Lipid Nanoparticle assembly & intracellular mRNA delivery
Before any nanoparticle can be assembled, the mRNA drug substance itself must be manufactured. This is done entirely outside a living cell — in a test tube reaction called In Vitro Transcription (IVT) — using only enzymes, [], and a DNA blueprint.
Messenger RNA (mRNA) is a transient copy of a gene — a single-stranded molecule that carries a protein-building recipe from the DNA nucleus to the ribosome factory. After delivering its message, natural mRNA is degraded within minutes to hours.
In vaccine and therapeutic applications, synthetic mRNA is used to temporarily instruct your own cells to produce a specific protein — for example, the Spike protein of SARS-CoV-2 — without ever introducing live virus, live bacteria, or DNA into the nucleus. This is why mRNA vaccines cannot alter your genome: the message never reaches the DNA.
Nobel Prize in Physiology or Medicine 2023: Katalin Karikó & Drew Weissman were awarded for their discovery of nucleoside base modifications that enabled the development of effective mRNA vaccines.
mRNA is produced in a cell-free enzymatic reaction with these components:
• Linear DNA template (plasmid or PCR product) — contains the gene of interest plus optimized UTR sequences • RNA Polymerase (T7 or SP6) — reads the DNA template 3'→5' and synthesizes mRNA 5'→3' at ~200 nt/sec • Nucleotide triphosphates: ATP, CTP, GTP, and crucially modified UTP (m1ΨTP instead of UTP) • Reaction buffer with Mg²⁺ ions, spermidine, and DTT
The reaction runs for 2–4 hours at 37 °C and yields milligram quantities of mRNA per mL. The DNA template is then destroyed with DNase I, and the mRNA is purified via HPLC to remove dsRNA by-products that would otherwise trigger a strong inflammatory response.
Unmodified mRNA made in the lab contains uridine (U), which is recognized as "foreign" by Toll-like receptors (TLR3, TLR7, TLR8) — innate immune sensors that evolved to detect viral RNA. When these sensors fire, they trigger rapid mRNA degradation and inflammatory cytokine storms.
Karikó & Weissman discovered that replacing uridine with N1-methylpseudouridine (m1Ψ) makes the mRNA molecularly "invisible" to TLRs. The structural difference is tiny (just a rotation of the uracil ring), but the functional impact is enormous:
• 10-fold increase in ribosomal translation efficiency • Dramatically reduced innate immune activation • 50–100× greater protein output per mRNA molecule • Extended functional half-life in vivo (hours vs. minutes)
The Pfizer-BioNTech (BNT162b2) and Moderna (mRNA-1273) COVID-19 vaccines both use full m1Ψ substitution — every single uridine in the sequence is replaced.
• 5' Cap (Cap1 structure — m7G(5')ppp(5')NmpNp): A methylated guanosine cap added co-transcriptionally. Protects against 5'→3' exonucleases, is essential for eIF4E ribosome docking, and signals "self" to immune sensors. Vaccines use "Cap1" (both first and second nucleotides 2'-O-methylated) for optimal stealth.
• 5' UTR: Engineered regulatory sequence between the cap and start codon. Derived from highly expressed human genes (α-globin, β-globin, CYBA). Optimized Kozak sequence context (GCCACCATG) ensures efficient ribosome scanning and AUG recognition.
• 3' UTR: Stabilizing element after the stop codon. Contains ARE-free sequences that resist deadenylation. Pfizer uses tandem 3'UTR elements from α-globin and AES to achieve exceptional mRNA stability.
• Poly-A Tail (100–150 nt): Added enzymatically or encoded in the template. Interacts with PABP (poly-A binding protein) to circularize mRNA for efficient recycling of ribosomes and protection from 3'→5' exonucleases. Longer poly-A = longer half-life.
Naked mRNA cannot enter cells on its own — it is too large (~2 MDa), highly negatively charged, and rapidly degraded by RNases in blood. The solution is encapsulating it in a lipid nanoparticle: a 60–120 nm synthetic vesicle that protects mRNA, evades the immune system, and fuses with cell membranes to deliver its cargo.
LNPs for mRNA delivery contain exactly four types of lipid, each with a specific role:
LNPs form spontaneously through rapid mixing of two streams in a microfluidic device:
• Aqueous stream (pH 4.0 acetate buffer): contains the mRNA cargo • Ethanol stream: contains all four lipids dissolved at precise molar ratios
When the streams collide at high flow rate (Reynolds number > 100), the ethanol rapidly dilutes below the lipid solubility threshold. Lipids self-assemble around the mRNA molecules through electrostatic attraction (ionizable lipids become positively charged at pH 4, mRNA is negatively charged) and hydrophobic forces.
The mixture is then dialyzed against PBS buffer at pH 7.4, which neutralizes the ionizable lipids, completing encapsulation and trapping mRNA inside the particle core.
The entire microfluidic mixing and LNP formation process takes less than 1 second — but getting the flow rates, lipid ratios, and pH precisely right requires months of formulation optimization.
Cryo-electron microscopy reveals that mRNA LNPs are not simple hollow vesicles like liposomes. Instead, they have an electron-dense inverse hexagonal or multilamellar core where mRNA strands are condensed and interleaved with ionizable lipid layers.
Key quality attributes measured after formulation:
• Particle size (Z-average, DLS): 60–120 nm — small enough to extravasate at fenestrated sinusoidal endothelium of the liver • PDI (Polydispersity Index): <0.1 — highly uniform distribution required for consistent dosing • Encapsulation efficiency (Ribogreen assay): >90% — percentage of input mRNA successfully captured inside particles • Zeta potential: near neutral (–5 to +5 mV) at pH 7.4 — prevents aggregation via charge-based repulsion • mRNA integrity (Agilent TapeStation): >90% intact full-length species
After intramuscular injection, LNPs must survive a hostile biological environment: nucleases, complement proteins, phagocytic cells, and physical shear forces. The PEG coating and protein corona determine where the drug goes, how long it survives, and which cells it ultimately reaches.
Polyethylene glycol (PEG) chains on the LNP surface create a dense hydrophilic brush layer that physically blocks proteins from adsorbing onto the particle. Without PEG:
• Complement proteins (C3b, C4b) opsonize particles within seconds → macrophage phagocytosis and rapid clearance • Particles aggregate due to unshielded hydrophobic surfaces • Liver Kupffer cells and spleen red pulp macrophages clear >90% of particles within 30 minutes
With 2% PEG-lipid: • Protein adsorption is reduced ~10-fold • Complement activation is dramatically suppressed • Plasma half-life extends from <30 min to ~6 hours • Particles remain monodisperse in circulation
Important caveat: PEG-lipid sheds from the particle over time as the acyl chains gradually dissociate, which actually helps the particle eventually fuse with the endosomal membrane after cell uptake.
PEG antibodies: ~7% of the general population has pre-existing anti-PEG antibodies (from cosmetics, laxatives, etc.), which can cause accelerated blood clearance (ABC phenomenon) and rare hypersensitivity reactions after repeat LNP dosing.
Even with PEG shielding, LNPs inevitably adsorb a layer of plasma proteins forming a "protein corona." Rather than being purely detrimental, for liver targeting this is intentional and exploited.
Apolipoprotein E (ApoE) — the same protein involved in Alzheimer's risk and lipid metabolism — adsorbs selectively onto LNP surfaces. ApoE is a natural ligand for LDL receptors (LDLR), which are highly overexpressed on hepatocytes (liver parenchymal cells).
The pathway: LNP → ApoE adsorption in bloodstream → ApoE-LDLR binding at hepatocyte surface → receptor-mediated endocytosis into liver cells.
This explains why current COVID vaccines primarily cause liver cells to produce spike protein (rather than muscle cells at the injection site), and why patients with hyperlipidemia or those on statins sometimes have altered vaccine pharmacokinetics.
Fluorescent lipid labeling and luciferase reporter mRNA studies in rodents show the following distribution of injected dose:
• Injection site (muscle/draining lymph nodes): ~25–40% — local dendritic cells take up LNPs, activating the adaptive immune response • Liver: ~50–70% — hepatocytes via ApoE/LDLR pathway, the dominant clearance organ • Spleen: ~10–20% — splenic antigen-presenting cells, important for germinal center reactions • Other organs (lung, heart, kidney, brain): <2% combined
For COVID vaccines (intramuscular injection), the goal is to maximize expression in muscle and draining lymph nodes where immune activation occurs, while minimizing systemic distribution. Liver accumulation is a side effect of the ApoE mechanism, not a target.
Getting mRNA inside the cell is only half the battle. The cell's endolysosomal system is designed to destroy foreign material. Only ~1–2% of endocytosed mRNA successfully escapes into the cytoplasm — making endosomal escape the critical rate-limiting bottleneck of the entire delivery process.
LNPs decorated with ApoE bind to LDL receptors (LDLR) on the hepatocyte surface. This triggers clathrin-mediated endocytosis:
1. LDL receptor clusters in clathrin-coated pits form at the plasma membrane 2. The membrane invaginates and pinches off, forming a clathrin-coated vesicle (~150 nm) 3. The clathrin coat is shed; the vesicle becomes an early endosome 4. Early endosome matures into late endosome: V-ATPase proton pumps acidify the lumen from pH 7.4 to pH 6.0 → 5.0 → 4.5 5. Late endosome can fuse with lysosomes (pH ~4.5) containing degradative enzymes (nucleases, lipases, proteases)
Without escape, the mRNA cargo is completely destroyed within 30–60 minutes.
This is the molecular magic of ionizable lipids and why their pKa is so carefully engineered (pKa 6.2–6.8):
• At physiological pH 7.4: ionizable lipids are neutral → no charge repulsion → stable, non-toxic particle • At endosomal pH ~5.5: protons protonate the tertiary amine headgroups → ionizable lipids become cationic • Cationic lipids in the endosome attract anionic phospholipids (mainly DOPE or analogues) from the endosomal membrane • This mixing induces a transition from a stable lamellar bilayer to an unstable inverted hexagonal (HII) phase • The destabilized membrane ruptures locally, creating a transient pore or full membrane fusion event • mRNA molecules escape through this disruption into the cytoplasm
The pKa sweet spot: if pKa is too high (>7.0), lipids are charged at neutral pH → systemic toxicity. If too low (<5.5), lipids remain neutral in endosomes → no escape.
Only ~1–2% of LNP-encapsulated mRNA successfully escapes the endosome. Improving this bottleneck even to 5–10% would allow a 5–10× dose reduction — a major area of current LNP research.
A secondary mechanism of endosomal disruption involves buffering capacity. Ionizable lipids can buffer the endosomal pH drop (the "proton sponge hypothesis"):
1. V-ATPase continuously pumps H⁺ ions into the endosome to acidify it 2. Ionizable lipids absorb these protons (being protonated) — buffering the pH change 3. To compensate, the cell pumps in MORE H⁺, which requires counter-ion (Cl⁻) influx 4. Osmotic pressure builds up as ion concentration increases 5. Water flows in by osmosis → endosome swells → membrane ruptures → cargo release
This mechanism works cooperatively with direct membrane fusion to maximize the small fraction of successful escape events.
Once mRNA escapes into the cytoplasm, your cell's own protein synthesis machinery takes over. Within hours, millions of spike protein molecules are produced, displayed on the cell surface, recognized by the adaptive immune system, and a lasting immunological memory is established.
The eukaryotic ribosome (80S) consists of two asymmetric subunits:
• Small subunit (40S): contains 18S rRNA + ~33 proteins. Responsible for mRNA decoding — reads the three-nucleotide codons and matches them to incoming aminoacyl-tRNAs • Large subunit (60S): contains 28S, 5.8S, 5S rRNA + ~49 proteins. Contains the peptidyl transferase center (PTC) — the catalytic heart that forms peptide bonds
Translation proceeds in three phases: 1. Initiation: 40S subunit scans from 5' cap, recognizes Kozak-AUG start codon, 60S joins → 80S initiation complex 2. Elongation: tRNAs deliver amino acids one by one at ~5–6 aa/sec; peptide chain grows from N-terminus to C-terminus 3. Termination: stop codon (UAA, UAG, UGA) recognized by release factors → polypeptide released → ribosome dissociates
The SARS-CoV-2 Spike (S) protein is a 1,273 amino acid type I transmembrane glycoprotein:
1. Co-translational insertion: as the ribosome synthesizes the N-terminal signal peptide, it is recognized by SRP (Signal Recognition Particle) → ribosome docked to rough ER membrane 2. The growing polypeptide is threaded into the ER lumen as it is synthesized 3. N-linked glycosylation: N-acetylglucosamine (GlcNAc) is added by oligosaccharyltransferase (OST) at ~22 asparagine (N-X-S/T) sequons — glycosylation is essential for correct folding 4. Calnexin/calreticulin chaperones assist folding in the ER lumen 5. Furin cleavage site (S1/S2) and the transmembrane anchor are encoded in the Spike sequence 6. Vaccine-encoded Spike contains 2x proline substitutions (K986P, V987P) that lock it in the prefusion conformation — more immunogenic than the post-fusion form 7. Fully folded trimeric Spike (three S monomers form a homotrimer) traffics through Golgi → plasma membrane display
The "2P" proline mutations in mRNA vaccine Spike protein were designed by Barney Graham and Jason McLellan (UT Austin) — a pre-existing structural biology innovation that made COVID vaccines dramatically more effective.
Once Spike protein appears on the cell surface, the adaptive immune system activates in a coordinated cascade:
• MHC-I presentation: proteasome degrades cytoplasmic Spike fragments → 8–10 aa peptides loaded onto MHC class I → displayed to CD8⁺ cytotoxic T cells → CTL response (kill infected cells)
• MHC-II presentation (in dendritic cells): endosomal Spike fragments loaded onto MHC class II → displayed to CD4⁺ helper T cells → Th1 cytokine release (IFN-γ, TNF-α) driving cellular immunity
• B cell activation: Free spike protein (secreted or shed from surface) binds B cell receptors (BCRs) → B cells internalize antigen → Th cell help → germinal center reaction in lymph nodes → somatic hypermutation → affinity maturation → high-affinity IgG antibodies
• Memory formation: Long-lived plasma cells in bone marrow continuously secrete antibodies for years. Memory B cells and memory T cells persist and respond rapidly (<24 h) upon future virus encounter.
One of the key safety features of mRNA vaccines is that the mRNA is inherently temporary:
• Functional lifetime in vivo: ~24–48 hours after LNP uptake (even with m1Ψ modification) • Degradation pathway: cellular RNases (XRN1, exosome complex) progressively digest from 3' and 5' ends after deadenylation • No integration into genome: mRNA cannot be reverse-transcribed and inserted into DNA without specialized retroviral machinery (reverse transcriptase + integrase), which human cells do not possess • No nuclear entry: mRNA stays in the cytoplasm and never approaches chromosomal DNA
The transient expression window (~48 h of peak protein, with tail end up to 1 week) is sufficient to prime a full adaptive immune response that provides months to years of protection.