Delayed reinforcement of costimulation improves the efficacy of mRNA vaccines in mice

Abstract

mRNA vaccines have demonstrated efficacy during the COVID-19 pandemic and are now being investigated for multiple diseases. However, concerns linger about the durability of immune responses, and the high incidence of breakthrough infections among vaccinated individuals highlights the need for improved mRNA vaccines. In this study, we investigated the effects of reinforcing costimulation via 4-1BB, a member of the TNF receptor superfamily, on immune responses elicited by mRNA vaccines. We first immunized mice with mRNA vaccines, followed by treatment with 4-1BB costimulatory antibodies to reinforce the 4-1BB pathway at different time points after vaccination. Consistent with prior studies, reinforcing 4-1BB costimulation on the day of vaccination did not result in a substantial improvement in vaccine responses. However, reinforcing 4-1BB costimulation on day 4 after vaccination, when 4-1BB expression levels were highest, resulted in a profound improvement in CD8+ T cell responses associated with enhanced protection against pathogen challenges. A similar clinical benefit was observed in a therapeutic cancer vaccine model. We also report time-dependent effects with OX40, another costimulatory molecule of the TNF receptor superfamily. These findings demonstrate that delayed reinforcement of costimulation may exert an immunologic benefit, providing insights for the development of more effective mRNA vaccines for infectious diseases and cancer.

Keywords: Adaptive immunity; Immunology.

Figure 1. Reinforcing 4-1BB costimulation on day 4 after vaccination increases the number and durability of CD8⁺ T cell responses.

(A) Experimental outline for evaluating whether treatment with α4-1BB on day 4 improves immune responses. Mice were immunized with 3 μg of an mRNA-spike vaccine followed by treatment with 50 μg of α4-1BB or control antibodies on day 4. (B) Summary of virus-specific CD8⁺ T cells. (C) Representative FACS plots of virus-specific CD8⁺ T cells. Data are from PBMCs. K^bVL8 (shown in the y axis) is an MHC I tetramer used to detect SARS-CoV-2 spike–specific CD8⁺ T cells. Data are from 1 experiment, n = 4–5 per group/experiment; experiment was performed twice with similar results. Indicated P value in B was calculated with the Mann-Whitney test at the last time point.

Figure 2. CD8⁺ T cell subset differentiation after reinforcing 4-1BB costimulation.

Experimental outline was similar to that in Figure 1A. On day 7 after vaccination, splenic CD8⁺ T cells were MACS sorted. Subsequently, live CD8⁺CD44⁺K^bVL8 tetramer⁺ cells were FACS-purified to approximately 99% purity and used for bulk RNA-seq. (A) PCA shows transcriptional clustering. (B) Heatmap showing row-standardized expression of selected proliferation and apoptotic genes. (C) Heatmap showing row-standardized expression of selected cell cycle (top) and kinesins (bottom) genes. (D) Heatmap showing row-standardized expression of selected activation genes. (E) Heatmap showing row-standardized expression of selected effector genes. (F) GSEA plot showing enrichment of effector genes. (G) Validation of gene expression results at the protein level. Representative FACS plots showing the frequencies of virus-specific CD8⁺ T cells (K^bVL8⁺) that differentiate into effector, effector memory, and central memory T cell subsets. (H) Pie diagrams showing CD8⁺ T cell subsets. (I–K) Numbers of central memory, effector memory, and effector CD8⁺ T cells. All data are from tetramer⁺ (K^bVL8⁺) cells from spleen. RNA-seq data are from 1 experiment, with n = 4 per group. Data in panel H are from 1 representative experiment, with n = 4 per group; the experiment was performed twice with similar results. All other data are from 2 experiments, with n = 4–5 per group/experiment. Indicated P values in I–K were calculated by the Mann-Whitney test.

Figure 3. Generalizability to other mRNA vaccines.

Mice were immunized with 3 μg of each respective mRNA vaccine followed by treatment with 50 μg of α4-1BB or control antibodies on day 4. (A) Summary of LCMV-specific CD8⁺ T cell responses. (B) Representative FACS plots of LCMV-specific CD8⁺ T cells. (C) Pie diagrams showing CD8⁺ T cell subsets (gated on LCMV-specific CD8⁺ T cells). (D) Summary of OC43 spike–specific CD8⁺ T cell responses. (E) Summary of HIV env–specific CD8⁺ T cell responses. (F) Summary of OVA-specific CD8⁺ T cell responses. Data from A–C and F are after tetramer staining; data from D and E are after intracellular cytokine stimulation using overlapping peptide pools (IFN-γ⁺). Data from A–F are from day 14 after vaccination, and are from 2 experiments, one with n = 5 per group/experiment and one with n = 2–5 per group/experiment. (G) Experimental outline for measuring 4-1BB following mRNA vaccination. P14 cells were transferred into C57BL/6 mice. One day after transfer, recipient mice were immunized with 3 μg of an mRNA-LCMV GP vaccine, and 4-1BB was measured on P14 cells at various time points. (H) 4-1BB on P14 cells after mRNA vaccination. Representative histograms showing 4-1BB expression on P14 cells. We utilized this P14 chimera model using a high number of P14 cells to allow us to detect 4-1BB expression on virus-specific CD8⁺ T cells at hyperacute points; endogenous virus-specific CD8⁺ T cells cannot be detected at hyperacute time points due to their low precursor frequency. Mean fluorescence intensity (MFI) is indicated on the x axis to denote “per-cell expression” of 4-1BB. This adoptive transfer experiment was performed 2 times, with n = 3 per group, showing similar results (peak of 4-1BB expression on day 4 after vaccination). All data are shown. Indicated P values in A and D–F were calculated by the Mann-Whitney test.

Figure 4. Reinforcing 4-1BB costimulation 4 days after mRNA vaccination induces sterilizing protection against pathogen challenges.

(A) Experimental outline to examine whether treatment with α4-1BB on day 4 improves immune protection conferred by an mRNA-LCMV vaccine. (B) Summary of LCMV Cl-13 loads in the spleen on day 7 after challenge. On day 14 after vaccination, mice were challenged i.v. with LCMV Cl-13 (2 × 10⁶ PFU) and viral loads were quantified in Vero-E6 monolayers. (C) Experimental outline to examine whether treatment with α4-1BB on day 4 improves immune protection conferred by an mRNA-OVA vaccine. (D) Summary of LM-OVA bacterial loads in the spleen on day 3 after challenge. On day 14 after vaccination, mRNA-OVA–vaccinated mice were challenged i.v. with a supralethal dose of LM-OVA (1 × 10⁷ CFU) and bacterial loads were quantified in agar plates. In the challenge experiments, mice were immunized with 3 μg of the respective vaccine followed by treatment with 50 μg of α4-1BB or control antibodies on day 4. LCMV Cl-13 challenge data are from 2 experiments, one with n = 5 per group/experiment and one with n = 4 per group/experiment. Data from the LM-OVA challenge experiment are from one experiment, n = 4–5 per group. The control vaccines were still able to confer partial protection, relative to no vaccination (mean LCMV Cl-13 viral loads in unvaccinated mice = 1.3 × 10⁷ PFU/g; mean LM-OVA loads in unvaccinated mice = 1.1 × 10⁶ CFU/g). Indicated P values in B and D were calculated by the Mann-Whitney test.

Figure 5. Delayed reinforcement of 4-1BB enhances the efficacy of a therapeutic cancer vaccine.

(A) Experimental outline to examine whether treatment with α4-1BB on day 4 improves immune protection by a therapeutic cancer vaccine. Mice were challenged s.c. with 2 × 10⁶ B16-OVA tumor cells. On day 10 after tumor challenge, mice were vaccinated intramuscularly with 3 μg of mRNA-OVA. Mice received either control antibodies or α4-1BB (50 μg on day 0 or day 4 after mRNA-OVA vaccination). (B) Tumor control. (C) Survival. (D) Representative FACS plots showing CD8⁺ T cell responses on day 9 after vaccination. (E) Summary of OVA-specific CD8⁺ T cell responses on day 9. (F–H) Central memory, effector memory, and effector CD8⁺ T cells (K^bSIINFEKL⁺ PBMCs) at 2 weeks after vaccination. Data are from 2 experiments, one with n = 6–7 per group and one with n = 8 per group. Indicated P value in C was calculated by the log-rank (Mantel-Cox) test; all other P values were calculated by 2-way ANOVA with the Holm-Šídák multiple-comparison test.

Figure 6. Generalizability to other costimulatory pathways: reinforcing OX40 costimulation on day 4 results in superior vaccine responses, relative to reinforcing OX40 costimulation on day 0.

(A) Experimental outline for evaluating OX40 expression following mRNA vaccination. We utilized the same adoptive transfer model from Figure 3G. (B) Kinetics of OX40 on virus-specific CD8⁺ T cells after mRNA vaccination. This adoptive transfer experiment was performed 2 times, with n = 3 per group, showing similar results (peak of OX40 expression on day 4 after vaccination). (C) Time-dependent effects of OX40 costimulation following mRNA vaccination. Mice were immunized with 3 μg of mRNA-spike vaccine, followed by treatment with OX40 costimulatory antibodies (200 μg of αOX40, clone OX-86) on day 0 or day 4 after vaccination. CD8⁺ T cell responses (D) and antibody responses (E) on day 15 after vaccination are shown. Data in D and E are from 3 experiments, with n = 5 per group. Indicated P values in D and E were calculated by Kruskal-Wallis test with Dunn’s multiple-comparison test.