organ failure, with death at 98 hours. At the AAV end, Zolgensma and Elevidys carry boxed warnings for liver injury. The pattern: systemic high-dose AAV loads the liver, while local delivery to a closed space does not. 738
Reported, not yet replicated
Case pair and the two others deep
Worked: Zolgensma, where 19 of 21 infants in the pivotal study were alive without permanent ventilation, and Luxturna, where 13 of 20 treated people against 0 of 9 controls passed the mobility test at 1 lux. Failed: the 1999 adenovirus trial, and mixed: retroviral SCID-X1 gene therapy, which corrected 9 of 10 patients while 4 of the 9 developed T-cell leukaemia 31 to 68 months later, with vector insertions near LMO2, BMI1 and CCND2. See the linked drug records for the sources of each number. 39710
Reported, not yet replicated
Limits: packaging, dilution, re-dosing deep
AAV vectors optimally accommodate genomes under 5.0 kb including the controls, so large genes need workarounds: the dystrophin coding sequence is 11.5 kb, hence shortened micro-dystrophin. Dilution in dividing tissue means durability must be shown over years, not months. After Zolgensma, anti-AAV9 titres rose to at least 1:102,400 in every patient in Study 2, and re-dosing is not evaluated. 13
Reported, not yet replicated
Amit's guess about persistence and promoters
He said: "if it's gene therapy, then you have this sort of persistence, and you want to engineer the expression levels. Now you have to get into how often the promoter is expressed and whether you have to co-express other things so that you're able to regulate this". 11
Amit's hypothesis
What the sources say about that guess more
Right in outline: persistence is characteristic of gene therapy, expression level can be engineered, and co-expressed elements exist (miRNA target sites in the transgene and a safety-switch gene in engineered T cells).
Two corrections from the cited sources. First, a promoter sets a rate in a given cell type, not a schedule of how often it is expressed. Second, for AAV the persistence is a property of the vector DNA form (episome versus integration) and of whether the cell divides, not of the promoter. Also, in the two approved examples the promoter is strong and ubiquitous and the selectivity comes from the capsid and route. This atlas found no clinical AAV with an off switch after dosing. 1346
Reported, not yet replicated
What it is
A viral-vector gene therapy is a particle: a protein shell (capsid) around a cassette of DNA holding a promoter, an enhancer, the therapeutic gene and signal sequences. AAV, the commonest in-body vector, is a small virus with about 4.7 kb of single-stranded DNA; the recombinant vector keeps the end sequences and swaps in the therapeutic cassette. Other vectors are adenovirus, and integrating retroviruses and lentiviruses used on cells outside the body. 1
Reported, not yet replicated
Intervenes at
Same as small molecule
The questions are the same as for any medicine: pick a disease target, show benefit on a clinical endpoint, manage safety. The drug is also tested by dose-response and a toxicity wall: both worked and failed cases below are dose-response curves with an immune-system wall. The 1999 adenovirus trial and Zolgensma differ in where that wall sits. 73
Our inference
Changes downstream
The unit of dose is vector genomes per kg, not milligrams (Zolgensma: 1.1 x 10^14 vg/kg in one infusion). The cell makes the drug, so potency is a cell-based property of the particle, and the titre assay itself matters: FDA called the phase 1 Zolgensma titre assay inaccurate and imprecise and the early doses were restated when re-measured (as cited by our own earlier report). There is no clearance to stop exposure: the dose cannot be lowered afterwards. Long-term follow-up for years is needed (the SCID-X1 leukaemias appeared about 3 years after treatment in the first two cases). 3125
Reported, not yet replicated
Delivery
Two things decide which cells take up the vector: the capsid and the route. AAV9 crosses the blood-brain barrier and is a leading capsid for central nervous system delivery after intravenous dosing; AAV8 and AAV9 reach several muscle types; almost all natural capsids transduce liver efficiently after systemic delivery; AAV2 is used in the retina (AAV review). Route matters as much: Luxturna is injected under the retina (0.3 mL per eye), keeping vector in a small closed space, while Zolgensma is infused into a vein, which sends vector everywhere including the liver. Anti-AAV antibodies are carried by 40 to 80% of people (10 to 40% neutralising for AAV8) per the review, which can block entry. 123
Reported, not yet replicated
Persistence
Persistence depends on the form of the vector DNA, not on the promoter. AAV DNA circularises into episomes; expression persists in non-dividing cells and genomes are gradually lost in dividing ones (AAV review). The Zolgensma label says the DNA persists largely in episomal form and reports random integration. Integrating retroviral and lentiviral vectors persist through cell division, with oncogenic risk, which is why they are used on dividing stem and T cells outside the body. Expression level and place are set by the promoter, enhancers, miRNA target sites in the transgene and drug-inducible elements. After dosing there is no off switch for a clinical AAV that this atlas found; the capsid antibodies that follow limit a second dose (the Zolgensma label says re-dosing is not evaluated). 134
Reported, not yet replicated
Examples
Teach this
In one sentence
Gene therapy gives cells the instruction rather than the protein, so what you must control is where the particle goes, how long its DNA stays and how strongly it is read. 1
Reported, not yet replicated
The five-minute version
Package a gene with a promoter in a viral shell. The shell and the route pick the cells, the vector's DNA form picks how long it lasts (a loose circle in non-dividing cells, or integrated into the chromosome), and the promoter and enhancer pick how much protein is made. Zolgensma and Luxturna worked in post-mitotic tissue with steroid cover; the 1999 adenovirus trial and the SCID-X1 trial show the two walls: the immune reaction to the shell and the effect of the integrated enhancer on neighbouring genes. 1375
Reported, not yet replicated
What a listener will ask
Can you turn it off after dosing?
Not for a clinical AAV that this atlas found; drug-inducible systems and caspase safety switches exist but are described for research and engineered T cells.
Why not dose twice?
Antibodies to the shell form after the first dose, and the Zolgensma label says re-dosing is not evaluated.
Is the promoter the selector?
In the two approved AAV examples, no: both use a strong ubiquitous promoter and the capsid and route supply selectivity.
The common wrong idea
Persistence and expression level are mainly set by the promoter, and "how often" it is on.
The promoter sets how strongly a gene is read in a given cell type; persistence comes from the vector DNA form and the cell's division; selectivity comes largely from the capsid and the route.
The one citation to hand over: 1
Still open here
- Can tissue-specific promoters and miRNA target sites make intravenous AAV safe enough that the liver stops being the limiting organ? The review lists the tools; no clinical result that settles it was found.
- Is the tumour signal in post-approval AAV reports a coincidence or a class effect of integration? The Zolgensma label says causality is not established.
- Can a clinical AAV be given an off switch after dosing, and can a second dose work after anti-capsid antibodies form?
- How long do effects last? No primary long-term durability paper beyond the regulator's 2-year statement for Luxturna was opened.
- Not covered: manufacturing (empty-capsid fraction, production system), cost, and dosing-schedule comparisons across products.
Linked from
Sources
- Wang D, Tai PWL, Gao G. Adeno-associated virus vector as a platform for gene therapy delivery. Nature Reviews Drug Discovery (2019). review; opened 2026-11-17 all records citing this
- US Food and Drug Administration. LUXTURNA (voretigene neparvovec-rzyl) prescribing information. FDA (2017). regulatory, not peer reviewed; opened 2026-11-17 all records citing this
- US Food and Drug Administration. ZOLGENSMA (onasemnogene abeparvovec-xioi) prescribing information, revision 6/2026. FDA (2026). regulatory, not peer reviewed; opened 2026-11-17 all records citing this
- Butterfield et al.. Gene regulation technologies for gene and cell therapy (Mol Ther 2025, 33(5); title taken from the file name of the hosted PDF). Molecular Therapy (2025). review; opened 2026-11-17 all records citing this
- Hacein-Bey-Abina S, Von Kalle C, Schmidt M, et al.. LMO2-associated clonal T cell proliferation in two patients after gene therapy for SCID-X1. Science (2003). paper; opened 2026-11-17 all records citing this
- US Food and Drug Administration. FDA Summary Basis for Regulatory Action: Luxturna, 18 December 2017. FDA (2017). regulatory, not peer reviewed; opened 2026-11-17 all records citing this
- Raper SE, Chirmule N, Lee FS, et al.. Fatal systemic inflammatory response syndrome in a ornithine transcarbamylase deficient patient following adenoviral gene transfer. Molecular Genetics and Metabolism (2003). paper; opened 2026-11-17 all records citing this
- US Food and Drug Administration. FDA takes action: new boxed warning for acute serious liver injury and acute liver failure following Elevidys (14 November 2025). FDA (2025). regulatory, not peer reviewed; opened 2026-11-17 all records citing this
- Russell S, Bennett J, Wellman JA, et al.. Efficacy and safety of voretigene neparvovec (AAV2-hRPE65v2) in patients with RPE65-mediated inherited retinal dystrophy: a randomised, controlled, open-label, phase 3 trial. The Lancet (2017). paper; opened 2026-11-17 all records citing this
- Hacein-Bey-Abina S, Garrigue A, Wang GP, et al.. Insertional oncogenesis in 4 patients after retrovirus-mediated gene therapy of SCID-X1. Journal of Clinical Investigation (2008). paper; opened 2026-11-17 all records citing this
- Amit's dictation of 2026-11-17: what the atlas must contain. Kept in the atlas repository; not published.
- US Food and Drug Administration. FDA Summary Basis for Regulatory Action: Zolgensma (2019). FDA (2019). regulatory, not peer reviewed; opened 2026-11-17 all records citing this
- Mercuri E, Muntoni F, Baranello G, et al.. Onasemnogene abeparvovec gene therapy for symptomatic infantile-onset spinal muscular atrophy type 1 (STR1VE-EU): an open-label, single-arm, multicentre, phase 3 trial. The Lancet Neurology (2021). paper; opened 2026-11-17 all records citing this
History of this record (7)
This record has been changed 7 times, on the dates below. The note kept against each change is written for the people who maintain the atlas, in their own shorthand, so it is not shown here; it is in the record itself.
- Status
- partly written
- Last updated
- 2026-11-19
- Record
- content/modalities/gene-therapy.json