mRNA platforms and LNP delivery

Messenger RNA is an appealing drug substance for a simple reason: it does not need to reach the nucleus and cannot integrate into the genome. Delivered to the cytosol, it is read by ribosomes already present, produces protein for a few days, and is degraded by ordinary cellular machinery. The therapeutic object is an instruction with a defined lifetime rather than a molecule that must persist. Two problems stood between that idea and a product, and neither is about the protein being encoded.

Making RNA that the cell does not fight

Innate immunity is built to detect foreign RNA. Endosomal TLR7 and TLR8 sense single-stranded RNA, TLR3 and cytosolic RIG-I and MDA5 sense double-stranded species, and the response — type I interferon, PKR activation, translational shutdown — destroys the very thing the drug is trying to do. The decisive finding, by Karikó and Weissman in 2005 and recognised by the 2023 Nobel Prize in Physiology or Medicine, was that substituting modified nucleosides for uridine suppresses this sensing; N1-methylpseudouridine, used in the approved COVID-19 vaccines, both blunts recognition and increases translation per transcript.

The rest of the construct is conventional molecular biology applied hard: a cap-1 structure so the transcript is not read as foreign and is efficiently initiated, untranslated regions chosen for ribosome loading and stability, a poly(A) tail of defined length, and codon choice tuned to the host tRNA pool. Purification matters as much as design, because in-vitro transcription generates double-stranded byproducts that are potent interferon inducers; removing them by chromatography restores expression by an order of magnitude.

The lipid nanoparticle is a pH machine

Naked mRNA is large, polyanionic and nuclease-sensitive, so it travels inside a lipid nanoparticle roughly 60–100 nm across. The functional component is the ionisable lipid, designed with an apparent pKa around 6.2–6.5. At blood pH it is essentially neutral, which keeps the particle from binding serum proteins indiscriminately and limits toxicity. Inside the acidifying endosome it protonates, becomes cationic, and ion-pairs with the anionic phospholipids of the endosomal membrane; the resulting lipid pair has a cone shape that favours a non-bilayer hexagonal phase, and the membrane destabilises. That transition is the whole delivery event.

It is also inefficient. Measurements of endosomal escape for these systems put it in the low single-digit percent — most of the dose is taken up, trafficked and degraded without ever reaching the cytosol. That number, more than any other, sets what the platform can and cannot do: it is compatible with vaccination, where a small amount of antigen suffices and local inflammation is useful, and it is a hard obstacle for protein-replacement indications requiring sustained, quantitative expression.

Where the constraints still bind

The PEG-lipid that keeps particles from aggregating also blocks cell uptake, so it must desorb in circulation; the same PEG raises anti-PEG antibody and complement-activation concerns on repeat dosing. Biodistribution is not neutral: conventional particles adsorb apolipoprotein E and are taken up by hepatocytes through the LDL receptor, which is why the liver is the default destination and why extrahepatic targeting needs deliberate change to the lipid composition or surface. And the RNA itself is chemically fragile — the 2′-hydroxyl makes hydrolysis intramolecular and metal-catalysed — which is the physical reason for ultracold storage rather than a logistics preference.