Prefusion Conformational Locking and Ionizable Lipids

Barney Graham, Katalin Karikó, Drew Weissman, and the Structural Engineering of mRNA Vaccines

Solving the forty-year mystery of the 1966 FI-RSV tragedy required visualizing viral surface glycoproteins at atomic resolution before and after they fuse with a host cell. Class I viral fusion proteins—including the F glycoprotein of Respiratory Syncytial Virus, influenza hemagglutinin, the HIV-1 Env trimer, and the coronavirus Spike (S) glycoprotein—exist on live virions in a kinetically trapped, high-energy metastable state known as the prefusion conformation. Like a set mousetrap, the prefusion trimer stores mechanical energy in a coiled heptad-repeat hinge. Upon binding a host receptor, the protein jackknifes outward, thrusts its hydrophobic fusion peptide into the target membrane, and collapses into an irreversible six-helix bundle called the postfusion hairpin. Because the prefusion state is metastable, purifying the soluble protein or treating virions with formalin spontaneously triggers the snap into the spent postfusion rod. That transition hides the apex epitopes (such as Antigenic Site $\varnothing$ on RSV F and the receptor-binding domain on coronavirus Spike) where the most potent neutralizing antibodies bind.

Between 2013 and 2017 at the NIH Vaccine Research Center and Dartmouth, structural biologist Jason McLellan and viral immunologist Barney Graham engineered a universal atomic pin to lock class I fusion proteins in their prefusion shape. First co-crystallizing the metastable RSV F prefusion trimer by trapping its apex with monoclonal antibody D25, McLellan and Graham introduced disulfide-bonded cysteines and cavity-filling phenylalanines (DS-Cav1). That modification raised neutralizing antibody titers more than tenfold and enabled the RSV prefusion subunit vaccines (Arexvy and Abrysvo) approved in 2023. Turning to the betacoronaviruses MERS-CoV, SARS-CoV-1, and HKU1 in 2016 and 2017 using cryo-electron microscopy, McLellan, Graham, Andrew Ward, and Nianshuang Wang identified the hinge loop inside the S2 subunit—between the first heptad repeat (HR1) and the central helix (CH)—that straightens into a continuous long $\alpha$-helix during fusion. Because proline possesses a rigid five-membered pyrrolidine ring bonded back to its own backbone nitrogen, proline sterically resists forming an $\alpha$-helix. Substituting two consecutive prolines (2P, residues K986P and V987P in SARS-CoV-2) at the turn between HR1 and the central helix acted as a molecular splint. It prevented helix elongation, locked the coronavirus Spike trimer in its prefusion conformation, and multiplied expression yields fifty-fold.

Delivering the genetic blueprint for a proline-locked Spike protein into human muscle cells required overcoming two biochemical barriers that had stalled synthetic messenger RNA (mRNA) since Robert Malone, Philip Felgner, and Inder Verma transfected liposome-encapsulated mRNA into cells in 1989. First, unmodified in vitro-transcribed mRNA is recognized inside endosomes by Toll-like receptors TLR7 and TLR8 and cytosolic sensors RIG-I and PKR. Those sensors unleash type I interferon inflammation while phosphorylating eukaryotic initiation factor $2\alpha$ (eIF2$\alpha$) to halt translation before viral antigen can be made. At the University of Pennsylvania in 2005, Katalin Karikó and Drew Weissman discovered that replacing unmodified uridine with pseudouridine ($\Psi$) or $N^1$-methylpseudouridine ($m^1\Psi$) silenced Toll-like receptor activation and eliminated PKR-mediated translational arrest, boosting protein output more than tenfold.

Second, because negatively charged mRNA chains are degraded by extracellular ribonucleases and cannot cross anionic membranes, Pieter Cullis and colleagues at the University of British Columbia engineered four-component lipid nanoparticles (LNPs). The decisive ingredient—alongside structural phospholipid, cholesterol, and a polyethylene-glycol (PEG)-lipid coat—is an ionizable cationic lipid (ALC-0315 or SM-102) with an apparent $\text{p}K_a$ around 6.1 to 6.7. At blood pH 7.4, the ionizable lipid remains electrically neutral, avoiding systemic toxicity. Once a cell internalizes the nanoparticle into an acidifying endosome (pH 5.0 to 6.0), the tertiary amine headgroup becomes positively protonated. It pairs electrostatically with anionic endosomal phospholipids to adopt an inverted hexagonal ($H_{\text{II}}$) phase and disrupts the endosome to release the $m^1\Psi$-modified mRNA into the cytosol. When the SARS-CoV-2 genome sequence was released on January 10, 2020, combining the 2P prefusion lock, $m^1\Psi$-modified mRNA, and ionizable lipid nanoparticles yielded clinical-grade vaccine vials within forty-two days and achieved 94 to 95 percent efficacy across more than 73,000 Phase 3 volunteers.