The siRNA was perfect. In cell culture it silenced the target oncogene by ninety percent. Cells died. The team injected it into tumor-bearing mice and waited. The tumors did not shrink. Their first explanation: wrong target. They spent months hunting for a better one. They were diagnosing the wrong failure. The sequence was right. The oncogene was right. The delivery failed at a single step — the cargo could not escape the endosome.

Two decades of this field's history are compressed into that opening story. The rate-limiting problem in nucleic acid therapeutics is not the sequence. It is the transport. A sequence can be designed in hours on a laptop. Getting that sequence to the cytosol of a tumor cell — in a living patient, after injection into the bloodstream — is the work. A nucleic acid drug is a passenger that cannot walk. This chapter is about every barrier between a needle and the cytosol, and how to tell a delivery failure from a biology failure before you spend months looking for a new target.

The vehicle that solved the delivery problem — not perfectly, but well enough to produce approved drugs and vaccines at global scale — is the lipid nanoparticle. A standard LNP contains four components. A structural phospholipid, cholesterol, a PEG-lipid that coats the surface and extends circulation time, and the functional heart of the entire mechanism: the ionizable lipid. An ionizable lipid has an amine head group with a pKa designed to sit between blood pH and endosomal pH. At blood pH of seven point four, the amine is neutral — the particle stays in circulation. When the endosome acidifies to pH five or six, the ionizable lipid gains a positive charge — and that charge switch is what triggers endosomal escape.

The LNP vehicle is cargo-agnostic — it can carry different nucleic acid payloads with different mechanisms and different durations of action. siRNA directs the cell's RNA interference machinery to destroy a specific messenger RNA. Silencing is transient and reversible — re-dosing is required. Onpattro, the first approved siRNA drug, is administered every three weeks for life. mRNA instructs ribosomes to synthesize a protein. The effect is transient and non-integrating, which makes the platform ideal for personalized neoantigen cancer vaccines. CRISPR-Cas9 cuts DNA at a specified location and produces a permanent edit — one successful editing event in a cell persists in all its daughters.

At the largest scale in history, LNP-delivered mRNA vaccines for COVID-19 demonstrated that the platform works at planetary scope. Billions of doses manufactured, administered, and monitored for safety. This was simultaneously a proof of the manufacturing technology and a massive demonstration that LNP-mRNA delivery is safe enough for prophylactic administration to healthy people. The infrastructure, regulatory experience, and manufacturing capacity this created directly accelerated cancer mRNA vaccine development. In oncology, the most active application is personalized neoantigen vaccines — the patient's tumor is sequenced, private neoantigens are identified, and mRNA encoding those neoantigens is formulated in LNPs and administered within weeks of tumor sequencing.

Lipid nanoparticles have no integration risk, are scalable, allow re-dosing, and carry flexible RNA payloads. They are not always the best choice. For applications requiring durable, long-term protein expression, transient mRNA delivery requires repeated administration indefinitely. Lentiviral vectors integrate permanently into the host genome — the property that makes them useful for CAR-T manufacturing, and the property that caused leukemia in the X-SCID trials. AAV persists as episomes in post-mitotic cells and carries a payload limit of about four point seven kilobases. Adenoviral vectors provoke strong immune responses — which ended their use in gene replacement therapy, but is actually a feature in oncolytic virotherapy, as in T-VEC for melanoma. The vehicle choice is application-specific, not a ranking from worse to better.

The pivot in nucleic acid therapeutics is endosomal escape. It is the narrowest point in the funnel, the step with the lowest efficiency, and the step most often responsible for the gap between in vitro results and in vivo outcomes. In vitro transfection bypasses the funnel almost entirely — commercial transfection reagents under optimized conditions deliver cargo to the cytosol far more efficiently than any systemically injected particle in a living animal. A result from a dish is evidence that the sequence can engage the target. It is not evidence that the vehicle can deliver the sequence in vivo. The two experiments are asking different questions. And confusing them is exactly how teams spend months looking for better targets when the problem was never the target.