Recombinant Subunit and Virus-Like Particle Vaccines

Recombinant subunit and virus-like particle (VLP) vaccines are molecularly defined vaccine platforms generated through recombinant DNA technology and heterologous protein expression. Rather than cultivating a live or killed pathogen—which is hazardous or impossible for non-cultivable viruses such as Hepatitis B virus (HBV) and Human Papillomavirus (HPV)—researchers clone the gene encoding the protective surface antigen into yeast, mammalian, insect, or bacterial expression vectors and purify the translated protein to homogeneity.

Within this architecture, soluble recombinant protein subunit vaccines deliver purified monomeric or trimeric viral glycoproteins (such as varicella-zoster virus glycoprotein E in Shingrix or influenza hemagglutinin in Flublok) formulated with potent vaccine adjuvants to overcome the weak intrinsic immunogenicity of soluble monomers. Conversely, Virus-Like Particle (VLP) vaccines exploit the biophysical self-assembly of viral structural capsid or envelope proteins into 20- to 100-nm icosahedral or spherical nanoparticles that lack any viral DNA or RNA. By displaying 60 to 360 copies of a conformational epitope in a rigid, quasi-crystalline geometric array spaced at 5 to 10 nm, VLPs directly cross-link surface B-cell receptors (BCRs) to elicit multi-decade neutralizing antibody titers.

Heterologous Expression Hosts: Yeast, Insect Sf9, and Mammalian CHO Cells

Because protective viral neutralizing epitopes are predominantly conformational and frequently depend on disulfide bond pairing and post-translational N-linked and O-linked glycosylation, selecting the heterologous expression host determines antigenic fidelity:

• Yeast (Saccharomyces cerevisiae, Pichia pastoris, Hansenula polymorpha): High-yield, rapid-growth eukaryotic fermenters capable of disulfide bond formation and ER lipid membrane assembly, though yeast adds high-mannose N-glycans rather than complex human sialylated glycans. Used for Hepatitis B surface antigen (HBsAg), HPV L1 VLPs (Gardasil 9), and the R21/Matrix-M malaria vaccine.
• Baculovirus–Insect Cell System (Spodoptera frugiperda Sf9 or Trichoplusia ni High Five): Recombinant Autographa californica multicapsid nucleopolyhedrovirus (AcMNPV) drives high-level expression under the polyhedrin promoter, folding complex oligomeric glycoproteins with paucimannose N-glycans. Used for Flublok (recombinant seasonal influenza hemagglutinin, which avoids the egg-adaptation mutations that distort antigenic site B in embryonated chicken eggs), Cervarix (HPV-16/18 VLP), and NVX-CoV2373 (Novavax).
• Mammalian Chinese Hamster Ovary (CHO) Cells: Provide human-like complex bi- and tri-antennary N-linked glycosylation, proper furin/protease processing, and chaperone-assisted folding of large viral envelope trimers.

Clinical Subunit Exemplars: Shingrix (RZV) and Novavax (NVX-CoV2373)

Soluble monomeric or trimeric protein subunits lack both microbial PAMPs and particulate multivalency, requiring pairing with advanced adjuvants:

• Recombinant Zoster Vaccine (RZV, Shingrix): Targets reactivation of latent varicella-zoster virus (VZV, herpes zoster / shingles). Whereas the older live-attenuated zoster vaccine (Zostavax) showed only 51.3% overall efficacy and declined to 37.6% efficacy in adults ≥70 years due to age-related T-cell immunosenescence, RZV pairs 50 μg of CHO-cell-expressed truncated VZV glycoprotein E (gE)—the most abundant viral envelope glycoprotein and primary target of VZV-specific CD4⁺ T cells—with the liposomal AS01_B adjuvant system (MPL + QS-21). In the ZOE-50 and ZOE-70 Phase III trials, two doses of RZV achieved 97.2% efficacy in adults ≥50 years and 91.3% efficacy in adults ≥70 years, maintaining >80% efficacy for over a decade by expanding polyfunctional IFN-γ⁺IL-2⁺TNF-α⁺CD40L⁺ CD4⁺ memory T cells.
• NVX-CoV2373 (Novavax SARS-CoV-2 Subunit Vaccine): Expresses full-length SARS-CoV-2 Spike glycoprotein in Sf9 insect cells engineered with both the 2P prefusion-stabilizing proline substitutions (K986P/V987P) and a mutated polybasic furin cleavage site (682-QQAQ-685). Detergent-extracted transmembrane Spike trimers spontaneously associate via their hydrophobic transmembrane tails around a polysorbate-80 micellar core into 27.2-nm rosette nanoparticles displaying up to 14 Spike trimers, co-formulated with the Matrix-M saponin cage adjuvant (achieving 89.7%–90.4% Phase III efficacy against ancestral/Alpha strains).

The Zinkernagel–Bachmann Antigenic Organization Principle

Why does a Virus-Like Particle induce neutralizing antibody titers 10- to 100-fold higher than an equal microgram mass of the identical protein administered as a soluble monomer? As established by Rolf Zinkernagel and Martin Bachmann, the vertebrate B-cell compartment evolved to discriminate self from viral non-self based on spatial epitope organization:

1. Optimal B-Cell Receptor (BCR) Cross-Linking Geometry: Few host cell-surface self-proteins are arranged in rigid crystalline arrays, so self-reactive B cells are not triggered by soluble monomers. Conversely, icosahedral and enveloped viral capsids display 60 to 360 identical epitopes spaced rigidly at 5 to 10 nm (50 to 100 Å)—a distance that matches the ~10- to 15-nm span between the two antigen-binding Fab arms of a single surface IgM or IgD B-cell receptor. When a VLP contacts a cognate naive B cell, it cross-links dozens of BCRs into dense microclusters within membrane lipid rafts, exceeding the signaling threshold for Syk and Bruton's tyrosine kinase (BTK) phosphorylation and cytosolic Ca²⁺ flux.
2. Size-Dependent Lymphatic Drainage (20–100 nm): Whereas soluble proteins (<10 nm) diffuse into blood capillaries or degrade locally, and large microparticles (>500 nm) remain trapped in the interstitial matrix until phagocytosed by peripheral macrophages, nanoparticles measuring 20 to 100 nm enter initial lymphatic vessels by convective interstitial flow and drain unimpeded into the subcapsular sinus of the lymph node within minutes.
3. Complement Fixation and Follicular Dendritic Cell (FDC) Retention: Repetitive surface carbohydrates and protein loops on VLPs bind pentameric natural IgM and mannose-binding lectin (MBL), activating the classical and lectin complement cascades to coat the particle in C1q, C3b, and C3d. Subcapsular sinus CD169⁺ macrophages capture these opsonized VLPs and shuttle them directly to Follicular Dendritic Cells (FDCs) via complement receptors CR1/CR2 (CD35/CD21). Unlike conventional dendritic cells that degrade internalized antigens within hours, non-degradative FDCs retain intact, native VLPs on their dendrites for weeks to months, continuously selecting high-affinity germinal center B-cell mutants and driving differentiation of long-lived bone marrow plasma cells.

Hepatitis B, HPV L1 Capsids, and Malaria Chimeric VLPs (RTS,S and R21)

• Hepatitis B Surface Antigen (HBsAg, 1986): Developed by Pablo Valenzuela, William Rutter, and Maurice Hilleman, cloning the HBV S gene into S. cerevisiae produced 24-kDa HBsAg polypeptides that spontaneously insert into the yeast endoplasmic reticulum membrane and bud into 22-nm spherical lipoprotein VLPs containing ~100 HBsAg monomers—becoming the first licensed recombinant human vaccine and the first anti-cancer vaccine by preventing chronic HBV-induced hepatocellular carcinoma.
• Human Papillomavirus (HPV) L1 VLPs: In 1991, Ian Frazer and Jian Zhou demonstrated that expressing the L1 major capsid protein of HPV alone in eukaryotic cells results in spontaneous self-assembly of five L1 monomers into a pentameric capsomere, and 72 pentamers (360 L1 monomers) into an empty 55-nm T = 7d icosahedral capsid structurally indistinguishable from authentic HPV virions. Because natural mucosal HPV infection sheds virions superficially without viremia, natural antibodies are weak; intramuscular immunization with adjuvanted HPV VLPs (Gardasil 9 covering types 6, 11, 16, 18, 31, 33, 45, 52, and 58) induces systemic IgG titers 10- to 100-fold higher than natural infection that transudate into the cervicovaginal mucus. Due to intense BCR cross-linking by the 360-mer array, even a single dose of HPV VLP vaccine maintains stable, plateauing neutralizing antibody titers for >15 years.
• Malaria Chimeric VLPs (RTS,S/AS01 and R21/Matrix-M): Because the pre-erythrocytic circumsporozoite protein (CSP) of Plasmodium falciparum does not self-assemble into particles, RTS,S (Mosquirix) fuses the central NANP B-cell repeat region and C-terminal T-cell domain of CSP to HBsAg, co-expressed in yeast with a 4-fold excess of unfused HBsAg monomers (so only ~20% of the VLP surface displays CSP, yielding ~36%–50% efficacy). To increase CSP epitope density, Adrian Hill and the Jenner Institute engineered R21/Matrix-M, expressed in Pichia pastoris such that every HBsAg monomer is fused to CSP (1:1 stoichiometry, 100% CSP display without unfused HBsAg dilution), achieving 75%–78% clinical efficacy in African children and receiving WHO recommendation in 2023.
• Computationally Designed Protein Nanoparticles (I53-50 and Ferritin): Pioneered by Neil King and David Baker, de novo computational protein design creates two-component icosahedral cages such as I53-50 (composed of 20 trimeric "A" subunits and 12 pentameric "B" subunits assembling into a 120-subunit 30-nm icosahedron). Fusing 60 copies of the SARS-CoV-2 Spike RBD to the I53-50A trimer produced SKYcovione (GBP510), the first clinically approved computationally designed protein nanoparticle vaccine (2022). Similarly, Pamela Bjorkman engineered 60-mer SpyCatcher003-mi3 nanoparticles co-displaying mosaic quartets or octets of heterotypic sarbecovirus RBDs: because adjacent RBDs on the same nanoparticle differ, only B cells recognizing conserved cross-reactive epitopes can bivalently cross-link their BCRs, focusing the germinal center response toward broad pan-coronavirus neutralization.

How can a single dose of the HPV L1 Virus-Like Particle vaccine confer stable antibody titers for over 15 years?

Unlike soluble monomeric proteins that induce transient plasmablast waves, the HPV L1 VLP presents 360 copies of the L1 capsid protein arranged in a rigid 72-pentamer icosahedral lattice with 5- to 10-nm epitope spacing. This dense, quasi-crystalline array cross-links dozens of B-cell receptors simultaneously and fixes complement C3d to remain trapped intact on follicular dendritic cells for months. The resulting prolonged germinal center reaction generates a high-affinity pool of bone-marrow-resident long-lived plasma cells whose antibody secretion plateaus at protective levels for more than 15 years without requiring booster doses.

Why did the R21/Matrix-M malaria vaccine achieve higher clinical efficacy than the earlier RTS,S/AS01 vaccine?

Both vaccines display the central NANP repeat and C-terminal domain of Plasmodium falciparum circumsporozoite protein (CSP) fused to the Hepatitis B surface antigen (HBsAg) VLP scaffold. However, RTS,S requires co-expression of 1 part CSP-HBsAg fusion protein with 4 parts unfused HBsAg monomer to assemble in Saccharomyces cerevisiae, meaning 80% of the particle surface consists of irrelevant HBsAg epitopes. R21, expressed in Pichia pastoris, self-assembles into VLPs composed solely of CSP-HBsAg fusion monomers (a 1:1 ratio with no free HBsAg), quadrupling the surface density of protective malaria CSP epitopes for B-cell receptor cross-linking.

How do mosaic RBD nanoparticles focus B-cell responses onto conserved pan-sarbecovirus epitopes?

On a homotypic nanoparticle displaying 60 identical SARS-CoV-2 RBDs, B cells targeting immunodominant, strain-specific variable loops can easily bind two adjacent identical RBDs with both Fab arms of their B-cell receptor (avidity advantage). On a mosaic nanoparticle randomly co-displaying 8 antigenically distinct sarbecovirus RBDs side-by-side, the probability that two adjacent RBDs belong to the same strain is low; consequently, only B cells whose receptors recognize conserved subdominant epitopes shared across all 8 sarbecoviruses can achieve bivalent avidity binding and dominate germinal center selection.

1. Bachmann, M. F., & Jennings, G. T. (2010). Vaccine delivery: A matter of size, geometry, kinetics and molecular patterns. Nature Reviews Immunology, 10(11), 787–796. https://doi.org/10.1038/nri2868
2. Lal, H., Cunningham, A. L., Godeaux, O., Chlibek, R., Diez-Domingo, J., Hwang, S. J., Levin, M. J., McElhaney, J. E., Poder, A., Puig-Barberà, J., Vesikari, T., Watanabe, D., Weckx, L., Zahaf, T., & Heineman, T. C. (2015). Efficacy of an adjuvanted herpes zoster subunit vaccine in older adults. New England Journal of Medicine, 372(22), 2087–2096. https://doi.org/10.1056/NEJMoa1501184
3. Datoo, M. S., Dicko, A., Tinto, H., Ouédraogo, J. B., Hamaluba, M., Olotu, A., Beaumont, E., Ramos Lopez, F., Natama, H. M., Weston, S., Chemba, M., Compaore, Y. D., Issiaka, D., Salou, D., Some, A.,urez, S., et al. (2024). Safety and efficacy of malaria vaccine candidate R21/Matrix-M in African children: A multicentre, double-blind, randomised, phase 3 trial. The Lancet, 403(10426), 533–544. https://doi.org/10.1016/S0140-6736(23)02511-402511-4)
4. Walls, A. C., Fiala, B., Schäfer, A., Wrenn, S., Pham, M. N., Murphy, M., Tse, L. V., Shehata, L., O'Connor, M. A., Chen, C., Navarro, M. J., Miranda, M. C., Prevent, M., Baric, R. S., King, N. P., & Veesler, D. (2020). Elicitation of potent neutralizing antibody responses by designed protein nanoparticle vaccines for SARS-CoV-2. Cell, 183(5), 1367–1382.e17. https://doi.org/10.1016/j.cell.2020.10.043