The mRNA part of lipid nanoparticle vaccines activates immune signals that may reduce the body's adaptive immune response

All animal experiments in this study were approved by the Institutional Animal Care and Use Committee (IACUC) of the Israel Institute for Biological Research (IIBR). Experimental procedures were performed under protocols M-12-22, M-31–23 and M-28-24. All mice used in this study were maintained according to the guidelines and regulations for animal experiments at the IIBR. Female C57BL/6J (#000664) and IFNAR^-/- (#032045) mice were purchased from the Jackson Laboratory (Bar Harbor, ME). Animals were age matched between groups and at age of 6–8 weeks at commence of experiments.

mRNA constructs were purchased from TriLink (San Diego, CA, USA) and included complete N1-methyl-pseudouridine (m1Ψ) nucleotide substitution. All mRNA constructs were codon optimized for expression in mice and included an initiator methionine and a Kozak consensus sequence (32, 33). mRNA was purified by cellulose purification, as previously described (34). Removal of double stranded RNA (dsRNA) contaminants was confirmed using dot blot, and endotoxin levels (measured by LAL Kinetic-QCL kit from Lonza) were <0.05EU/ml. For encapsulation, ionizable lipid (ALC0315, CAYMAN 34337), Cholesterol (Avanti Polar Lipids), distearoyl-sn-glycero-3-phosphocholine (DSPC, Avanti Polar Lipids), and dimyristoyl-rac-glycero-3-methoxypolyethylene glycol (DMG-PEG, Avanti Polar Lipids) were mixed at a molar ratio of 40:47.5:10.5:2 with absolute ethanol and cellulose-purified mRNA payloads were suspended in citrate buffer (50mm, pH 4.5). To create LNPs, a dual-syringe pump was used to transport the two solutions through the NanoAssemblr^® micromixer (Precision Nanosystems, Vancouver, British Columbia, Canada) at a total flow rate of 12ml/min. For empty LNPs preparation, citrate buffer was mixed with lipid formulation using the same mixing parameters as mentioned above. The particles were then transferred into dialysis overnight against PBS. Particles in PBS were analyzed for size and uniformity by dynamic light scattering (DLS) (Stunner^® by Unchained Labs). RNA encapsulation efficiency was confirmed using the RiboGreen Assay (ThermoFisher Scientific, Waltham, MA, USA) and surface charge was measured by Zetasizer analyzer (Malvern Panalytical Ltd, UK). LNP-mRNA formulations displayed comparable hydrodynamic size of 63nm ± 3.2 (RBD), 68.47nm ± 0.98 (F1), 69.5nm ± 0.66 (non-coding) and 59.5nm ± 1.6 (empty LNP). Polydispersity index (PDI) was 0.11 ± 0.01 (RBD), 0.2 ± 0.02 (F1), 0.13 ± 0.01 (non-coding) and 0.23 ± 0.01 (empty LNP). Encapsulation efficiency for all preparations was >93%.with zeta potential of -8.8mV ± 0.4.

Vaccines were administered by intramuscular injection of 50μl volume into each hind leg, for a total of 100μl. Mice were immunized with either 5μg LNP-mRNA, an equivalent dose of empty LNP or PBS. For IFNAR blocking, mice were injected intraperitoneally (IP) with 2.5mg anti IFNAR monoclonal antibodies (I-401-100, Leinco Technologies) 24hr prior immunization and 24hr post immunization.

Deucravacitinib treatment was adapted from previously published studies (35, 36). Briefly, Deucravacitinib (MedKoo Biosciences) was dissolved in DMSO to a concentration of 10mg/ml and stored at -20 °C. For injection, a mixture of PEG-300:Tween-80 (both from Sigma-Aldrich) was prepared at a ratio of 18:1. For each injection, 20μl of DMSO with or without the inhibitor was added to 120μl of the PEG300:Tween-80 solution, mixed and further added to 180μl PBS for a total of 300μl (total dose of 0.6mg Deucravacitinib). Mice were injected IP with Deucravacitinib or vehicle 24hr and 4hr before immunization with LNP-RBD.

At designated time points post immunization, mice were euthanized by IP injection of Pentobarbiton (300mg/kg) and organs were collected. Inguinal LNs and spleens were digested with 1mg/ml collagenase D (Roche, Mannheim, Germany) for 30min at 37 °C. The digested organs were further mechanically processed and filtered to generate single cell suspensions, followed by Red Blood Cells lysis (R7757, Merck) for splenic cells. For analysis of cellular responses at 3 weeks post vaccination, spleens were dissociated into single-cell suspensions by GentleMACS (Miltenyi, Bergisch Gladbach, Germany), filtered, separated on 1.084gr/ml Ficoll Paque premium (GE) and washed with medium.

The following mAb clones were used for staining: CD3 (145-2C11), CD8 (53-6.7), CD4 (GK1.5), CD11b (M1/70), MHCII (I-A/I-E), CD11c (N418), mPDCA1 (129c), Ly6C (HK1.4), CD103 (2E7), and IFNγ (XMG1.2). All antibodies were purchased from Thermo Fisher Scientific or Biolegend. All washing steps were done using FACS buffer (PBS + 0.5% FBS + 2nM EDTA). For Live/Dead staining, cells were stained with Aqua Fluorescent Reactive Dye (1:300 dilution; L34965, Invitrogen) and incubated for 30min at 4 °C. All samples were incubated with anti CD16/CD32 blocking antibody prior to extracellular staining. Following staining of extracellular markers, cells were fixed and permeabilized using a Cytofix/Cytoperm kit (BD Biosciences) according to the manufacturer's instructions prior to intracellular staining. For intracellular cytokine staining (ICS), cells were incubated for 5hr with SARS-CoV-2 S1 peptide pool (130-127, Miltenyi) at final concentration of 2µg/ml in the presence of protein transport inhibitor cocktail (eBioschience) as directed. Samples were acquired on LSRFortessa (BD Biosciences) and analyzed with FlowJo V.10 software (TreeStar, Ashland, OR).

At 6hr post immunization, blood was collected from the tail vein. Following an incubation period of 30min at room temperature serum was separated by centrifugation and the samples were stored at -20°C until further analysis. Serum levels of IFNα, CXCL10, IL-6 and CCL2 were determines by ELISA kits (IFN-α-MFNAS0, CXCL10-DY466, IL-6-DY406, CCL2-DY479; R&D Systems, USA) according to the manufacturer's instructions.

Direct anti-RBD ELISA was performed for the detection of RBD-specific IgG antibodies in mouse sera. Nunc MaxiSorp ELISA plates were coated with 2μg/ml of recombinant SARS-CoV-2 RBD in carbonate/bicarbonate solution (C3041, Sigma, Israel) overnight at 4 °C. Plates were blocked with PBST buffer (PBS + 0.05% + 2% BSA) for 60 min at 37 °C. Following blocking and washing with PBST buffer (PBS + 0.05% Tween 20), the plates were incubated with mouse serum for 1hr at 37 °C. Following washing, alkaline phosphatase-conjugated anti-mouse IgG was used (diluted 1:1000) as a secondary antibody (Jackson ImmunoResearch, USA). P-nitrophenyl phosphate substrate (N2770, Sigma, Israel) was added after washing, and the optical density was measured by microplate reader (SpectraMax 190, Molecular Devices, CA).

RBD levels in mice sera was determined using SARS-CoV-2 RBD ELISA Kit (Abcam 289833) according to the manufacturer's instructions.

Statistical analyses were performed using GraphPad software 8.1.1 (La Jolla, CA). Statistical differences between groups were evaluated by unpaired t tests (two groups) or one-way ANOVA (>2 groups) with Tukey corrected for multiple comparisons where p<0.05 = *, p<0.01 = **, p<0.001 =*** were considered significantly different among groups. All experimental data are presented as the mean ± standard error of the mean (SEM).

It was previously documented that vaccination with a formulation of lipid nanoparticles encapsulating the SARS-CoV-2 spike receptor binding domain (LNP-RBD) elicited antigen-specific T-cells and significant humoral responses which resulted in full protection against challenge with SARS-CoV-2 in a murine model for COVID19 (32). Accordingly, LNP-RBD served in the current study as a model for dissection of the immune response associated with LNP-mRNA vaccination. The effects of LNP-mRNA vaccine on different innate cell populations were initially interrogated in the dLNs by flow cytometry following intramuscular (IM) delivery. At different early time points (24, 48, 72 hours) post vaccination, the inguinal lymph nodes were collected for analysis. At 24hr after vaccination the number of resident dendritic cells (DCs, defined as CD11c^highMHCII⁺, referred as Population #1 in Figure 1A) decreased significantly compared to control unvaccinated mice (Figures 1A, B). This reduction was accompanied by significant elevation in the population of activated DCs which was identified by higher MHC-II expression (population #2, Figure 1A and Figure 1C), implying that resident DCs underwent activation. Notably, the number of migratory DCs (CD11c^intMHCII^high, population #3 in Figure 1A) in the dLNs was not affected by LNP-mRNA vaccination (Supplementary Figure S4). In addition, the number of monocytes (CD11b⁺Ly6C^high) increased significantly in the dLNs at 24hr (Figures 1D, E). This increase sustained for at least 72hr post immunization (Figure 1E). The number of plasmacytoid DCs (PDCs, CD11c^lowPDCA1⁺) decreased significantly in the dLNs during the first 48hr and returned to initial levels at 72hr (Figure 1F). Moreover, multiple innate immune cell populations including DCs (Figures 1G, H), PDCs (Figure 1I) and B cells (Figure 1J) were highly activated 24hr following vaccination with LNP-mRNA, as indicated by the enhanced expression of the co-stimulation marker CD86. While CD86 levels in both activated and migratory DCs (Figure 1G and Figure 1H, respectively) remained significantly high in comparison to control animals for at least 72hr post vaccination, they restored to basal level in PDCs and B-cells already after 48hr (Figures 1I, J).

To address the systemic response, modulations of cell populations in the spleen were examined. In a similar manner to the observations pertaining to dLNs, at 24hr post vaccination, a significant elevation in the number of DCs exhibiting higher levels of MHC-II (population #2 in Figure 2A) was recorded (Figures 2A, B). In contrast to the prolonged activation of DCs in the dLNs, the elevation in the number of activated DCs in the spleen was transient and returned to the initial base line level as early as 48hr post-vaccination. Moreover, the ratio between the frequencies of CD8⁺ DCs (cDC1 subset) and CD11b⁺ DCs (cDC2 subset) among the non-activated DC population (population #1 in Figure 2A) decreased significantly and was maintained low throughout the course of the experiment (Figure 2C). As documented above for dLNs, enhanced expression of CD86 was observed in multiple innate immune cells in the spleen including activated DCs, monocytes and pDCs, albeit this activation was transient and lasted during the first 24 hours post vaccination (Figures 2D–F). Furthermore, as early as 6 hours post vaccination significant elevation in the levels of IFNα, CXCL10, IL-6 and CCL2 in the serum of vaccinated animals was observed (Figures 2G–J). Taken together these data establish that in the current system, LNP-mRNA vaccination induces massive innate immune activation both locally and systemically.

As explained in the introduction, the rapid activation of the innate immune system following immunization with LNP-RBD, may be the result of response to LNPs, to the mRNA it encapsulates, or to both. Additionally, the magnitude of the response may be affected by the nature of the expressed antigen encoded by the mRNA molecule. To discern among these possibilities, mice were vaccinated with LNP-mRNA that encode to either RBD as described above, Yersinia-pestis capsular antigen F1 (33), or non-coding mRNA. Additional control experimental groups were injected with empty LNPs or PBS. Mice were vaccinated and analysis was performed at 24hr post immunization, the time-point of maximal response as described above (Figure 1 and Figure 2). A significant elevation in the numbers of activated DC was observed in both dLNs (Figure 3A) and spleens (Figure 3B) of mice vaccinated with mRNA-containing particles. It is important to note that the similar increase in activated DCs was observed for both RBD and F1-coding mRNA, as well as for the non-coding molecule. Most notably, empty LNPs did not activate DCs in-vivo neither in the dLNs nor in the spleen. These results imply that the presence of the mRNA itself in the LNP-mRNA vaccine is necessary for the robust activation of DCs following vaccination. This inherent ability of mRNA to activate the innate immune system was further confirmed by the rapid recruitment of monocytes to the draining lymph nodes (Figure 3C) and spleens (Figure 3D) following vaccination with the different mRNA-containing particles, including non-coding mRNA. CD86 expression on DCs, monocytes, PDC and B-cells was similarly elevated solely in the groups of animals immunized with mRNA-containing LNPs (Figures 3E–H). In spleens, the described above decrease in cDC1 to cDC2 subsets ratio (Figure 2C), was observed only as a result of mRNA presence in the vaccine (Figure 3I).

The intrinsic capacity of mRNA to activate the innate immune system in-vivo was substantiated by the systemic release of IFNα, CXCL10, IL-6 and CCL2. Thus, similar amounts of these cytokines were induced by all mRNAs including the non-coding mRNA, indicating that the mRNA molecule itself and not the encoded protein is essential for the robust activation of the innate immune system. Of note, empty LNPs did not induce IFNα, CXCL10, IL-6 and CCL2 release, indicating that the activation of innate immune response was specific for mRNA presence (Figures 3J–M).

To conclude, the data strongly support the notion that the mRNA component of LNP-mRNA is crucial for the massive innate activation observed early after vaccination. This mRNA-dependent activation was manifested by elevation in MHC-II and CD86 expression, increased numbers of antigen presenting cells and inflammatory cytokines secretion.

The immune stimulatory effect of foreign mRNA is known to involve the type I IFN receptor (IFNAR) signaling pathway (37). To determine to what extent IFNAR signaling is required for innate immune activation following LNP-mRNA vaccination, WT or IFNAR^-/- mice were vaccinated with LNP-mRNA and the innate immune response was analyzed 24hr later, as shown above. While significant elevation in activated DCs was observed in the dLN and spleen of WT mice, augmentation of activated DCs could not be detected, neither in dLNs nor in spleens of IFNAR^-/- mice (Figures 4A, B). In a similar manner, there was no elevation in monocyte migration to the dLN in IFNAR^-/- mice (Figure 4C). In addition, the significant elevation in CD86 expression on different innate immune cells in the spleens and dLNs of WT mice, was not apparent in IFNAR^-/- mice (Figures 4D–G). Furthermore, in contrast to WT mice, in IFNAR^-/- mice no induction of the cytokines IFNα, CXCL10, IL-6 and CCL2 was detected (Figures 4H–K). These results strongly suggest that the IFNAR signaling pathway is essential for the mRNA-dependent activation of the innate immune response following LNP-mRNA vaccination. This IFNAR-dependent activation was further characterized to be mediated by the Signal transducer and activator of transcription 1 (STAT1). LNP-mRNA vaccinated mice were pre-treated with Deucravacitinib, an approved TYK2 inhibitor that blocks STAT1/2 signaling (38). In contrast to the robust systemic activation observed following LNP–mRNA vaccination, Deucravacitinib treatment of immunized mice resulted in a marked suppression of the innate immune response. This was evidenced by reduced numbers of activated dendritic cells and decreased expression of the co-stimulatory receptor CD86 across several cell types, including dendritic cells, monocytes, and B-cells(Supplementary Figure S5).

Efficient and robust activation of the innate immune response is a pivotal step for an effective adaptive response which further results in the elicitation of antigen-specific humoral as well as T-cell responses. Therefore, we addressed the assumption that the absence of innate activation (observed in IFNAR^-/- mice) would affect the induction of adaptive immune responses. At 3 weeks post vaccination, RBD-specific antibody levels in the circulation were measured by ELISA and RBD-specific T cells were quantified by intracellular staining (ICS). Surprisingly, the data evidenced similar titers of specific antibodies in IFNAR-deficient and WT mice (Figure 5A) and a significant elevation in the frequencies of RBD-specific CD8-T cells in the spleens of IFNAR^-/- mice (Figure 5B).

Complete depletion of IFNAR signaling in IFNAR-/- mice may affect T-cell homeostasis and may possess compensatory signaling pathways affecting adaptive immune response (39 –41). Accordingly, an alternative experimental approach was applied, consisting of transient short-term IFNAR-blockage by an antagonistic antibody (Figure 6A). Twenty-four hours post vaccination, the immune response was assessed by various parameters, as described above, which established that following transient IFNAR inhibition, both local and systemic innate response, including DC activation, monocyte recruitment, upregulation of co-stimulatory receptors and cytokine secretion were completely abolished (Supplementary Figure S6).

The role of IFNAR signaling using the antibody-mediated blockage system, was further used to assess its impact on the development of anti-RBD-specific antibodies and RBD-specific T-cells. Most notably, transient inhibition of the IFNAR signaling pathway resulted in significant elevation both in RBD-specific binding antibodies (Figure 6B) and RBD-specific CD8⁺ T-cells (Figure 6C), suggesting an unexpected apparent negative effect of type I IFN on the response.

To further investigate the mechanism by which IFNAR blockade enhances antigen-specific T-cell and antibody responses, we hypothesized that inhibition of IFNAR signaling may increase antigen expression from the LNP-RBD vaccine, as previously suggested by others (42). Consistent with this hypothesis, serum RBD levels were elevated approximately tenfold in anti-IFNAR–treated animals compared with vaccinated, untreated controls (Figure 6D). These findings indicate that IFNAR signaling strongly suppresses the expression of mRNA-encoded vaccine antigens. When this inhibitory effect is alleviated through IFNAR blockade, antigen expression is markedly increased, thereby driving enhanced induction of antigen-specific T cells and antibodies.