Characterization of humoral and cellular immunity induced by mRNA vaccines expressing norovirus VP1 proteins in mice and NHPs

Abstract

Noroviruses are a leading cause of acute gastroenteritis across all age groups, associated with ∼18% of diarrheal disease globally. Infections span a broad clinical spectrum, ranging from asymptomatic and self-limiting illnesses to hospitalizations and deaths. No preventive vaccines or targeted therapies are currently available. In this study, we designed vaccine candidates containing mRNAs encoding norovirus VP1 proteins of genotypes GI.1, GII.2, GII.3, GII.4, and GII.6, and evaluated their immunogenicity in mice and nonhuman primates (NHPs). In mice, all five mRNAs, dosed separately or together, elicited serum IgG antibodies that bound to norovirus virus-like particles (VLPs), serum antibodies that blocked the binding of VLPs to histo-blood group antigens (HBGAs), and norovirus-specific T cell responses. In NHPs, a pentavalent vaccine candidate induced strong humoral and cellular immune responses against each genotype included in the vaccine. Furthermore, compared to VLPs, mRNAs induced similar levels of humoral and cellular responses and comparable levels of durability, as measured by serum antibody titers. Our results indicate that a multivalent mRNA vaccine candidate encoding norovirus VP1 proteins is immunogenic in preclinical models and is a promising candidate to be evaluated in clinical trials.

Graphical abstract

Norovirus is a major cause of gastroenteritis, with no vaccine available. We developed a multivalent mRNA vaccine encoding norovirus VP1 proteins from five genotypes. This vaccine induced robust humoral and cellular immunity in mice and nonhuman primates, offering a promising strategy for broad-spectrum norovirus prevention adaptable to viral evolution.

Introduction

Noroviruses are a leading cause of sporadic cases and outbreaks of acute gastroenteritis, affecting all age groups and contributing to ∼18% of diarrheal disease globally.¹ Norovirus infections have a broad clinical spectrum that ranges from asymptomatic and self-limiting illnesses to severe outcomes, including hospitalizations and deaths. The vulnerable groups include younger children, older adults, and immunocompromised individuals. It is estimated that noroviruses cause ∼699 million illnesses and 219,000 deaths each year worldwide. Children younger than 5 years of age account for the majority of diseases (∼453 million) and a substantial number of deaths (∼70,000).² In developed countries with effective rotavirus vaccination coverage, norovirus has surpassed rotavirus as the most common cause of pediatric gastroenteritis requiring medical care.¹ Worldwide, the total economic impact of noroviruses is estimated to be $64.5 billion annually.² Despite the human and economic burden, there are currently no preventive vaccines or targeted therapies for norovirus. A safe and effective vaccine against norovirus acute gastroenteritis could significantly impact public health.

An effective norovirus vaccine likely requires a multivalent approach and periodic compositional updates because noroviruses are genetically and antigenically diverse, with multiple genogroups and genotypes cocirculating and evolving. Noroviruses are nonenveloped, single-stranded RNA viruses from the Caliciviridae family, classified into 10 genogroups (GI–GX) and 48 genotypes.³ GI and GII genogroups are responsible for up to 90% of the global outbreaks, with GII.4 being the most detected genotype (>50%) in all age groups.^4,5,6 GII.4 is also associated with more severe disease outcomes.⁷ New GII.4 variants emerged every 2–4 years between 2002 and 2012, but no new variants have caused widespread infection since 2012.⁸ Other GII genotypes, such as GII.2, GII.6, and GII.3, are also frequently detected.⁹ Studies have demonstrated limited cross-protective immunity among genotypes, indicating the need for a multivalent approach.^10,11,12,13

Developing a norovirus vaccine that adapts to changing epidemiology, including the emergence of new GII.4 sequence variants or the dominance of a new genotype, can benefit from the speed and flexibility offered by mRNA technology. Thus far, three US Food and Drug Administration-approved vaccines have utilized the mRNA platform, including Comirnaty and Spikevax for SARS-CoV-2 and mRESVIA for respiratory syncytial virus (RSV). The rapid development of the COVID-19 vaccine using mRNA technology and its regular composition updates demonstrate the adaptability of the platform.

Several lines of evidence support the selection of the major capsid protein VP1 as the vaccine antigen. The human norovirus genome contains three open reading frames, encoding six nonstructural proteins and two capsid proteins, VP1 and VP2. When VP1 is recombinantly expressed in vitro, 90 dimers self-assemble into virus-like particles (VLPs) that mimic native virions morphologically and antigenically.¹⁴ VP1 contains two major domains, the shell (S) domain and the protruding (P) arm, with the latter critical for norovirus infection by interacting with carbohydrates present on various histo-blood group antigens (HBGAs) in the host.¹⁵ Although immune correlates of protection against norovirus infection and disease are not definitively established, human challenge studies have identified HBGA-blocking antibodies as a potential correlate of protection.^16,17,18 These antibodies prevent VLPs and virions from binding to HBGA carbohydrates and interfere with the association of virions with their susceptible cells. VP1-based vaccine candidates, including VLPs, P particles, and recombinant adenoviruses expressing VP1s, elicit HBGA-blocking antibodies and stimulate other immunologic responses for potential protection against norovirus disease and infection, such as serum immunoglobulin G (IgG), serum IgA, fecal IgA, salivary IgA, and norovirus-specific IgG memory B cells.^19,20,21,22 Compared to humoral immunity, the cell-mediated immune response to norovirus is less known. Studies have shown that norovirus-specific T cells are present in healthy adults and children younger than 5 years of age,^23,24 and CD8 T cells with diverse differentiation states are present in human blood and tissues,²⁵ but the role of cell-mediated immunity in protecting against norovirus infections is still under investigation. During the preparation of our manuscript, data for a bivalent (GI.1 and GII.4) VP1-based norovirus mRNA vaccine candidate were published, demonstrating robust cellular and humoral responses in mice,²⁶ serving as a proof of concept.

In this study, we designed a multivalent mRNA vaccine encoding VP1 proteins from five genotypes—GII.4, GI.1, GII.2, GII.3, and GII.6—each encapsulated in lipid nanoparticles (LNPs). In mice, all five mRNAs, administered separately or together, elicited serum IgG antibodies that bound to VLPs, serum HBGA-blocking antibodies, and norovirus-specific T cell responses. In nonhuman primates (NHPs), a pentavalent vaccine candidate induced robust humoral and cellular immune responses against each genotype included in the vaccine. Furthermore, compared to a VLP-based vaccine, the mRNA vaccine induced similar characteristics of immune responses, including serum antibody durability. We thus present a multivalent mRNA vaccine candidate that is immunogenic in preclinical models and warrants further evaluation in clinical trials.

Results

Monovalent VP1 mRNA/LNP vaccines elicit norovirus-specific humoral responses in mice

Despite the considerable disease burden associated with norovirus, no vaccine is available. We aimed to develop a multivalent norovirus vaccine utilizing the mRNA platform. The safety and effectiveness of this platform have been demonstrated by the SARS-CoV-2 vaccines Comirnaty and Spikevax, as well as the recently approved RSV vaccine mRESVIA. All three vaccines express transmembrane glycoproteins with intact signal peptides, presenting antigens on cell surfaces. In contrast, the primary norovirus antigen, VP1, is expressed intracellularly and released from infected cells as assembled virions.²⁷ Given the difference in biology, we explored the impact of adding a signal peptide to VP1 on immunogenicity.

We utilized in vitro expression assays to assess the differences between VP1 and VP1 with a signal peptide first. We constructed three mRNAs encoding native GI.1 VP1 protein (No SP-GI.1), VP1 protein fused with the signal peptide of mouse Igκ at its N terminus (SP1-GI.1), or VP1 protein fused with the signal peptide of mouse Igκ and one-amino acid glutamic acid in between to facilitate cleavage (SP2-GI.1) (Figures 1A and S1A). Using a HiBiT tag fused at the C terminus of VP1, we first assessed VP1 expression by HiBiT real-time detection assay in three cell lines: HEK293 (human embryonic kidney), C2C12 (mouse myoblast), and SJCRH30 (human muscle) (Figure S1). Following mRNA transfection, real-time HiBiT detection assays revealed that the presence of the signal peptide resulted in only a slight increase in VP1 levels in the culture supernatant, particularly at early time points (Figures S1B–S1D, center panels). However, it caused a substantial decrease in total VP1 levels (Figures S1B–S1D, left panels). Quantification showed that while the signal peptide increased the secreted fraction (Figures S1B–S1D, right panels), it reduced the overall protein expression (Figure S1E). Notably, we did not observe such a dramatic decrease in western blot (Figure S1F); this could be a result of HiBiT-tagged SP1-VP1 trapped within the endomembrane system that the cytoplasmic LgBiT protein cannot access. Interestingly, the addition of signal peptide increased the apparent molecular weight of VP1 (Figure S1F). The increase cannot be explained by the molecular weight of the signal peptide alone and was reduced by endoglycosidase H (Endo H) treatment, indicating that the increase resulted from glycosylation. In contrast, Endo H treatment did not apparently shift the molecular weight of VP1 without signal peptides. These observations suggest that VP1 without signal peptides is translated by free ribosomes in the cytosol with minimal glycosylation. Conversely, signal peptides direct VP1 into the secretory pathway, where the endoplasmic reticulum and Golgi apparatus decorate VP1 with glycans. It is known that glycosylation can decrease immunogenicity by masking antigens²⁸; we then assessed the impact of signal peptides on immunogenicity in mice.

Figure 1.

Norovirus GI.1 VP1 mRNA vaccines elicit serum antibody titers in mice

(A) Balb/c mice (n = 8 per group) were injected intramuscularly at weeks 0 and 4 with lipid nanoparticle (LNP)-encapsulated mRNAs encoding GI.1 VP1 or GI.1 VP1 with two signal peptide designs at the N terminus (No SP-GI.1 mRNA/LNP, SP1-GI.1 mRNA/LNP, and SP2-GI.1 mRNA/LNP, respectively). Control groups included mice vaccinated with GI.1 virus-like particles (VLPs) formulated with amorphous aluminum hydroxyphosphate sulfate (AAHS) or empty LNPs. Doses are indicated in the table. Serum was collected at weeks 2 and 6 to measure vaccine-induced humoral responses. (B) Serum IgG was assayed for binding to GI.1 VLPs by ELISA with interpolated titers shown for each animal. The center bar indicates the geometric mean, and the error bars show the 95% confidence interval. The dotted line at 50 indicates the limit of detection for the assay. Samples below the limit of detection were given a placeholder titer of 25. ∗∗∗∗p < 0.0001 by ordinary one-way analysis of variance (ANOVA) with Tukey’s multiple comparisons. (C) Week 6 serum was assayed for its ability to block the binding of VLPs to porcine gastric mucin (PGM) in an HBGA-blocking assay. The 50% blocking titer (BT₅₀) was defined as the reciprocal of the serum dilution required to block 50% of binding as determined by a four-parameter sigmoidal curve fit. Measurements for each animal are indicated. The center bar indicates the geometric mean, and the error bars show the 95% confidence interval. The dotted line at 20 indicates the limit of detection for the assay. Samples below the limit of detection were given a placeholder titer of 10. ∗∗∗∗p < 0.0001 by two-sided unpaired t test with Welch’s correction. ∗∗∗p < 0.001 and ∗∗∗∗p < 0.0001 by Welch’s ANOVA with Dunnett’s T3 multiple comparisons.

To determine the immunogenicity difference, we injected BALB/c mice intramuscularly twice at weeks 0 and 4 with mRNAs encapsulated within LNPs at the 0.5- or 2-μg dose level and collected sera 2 weeks after each immunization at weeks 2 and 6. 10 μg of GI.1 VLPs formulated with amorphous aluminum hydroxyphosphate sulfate (AAHS) served as a positive control and allowed a comparison between the VLP- and mRNA-based vaccines. Empty LNPs without mRNAs were used as a negative control (Figure 1A).

We evaluated serum IgG antibody binding to GI.1 VLPs in an ELISA assay (Figure 1B) using sera from weeks 2 and 6 and functional antibodies that blocked the binding of GI.1 VLPs to HBGA carbohydrates in an HBGA blocking assay using sera from week 6 (Figure 1C). As expected, both assays showed that GI.1 VLP with AAHS elicited antibody titers, while the empty LNP did not induce any titers. All three mRNA/LNP vaccines elicited dose-dependent VLP binding and HBGA-blocking titers, with high-dose groups inducing higher titers than the corresponding low-dose groups. The second immunization boosted titers in all vaccination groups. Among the three mRNA/LNP vaccines tested, VP1 without signal peptide elicited the highest antibody titers in both assays. Compared to 10 μg VLP with AAHS, 2 μg mRNA encoding VP1 without signal peptide induced higher antibody titers of both types. The results demonstrate that an mRNA/LNP vaccine encoding native norovirus VP1 protein can stimulate robust antibody responses in mice. Thus, for subsequent experiments, the native VP1 sequence was used without a signal peptide.

Structural studies have shown that the HBGA binding site is formed at the interface of two P2 domains within assembled VP1 protein dimers.²⁹ The strong HBGA blocking titers observed following mRNA vaccination (Figure 1C) are consistent with the correct assembly of VP1 into VLPs. To further confirm VLP formation post-mRNA vaccination, we examined VLP production in vitro using both unformulated mRNAs (Figures S2A–S2D) and LNP-formulated mRNAs (Figures S2E–S2G). Consistent with a previously report,²⁶ transfection of mRNAs into Expi293F cells resulted in VLP production (Figures S2A–S2D). Immunoblot analysis detected VP1 protein expression in both supernatants and cell lysates as early as 6 h post-transfection (Figures S2A and S2B). Sucrose gradient ultracentrifugation demonstrated VLP formation with norovirus VLPs eluted near the center of a 15%–36% gradient (fractions 5–8 in Figure S2C). Purified VLPs, obtained by pooling these fractions, exhibited an icosahedral-like structure as visualized by electron microscopy (Figure S2D). Furthermore, we confirmed that LNP-encapsulated mRNA can generate VP1 protein, as demonstrated by western blot (Figure S2E), and produce VLPs, as evidenced by sucrose gradient fractionation (Figure S1F) and electron microscopy (Figure S1G) in human hepatocellular carcinoma HepG2 cells.

Multivalent VP1 mRNA/LNP vaccines elicit norovirus-specific humoral responses in mice

Besides GI.1, we considered designing vaccines against GII.4, the leading cause of infection in adults and children globally^4,5,6 and GII.2, GII.6, and GII.3.

An advantage of the mRNA platform is the ready inclusion of multiple antigens, covering and adapting to genotype changes; however, it is essential to understand the immunogenicity and potential immune interference of each component of multivalent vaccines. We thus designed a mouse study (Figure 2) with monovalent groups for five norovirus genotypes GI.1, GII.4, GII.2, GII.3, and GII.6 (groups 2 to 6), combinations of key genotypes GI.1 and GII.4 with GII.2, GII.3, and GII.6 (groups 7 to 13), and a group that received an equal amount of mRNA/LNPs of all five genotypes (group 14). Because lower titers of GII.4 compared to GI.1 were observed with a bivalent VLP vaccine candidate,¹⁹ we also included a group with a higher amount of GII.4 (2 μg) than other genotypes (1 μg) (group 15). We evaluated serum antibody binding titers (Figures 2B and S3), serum antibody HBGA blocking titers (Figures 2C and S4), and T cell responses (Figure 3) for all groups.

Figure 2.

Norovirus VP1 mRNA vaccines of multiple genotypes elicit serum antibody titers in mice

(A) Balb/c mice (n = 8 per group) were injected intramuscularly with mRNA vaccines indicated in the table at weeks 0 and 4. (B) Serum from each animal collected at weeks 2 and 6 was assayed for binding to VLPs of five genotypes by ELISA, with interpolated titers shown for each animal. The VLP used in the ELISA assay is indicated at the top of each graph. Homotypic results of monovalent and pentavalent groups are plotted. Other groups are shown in Figure S3. The center bar indicates the geometric mean, and the error bars show the 95% confidence interval. The dotted line indicates the limit of detection for the assay. (C) Week 6 serum from each animal was assayed for its ability to block the binding of VLPs to PGM in an HBGA-blocking assay. The BT₅₀ was defined as the reciprocal of the serum dilution required to block 50% of binding as determined by a four-parameter sigmoidal curve fit. The VLP used in the assay is indicated at the top of each graph. Homotypic results of monovalent and pentavalent groups are plotted. Other groups are shown in Figure S4. The center bar indicates the geometric mean, and the error bars show the 95% confidence interval. The dotted line indicates the limit of detection for the assay. ∗p < 0.05, ∗∗p < 0.005, and ∗∗∗p < 0.001 by ordinary one-way ANOVA with Tukey’s multiple comparisons or nonparametric Kruskal-Wallis test with Dunn’s multiple comparisons.

Figure 3.

Norovirus VP1 mRNA vaccines of multiple genotypes elicit cellular immune responses in mice

(A) Splenocytes from Balb/c mice (n = 8 per group) were collected at week 6 after mRNA/LNP vaccination at weeks 0 and 4. Whole splenocytes from each group were pooled, and the T cell cytokine responses to the five norovirus genotypes were evaluated by a fluorospot assay, specifically for IFN-γ, IL-2, and TNF-α. The number of spots per well representing cytokine-producing cells was quantified using an automated fluorospot reader, and the total spots per group are shown as the average counts of duplicate wells per condition. (B) Spots positive for two cytokines, representative of double-cytokine-producing cells, were quantified and shown as the average of duplicate wells per condition. (C) The number of triple-cytokine-producing cells is shown as average spot counts of duplicate wells per condition.

In general, antibody responses in mice were as expected. Homotypic antibody responses were comparable across groups. In some instances, we observed small genotype-specific differences in ELISA titers between monovalent and pentavalent groups (Figure 2B) that were not replicated in the functional blockage assay (Figure 2C). For GI.1 and GII.2, the pentavalent groups induced slightly higher binding titers at both weeks 2 and 6, while the differences were not detected in the blockage assay at week 6 (Figures 2B and 2C). For GII.6 and GII.3, binding and blocking titers of the pentavalent groups trended lower than those of monovalent groups at week 6. Pentavalent group 15, containing 2 μg GII.4 mRNA/LNP, elicited significantly higher GII.4 binding titers than the monovalent group containing 1 μg GII.4 mRNA/LNP, but the difference in blockage titer was not statistically significant. Notably, compared to the GII.3 monovalent group, group 15 elicited lower binding and blocking titer for GII.3 at week 6, the only trend that achieved statistical significance in both assays. Other multivalent (bi-, tri-, and quad-) groups also induced comparable homotypic titers at week 6 (Figures S3A and S4A).

We also assessed heterotypic antibody responses and detected cross-reactive antibody titers, especially GII cross-binding antibodies (Figure S3B). The multivalent groups elicited binding titers as high as 10⁵ to heterotypic VLPs (Figure S3B), but the blockage assay showed minimal cross-functional antibody titers (Figure S4B). The serology results indicate that the pentavalent mRNA vaccine groups are immunogenic in mice, with minimal immune interference among the tested genotypes.

Multivalent VP1 mRNA/LNP vaccines elicit norovirus-specific T cell immunity in mice

We collected splenocytes 2 weeks after the second immunization with mRNA/LNP vaccines to evaluate T cell cytokine production after in vitro peptide stimulation with GI.1, GII.2, GII.3, GII.4, or GII.6 in a fluorospot assay (Figure 3). We detected interferon-γ (IFN-γ), interleukin-2 (IL-2), and tumor necrosis factor α (TNF-α) production in response to all five genotypes, and the most robust responses were observed in multivalent vaccine groups after stimulation with peptide pools from GII.2, GII.3, and GII.6 (Figure 3A). We also noted cross-reactive responses to GII.2, GII.3, and GII.6 peptide stimulation from groups that received the monovalent GII.2, GII.3, and GII.6 mRNA/LNP vaccines. Vaccinated mice also presented polyfunctional T cells, as characterized by the capability of T cells to produce multiple cytokines simultaneously in response to peptide stimulation (Figures 3B and 3C). In particular, we detected higher double producers (IFN-γ and IL-2, IFN-γ and TNF-α) and triple producers (IFN-γ, IL-2, TNF-α) in response to GII.2, GII.3, and GII.6 peptides, compared to GI.1 and GII.4 peptides. Collectively, the data support genotype-specific, differentiated T cell responses, but the underlying reason(s) for these differences is not clear.

VP1 mRNA/LNP vaccines elicit norovirus-specific humoral responses in NHPs

We further evaluated the immunogenicity of norovirus VP1 mRNA vaccines in rhesus macaques. We injected mRNA/LNPs intramuscularly at weeks 0 and 4 and collected serum and peripheral blood mononuclear cells (PBMCs) at time points shown in Figure 4A. The pentavalent composition with an equal amount of mRNAs of each genotype was tested at two dose levels (10 and 40 μg/genotype). An empty LNP group and a GII.4 monovalent group (40 μg) served as controls.

Figure 4.

Norovirus VP1 mRNA vaccines elicit serum antibody titers in NHPs

(A) Rhesus monkeys (n = 4 per group) were injected intramuscularly with mRNA vaccines indicated in the table at weeks 0 and 4. Serum was collected to measure vaccine-induced humoral responses, and PBMCs were collected to measure cellular responses. (B) Serum from each animal was assayed for binding to VLPs by ELISA. Interpolated ELISA titers are shown, with the center bar indicating the geometric mean and error bars showing the 95% confidence interval. The limit of detection for the assay is 50. Samples below the limit of detection were given a placeholder titer of 25. (C) Serum from each animal was assayed for the ability to block the binding of VLPs to carbohydrates in an HBGA blocking assay. The BT₅₀ was defined as the reciprocal of the serum dilution required to block 50% of binding as determined by a four-parameter sigmoidal curve fit. The center bar indicates the geometric mean, and the error bars show the 95% confidence interval. The limit of detection for the assay is 20. Samples below the limit of detection were given a placeholder titer of 10.

All vaccines elicited strong binding and blockage titers (Figures 4B and 4C) that peaked around weeks 5 and 6 before slowly declining. The high-dose pentavalent group induced slightly higher but not statistically different titers against all genotypes compared to the low-dose group. The monovalent group containing 40 μg GII.4 elicited GII.4 titers similar to those of the high-dose group. The monovalent group also generated cross-binding ELISA titers against all other genotypes but no cross-blocking titers.

Notably, both serological readouts showed differences among genotypes. The differences in binding titers between the control and vaccinated animals were more pronounced for GI.1 and GII.4 than other genotypes (Figure 4B). Given that the peak binding titers of all genotypes are similar (around 10⁶), this discrepancy may be due to lower preexisting binding titers against GI.1 and GII.4 in rhesus monkeys. The immunogenicity assessed by the HBGA blocking assay also differed in magnitude for each genotype. Peak titers varied, and by week 12, GI.1 and GII.2 titers had decreased to levels comparable to those of the LNP control group, while GII.3, GII.4, and GII.6 titers maintained discernible differences. This difference could result from differing responses in NHPs or assay variations.

We next evaluated vaccine-elicited antibody responses in NHPs against genotypes not included in the vaccine—GI.3, GII.5, and GII.17—as well as multiple GII.4 strains (GII.4 Sydney, GII.4 San Francisco, and GII.4 Farmington Hills) (Figure S5). Consistent with previous reports in mice,²⁶ although the multivalent vaccine elicited cross-reactive ELISA binding antibodies against these heterologous genotypes (Figure S5A, left panel), the cross-reactive functional antibody titers were largely diminished (Figure S5A, right panel). Interestingly, we detected low but above background HBGA blocking titers against GII.5, indicating that heterotypic functional antibodies can be generated depending on the genotype. As expected, antibody responses showed significantly lower cross-reactivity with GII.4 variants other than GII.4 Sydney, reflecting the inclusion of the Sydney strain in the vaccine and the influence of amino acid sequence similarity on antigenic differences (Figures S5B and S5C).

VP1 mRNA/LNP vaccines elicit norovirus-specific T cell immunity in NHPs

We also evaluated the T cell responses to norovirus peptides in PBMCs collected from NHPs at baseline and weeks 4, 8, and 12 (Figure 5). Compared to baseline levels, we observed a noticeable increase in IFN-γ and IL-2 production in response to GII.2 and GII.3 peptides at week 4, particularly in the pentavalent vaccine groups (Figure 5). Cytokine production of T cells further increased in response to all five norovirus genotypes, reaching statistical significance for most groups by week 8. T cells from the pentavalent groups continued to show elevated IFN-γ production at week 12, while there was no discernible decrease in IL-2 production at week 12 compared to earlier time points (Figure 5). The T cell results are consistent with previous observations that PBMCs predominantly produce IFN-γ after norovirus challenge.^30,31

Figure 5.

Norovirus VP1 mRNA vaccines elicit cellular immune responses in NHPs

Rhesus monkeys (n = 4 per group) were immunized with LNP or a combination of mRNA/LNP vaccines, as indicated in Figure 4A. PBMCs were collected at weeks 0, 4, 8, and 12 and evaluated for T cell responses to the five norovirus genotypes by quantifying the production of IFN-γ and IL-2 via a fluorospot assay. Total spot counts per well representing cytokine-producing cells were quantified by an automated fluorospot reader and shown as the average spot counts of all animals in each treatment group. Dotted lines indicate the baseline spot counts as determined by the average counts of the DMSO negative controls for each cytokine. ∗p < 0.05 by nonparametric Kruskal-Wallis test with Dunn’s multiple comparisons.

mRNA/LNP vaccines elicit similar antibody and T cell responses compared to VLP vaccines in NHPs

To directly compare the two vaccine platforms, we administered either VLPs or mRNA/LNPs to NHPs intramuscularly twice at weeks 0 and 4 and collected serum and PBMCs at designated time points to assess immunogenicity (Figure 6A). We used 50 μg GII.4 VLPs with AAHS as an adjuvant or 50 μg mRNAs for each of the four genotypes (GII.2, GII.3, GII.4, and GII.6) to simulate potential human doses. We measured GII.4-specific antibody responses (Figures 6B and 6C) and observed that both binding and blocking titers elicited by the mRNA vaccine group trended higher than the VLP vaccine group. Notably, the decay kinetics of the antibody titers were similar, suggesting a comparable durability of antibody responses elicited by VLPs and mRNAs.

Figure 6.

Norovirus VP1 mRNA and VLP vaccines elicit similar immune responses in NHPs

(A) Rhesus monkeys (n = 4 per group) were injected intramuscularly with an mRNA vaccine or VLPs formulated with AAHS at weeks 0 and 4. Serum and PBMCs were collected to measure vaccine-induced humoral and cellular responses. (B) Serum from each animal was assayed for binding to GII.4 VLPs by ELISA. Interpolated ELISA titers are shown, with the center bar indicating the geometric mean and error bars showing the 95% confidence interval. (C) Serum from each animal was assayed for the ability to block the binding of GII.4 VLPs to PGM in an HBGA blocking assay. The BT₅₀ was defined as the reciprocal of the serum dilution required to block 50% of binding as determined by a four-parameter sigmoidal curve fit. The center bar indicates the geometric mean, and the error bars show the 95% confidence interval. (D) PBMCs from each animal were collected at weeks 0 and 6. T cell production of IFN-γ and IL-2 in response to stimulation with GII.4 was evaluated by a fluorospot assay. The spot counts, which represent the number of IFN-γ, IL-2, and double-producing cells, are shown as the average counts of all animals in each treatment group. ∗p < 0.05 by nonparametric Mann-Whitney test.

We also evaluated antibody responses against GII.3, a genotype included only in the multivalent mRNA vaccine but absent from the monovalent VLP vaccine. As expected, we observed functional antibody titers only in animals that received the multivalent mRNA vaccine (Figure S6, upper panels). Although we detected cross-reactive binding titers against GI.1, a genotype not present in either vaccine, GI.1-specific HBGA blocking antibody titers remained below the limit of detection (Figure S6, lower panels). Combined with the cross-reactive antibody responses shown in Figure S5, these findings underscore the necessity for a multivalent vaccine to effectively address norovirus diversity.

We used a fluorospot assay to assess GII.4-specific T cell responses. Both groups showed a considerable increase in cytokine production at week 6 compared to baseline, with comparable levels of IFN-γ and IL-2-producing T cells and polyfunctional T cells (Figure 6D).

Taken together, our findings indicate that mRNAs encoding norovirus VP1 proteins of multiple genotypes can stimulate robust humoral and cellular responses in mice and NHPs. Therefore, the pentavalent norovirus mRNA/LNP vaccine is a promising vaccine candidate for further evaluation in human trials.

Discussion

It has been challenging to develop norovirus vaccines partially due to the limited evidence of heterotypic protection and the evolving epidemiology. The mRNA vaccine platform offers the flexibility to rapidly update antigens to address emerging variants. The pentavalent vaccine consisting of mRNAs encoding native VP1 of five norovirus genotypes described in this paper elicited strong humoral and cellular responses in mice and NHPs. It demonstrated the application of the mRNA platform to target a nonenveloped virus and the advantage of a multivalent vaccine. This vaccine is a promising candidate to reduce the burden of acute gastroenteritis caused by norovirus infection.

Immune interference is a risk with multivalent vaccines and must be closely monitored for norovirus vaccines, as a balanced response against all components is desired. Previously, a bivalent VLP-based norovirus vaccine candidate elicited a more robust response to GI.1 than to GII.4c in humans,³² and a 3:1 ratio of GII.4c to GI.1 was selected for further clinical development. To determine whether a similar strategy would be necessary for the pentavalent mRNA vaccine, we explored the addition of a higher GII.4 dose level in mouse studies but did not observe a clear advantage. Pentavalent vaccines with equal or higher amounts of GII.4 than other genotypes elicited similar antibody responses compared to monovalent vaccines with one exception: a higher dose of GII.4 dampened the immune responses against GII.3, resulting in minor but statistically significant decreases in titers detected in both ELISA and HBGA assays. Moreover, the total mRNA/LNP level influences tolerability and reactogenicity in human, and increasing the amount of GII.4 mRNA may require a reduction in the amount of other components. Consequently, we decided to move the pentavalent candidate with an equal amount of mRNA forward into NHP studies.

A previous study²⁶ demonstrated that a bivalent mRNA-LNP vaccine targeting GI.1 and GII.4 genotypes induced potent neutralizing antibodies and robust cellular immunity in mice, with functional protection confirmed in human intestinal enteroids. The current multivalent candidate expands coverage to five genotypes, maintaining strong immune responses against all targets with minimal immune interference in both mice and NHPs. Together, these studies confirm the flexibility of the mRNA platform and highlight its potential for norovirus vaccine development. Since functional cross-blocking antibodies tend to be genotype specific, the pentavalent approach represents a major advance toward broad-spectrum norovirus prevention.

Serum HBGA blocking antibody has been identified as a potential correlate of protection against norovirus infection and illness³³; however, HBGA assays are not standardized in the field. The assay components (e.g., HBGA carbohydrates, VLPs, genotype-specific detection antibodies) and the experimental conditions (e.g., incubation time, amount of VLP used, binding specificity of the detection antibodies) vary widely among labs and are critical for determining the blocking titer. As a result, the magnitude of HBGA blocking titers cannot be compared between labs or genotypes. These values must be considered relative to other parameters, such as serum IgG binding titers, to better evaluate vaccine response. For example, in the NHP study shown in Figure 4, the pentavalent vaccine elicited similar serum IgG ELISA binding titer patterns against GII.2 and GII.3, with comparable levels of preexisting titers and magnitudes of titer change. In contrast, the HBGA titers were noticeably different, with the magnitude of the titer change for GII.3 (peak titer ∼10⁴) appearing higher than GII.2 (peak titer <10³). It is possible that differences in the sensitivity of these two HBGA assays could confound the interpretation of vaccine responses such as durability, as measured by titers at week 12, as well as the previously noted concern about immune interference. In humans, preexisting immunity to different genotypes could also impact immunogenicity and efficacy. Ultimately, the magnitude and balance of the immune response to this pentavalent mRNA vaccine must be measured in clinical trials and interpreted through comparison to efficacy data.

The requirement for durability of a norovirus vaccine should be informed by clinical outcome data. Because approximately 70% of pediatric norovirus cases occur between 6 and 23 months of age,³⁴ a vaccine with 1- or 2-year durability may be sufficient to alleviate norovirus disease burdens significantly. A memory probe 1 year after immunization with a bivalent (GI.1 and GII.4c) VLP-based vaccine candidate elicited similar levels of GII.4c HBGA antibodies in the vaccine group and the placebo group, raising questions about the durability of this vaccine image.²¹ A comparison of the pentavalent mRNA vaccine and an adjuvanted, monovalent (GII.4) VLP vaccine in NHPs demonstrated similar antibody peak levels and decay kinetics, suggesting similar durability. Nevertheless, it is unclear how T and B cell recall responses will differ after VLP and mRNA vaccinations, the degree to which these biomarkers will translate from NHPs to humans, and whether any of these biomarkers will influence the severity of norovirus diseases.

The durability of a norovirus vaccine will also hinge on the frequency of emergence of new genotypes with unique antigenic presentation. mRNA-based vaccines can be readily updated to accommodate viral antigenic evolution, as demonstrated by the clinical experience with vaccines for SARS-CoV2.^35,36 More durable immune responses may be addressable through the use of adjuvants. VLP-based vaccines could benefit from being formulated with modern, potent adjuvants. The advantage of the addition of adjuvants to mRNA vaccines has not been demonstrated clinically, but data from animal studies are encouraging. Modifications to the LNP and the inclusion of mRNA-encoded molecular adjuvants have enhanced vaccine responses.^37,38 Further investigation is warranted to determine whether different adjuvant strategies can improve the magnitude, range, and/or duration of the immunogenicity to this pentavalent norovirus mRNA vaccine.

Materials and methods

Vaccine production and formulation

mRNAs encoding norovirus VP1 of genotypes GI.1, GII.2, GII.3, GII.4 and GII.6 (GenBank: NP_056821, YP_009518839, QJF54133, QBW96040, and AWT08315) were generated by Trilink with N1-methyl-pseudouridine triphosphate modification and clean-cap. LNPs encapsulating mRNAs were prepared by the rapid precipitation process as previously described.³⁹ The lipid components of the LNP comprised an asymmetric ionizable amino lipid, 1,2-distearoyl-sn-glycero-3-phosphocholine, cholesterol, and poly(ethylene glycol)2000-dimyristoylglycerol in a molar ratio of 58:30:10:2, respectively.

VLP generation and purification

VP1 protein sequences were mammalian codon optimized and subcloned into a eukaryotic expression vector under the control of a cytomegalovirus promotor. GII.2 and GII.3 contain amino acid mutations designed to increase VLP stability (data not shown). In addition to GI.1, GII.2, GII.3, GII.4 and GII.6, other VLPs used were GI.3, GII.5, GII.17, GII.4 Sydney, GII.4 San Francisco, and GII.4 Farmington Hills (GenBank: QIC52023, QNS37108, AWW21424, AGJ52172, WKD84242, and AFN70960). Plasmids were transiently transfected into Expi 293F cells (Thermo Fisher) using Expifectamine (Thermo Fisher) following the manufacturer’s recommended protocol. Bulk cell cultures were harvested 72 h post-transfection and frozen at −70°C. For GI.1 and GII.4 VLPs, the frozen cell culture was thawed, adjusted to pH 4.0 with citric acid, and centrifuged at 5,000 × g for 10 min. The supernatant was clarified through a Sartopure GF+ (Sartorius) depth filter and purified by cation exchange chromatography (CEX) using a Sartobind S membrane (Sartorius). The VLPs were eluted from the CEX membrane in a neutral pH buffer and filtered through a 0.2-μm polyvinylidene fluoride (PVDF) membrane (Millipore). The filtered CEX eluate was further purified by size-exclusion chromatography (SEC) using a Sepharose 6 FF (Cytiva) column. Selected SEC fractions were pooled, concentrated, and formulated with cryoprotectant and filtered through a 0.2-μm PVDF membrane (Millipore). For GII.2, GII.3, and GII.6 VLPs, the frozen cell culture was thawed and clarified through a Sartopure GF+ (Sartorius) depth filter. The clarified VLPs were precipitated using 1.5 M ammonium sulfate (AS) and pelleted by centrifugation at 5,000 × g for 10 min. The pellet was resuspended in an anion exchange chromatography (AEX) loading buffer and filtered through a 0.2-μm PVDF membrane (Millipore). The filtered AS precipitation product was purified by AEX using a Sartobind Q membrane (Sartorius), and the VLPs were eluted in an elevated ionic strength buffer. The AEX eluates for GII.2, GII.3, and GII.6 VLPs were purified using the same SEC, formulation, and filtration procedure as described for GI.1 and GII.4 VLPs. All purified VLPs were frozen and stored at −70°C. The purified VLP concentrations were about 1 mg/mL as measured by the Bradford assay (Thermo Fisher) using the manufacturer’s recommended protocol.

Signal peptide evaluation in vitro

SJCRH30 or C2C12 cells were seeded in a 24-well plate and transfected with mRNAs encoding the indicated GI.1-VP1 sequences using Lipofectamine MessengerMax (Life Technologies). Cells were directly lysed 24 h post-transfection in 100 μL 1× NuPAGE LDS sample buffer (Thermo Fisher) and denatured by heating at 70°C for 10 min. Samples were subsequently cooled, and two 40-μL aliquots were taken. One was treated with 500 U (1 μL) of Endo H (NEB, catalog no. P0702) for 1 h at 37°C; the other received 1 μL sterile water and was incubated as above. After Endo H treatment, samples were again heated to 70°C and separated on a 4%–12% NuPAGE Bis-Tris protein gel (Thermo Fisher). Samples were blotted to a nitrocellulose membrane, which was subsequently probed using the Nano-Glo HiBiT blotting kit (Promega, catalog no. N2410) to detect the VP1 constructs and anti-HSP90 (Cell Signaling Technology, catalog no. 5174) to control for equal loading.

For the real-time NanoBit detection assay, Lipofectamine MessengerMAX (Life Technologies) was diluted in OptiMEM (minimum essential medium) to the desired working concentration and dispensed onto 384-well white assay plates (Greiner, catalog no. 781080). A source plate (Labcyte, catalog no. P-0200) containing serial dilutions of the indicated mRNAs was prepared using the Bravo liquid handler (Agilent), and a 10-point 2-fold dose-titration of each mRNA was dispensed onto the assay plate using Echo (Labcyte). After a 10 min incubation, HEK293-LgBit, SJCRH30-LgBit, and C2C12-LgBit cells or the corresponding parental cells without LgBit were added per well, followed by 20 μM Endurazine (Promega), an extended time-released live cell substrate. Total HiBit or extracellular HiBit detection was conducted in the LgBit cell line or in the presence of LgBit protein (Promega), respectively, as recommended by the manufacturer in the parental cell line. Luminescence was measured continuously at 1 h intervals for 48 h on the Tecan Spark 20M set to 37°C, 5% CO₂. Cell lines with constitutive expression of LgBit were generated using lentiviral-mediated integration.

mRNA and mRNA/LNP transfection

Expi293F cells (Thermo Fisher) were cultured in suspension with Expi293 expression media (Thermo Fisher) at 37°C with 8% CO₂, shaking at 110 rpm. Cells were seeded in 125-mL nonbaffled shaker flasks at a density of 2.2 × 10⁶ cells/mL in 30 mL for transfection. Transfections were performed using the ExpiFectamine 293 transfection kit (Thermo Fisher) according to the manufacturer’s recommendations, with 30 μg mRNA and 80 μL of reagent in a total volume of 3 mL OptiMEM media. At 18 h post-transfection, cells were supplemented with ExpiFectamine 293 Transfection Enhancers (Thermo Fisher). Samples were collected by withdrawing 5 mL suspended cells at the indicated time points (6, 24, 48, and 72 h post-transfection). From each sample, 1 mL was centrifuged at 13,000 × g for 5 min to separate supernatant and cell pellet. The pellet was resuspended in 350 μL lysis buffer (0.5% Triton X-100, 40 mM Tris-HCl pH 7.4, 120 mM NaCl, and 1× protease inhibitor EDTA-free), and lysed at 4°C for 30 min. Lysates were clarified through centrifugation at 14,000 × g for 15 min at 4°C. Clarified supernatants and lysates were used for immunoblot analysis of pre-purification samples. The remaining 4 mL of harvest underwent two freeze-thaw cycles prior to sucrose gradient purification. The total protein concentration was determined using the Pierce Bradford Plus Protein assay kit (Thermo Scientific).

HepG2 cells were maintained in Eagle’s MEM (EMEM) (American Type Culture Collection) supplemented with 10% fetal bovine serum (FBS) and cultured at 37°C with 5% CO₂. Cells were seeded in T75 vented flasks in 12 mL media at 8 × 10⁵ cells/mL. After 24 h, media was removed, and the cells were overlaid with 3 μg/mL LNP-encapsulated GII.4 VP1 mRNA in 3 mL OptiMEM media (Thermo Fisher). Cells were incubated for 1 h at 37°C 5% CO₂, after which fresh EMEM 10% FBS media was added. Cells were harvested after 24 h by scraping, and VLPs were released via two freeze-thaw cycles followed by centrifugation at 3,000 × g for 15 min. Total protein concentration was determined using the Pierce Bradford Plus Protein assay kit (Thermo Scientific).

Sucrose gradient purification

A 16-mL sucrose gradient (15%–36% w/v) was prepared using a Hoefer SG15 gradient maker and a peristaltic pump at 3.6 mL/min. Harvested samples were concentrated to 1 mL using a 100 kDa molecular weight cutoff (MWCO) Amicon Ultra centrifugal filter (Millipore) and overlayed onto the gradient. Samples were ultracentrifuged at 150,000 × g for 3 h at 4°C. Following centrifugation, samples were fractionated into 12 aliquots, 1.34 mL, and analyzed by capillary immunoblotting. Fractions containing purified VLPs (fractions 5–8) were pooled and dialyzed overnight against storage buffer (25 mM Tris pH 7.5, 150 mM NaCl, 0.02% PS-80, 5% sucrose) using a 10-kDa MWCO Slide-A-Lyzer G3 dialysis cassette (Thermo Scientific). Protein concentration was measured using the Pierce Bradford Plus Protein assay kit (Thermo Scientific).

Capillary western blot

Capillary western blots were performed using the Protein Simple Jess Automated Western Blot system (Bio-Techne). Samples were diluted in 2× fluorescent master mix, combined with 5× DTT (final 1×) and sample buffer, and denatured at 95°C for 5 min. Samples were separated and analyzed using a chemiluminescent 12–230 kDa separation module (Protein Simple) with the provided biotinylated ladder. VP1 was detected using a cocktail of anti-norovirus capsid VP1 antibodies (MyBioSource, catalog no. MBS832466, and Abcam, catalog no. ab272687, 1:1,000 dilution). Anti-actin antibody (R&D Systems, catalog no. MAB8929) was included as a loading control using the RePlex module (Protein Simple). Secondary detection used the anti-mouse detection module (Protein Simple). All procedures followed the manufacturer’s protocol. Data were analyzed using Compass for SW software (version 6.2.0).

Negative staining electron microscopy and imaging analysis

Transmission electron microscopy with negative staining was performed by the Electron Microscopy Core Facility at Alpha Nano Tech LLC. Purified VLPs in storage buffer were prepared for imaging. Formvar/carbon-supported copper grids were glow-discharged immediately prior to sample addition. Grids were floated on a drop of sample for 5 min and then washed twice by floating on filtered water for 30 s each time. Excess liquid was blotted with Whatman filter paper sheets. Grids were then placed in 2% uranyl acetate for 30 s, and the excess was removed by blotting until dry. Samples were then imaged on a Jeol JEM-1400 electron microscope with an AMT NanoSprint 61 ActiveVu camera. An operating voltage of 80 kV was used for image acquisition. Image processing was performed using ImageJ software (version 1.47t).

Serum IgG ELISA assays

Maxisorp 384-well plates were coated with 50 ng/well of VLPs diluted in phosphate-buffered saline (PBS). After overnight incubation at 4°C, assay plates were washed six times with PBS-T, blocked with 3% nonfat powdered dry milk in PBS with 0.05% Tween 20 (NFPDM PBS-T) for 30 min, and washed six times with PBS-T again. Sera was diluted 10 times in 4-fold serial dilutions in 3% NFPDM PBS-T, transferred to precoated assay plates, and incubated for 2 h in a humidified incubator (21°C, 80% relative humidity). Unbound antibodies were removed by washing with PBS-T, and goat anti-mouse IgG (Fc)-horseradish peroxidase (HRP) (1:10,000 in 3% NFPDM PBS-T) (Jackson ImmunoResearch) was added to assay plates and incubated for 1 h. After six washes with PBS-T, luminescent signals from plates were developed with West Pico PLUS Chemiluminescent Substrate (Thermo Fisher) and read on an EnVision plate reader (PerkinElmer). An interpolated endpoint titer was calculated for each serum sample using the luminescence relative light unit (RLU) values and the following formula: (starting fold dilution of the sample/series dilution factor) × (series dilution factor^t), where t = x − [(cutoff − L)/(H − L)]. The cutoff value was designated as 50,000. H is the high well RLU value (the RLU value of the first titration point above 50,000), L is the low well RLU value (the RLU of the first titration point below 50,000), and x is the low well number (the number in the titration series of L where the first dilution in the titration series was 1 and the highest serum dilution of the titration series was 10). Samples without a dilution that crossed the threshold were given a placeholder titer of half of the initial starting fold dilution, or “25.”

HBGA blocking assays

Maxisorp 384-well assay plates were coated with human saliva (Precision for Medicine) diluted 1:500 in Dulbecco’s PBS or 10 μg/mL porcine gastric mucin (PGM) in bicarbonate buffer with 34.8 mM NaHCO₃ and 15 mM Na₂CO₃, incubated overnight at 4°C, washed six times with PBS-T, and blocked with 5% NFPDM PBS-T for 30 min. Ten 2-fold serial dilutions of animal study sera were prepared in 5% NFPDM PBS-T in a 384-well plate (Greiner) with a starting dilution of 1:10. VLPs were prepared at 1 μg/mL, mixed with diluted animal study sera, and incubated for 1 h before being added to assay plates. VLP mouse sera premix was incubated overnight in a humidified incubator (4°C, 80% relative humidity) and transferred to assay plates on the assay day. VLP NHP sera premix was incubated for 15 min in a humidified incubator (21°C, 80% relative humidity) and transferred to the assay plates. Assay plates were then washed six times with PBS-T, incubated with genotype-specific polyclonal rabbit sera (1:200–1:3,000 in 5% NFPDM PBS-T depending on the genotype) for 1 h, washed six times with PBS-T, incubated with goat anti-rabbit-HRP (Fc)-IgG (1:5,000 in 5% NFPDM PBS-T) (Jackson ImmunoResearch) for 1 h, and washed six times with PBS-T. Luminescent signals were developed for 15 min at room temperature with a luminescent substrate (West Pico PLUS, Thermo Scientific, Pierce). Blocking titer (BT)₅₀ values were defined as the titer at which luminescence readings were 50% of the positive control. A value of 10 was assigned to samples with a BT₅₀ less than the starting dilution of 20.

Cell-mediated immunity assays

To evaluate the cytokine production of norovirus-specific T cells after immunization in mice and NHPs, mouse-specific or NHP-specific fluorospot assay kits (Mabtech), which can simultaneously detect the production of IFN-γ, IL-2, and TNF-α, were used. The manufacturer’s protocol for each kit was followed except for a few modifications. For the NHP fluorospot assay, frozen NHP PBMCs were initially thawed, and cells were allowed to rest overnight in R10 medium (RPMI 1640 + 10% FBS, 2-mercaptoethanol, l-glutamine, HEPES, MEM sodium pyruvate, and penicillin/streptomycin) at 37°C, 5% CO₂. After overnight culture, cells were seeded at 2.5 × 10⁵ cells/well in antibody-coated, 96-well plates (kit component). For the mouse fluorospot assay, fresh splenocytes were seeded directly in the fluorospot plates at 2.5 × 10⁵ cells/well density in R10 medium. NHP and mouse T cells were activated with 0.1 μg/mL anti-CD28 (kit component) and 1.25 μg/mL peptide pools composed of 15-mer sequences with 11 amino acid overlap covering the VP1 of norovirus genotypes GI.1, GII.2, GII.3, GII.4, or GII.6 (JPT). After stimulation for 20–24 h at 37°C, 5% CO₂, cells were subsequently washed five times with PBS (GIBCO), and wells were incubated for 2 h with primary antibodies (kit component) against the cytokines of interest, followed by a 1 h incubation with the appropriate fluorophore-conjugated secondary antibodies (kit component). The wells were incubated next with the fluorophore enhancer (kit component) as instructed by the manufacturer, and the plates were washed and dried overnight. The spots in each well were analyzed and quantified using Iris (Mabtech), an automated fluorospot reader equipped with the appropriate filters for the fluorophores used.

Rabbit polyclonal sera generation

The Institutional Animal Care and Use Committee at Merck & Co., Inc., Rahway, NJ, USA approved the rabbit studies. Three female New Zealand white rabbits (∼4 months old at receipt) from The Jackson Laboratory were vaccinated per genotype. Animals were inoculated three times (2 weeks between doses) intramuscularly in the quadriceps bilaterally. Inoculations comprised 50 μg selected VLP and 225 μg AAHS in 250 μL total volume of PBS. Blood draws were performed at weeks 1, 3, 5, and 6 after the first injection. For the first three time points, 2 mL blood was collected from the ear artery into BD serum separator tubes and left at room temperature until clotted. Tubes were then spun at 4,000 × g for 10 min, and serum was aliquoted into 2-mL tubes for analysis. The final blood collection was a terminal bleed. Animals were deeply anesthetized and had blood collected via intracardiac puncture. Blood from this time point was used as a reagent in HBGA-blocking assays.

Mouse immunogenicity studies

The Institutional Animal Care and Use Committee at Merck & Co., Inc., Rahway, NJ, USA approved the mouse studies. BALB/c mice (Charles River Laboratories) aged 6–8 weeks were immunized intramuscularly with VLPs or candidate mRNA/LNP vaccines. Blood was drawn for serological assays 2 weeks following each immunization, as indicated in Figure 1A. At week 6, all mice were euthanized, and cardiac puncture was used to collect blood, followed by the removal of spleens. Blood was collected into Sarstedt serum separator tubes (Sarstedt AG KG Microvette 500 Z-Gel, catalog no. 20.1344), which were left at room temperature until clotted. The blood was then centrifuged at 10,000 × g for 5 min to separate serum, which was aliquoted into a 96-well plate and stored at 4°C until it was used for assays. Spleens from four animals of a single group were harvested. Single-cell suspension of splenocytes was generated via mechanical dissociation of mouse spleens.

NHP immunogenicity studies

All NHP studies were approved by the Institutional Animal Care and Use Committee at Merck & Co., Inc., Rahway, NJ, USA, and the New Iberia Research Center (NIRC) at the University of Louisiana at Lafayette. Only animals confirmed to be in good health via veterinary physical examination within 2 months prior to study initiation were eligible. Food was withheld on the morning of scheduled vaccinations and blood collections to ensure a minimum fasting period of 2–3 h prior to sedation. The 8773 Teklad NIB Diet Monkey Chow (Harlan) or an equivalent diet was provided daily in amounts appropriate for the size of the animal. Tap water was provided ad libitum via an automatic watering device. No contaminants known to interfere with study outcomes were present in the food or water. Animals were housed according to NIRC standards at animal biosafety level 2 throughout the study. Enrichment devices were provided for manipulating and foraging opportunities, enabling species-specific behaviors. Small quantities of novel fruits, nuts, or other foodstuffs were also offered for enrichment purposes, in addition to manipulanda. Detailed information on the age, sex, and weight of each animal is provided in Table S1. Rhesus macaques were vaccinated intramuscularly with 0.5 mL VLPs or mRNA/LNPs in the right deltoid. Serum and PBMCs were collected at the time points indicated in Figures 4A and 6A.

Statistical analysis

Data visualization and statistical analyses were performed using GraphPad Prism, version 10. For mouse studies, normality was assessed using the Shapiro-Wilk test and a quantile-quantile plot, while homogeneity of variances was evaluated by Bartlett’s test. When data were normally distributed with equal variances, one-way analysis of variance (ANOVA) with Tukey’s multiple comparisons was performed. Welch’s ANOVA with Dunnett’s T3 multiple comparisons was used when variances were unequal. Kruskal-Wallis with Dunn’s multiple comparisons was performed for non-normally distributed data. Due to the small sample sizes in NHP studies, data were analyzed using the nonparametric Mann-Whitney test for two groups or the nonparametric Kruskal-Wallis test with Dunn’s multiple comparisons for multiple groups.

Data and code availability

Data generated in the study are included in this published article and its supplemental information files. Noncommercial reagents and materials are available upon request.

Acknowledgments

The authors would like to acknowledge Michael Citron, Daniel Freed, and the New Iberia Research Center for conducting the NHP studies and our colleagues in laboratory animal resources for assistance with the mouse and rabbit studies. We also thank Qiansheng Zhu for valuable input on statistical analysis and Lisa Plitnick for her critical review of this manuscript. The study was funded by Merck & Co., Inc., Rahway, NJ, USA.

Author contributions

J.W., A.J.B., and L.Z. conceived the concept of the study. J.W., K.B., L.A., S.L., J.G., P.G., J.M.R.B., J.D.G., C.W., A.S., C.D., U.J., E.G.-F., L.A.D., G.G., A.L., S.R., C.D.M.F., P.S., A.F., J.G., A.P., Z.W., G.O., and L.A. performed the experiments and analyzed the data. J.W. drafted the manuscript and supervised all other activities. J.W., D.M., P.K., Y.L., A.J.B., and L.Z. were responsible for project administration. All authors have read and approved the manuscript.

Declaration of interests

The authors declare the following financial interests/personal relationships that may be considered potential competing interests: all authors were employees of Merck & Co., Inc., Rahway, NJ, USA, during the study and may own stock or hold stock options in these companies. L.A., A.J.B., A.F., J.G., and J.W. have filed a Patent Cooperation Treaty international patent application, “Polynucleotides Encoding Norovirus VP1 Antigens and Uses Thereof.”