Lymph node macrophages drive innate immune responses to enhance the anti-tumor efficacy of mRNA vaccines

Abstract

mRNA vaccines are promising for cancer treatment. Efficient delivery of mRNAs encoding tumor antigens to antigen-presenting cells (APCs) is critical to elicit anti-tumor immunity. Herein, we identified a novel lipid nanoparticle (LNP) formulation, L17-F05, for mRNA vaccines by screening 34 ionizable lipids and 28 LNP formulations using human primary APCs. Subcutaneous delivery of L17-F05 mRNA vaccine encoding Gp100 and Trp2 inhibited tumor growth and prolonged the survival of mice bearing B16F10 melanoma. L17-F05 efficiently delivered mRNAs to conventional dendritic cells (cDCs) and macrophages in draining lymph nodes (dLNs). cDCs functioned as the main APCs by presenting antigens along with enhanced expression of co-stimulatory molecules. Macrophages triggered innate immune responses centered on type-I interferon (IFN-I) in dLNs. Lymph node (LN) macrophage depletion attenuated APC maturation and anti-tumor activity of L17-F05 mRNA vaccines. Loss-of-function studies revealed that L17-F05 works as a self-adjuvant by activating the stimulator of interferon genes (STING) pathway in macrophages. Collectively, the self-adjuvanticity of L17-F05 triggered innate immune responses in LN macrophages via the STING–IFN-I pathway, contributing to APC maturation and potent anti-tumor activity of L17-F05 mRNA vaccines. Our findings provide strategies for further optimization of mRNA vaccines based on the innate immune response driven by LN macrophages.

Graphical abstract

Kubara and colleagues identified a novel LNP formulation, L17-F05, for an mRNA vaccine and revealed that the L17-F05 mRNA vaccine has robust anti-tumor effects against the melanoma model along with innate immune response triggered by lymph node macrophages via STING–IFN-I pathway.

Introduction

mRNA vaccines are promising therapeutics for treatments against infectious diseases and cancers.^1,2 They are superior to conventional vaccines, including inactivated, DNA, and peptide vaccines, owing to high potency, rapid development, low-cost manufacturing, and safe administration.³ mRNA vaccines for preventing coronavirus disease 2019 (COVID-19) achieved great success and have shown strong potential against infectious diseases.^4,5 In addition, mRNA vaccines for cancer treatment are being tested in the clinical stage.⁶

Lipid nanoparticle (LNP) is widely used as a delivery system for mRNA therapeutics.^4,5 LNPs protect mRNAs from in vivo degradation and efficiently deliver mRNAs inside cells. Ionizable lipids are a main component in LNPs, helping the endosomal escape of mRNA due to their cationic property at low pH in the endosomes of target cells. Thus, mRNA delivery efficacy largely depends on the characteristics of ionizable lipids in LNPs. Studies have screened synthetically designed ionizable lipids, identifying several LNP types for generating mRNA vaccines.^6,7 Efficient delivery of antigen mRNAs to antigen-presenting cells (APCs) in peripheral lymphoid organs, e.g., the draining lymph nodes (dLNs) or spleen, is critical to potentiate the anti-tumor efficacy of mRNA vaccines.^1,2,8

mRNA vaccines exhibit intrinsic immunogenicity by activating pattern-recognition receptor (PRR) families and acting as self-adjuvants. The involvement of Toll-like receptors (TLRs), TLR7 and TLR8, and retinoic acid-inducible gene I (RIG-I) in recognizing unmodified mRNA and double-stranded RNA (dsRNA) contaminants in mRNA vaccines has been elucidated.^1,9 Self-adjuvant effects of LNP have been demonstrated for lipidoids and heterocyclic lipids by activating TLR4 and stimulator of interferon genes (STING), respectively.^10,11,12,13 Activation of the innate immune sensors leads to the release of interferons (IFNs; e.g., IFN-β and IFN-γ), interleukins (ILs; e.g., IL-1β and IL-12), and proinflammatory cytokines (e.g., IL-6 and tumor necrosis factor-α [TNFα]), resulting in an enhancement of the efficacy of mRNA vaccines.^1,13,14,15 More recently, the success of mRNA vaccines for severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) has led to an emerging understanding of the mechanisms underlying the innate immune activation induced by mRNA vaccines. The maturation of germinal center B cells and follicular helper T (T_FH) cells in dLNs post inoculation with mRNA vaccines induces robust Th-1-biased humoral immunity.^16,17 Li et al. examined immune responses to the BNT162b2 mRNA vaccine of Pfizer (New York City, NY, USA) against COVID-19 using single-cell RNA sequencing (RNA-seq) and knockout mice lacking a series of innate immune sensors and demonstrated the landscape of immune cell responses to BNT162b2, showing that CD8⁺ T cell responses induced by BNT162b2 are dependent on IFN-I via MDA5 signaling.¹⁸ Tahtinen et al. investigated the immune responses to two types of mRNA vaccines, BioNTech’s (Mainz, Germany) RNA-Lipoplex and Moderna’s (Cambridge, MA, USA) ionizable lipid SM102 used in the mRNA-1273 vaccine against COVID-19, and they revealed that the IL-1 pathway is essential in triggering innate immune responses induced by both types of mRNA vaccines.¹⁹ However, immunological mechanisms underlying mRNA-vaccine-elicited robust immune responses remain unelucidated.

Herein, we identified a novel mRNA vaccine formulation, L17-F05, by screening 34 ionizable lipids and 28 formulations using human APCs. The L17-F05 mRNA vaccine exhibited potent anti-tumor effects in melanoma-model mice. To address its mechanisms of action, we analyzed immunophenotypes in dLNs post L17-F05 mRNA vaccine administration. We provide immunological insights into mechanisms underlying the efficacy of the L17-F05 mRNA vaccine.

Results

Screening of ionizable lipids and LNP formulations and optimization using human APCs

To explore novel LNP formulations for mRNA vaccines against tumors, we screened 34 ionizable lipids by optimizing their head and lipid tail structures and 28 formulations with varying phospholipids and molar ratio in LNP (Figure 1A; Tables S1 and S2). For screening, we employed APCs, human bone marrow-derived macrophages (BMDMs), and bone marrow-derived dendritic cells (BMDCs), which were differentiated from human bone marrow-derived hematopoietic progenitor cells (Figures S1A and S1B). Based on our previous works on LNP,^17,20 we prepared 10 ionizable lipids (L01–L10) with varied head structures and evaluated the translation efficacy of firefly luciferase (Fluc) mRNA expression in human BMDMs and BMDCs (Figure 1B). Among the head groups tested, L01 exhibited the highest Fluc expression in BMDCs and BMDMs. L06, which has the same head as L01, also showed high Fluc expression, suggesting that the six-membered head structure is highly active (Figure 1C). Interestingly, two stereoisomer types of the head group showed different Fluc expression activity, such as L02 vs. L03, and L07 vs. L08. The pKa of the highly active heads (L01, L06, L03, and L07) ranged from 6.0 to 7.0 (Figure 1D). We selected the six-membered ring head as the main target structure and the five-membered ring head as a similar head structure for further optimization, resulting in the design of 24 ionizable lipids (L11–L34) with varied lipid tail structures (Figure 2A). Tail optimization demonstrated that L16 and L17 showed the highest Fluc expression activity increased by 2- to 3-fold compared with that of L01, the top compound in the head optimization step (Figure 2B). The screening results of all 34 ionizable lipids indicated that BMDCs and BMDMs exhibited a strong correlation in Fluc expression activity (R² = 0.73), and branched tail lipids exhibited higher activity than linear tail lipids (Figure S2A). For highly active LNPs, pKa = 6.9–7.0 and Z average = ∼100 nm (Figures S2B and S2C). Both L16 and L17 exhibited high mRNA encapsulation efficiencies (>90%) (Figure 2C). LNP stability, indicated by size change during storage as a liquid at 4°C, was higher for L17 (green) than for L16 (pink) (Figure 2D). Therefore, we selected L17 as a lead ionizable lipid and optimized LNP formulation (Figure 2E). We employed three types of phospholipids (1,2-distearoyl-sn-glycero-3-phosphocholine [DSPC], 1,2-dioleoyl-sn-glycero-3-phosphocholine [DOPC], and 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine [DOPE]) and changed the molar ratio of L17 (40%–70%), phospholipid (0%–23%), and cholesterol (17%–59%) with a fixed 1.5 mol% of 1,2-dimyristoyl-rac-glycero-3-methoxypolyethylene glycol-2000 (DMG-PEG) (Table S2). Of the 28 formulations (F01–F28), DSPC-based F05 exhibited the highest Fluc expression activity in both BMDCs and BMDMs (Figure 2F). The lead L17-F05 formulation showed multilamellar LNPs in cryogenic electron microscopy (Figure 2G). L17-F05 showed higher Fluc expression in cells than control LNP using D-Lin-MC3-DMA, the ionizable lipid used in the Food and Drug Administration (FDA)-approved LNP-based medicine²¹ (Figure S2D). A strong correlation was observed between human and mouse APCs in Fluc expression among selected LNPs (Figure S3A). The order of in vivo Fluc expression in representative LNPs agreed with that of in vitro Fluc expression (L17 > L03 > L15) (Figure S3B).

Figure 1.

Lipid head optimization in LNP screening for mRNA delivery

(A) Schematic illustration of ionizable lipid screening and formulation optimization. (B) Structures of head groups in the screening for lipid head optimization. (C) Fluc expression in lipid head optimization. n = 3 independent experiments; mean ± SD. (D) Relationship between Fluc expression and pKa in lipid head optimization.

Figure 2.

Lipid tail optimization in LNP screening for mRNA delivery

(A) Structures of tail groups in the screening for lipid tail optimization. (B) Fluc expression in lipid tail optimization. n = 3 independent experiments; mean ± SD. (C and D) mRNA encapsulation efficacy (C) and size change during storage (D) of 34 types of ionizable lipid in LNP screening. (E) Parameters for optimizing L17 formulation. (F) Fluc expression of 28 LNPs for optimizing L17 formulation. n = 3 independent experiments; mean ± SD. (G) Cryogenic electron microscopy image of L17-F05. Bar, 100 nm.

L17-F05 mRNA vaccine elicited robust cellular immunity and therapeutic effects against B16 melanoma models

To investigate the potency of the L17-F05 mRNA vaccine for inducing cellular immunity, L17-F05 encapsulating chicken ovalbumin (OVA) mRNA (OVA mRNA/L17-F05) was subcutaneously injected into the back of mice. Then, we monitored OVA-specific T cells in peripheral blood using an OVA tetramer on days 4, 7, and 13. Post vaccination with OVA mRNA/L17-F05, OVA-specific CD8⁺ T cells emerged in peripheral blood, and their number reached a peak on day 7 in a dose-dependent manner (Figure 3A). IFN-γ enzyme-linked immunosorbent spot (ELISPOT) assays showed that splenocytes stimulated by OVA peptide (257–264 amino acids [a.a.]) for CD8⁺ T cells showed significant secretion of IFN-γ on day 7 post vaccination (p < 0.001) (Figure 3B). Splenocytes from vaccinated mice hardly secreted IFN-γ when stimulated by OVA peptide (323–339 a.a.) for CD4⁺ T cells.

Figure 3.

Cellular immunity and anti-tumor therapy effects induced by the L17-F05 mRNA vaccine

(A) Flow cytometry analysis of OVA-specific CD8 T cells in peripheral blood over time post injection with OVA mRNA/L17-F05. Representative flow cytometry plot and percentage of OVA-specific CD8⁺ T cells on days 4, 7, and 13 post vaccination. n = 4 independent experiments; mean ± SEM. ∗p < 0.05, ∗∗∗∗p < 0.0001; repeated-measures ANOVA followed by Dunnett’s multiple comparison test compared with the saline group. (B) IFN-γ ELISPOT of splenocytes from mice vaccinated with OVA mRNA/L17-F05. Representative ELISPOT images and quantification of spots per well. n = 4 independent experiments; mean ± SEM.; ∗∗∗p < 0.001; two-way ANOVA followed by Bonferroni’s multiple comparison test. (C) Schedule of anti-tumor therapy experiments using bilateral B16F10-bearing model (upper). B16F10-empty and B16F10-OVA were inoculated in the right and left flanks, respectively. OVA mRNA/L17-F05 was subcutaneously injected in the center of the back at 20 μg/mouse on days 3, 7, and 10 after tumor inoculation. Lower images show representative mice on day 21. (D) Tumor growth curves in (C). n = 5–6 independent experiments; mean ± SEM. ∗∗∗∗p < 0.0001; repeated-measures ANOVA followed by Bonferroni’s multiple comparison test. (E) Tumor growth curves of B16F10-OVA-bearing syngeneic mice treated with L17-F05 or ALC-0315 vaccine encapsulating Fluc mRNA (left: 10 μg) or OVA mRNA (right: 4 or 10 μg). Black arrows indicate administration of mRNA vaccines on days 7, 10, and 13 after tumor inoculation. n = 5 independent experiments; mean ± SEM. ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001; repeated-measures ANOVA followed by Dunnett’s multiple comparison test compared with the saline group. (F) Survival curves in (E). n = 5 independent experiments; ∗p < 0.05, ∗∗p < 0.01; log rank test compared with the saline group. (G) Flow cytometry analysis of OVA-specific CD8 T cells in peripheral blood on day 15 after tumor inoculation. n = 5 independent experiments; mean ± SEM. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001; one-way ANOVA followed by Dunnett’s multiple comparison test compared with the saline group. (H) Tumor growth curves of B16F10-bearing syngeneic mice treated with L17-F05 vaccine encapsulating Gp100-Trp2 mRNA (10 μg). Black arrows indicate administration of mRNA vaccines on days 3, 7, and 10 after tumor inoculation. n = 6 independent experiments; mean ± SEM. ∗∗∗∗p < 0.0001; repeated-measures ANOVA followed by Dunnett’s multiple comparison test compared with the saline group. (I) Survival curves in (H). n = 6 independent experiments; ∗∗∗p < 0.001; log rank test.

To test whether the L17-F05 mRNA vaccine exhibited anti-tumor effects, we examined OVA mRNA/L17-F05 in a syngeneic melanoma mouse model inoculated with B16F10 (B16) cells stably expressing OVA (B16-OVA). We tested L17-F05 in a bilateral tumor model, wherein B16-OVA and B16-control cells were inoculated at the right and left flanks, respectively, to demonstrate OVA antigen-specific tumor growth inhibition with OVA mRNA vaccine (Figure 3C, upper). Mice treated with OVA mRNA/L17-F05 exhibited a significant tumor growth inhibition in a B16-OVA-specific manner, as revealed in the appearances of treated mice on day 21 (Figure 3C, lower) and tumor growth curve over time (Figure 3D). Anti-tumor effects were confirmed using a unilateral B16-OVA-bearing mice model. We compared the responses to L17-F05 with those of ALC-0315, the ionizable lipid employed in an FDA-approved mRNA vaccine produced by BioNTech.²² We observed a dose-dependent response of mRNA vaccine in tumor growth inhibition and survival extension caused by the treatment with L17-F05 and ALC-0315 (benchmark LNP) vaccines encapsulating OVA mRNA but not the control treatment with free OVA mRNA or irrelevant Fluc mRNA vaccines (Figures 3E and 3F). To address T cell function in the treatment of the B16-OVA model, we monitored OVA-specific T cells in peripheral blood using an OVA tetramer on day 15 post inoculation. We found OVA-specific CD8⁺ T cells in peripheral blood in a dose-dependent manner, confirming the anti-tumor effects (Figure 3G). We evaluated the efficacy of the L17-F05 mRNA vaccine for self-antigen in melanoma using Gp100 and TRP2. In a B16-bearing mouse model, the L17-F05 encapsulating Gp100 and Trp2 mRNA showed significant tumor growth inhibition and prolonged survival compared with that in the control groups treated with saline or irrelevant OVA mRNA (Figure 3H). These results demonstrate that the L17-F05 mRNA vaccine elicited robust cytotoxic CD8⁺ T cell induction and therapeutic effects against B16 melanoma models.

L17-F05 allowed efficient delivery of mRNA to myeloid cells in dLNs without any significant systemic toxicity

To evaluate the in vivo organ distribution of L17-F05, we formulated Fluc mRNA in L17-F05 and injected it into mice subcutaneously at the back or footpad. Organ bioluminescence was measured at 4, 24, and 48 h post injection. Bioluminescence was mainly detected in lymph nodes (LNs) and peaked at 4 h but was almost undetectable in other organs, including the liver, spleen, lung, kidney, and heart (Figure 4A). LNP injection into the center of the back efficiently delivered LNP to brachial LNs on both sides (right and left), whereas that into the footpad enabled specific delivery of LNP to popliteal LNs on the injected side (right). We subsequently formulated enhanced green fluorescent protein (EGFP) mRNA hybridized with the fluorophore-conjugated probe in L17-F05 and injected it subcutaneously in the footpad of mice to analyze L17-F05 distribution and mRNA translation simultaneously in draining popliteal LNs. Popliteal dLNs were collected 4 and 24 h post injection, and LNP distribution (AZDye647) and RNA translation (EGFP) were analyzed at a single-cell level using flow cytometry (FCM). The FCM gating strategy is shown in Figure S4. FCM analysis showed that myeloid cells largely underwent LNP uptake and mRNA translation, whereas lymphoid cells did not. L17-F05 was efficiently taken up by APCs, such as cDCs, macrophages, and monocytes, and different levels of LNP uptake and mRNA translation were observed among cell types (Figures 4B—4D). In cDC subsets, L17-F05 was efficiently distributed to CD103⁺ cDC compared to other cDC subsets, although EGFP mRNA translation was higher in CD11b⁺ cDC than in CD103⁺ cDCs. These data demonstrate that L17-F05 could efficiently deliver mRNA to myeloid cells in dLNs via subcutaneous injection. The efficient targeting to LNs without systemic exposure should contribute to the safety of mRNA vaccines. To assess the safety profiles of L17-F05, we evaluated body and organ weights, plasma biochemistry, and hematology after repeated dosing with LNPs (Figure S5A). The body and organ weights of the LNP-treated group were not significantly different from those of the saline group, except for the weight of the spleen (Figures S5B and S5C). The plasma biochemistry and hematological data indicated no significant toxicity (e.g., aspartate transaminase [AST] and alanine aminotransferase [ALT]) in the LNP-treated group, although some markers related to inflammation (e.g., globulin and neutrophil count) changed after LNP treatment (Tables S3 and S4). These results indicate that F17-F05 vaccines were well tolerated but had mild immunogenicity.

Figure 4.

Biodistribution of L17-F05 mRNA vaccine

(A and B) Bioluminescence of organs in mice 4, 24, or 72 h after the subcutaneous injection of Fluc mRNA/L17-F05 into the back at 20 μg/mouse (A) or footpad at 5 μg/mouse (B). iLNs, inguinal LNs; bLNs, brachial LNs; pLNs, popliteal LNs; L, left; R, right. n = 3 independent experiments; mean ± SEM. (C–E) Flow cytometry analysis indicating LNP distribution (AZDye647) and EGFP mRNA translation in immune cells in popliteal lymph nodes from mice 4 or 24 h after subcutaneous injection of EGFP mRNA-AZDye647/L17-F05 into footpad at 5 μg/mouse. Representative FACS plots (C), ΔMFI of AZDye647 (D), and ΔMFI of EGFP (E). Med, medullary; SCS, subcapsular sinus. n = 3 independent experiments; mean ± SEM.

L17-F05 mRNA vaccine drove innate immune responses in dLNs

To investigate innate immune responses in dLNs after L17-F05 mRNA vaccination, we performed FCM analysis of popliteal dLNs post injection of OVA mRNA/L17-F05 into the footpad. Uniform manifold approximation and projection (UMAP) analysis demonstrated a change of cell clusters in CD45⁺ cells of dLNs post vaccination with the L17-F05 mRNA vaccine (Figure 5A). Changes in cell population within T and B cell clusters were detected on day 1 post vaccination, which were normalized on days 4 and 7. Myeloid clusters in CD45⁺ cells appeared to change their populations over time post vaccination. Thus, the myeloid cells were enriched and re-embedded on UMAP, with a strikingly increased number of neutrophils and monocytes 1-day post vaccination (Figure 5B). Cell number counting demonstrated that CD45⁺ cells and all subsets except for subcapsular sinus (SCS) macrophages showed increased cell numbers in dLNs within 7 days post vaccination, although there was a transient decrease in some subsets. (Figure 5C). Monocytes and neutrophils showed >10-fold increase in cell number post vaccination. We next focused on marker expression related to antigen presentation. Co-stimulatory molecules, e.g., CD40, CD80, and CD86, which are APC maturation markers, are essential for T cell priming. In the FCM data, CD40, CD80, and CD86 were upregulated in cDC subsets, and the expression reached a peak on day 1 post administration of the L17-F05 mRNA vaccine (Figure 5D). cDCs displayed higher levels of OVA antigen on H2Kb than other myeloid cells, including plasmacytoid DCs (pDCs), macrophages, and monocytes (Figure 5E).

Figure 5.

Innate immune response in dLNs post administration of L17-F05 mRNA vaccine

(A and B) UMAP analysis based on multi-color flow cytometry of CD45⁺ cells (A) and myeloid cells (B) in popliteal LNs from mice 24 h post administration of OVA mRNA/L17-F05 into the footpad. Med-Mac, medullary macrophages; SCS-Mac, subcapsular sinus macrophages. (C–E) Cell number analysis (C) and expression analysis for APC maturation markers (D) and OVA-bound H2b (E) in the immune cells in popliteal LNs from mice 24 h post administration of OVA mRNA/L17-F05 into the footpad. n = 3 independent experiments; mean ± SEM.

Transcriptional signatures related to IFNs and IFN-stimulated genes were upregulated by L17-F05 mRNA vaccine treatment

To investigate transcriptional signatures of myeloid cells in dLNs post administration of the L17-F05 mRNA vaccine, we collected dLNs from mice 24 h post vaccination, sorted six types of myeloid subsets using fluorescence-activated cell sorting (FACS), and performed RNA-seq analysis (Figure 6A). FACS gating strategy and expression patterns of linage markers in RNA-seq analysis are shown in Figures S6A and S6B, respectively. Heatmap and hierarchical clustering analysis based on RNA-seq data showed increased expression of markers related to proinflammation and antigen presentation throughout myeloid subsets in dLNs post administration of the L17-F05 mRNA vaccine (Figure 6B). Quantitative reverse transcriptase polymerase chain reaction (RT-qPCR) analysis exhibited significant upregulation of IFN-I (IFNa4 and IFNb1), IFN-stimulated genes (ISGs; Irf7 and Isg15), chemokines (Ccl2 and Cxcl10), and co-stimulatory molecules (Cd80 and Cd86) in myeloid subsets post vaccination (Figure 6C). IFN-I (Ifna4 and Ifnb1) and Ccl2 were upregulated in limited subsets, including macrophages, monocytes, and pDCs, whereas ISGs (Irf7 and Isg15) were upregulated in all myeloid subsets. Co-stimulatory molecules, Cd80 and Cd86, showed the highest expression in cDCs, consistent with the results of protein-level analysis using FCM. Ingenuity pathway analysis (IPA) based on RNA-seq data also revealed the activated or inhibited status of signaling pathways post administration of the L17-F05 mRNA vaccine in each myeloid subset. In upstream regulators, IFNs, including IFNB1 and IFNG, and ISGs, including IRF3 and IRF7, were predicted as activated upstream regulators (Figure 6D). In canonical pathways, proinflammatory pathways, including TREM1 signaling and the macrophage classical activation signal pathway (M1), were activated, particularly in macrophages and pDCs, whereas anti-inflammatory pathways, including IL-4 signaling and the macrophage alternative activation signaling pathway (M2), were inhibited in all myeloid cells (Figure 6E). IPA graphical summary denoted activation of inflammatory signal networks, including IFNB1, IRF3, and IRF7, in SCS macrophages and cDCs (Figure S6C). These data indicate that transcriptional signatures related to IFN signaling pathways were upregulated in myeloid cells in dLNs post administration of the L17-F05 mRNA vaccine.

Figure 6.

RNA-seq analysis of LN myeloid cells post administration of L17-F05 mRNA vaccine

(A) Schematic overview depicting bulk-RNA-seq post sorting myeloid cells in brachial LN from mice 24 h post administration of OVA mRNA/L17-F05. (B) Heatmap of proinflammatory and antigen-presentation-related genes in the RNA-seq analysis of myeloid cells. (C) RT-qPCR analysis of representative genes in myeloid cells. n = 3 independent experiments; mean ± SEM. (D and E) Heatmap of the activation Z score of significantly enriched upstream regulators (D) and canonical pathways (E) in IPA using differentially expressed genes obtained from RNA-seq data.

Depletion of LN macrophages attenuated proinflammatory responses in dLNs and anti-tumor effects induced by L17-F05 mRNA vaccine treatment

Among the myeloid cells in LNs, macrophages have been reported to be essential for the immune response to different types of vaccines.^23,24,25 However, the involvement of LN macrophages in the immune response to mRNA vaccine has not been elucidated yet. To study the role of LN macrophages in the proinflammatory response after L17-F05 mRNA vaccination, we performed macrophage-depletion experiments using chlodronate liposome (CLL) and control phosphate-buffered saline (PBS) liposome (PBSL). We injected CLL into the footpad for drainage to the popliteal LNs to deplete macrophages. FCM analysis of popliteal LNs on day 3 post injection demonstrated that CLL specifically depleted macrophages but not any other cell population in LNs (Figure S7). RT-qPCR analysis revealed that upregulated expression of proinflammatory genes in dLNs post vaccination was significantly suppressed by pre-treatment with CLL (Figure 7A). Recruitment of immune cells, including monocytes, neutrophils, and cDCs, post vaccination was significantly reduced by pre-treatment with CLL (Figure 7B). FCM analysis showed that the enhanced expression of CD80 and CD86 in cDCs post vaccination was lowered by the pre-depletion of LN macrophages with CLL (Figure 7C). To test whether macrophage depletion affected anti-tumor effects of the L17-F05 mRNA vaccine, we performed anti-tumor therapy experiments using B16-OVA mice along with the pre-depletion of macrophages following the regimen depicted in Figure 7D. Tumor growth curves revealed that macrophage depletion significantly suppressed tumor growth inhibition caused by the OVA mRNA/L17-F05 (p < 0.001 between the PBSL→OVA mRNA/LNP and CLL→OVA mRNA/LNP groups) (Figure 7E). Production of peripheral OVA-specific CD8⁺ T cells decreased significantly in the CLL-pretreated group compared with that in the control saline- (p < 0.05) and PBSL-pretreated groups (p < 0.001) (Figure 7F). These findings suggest that LN macrophages can trigger innate immune responses and potentiate anti-tumor effects induced post administration of the L17-F05 mRNA vaccine.

Figure 7.

Impact of macrophage depletion on inflammatory response in dLNs and anti-tumor effects induced by L17-F05 mRNA vaccine

(A) RT-qPCR analysis of proinflammatory markers in whole popliteal LN 24 h post administration of OVA mRNA/L17-F05 into the footpad of mice pretreated with PBSL or CLL. PBSL, PBS liposome; CLL, chlodronate liposome. n = 3 independent experiments; mean ± SEM. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001; one-way ANOVA followed by Tukey’s multiple comparison test. (B) Cell population change on day 3 after administration of OVA mRNA/L17-F05 into the footpad of mice pretreated with PBSL or CLL. n = 3 independent experiments; mean ± SEM. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001; one-way ANOVA followed by Tukey’s multiple comparison test. (C) Flow cytometry analysis of APC maturation markers on the surface of cDC on day 3 after injection of OVA mRNA/L17-F05 into the footpad of mice pretreated with PBSL or CLL. n = 3 independent experiments; mean ± SEM. ∗p < 0.05; Student’s t test. (D) Schedule of anti-tumor therapy experiments. Macrophages were depleted by treatment with CLL on day 4 after tumor inoculation, and OVA mRNA/L17-F05 was injected on days 7 and day 11. (E) Tumor growth curve of B16-OVA-bearing mice treated as indicated in (D). Red and blue arrows indicate time points of pre-treatment and vaccination, respectively. n = 8 independent experiments; mean ± SEM. ∗∗p < 0.01, ∗∗∗∗p < 0.0001; repeated-measures ANOVA followed by Bonferroni’s multiple comparison test. (F) OVA-specific CD8 T cells in peripheral blood on day 15 post inoculation. n = 8 independent experiments; mean ± SEM. ∗p < 0.05, ∗∗∗∗p < 0.0001; two-way ANOVA followed by Bonferroni’s multiple comparison test.

Self-adjuvant effects of the L17-F05 mRNA vaccine depend on the STING–IFN-I pathway

To understand detailed mechanisms underlying the adjuvant effects of L17-F05 against macrophages, we examined the involvement of TLRs 2/4/9 and cGAS–STING pathways using mouse BMDMs. RT-qPCR analysis showed that the L17-F05 mRNA vaccine (mRNA-LNP) induced the upregulation of proinflammatory genes, including Ifnb1, Il-6, and Tnf, in BMDMs, also with L17-F05 LNP formulated without mRNA (empty LNP) (Figure 8A). Ligands for TLRs (Pam3CSK4, TLR2 agonist; monophosphoryl lipid A [MPLA], TLR4 agonist; ODN1826, TLR9 agonist) and STING (2′,3′-cyclic guanosine monophosphate [cGAMP], a STING agonist) also promoted the expression of proinflammatory genes in BMDMs. Correlation analysis based on RT-qPCR data demonstrated that gene expression signature of 2′,3′-cGAMP-treated cells was highly similar to that of cells in the L17-F05-treated groups (empty LNP, r = 0.85; mRNA-LNP, r = 0.75), whereas TLR ligands exhibited weak correlation against 2′,3′-cGAMP and L17-F05 in the gene expression profiles of BMDMs (r < 0.3) (Figure 8B).

Figure 8.

Self-adjuvant effects of L17-F05 LNP via STING and IFN-I pathways

(A and B) Heatmap of proinflammatory gene expression (A) and correlation analysis of gene expression signature (B) determined by RT-qPCR of mouse BMDMs 24 h after treatment with OVA mRNA/LNP (L17-F05) or indicated PRR ligands, Pam3CSK4 (TLR2 agonist), MPLA (TLR4 agonist), ODN1826 (TLR9 agonist), and 2′,3′-cGAMP (STING agonist). (C–E) RT-qPCR analysis of Ifnb1, Il6, Mx1, and Isg15 (C), ELISA of IFN-β in culture supernatants (D), and flow cytometry analysis of APC maturation markers (E) of mouse BMDMs derived from WT, Sting1^gt/gt, and Ifnar1^−/− mice 24 h after treatment with empty LNP (L17-F05), OVA mRNA/LNP (L17-F05), or 2′,3′-cGAMP. n = 3 independent experiments; mean ± SD. ∗p < 0.05, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001; two-way ANOVA followed by Tukey’s multiple comparison test. (F) Western blot analysis of pTBK1, TBK1, and GAPDH (loading control) in BMDMs from WT mice over time after treatment with empty L17-F05, Fluc mRNA/L17-F05, empty ALC-0315, Fluc mRNA/ALC-0315, or 2′,3′-cGAMP. (G and H) ELISA of IFN-β in culture supernatants (G) and flow cytometry analysis of APC maturation markers (H) of BMDMs from WT mice 48 h after treatment with empty L17-F05, Fluc mRNA/L17-F05, empty ALC-0315, Fluc mRNA/ALC-0315, or 2′,3′-cGAMP. n = 3 independent experiments; mean ± SD. ∗∗∗∗p < 0.0001; two-way ANOVA followed by Tukey’s multiple comparison test.

To explore the pathways responsible for the adjuvant effects of the L17-F05 mRNA vaccine, we employed BMDMs from Tlr2^−/−, Tlr4^−/−, Tlr9^−/−, Sting1^gt/gt, and Ifnar1^−/− mice for loss-of-function studies. In RT-qPCR analysis, Ifnb1 and Il6 were upregulated by treatment with mRNA-LNP in BMDMs from wild-type homozygous (WT) mice, but the upregulation was significantly reduced in BMDMs derived from Sting1^gt/gt (p < 0.001) and Ifnar1^−/− mice (p < 0.001) (Figure 8C). ISGs, Mx1 and Isg15, were upregulated in WT BMDMs treated with mRNA-LNP and empty LNP, whereas the induced expression of these genes was significantly attenuated in Sting1^gt/gt (p < 0.0001) and Ifnar1^−/− BMDMs (p < 0.001). Upregulation of ISGs induced by empty LNP in WT BMDMs was absent in both Sting1^gt/gt and Ifnar1^−/− BMDMs, but that of Ifnb1, Il6, and ISGs induced by mRNA-LNP in WT BMDMs persisted partially in Sting1^gt/gt BMDMs and was completely lost in Ifnar1^−/− BMDMs. Similar to the results of RT-qPCR, ELISA data showed that IFN-β was released in culture supernatants of WT BMDMs post treatment with empty LNP or mRNA-LNP, but the release of IFN-β was significantly suppressed in Sting1^gt/gt and Ifnar1^−/− BMDMs (p < 0.001) (Figure 8D). FCM analysis demonstrated that L17-F05 with/without mRNA enhanced the expression of APC maturation markers (CD80 and CD86) in WT BMDMs but significantly abrogated in Sting1^gt/gt and Ifnar1^−/− BMDMs treated with empty LNP. It remained high in Sting1^gt/gt BMDMs but was absent in Ifnar1^−/− BMDMs treated with mRNA-LNP (Figure 8E). Knockout of Tlr2, Tlr4, or Tlr9 partially inhibited the induction of proinflammatory genes, IFN-β release, and APC maturation post treatment with L17-F05 vaccines; however, the magnitude of inhibition in Tlr2^−/−, Tlr4^−/−, Tlr9^−/− BMDMs was marginal compared with that in Sting1^gt/gt and Ifnar1^−/−BMDMs, indicating the pivotal role of the STING-IFN-I pathway compared with that of TLR2/4/9 pathways in the adjuvant effects of L17-F05 (Figures S8–S10). Western blotting of WT BMDMs demonstrated that L17-F05 vaccines efficiently activated TANK-binding kinase 1 (TBK1), a downstream signal molecule of STING, earlier than 2′,3′-cGAMP, whereas ALC-0315 vaccines did not (Figure 8F). Consistently, IFN-β was secreted by BMDMs after treatment with L17-F05 but not ALC-0315 vaccines (Figure 8G). Moreover, L17-F05 vaccines enhanced the expression of APC maturation markers (CD80 and CD86) more efficiently than ALC-0315 (Figure 8H). These studies suggested that L17-F05 has self-adjuvant effects through the STING–IFN-I pathway in BMDMs.

Discussion

Despite the emerging evidence indicating the efficacy of anti-tumor therapy using mRNA vaccines,^1,2,13 detailed mechanisms underlying the anti-tumor effects of mRNA vaccines are largely unknown. Herein, we found a novel L17-F05 LNP formulation for an mRNA vaccine with robust anti-tumor effects in a melanoma model. We investigated the molecular mechanisms potentiating the anti-tumor effects of L17-F05.

We first screened 34 ionizable lipids based on mRNA translation in human APCs for mimicking antigen presentation. Screening results revealed that the amine head and lipid tail largely affected protein expression. Expression was maximized at pKa values 6.9–7.0 of lipids using various head groups, consistent with previous reports of ionizable lipids on small interfering RNA (siRNA) therapeutics^21,26 and mRNA therapeutics,²⁷ indicating that the critical step in RNA delivery is endosomal escape regardless of RNA length. Among the head group, distinct Fluc expression activity was observed between stereoisomer heads (L02 vs. L03; L07 vs. L08). Previous reports have described different activity between stereoisomer lipids C12-S200 for mRNA delivery or 1,2-dioleoyl-3-trimethylammonium-propane (DOTAP) for siRNA delivery.^28,29 We need to consider whether a stereopure head can improve the delivery efficacy of L17-F05 in future works. Among the lipid tail group, the asymmetric lipid tail (>17 carbons: <10 carbons) exhibited greater mRNA expression activity than the symmetric lipid tail (>17 carbons: >17 carbons), wherein efficacies were enhanced when short branch (<9 carbons) existed at the tail end. This highlighted the importance of fine-tuning lipid tail structures for mRNA delivery, also suggested by another group.³⁰ We screened 28 LNP formulations, revealing that LNPs using DSPC-based LNPs showed higher Fluc expression than that of DOPC- and DOPE-based LNPs. The role of phospholipids has been suggested in LNP formulation for mRNA delivery.^31,32,33 Certain vehicles performed better with DSPC than with DOPE. Other vehicles (e.g., lipidoids) performed better with DOPE than with DSPC, suggesting that each LNP system has optimal phospholipids dependent on its ionizable lipid type.^34,35,36 Within the same molar ratio of four lipids, our DSPC-based L17-F05 showed 29-fold higher expression in BMDMs than DOPE-based L17-F23, confirming that phospholipids and ionizable lipids exhibit a profound impact on mRNA delivery.

In the organ distribution study, we demonstrated that subcutaneous administration efficiently delivered L17-F05 to dLNs with little distribution to systemic organs, e.g., livers, indicating that L17-F05 enabled LN-targeting delivery via subcutaneous administration as previously reported.^2,8 In the cellular distribution study, we demonstrated efficient LNP uptake and mRNA translation in myeloid cells, including macrophages and cDCs, but not in lymphoid cells. Similar to our observation, previous works have reported efficient myeloid cell targeting by LNPs.¹⁸ However, detailed cellular distribution and functional marker analysis focusing on myeloid subsets have rarely been reported. Herein, we observed different levels of LNP uptake and mRNA translation among cDC subsets in dLNs, suggesting that each subset exhibits distinct mechanisms underlying LNP uptake (e.g., endocytosis, phagocytosis, or micropinocytosis) and mRNA release from the endosome. For the antigen-presentation process, efficient antigen loading onto major histocompatibility complexes (MHCs) and the upregulation of co-stimulatory molecules in APCs are critical for antigen-specific T cell priming.^37,38 Among the myeloid subsets, cDCs most efficiently loaded OVA antigens onto H2Kb along with enhanced expression of CD80 and CD86 compared with that of other cell subsets post treatment with the L17-F05 mRNA vaccine, indicating that cDCs function as main APCs in comparison with the other APCs, e.g., macrophages, monocytes, and pDCs. Notably, MHC-II^hiCD103⁺ cDCs showed the highest potential for antigen presentation with enhanced expression of CD80/CD86. CD103⁺ cDCs are characterized as the cDC1 subset, specialized for cross-presenting antigens to CD8⁺ T cells in anti-tumor responses.³⁹ Herein, the L17-F05 mRNA vaccine exhibited robust induction of OVA-specific IFN-γ⁺CD8⁺ T cells in ELISPOT assays. However, we did not find much induction of OVA-specific IFN-γ⁺CD4⁺ T cells in the ELISPOT assays. Oberli et al. also did not detect OVA-specific CD4⁺ T cells post treatment with their OVA mRNA vaccine.² However, SARS-CoV-2 mRNA vaccines induce SARS-CoV-2-specific CD4⁺ T_FH cells, indicating that mRNA vaccines exhibit the potential to prime CD4⁺ T cells.¹⁶ The potential of our mRNA vaccine for CD4⁺ T cell priming warrants further testing using other antigens.

We revealed that proinflammatory immune responses related to IFN-I signaling were strikingly induced in myeloid cells in dLNs, with a potent gene expression signature of ISGs, including IRF3/7, post administration of the L17-F05 mRNA vaccine. Activation of the IFN-I pathway in myeloid cells is a hallmark of immune response to the BNT162b1 mRNA vaccine.¹⁸ However, the detailed roles of each myeloid subset are largely unknown. Herein, we confirmed the expression of IFN-I, Ifna4 and Ifnb1, in limited subsets, including macrophages, monocytes, and pDCs in dLNs. The highest Ifnb1 expression was observed in pDC post administration of the L17-F05 mRNA vaccine. However, the Ifnb1 expression in dLNs post vaccination was almost diminished by macrophage depletion with CLL, indicating that LN macrophages likely trigger IFN-I production in dLNs, although the main IFN-I-producing cells seemed to be pDCs. Previous studies have shown that pDCs are specialized for IFN-I production, and a positive feedback model has been proposed to explain robust IFN production from limited early IFN production.^40,41 Together, LN macrophages may contribute to early IFN-I production post vaccination, leading to subsequent robust IFN-I production in pDCs through positive feedback in the late phase. Our data also suggest that LN macrophages are responsible for enhanced expression of chemokines, Ccl2 and Cxcl10, post treatment with the L17-F05 mRNA vaccine. Ccl2 and Cxcl10 are involved in the trafficking of monocytes, DCs, and T cells to inflammatory sites.^42,43 FCM data demonstrated that all cell subsets, including lymphoid and myeloid, showed strikingly increased cell numbers in dLNs within 7 days post treatment with the L17-F05 mRNA vaccine. Thus, our data suggest that LN macrophages orchestrate IFN-I and chemokine signals in dLNs post treatment with the L17-F05 mRNA vaccine, leading to the recruitment of immune cells, maturation of APCs, and subsequent potent T cell priming. Among LN macrophages, SCS macrophages are localized in the frontline of LNs, i.e., SCS, and directly exposed to afferent lymph, allowing for a quick response to viruses and bacteria.⁴⁴ Quick response to mRNA-LNP induced by LN macrophages could be attributed to the localization of SCS macrophages in dLNs. In response to viruses and bacteria, LN macrophages produce IFN-I and recruit pDCs to SCS to further produce IFN-I in dLNs, and the depletion of LN macrophages lowers IFN-I and cytokine levels in dLNs, decreasing the recruitment of immune cells to dLNs and survival rate post infection.^45,46,47 Therefore, the innate immune responses of LN macrophages to our mRNA vaccine seem comparable with those of viruses and bacteria. In addition, FCM analysis demonstrated that the number of SCS macrophages in dLNs decreased after vaccination with L17-F05. LN macrophages are known to undergo necrosis soon after the administration of different types of vaccines.^23,24 This cell death can trigger the release of pre-stored cytokines or damage-associated molecular patterns (DAMPs) by LN macrophages, leading to an immediate immune response post vaccination. Similarly, the clearance of SCS macrophages after treatment with our mRNA vaccine might accompany the immediate immune response post vaccination, which could boost cytokine production in dLNs by releasing cytokines and DAMPs. Thus, the decrease of SCS macrophages would be a sign of a robust immune response by our mRNA vaccine.

This study also revealed that L17-F05 LNP induced the production of IFN-β by activating the STING pathway rather than TLR2/4/9 pathways. Regarding molecular mechanisms underlying IFN-I production induced by mRNA vaccines, the involvement of intrinsic immunogenicity of mRNAs or LNPs is known. mRNAs could be sensed by TLR3/7/8, RIG-I, and MDA5 through recognition of unmodified mRNA and dsRNA contaminants in mRNA vaccines.^1,9,18 Ionizable lipids in LNPs are sensed by the TLR4 and STING pathways through lipidoid structure and heterocyclic head structure, respectively.^10,11,12,13 Activation of these pathways, followed by the production of IFN-I, enhances immune responses of mRNA-LNP vaccines. Herein, Sting loss-of-function mutation (gt/gt) abrogated the adjuvant effects of L17-F05 without mRNA (empty LNP) but did not completely abrogate those of L17-F05 containing mRNAs (mRNA-LNP), suggesting that immune responses caused by L17-F05 are dependent on the STING pathway, but mRNAs in L17-F05 also stimulate STING-independent RNA sensors, e.g., TLR3/7/8, RIG-I, and MDA-5. Miao et al. reported that LNPs containing ionizable lipids with hetero cyclic amine head stimulate the STING pathway and enhance anti-tumor effects.¹³ Luo et al. reported that a polymeric nanoparticle with a cyclic side chain containing a seven-membered cyclic amine head activates the STING pathway.⁴⁸ L17-F05 ionizable lipid has a six-membered cyclic amine head, likely activating the STING pathway similar to that in the aforementioned reports. Interestingly, ALC-0315, which has no cyclic head, did not activate TBK1 signaling enough for BMDMs to secrete IFN-β, suggesting the possibility that the cyclic head of L17-F05 is a core structure that activates the STING pathway. The mechanisms of L17-F05 on the STING pathway warrant further evaluation.

In summary, our study provides insights into mechanisms underlying the anti-tumor effects of the L17-F05 mRNA vaccine. The self-adjuvanticity of L17-F05 via the STING–IFN-I pathway triggered by LN macrophages enhanced the maturation of cDCs and potentiated the anti-tumor efficacy of the L17-F05 mRNA vaccine. It is important to mention the limitations of this work for clinical translation. First, we performed most studies using mice; evaluation in non-human primates and humanized models is needed. Second, we selected subcutaneous administration. We hope to test other clinically relevant injection roots (intramuscular and intravenous) in future work. Third, we highlighted the efficacy of mRNA vaccines based on STING activation. Adverse effects through STING activation warrant consideration. Thus, fine-tuning of LNP formulations while balancing efficacy and safety is needed. Despite these limitations, we believe our findings are helpful for further optimization and rational design of mRNA-LNP vaccines for clinical application.

Materials and methods

Materials

Materials used in LNP preparation were purchased as follows: DSPC, DOPC, DOPE, and cholesterol from Nippon Fine Chemical (Osaka, Japan); DMG-PEG from NOF (Tokyo, Japan); sodium acetate, PBS powder, ethanol, and Triton X-100 from FUJIFILM Wako (Osaka, Japan). Ionizable lipids were synthesized as per our previous patents (WO2017222016A1 and WO2019131580A1).

Establishment of B16F10 cells stably expressing chicken egg ovalbumin

B16F10 (B16) cells were obtained from ATCC (CRL-6475; Manassas, VA) and cultured in DMEM (FUJIFILM Wako) containing 10% fetal bovine serum (FBS; GIBCO, Grand Island, NY, USA), 1.0 mM sodium pyruvate (GIBCO), and 1× penicillin-streptomycin (P/S) (FUJIFILM Wako) at 37°C in 5% CO₂. B16-OVA cells were established using a piggyBac transposon vector system (System Biosciences, Palo Alto, CA, USA). Codon-optimized OVA was cloned into a piggyBac vector with a puromycin-resistant gene, resulting in pPB-CMV-OVA-Puro. Transposase expression vector and pPB-CMV-OVA-Puro or pPB-CMV-Puro (control vector) were co-transfected into B16 cells. On day 7 post transfection, cells were selected with 1 μg/mL puromycin (Thermo Fisher Scientific, Waltham, MA, USA) for 2 weeks, resulting in B16-OVA and B16-control cells.

Mice

Female C57BL/6J (000664), C57BL/6J-Sting1^gt/J (017537) (Sting1^gt/gt),⁴⁹ and B6(Cg)-Ifnar1^tm1.2Ees/J mice (028288) (Ifnar1^−/−)⁵⁰ were obtained from The Jackson Laboratory (Bar Harbor, ME, USA). Female B6.129P2-Tlr2^tm1Aki/Bos (Tlr2^−/−),⁵¹ B6.129P2-Tlr4^tm1Aki/jobs (Tlr4^−/−),⁵² and B6.129P2-Tlr9^tm1Aki/Obs mice (Tlr9^−/−)⁵³ were purchased from Oriental BioService (Kyoto, Japan). Animal care and experimental procedures were performed in an animal facility accredited by the Health Science Center for Accreditation of Laboratory Animal Care and Use of the Japan Health Sciences Foundation. All protocols were approved by the Institutional Animal Care and Use Committee of Eisai (Tokyo, Japan) and performed following the Animal Experimentation Regulations.

Preparation of human BMDMs and BMDCs

Human BMDMs and BMDCs were prepared from bone marrow CD34⁺ cells (Lonza, Basel, Switzerland). Bone marrow CD34⁺ cells were cultured in StemSpan II (Stem Cell Technologies, Vancouver, Canada) containing 1× StemSpan Myeloid Expansion Supplement (Stem Cell Technologies) and 1× P/S for 14 days, followed by a 7-day culture in StemSpan II containing 50 ng/mL recombinant human M-CSF (rhM-CSF; FUJIFILM Wako), 25 ng/mL recombinant human granulocyte-macrophage colony-stimulating factor (GM-CSF) (rhGM-CSF; FUJIFILM Wako), and 1× P/S for myeloid differentiation. Harvested cells were cultured in the basal medium (RPMI-1640 [FUJIFILM Wako], 10% FBS, and 1× P/S) containing 50 ng/mL rhM-CSF for 7 days for BMDM preparation or in basal medium containing 50 ng/mL rhGM-CSF and 20 ng/mL recombinant human interleukin-4 (rhIL-4; FUJIFILM Wako) for 7 days for BMDC preparation.

Preparation of murine BMDMs and BMDCs

BMDMs were differentiated from bone marrow cells of B6 (WT), Sting1^gt/gt, Infa1r^−/−, Tlr2^−/−, Tlr4^−/−, and Tlr9^−/− mice as follows: mice were euthanized by cervical translocation, and the femurs were collected. Bone marrow cells were harvested from the femurs and cultured in RPMI-1640 containing 10% FBS, 1× P/S, and recombinant mouse M-CSF (40 ng/mL; FUJIFILM Wako) for 7 days for macrophage differentiation. On day 7, BMDMs were re-seeded onto 24-well plates at 1.2–1.5 × 10⁵ cells/well and treated with LNPs or ligands the following day. BMDCs were prepared using the same protocol as that for BMDMs, whereby recombinant mouse GM-CSF (20 ng/mL; FUJIFILM Wako) and recombinant mouse IL-4 (20 ng/mL; FUJIFILM Wako) were used for differentiation.

Preparation of mRNA

mRNAs used in this study were purchased from TriLink BioTechnologies (San Diego, CA) or prepared as previously described.⁵⁴ EGFP mRNA was hybridized with fluorescence-labeled oligonucleotides. Four oligonucleotide probes were designed to hybridize at the 3′ UTR of EGFP mRNA (Table S3). Oligonucleotides were prepared as 2′-O-methyl RNA 17–18 nucleotides with C6-amino modifier at the 5′ and 3′ ends (GeneDesign, Osaka, Japan). Both 5′ and 3′ ends of oligonucleotides were labeled with AZDye 647 TFP Ester (Fluoroprobes, Scottsdale, AZ) following the manufacturer’s protocols, and excess dye was removed using a 3-kDa ultrafiltration unit (Millipore, Burlington, MA, USA). EGFP mRNA was mixed with four fluorescence-labeled probes in a 1:7 molar ratio and heated at 70°C for 10 min. Post incubation at room temperature for 10 min, the free fluorescence-labeled oligonucleotides were removed with 100-kDa ultrafiltration. mRNAs with probes were concentrated and used for LNP preparation.

Preparation of LNP

mRNA-LNP was prepared as described earlier.^17,54,55 mRNA was dissolved in 50 mM sodium citrate, pH 4.0, whereas ionizable lipids, DSPC, cholesterol, and mPEG2000-DMG were dissolved in ethanol. For formulation optimization, mole ratios of cationic lipids, phospholipids (DSPC/DOPC/DOPE), and cholesterol were varied. mRNA and lipid solutions were mixed in a 3:1 ratio (v/v). Resultant LNPs were dialyzed against PBS (pH 7.4) in 100-kDa dialysis tubes at 4°C overnight, followed by filtration with a 0.22-μm filter before using for further experiments. For cryopreservation, dialysis was also performed against 8% (w/v) sucrose with 20 mM Tris-HCl (pH 7.5) in 100-kDa dialysis tubes at 4°C overnight, and LNPs were stocked at −70°C before use.

LNP characterization

Particle sizes and polydispersity indexes were determined by dynamic light scattering using a Zeta Sizer (Malvern Panalytical, Malvern, UK). Total mRNA concentrations in LNPs were determined using a Quant-iT RiboGreen RNA assay kit (Thermo Fisher Scientific) as previously described.⁵⁶ Encapsulation efficiency (EE; %), was calculated as follows: EE (%) = (1 − free mRNA concentration/total mRNA concentration) × 100. pKa values of cationic lipids were measured using a 2-(p-toluidino)-6-napthalene sulfonic acid (TNS) fluorescent assay as previously described.²¹

Cryogenic electron microscopy

A 3-μL droplet of mRNA/LNP solution was applied to a glow-discharged holey carbon grid (Cu R1.2/1.3, 300 mesh, #M2955C-1–300; Quantifoil Micro Tools, Jena, Germany). The grid was blotted for 15 s with a blot force of 0 and flash-frozen in liquid ethane using Vitrobot Mark IV (Thermo Fisher Scientific) at 18°C and 100% humidity. Data were collected on a Talos Arctica electron microscope (Thermo Fisher Scientific) equipped with a Falcon 3 direct electron detector at 200 kV. Images were acquired at a defocusing of −1 μm with a nominal magnification of 92,000 at a pixel size of 1.13 Å.

Screening of ionizable lipids using human BMDMs and BMDCs

Fluc mRNA/LNPs (50 ng) were added to 96-well plates pre-seeded with BMDMs or BMDCs at 1 × 10⁴ cells/well. After 24-h incubation, Fluc expression and cell viability were measured using Steady-Glo (Promega, Madison, WI, USA) and Cell-Titer Glo (Promega), respectively. Luminescence was quantified using a Nivo multiplate reader (PerkinElmer, Waltham, MA, USA). Fluc expression was normalized by cell viability.

Biodistribution analysis

Fluc mRNA/LNPs were used for organ distribution analysis. C57BL/6J mice were injected with Fluc-LNPs subcutaneously in the back (20 μg) or footpad (5 μg). After 4, 24, and 72 h, organs (liver, kidney, spleen, lung, heart, and LN) were collected and homogenized in Glo Lysis buffer (Promega) using a TissueLyzser II (Qiagen, Hilden, Germany). Fluc expression and total protein amount were measured using Steady-Glo (Promega) and Pierce 660-nm protein assay (Thermo Fisher Scientific), respectively. Luminescence was quantified using a Nivo multiplate reader (PerkinElmer). Absorbance was quantified using a CLARIOstar microplate reader (BMG LABTECH, Ortenberg, Germany). Fluc expression was normalized by total protein amount.

LNPs encapsulating EGFP mRNA with an AZDye 647 probe (EGFP-AZDye647/LNPs) was used for simultaneous analysis of cellular distribution and mRNA translation. C57BL/6J mice were injected with EGFP-AZDye 647/LNPs subcutaneously in the right footpad (5 μg). After 4 and 24 h, right popliteal LNs were collected and analyzed using flow cytometry (described in the “flow cytometry” section).

Safety study

Female C57BL/6J mice were injected with Fluc mRNA/LNPs subcutaneously in the back at a 0.5-mg/kg dose on days 0, 3, and 6. On day 7, blood was collected by cardiac puncture under isoflurane anesthesia for plasma biochemical and hematological analyses (n = 5/group). We measured plasma biochemistry using a 7180 Clinical Analyzer (Hitachi High-Technologies, Tokyo, Japan) and hematological items using an automated hematology analyzer (XN-2000V; Sysmex, Kobe, Hyogo, Japan). Histopathological analysis was performed for the liver, spleen, and kidney.

Tumor therapy experiments

C57BL/6J mice aged 6–8 weeks were inoculated subcutaneously with B16 cells at 5 × 10⁵ cells/spot. For a bilateral tumor model, B16-OVA and B16-control cells were inoculated into the right and left flanks, respectively. OVA mRNA/LNPs were injected into the center of the back (20 μg) on days 3, 7, and 10. For a unilateral B16-OVA or B16 model, the cells were inoculated into the right flank, and mice were randomized by tumor volumes (50–100 mm³) on day 7. Control or target mRNA/LNP vaccines were injected into the back (10 μg) on days 7, 10, and 13. Tumor volumes were measured twice a week using a digital caliper and calculated as (a² × b)/2 (a, width; b, length). Mice were euthanized when exhibiting signs of impaired health or tumor volume exceeding 2,000 mm³.

Preparation of RNA-seq libraries of myeloid cells in LN

C57BL/6J mice were injected with OVA mRNA/L17-F05 subcutaneously at the back (20 μg). After 24 h, brachial LNs were collected and pooled from four mice, digested with RPMI1640 containing 0.3 mg/mL Liberase DL (Roche, Basel, Switzerland) and 0.2 mg/mL DNase I (Roche) at 37°C for 30 min, and filtered through a 70-μm-mesh cell strainer. Cells were pre-incubated with mouse BD Fc Block (BD Biosciences, San Jose, CA, USA) for 10 min and stained with fluorophore-conjugated antibodies as follows: anti-CD45 (3F-11), anti-CD11b (M1/70), anti-CD11c (N418), anti-IA/IE (M5/114.15.2), anti-CD317 (129C1), anti-Ly6C (HK1.4), anti-Ly6G (1A8), anti-CD169 (3D6.112), and anti-F4/80 (BM8). All antibodies were obtained from BioLegend (San Diego, CA, USA) or BD Biosciences. Myeloid cell subsets (SCS macrophages, Med macrophages, monocytes, neutrophils, cDCs, and pDCs) were sorted using SH800 (SONY, Tokyo, Japan). Total RNA was extracted and purified from the collected cells (>1,000 cells) using an RNeasy micro kit (Qiagen) following the manufacturer’s instructions and converted into cDNA using an SMART-Seq HT kit (Takara, Kusatsu, Japan). Sequencing libraries were generated using Nextera XT DNA Library Preparation Kit (Illumina, San Diego, CA, USA). The fragment length and quality of the cDNA library were assessed with a 2100 Bioanalyzer (Agilent, Santa Clara, CA, USA).

RNA-seq analysis

Paired-end sequencing of 2 × 75 bp was conducted using NextSeq 550 (Illumina) following the manufacturer’s instructions. Sequencing data were processed as previously described.⁵⁷ Raw reads were trimmed for raw quality reads using Trimmomatic (http://www.usadellab.org/cms/?page=trimmomatic). Reads were mapped to mouse transcriptome annotations and reference genome sequences (Mm10 and Ensembl 84) using STAR. RSEM was used for estimating gene-level expression reported as transcripts per million (TPM). RNA-seq data have been deposited in the NCBI Gene Expression Omnibus under accession number GSE239881. Gene expression with TPM > 0 in at least one sample was selected and transformed to log₂ (TPM + 0.001). Pathway analysis was conducted using the IPA platform (Qiagen). Differentially expressed genes were defined by |log₂-fold change (FC)| > 2 between saline and treated groups in each subset. Canonical pathways and upstream regulators were filtered by |activation Z score| >3 (p < 0.01) and |activation Z score| >5 (p < 0.01), respectively.

Macrophage depletion

To deplete popliteal LN macrophages, B57BL/6J mice were injected with 30 μL of 5.0 mg/mL CLL (LIPOSOMA, Amsterdam, Netherlands) into the footpad on day 3 pre-vaccination (day −3). PBS liposome (LIPOSOMA) was used as a control. For RT-qPCR and flow cytometry analysis of LN, mice were vaccinated with OVA mRNA/L17-F05 into the footpad on day 0. Popliteal LNs were collected and proceeded to RT-qPCR analysis on day 1 and used for flow cytometry analysis on day 3. For tumor therapy experiments, mice were inoculated with B16-OVA cells into the right flank on day 0 and injected with CLL or PBSL into the footpad on day 4. On day 7, mice bearing tumors were randomized by tumor volumes (50–100 mm³) and received vaccination via injection of OVA mRNA/L17-F05 into the right footpad at 10 μg/mouse on days 7 and 11. Tumor volumes were measured twice a week using a digital caliper and calculated as (a² × b)/2 (a, width; b, length).

Flow cytometry

To analyze peripheral blood cells, blood was collected from the caudal vein, mixed with saline (Otsuka Pharmaceutical, Tokyo, Japan) containing heparin (10 units/mL; AY Pharmaceuticals, Tokyo, Japan), and treated with BD Pharm Lyse buffer (BD Biosciences) for removing erythrocytes. For LN analysis, LNs were collected, digested with RPMI1640 containing 0.3 mg/mL Liberase DL (Roche) and 0.2 mg/mL DNase I (Roche) at 37°C for 30 min, and filtered through a 70-μm-mesh cell strainer. Cells were pre-incubated with mouse BD Fc Block (BD Biosciences) for 10 min and stained with fluorophore-conjugated antibodies as follows: anti-CD45 (3F-11), anti-B220 (RA3-6B2), anti-CD3ε (145-2C11), anti-CD4 (RM4-5), anti-CD8 (53–6.7), anti-NK1.1 (PK136), anti-CD49b (DX5), anti-CD11b (M1/70), anti-CD11c (N418), anti-IA/IE (M5/114.15.2), anti-CD317 (129C1), anti-Ly6C (HK1.4), anti-Ly6G (1A8), anti-CD169 (3D6.112), anti-Siglec F (E50-2440), anti-F4/80 (BM8), anti-CD103 (2E7), anti-CD40 (3/23), anti-CD80 (16-10A1), and anti-CD86 (GL-1). All antibodies were obtained from BioLegend or BD Biosciences. Stained cells were analyzed using a BD LSRFortessa Flow Cytometer (BD Biosciences) or a BD FACSymphony Flow Cytometer (BD Biosciences). Detailed data analysis was performed using FlowJo software (Tree Star, Ashland, OR, USA).

RT-qPCR

For LN analysis, whole LNs were collected and homogenized using a TissueLyzser II (Qiagen), and total RNA was purified using a Maxwell RSC system (Promega). For analyzing cultured cells, total RNA was extracted and purified using an RNeasy kit (Qiagen) following the manufacturer’s instructions. Reverse transcription was performed with 100 ng of total RNA using the PrimeScript RT kit (Takara). For RT-qPCR, each reaction mix of a total of 10 μL contained 0.5 μL of TaqMan probe, 5.0 μL of TaqMan Gene Expression Master Mix (Thermo Fisher Scientific), and 4.5 μL of cDNA. RT-qPCR reactions were performed using a ViiA7 Real-Time PCR System (Applied Biosystems, Waltham, MA, USA) at 95°C for 10 min, followed by 45 cycles of 95°C for 15 s and 60°C for 1 min. TaqMan probes were purchased from Thermo Fisher Scientific and included Il1b (Mm00434228_m1), Il6 (Mm00446190_m1), Ifnb1 (Mm00439552_s1), Ifna4 (Mm00833969_s1), Tnf (Mm00443258_m1), Ccl2 (Mm00441242_m1), Cxcl10 (Mm00445235_m1), Cd80 (Mm00711660_m1), Cd86 (Mm00444543_m1), Irf3 (Mm00516784_m1), Irf7 (Mm00516793_g1), Mx1 (Mm00487796_m1), Isg15 (Mm01705338_s1), and Gapdh (Mm99999915_g1).

IFN-β ELISA

IFN-β levels in culture supernatants of mouse BMDMs treated with STING ligands, L17-F05, and ALC-0315 were measured using mouse IFN-β ELISA kit (R&D Systems, Minneapolis, MN, USA) following the manufacturer’s protocol. Plates were scanned at 450 nm using a Nivo multiplate reader (PerkinElmer).

IFN-γ ELISPOT assay

ELISPOT was performed with an IFN-γ mouse ELISPOT kit (R&D Systems) following the manufacturer’s instructions. C57BL/6J mice were immunized with OVA mRNA/L17-F05 (20 μg) subcutaneously. On day 7, splenocytes were collected from mice, seeded onto a 96-well ELISPOT plate at 2 × 10⁵ cells/well, and pulsed with 2.0 μg/mL OVA peptide (OVA^257-264, SIINFEKL, or OVA^323-399, ISQAVHAAHAEINEAGR; MBL, Tokyo, Japan) at 37°C for 24 h. Plate images were acquired using BZ-X810 (KEYENCE, Osaka, Japan), and spots were counted with ImageJ (https://imagej.nih.gov/ij/) software.

Western blotting

Cells were lysed in sample buffer (BioRad, Hercules, CA, USA) containing PhosSTOP (Roche), protease inhibitor cocktail (Roche), and 20 mM dithiothreitol (Sigma-Aldrich, St. Louis, MO, USA) followed by heating at 95°C for 5 min. Samples were resolved by SDS-PAGE and western blotting following standard protocols. Antibodies used for the analysis of BMDMs included anti-pTBK1 (1:1,000), anti-TBK1 (1:1,000), and anti-GAPDH (1:5,000), purchased from Cell Signaling Technology (Danvers, MA, USA). Horseradish peroxidase (HRP)-conjugated antibodies were used as second antibodies, i.e., anti-rabbit IgG-HRP (1:2,500; Cytiva, Tokyo, Japan). The signal was visualized by chemiluminescence using LuminataForte HRP substrate (Millipore) with a FUSION FX (VILBER, Marne-la-Vallee, France).

Treatment of mouse BMDMs with LNPs or TLR/STING ligands

To examine adjuvant effects of LNPs, mouse BMDMs were treated with L17-F05 (1.0 μg/mL), ALC-0315 (1.0 μg/mL), Pam3CSK4 (TLR2 agonist; 0.1 μg/mL), MPLA (TLR4 agonist; 0.1 μg/mL), ODN1826 (TLR9 agonist; 1.0 μg/mL), or 2′,3′-cGAMP (STING agonist; 5 or 10 μg/mL). After incubation, cells were harvested and proceeded to RT-qPCR, flow cytometry, and western blotting. Culture supernatants were used for IFN-β ELISA.

Statistical analysis

GraphPad Prism 9.2 (GraphPad Software, Boston, MA, USA) was used for preparing graphs and statistical analysis. Student’s t test was performed when comparing two groups. For comparing >2 groups, one-way or two-way analysis of variance (ANOVA) was applied. Repeated-measures ANOVA was conducted to analyze repeated-measures data. Survival rates were analyzed using log rank tests. The difference was considered significant at p < 0.05.

Data and code availability

The authors confirm that the data supporting the findings of this study are available within the article and its supplementary material.