Science Platforms

A circular piece of DNA carrying the gene of interest, delivered without a virus. Because there is no viral capsid, there is no anti-capsid immune response, and the same patient can be dosed again and again — which is the single most important property for any therapy intended to be maintained over a lifetime rather than given once. The trade-off is efficiency: naked plasmid enters cells poorly and needs a physical or chemical assist, and expression falls away as the DNA is diluted and silenced.

A plasmid with the bacterial backbone removed, leaving only the expression cassette. Stripping out the bacterial sequence and the antibiotic-resistance marker removes the main trigger for transcriptional silencing, so expression lasts substantially longer than from a conventional plasmid while keeping the non-viral advantages of redosing and no integration. The cost is manufacturing: minicircles require an additional in-vivo recombination and purification step, which reduces yield and raises cost per dose.

Very small, supercoiled DNA circles reduced to close to the minimum functional sequence. Their size and topology let them resist the shear forces of injection and nebulisation and penetrate tissue that larger constructs do not reach, which makes them useful where the barrier is physical rather than immunological. Payload capacity is the limitation — they carry short cassettes, not large genes.

Lipid particles carrying mRNA or DNA across the cell membrane. Nothing integrates and nothing persists, so the safety profile is defined by the dose given rather than by anything left behind, and the same patient can be redosed indefinitely. Manufacturing is fast and the formulation is swappable, which makes LNP the most iterable platform available. Untargeted particles concentrate in the liver by default; reaching any other tissue reliably requires either local administration or engineered targeting, and the material requires cold chain.

The most clinically validated vector in gene therapy, with multiple approved products and two decades of human follow-up. AAV persists as an episome rather than integrating into the genome, and in tissue that does not divide — neurons, retina, muscle, cardiac muscle — a single administration can express for years. Three constraints govern its use. Payload is capped at roughly 4.7 kilobases. A large fraction of adults carry pre-existing neutralising antibodies that exclude them from treatment. And once dosed, a patient's anti-capsid response generally prevents redosing with the same serotype.

An integrating vector that writes its payload into the host genome, so the modification persists through cell division and is passed to every descendant of the treated cell. That makes it the standard for engineering cells that must expand after treatment, and it carries a larger payload than AAV. Integration is also its risk: insertion near a growth-regulatory gene can contribute to malignancy, which is why current practice keeps lentiviral work outside the body wherever the biology allows.

A high-efficiency vector with a large payload capacity that transduces both dividing and non-dividing cells. It provokes a strong innate and adaptive immune response, which ends expression quickly. That response is a liability for chronic replacement and an asset where immune recruitment is the therapeutic mechanism.

Where the other platforms add a function, RNAi removes one: short RNA sequences direct the cell to degrade a specific transcript before it is translated. The therapeutic argument is control rather than power. The effect is graded by dose, wears off when treatment stops, and can be reversed by stopping — the opposite risk profile to an integrated permanent edit. The corollary is that treatment is continuous, not one-time.

Cells are removed, edited in the laboratory, tested, and only then returned. The advantage is quality control: editing accuracy, off-target sequencing, sterility, and cell identity are all confirmed before anything is administered, which is not possible when editing happens inside the body. The limitation is anatomical — it applies only to cells that can be taken out and put back.

Editing applied to stem and progenitor cells rather than to mature cells. Because the edited population renews itself, the correction persists as tissue turns over instead of fading as treated cells are replaced. This is the difference between correcting a tissue and maintaining a correction in one, and it is why stem-cell editing is pursued despite being the more demanding manufacturing route.

Engineered multilineage stress-enduring cells carrying an added or corrected gene, so that the delivered cell performs the therapeutic function itself rather than acting only as a carrier. A characterised, tested cell population can be released against identity and potency specifications before administration — a different and, for some indications, a stronger safety position than introducing a vector into the body and measuring what happens afterwards.

Small-molecule-inducible switches (doxycycline, rapamycin analogues) allowing transgene transcription to be turned on, modulated, or completely arrested post-administration.

Overcoming AAV packaging size constraints (4.7 kb) through split-intein and dual-AAV complementation constructs.

MicroRNA-target sequences engineered into the 3’ UTR of expression cassettes to suppress off-target transgene expression in unintended organs such as dorsal root ganglia or hepatic parenchymal cells.