Intratumoral IL12 mRNA Therapy Promotes TH1 Transformation of the Tumor Microenvironment

This article is featured in Highlights of This Issue, p. 6075

Patients who respond to programmed death-1/programmed cell death 1 ligand (PD-1/PD-L1) immune checkpoint blockade (ICB) tend to have an inflamed, TH1-polarized tumor microenvironment (TME), characterized by expression of IFNγ and PD-L1 (1). Novel therapies that induce TH1 transformation of the patient TME therefore have the potential to enhance antitumor responses in patients that are currently nonresponders to ICB.

As a central mediator of TH1 immune responses, IL12 is known to play a key role in driving antitumor immunity. IL12 guides the differentiation of TH1 T cells and enhances the activation and cytotoxic activity of natural killer cell (NK), natural killer T cells (NKT), and cytotoxic T cells (CTL; ref. 2). Many of the downstream effects of IL12 are mediated via induction of the key TH1 cytokine, IFNγ, by innate and adaptive immune cells. IL12 acts both directly and via IFNγ to promote antigen presentation, reduce myeloid immunosuppression, and to enhance T-cell recruitment through production of other TH1 chemokines including IFNγ-inducible protein 10 (IP-10; refs. 3, 4). IL12 also exerts antiangiogenic effects through IP-10 (5).

Systemic recombinant IL12 (recIL12) promotes IFNγ- and CTL-dependent antitumor immunity in a wide variety of mouse syngeneic tumor models (6, 7). NK and NKT cells also play a key role in driving antitumor activity of IL12 in some mouse models (8–10). However, systemic recIL12 was poorly tolerated in early clinical trials, and efficacy of systemic recIL12 has been limited at tolerated doses (11, 12).

The narrow therapeutic margin of systemic recIL12 has led to the development of alternative local strategies for direct delivery of IL12 to the TME. IL12 plasmid intratumoral injection with electroporation drives local IL12 production and cytotoxic T-cell–dependent antitumor immunity in mice bearing various syngeneic tumors (13, 14). Intratumoral administration of IL12-encoding adenovirus and oncolytic viruses promotes antitumor immune responses in preclinical models, and this activity is enhanced by combination with ICB therapy (15–17). Intratumoral delivery of dendritic cells or of antitumor CD8⁺ T cells engineered to express IL12 also drives tumor regression in preclinical models (18, 19). IL12 plasmid (tavokinogene telseplasmid) intratumoral delivery by electroporation demonstrated clinical activity in patients with metastatic melanoma, inducing regression of both treated and untreated lesions (20). However, the requirement for an electroporation device limits tavokinogene telseplasmid utility to patients with superficial lesions.

Here we report on the development of a novel lipid nanoparticle (LNP)-formulated, IL12 mRNA-based therapy designed for intratumoral injection in patients with superficial and deep-seated lesions. We found that a single intratumoral dose of mouse IL12 mRNA induced tumor regression in multiple tumor models. This was both CD8⁺ T-cell and IFNγ dependent and correlated with upregulation of signature TH1 immune response genes, thus TH1 TME transformation. In a PD-L1–resistant model, antitumor immunity induced by mouse IL12 mRNA was enhanced by PD-L1 blockade. Mouse IL12 mRNA also drove regression of uninjected distal lesions, and anti–PD-L1 potentiated this response. Intratumoral delivery of mRNA encoding membrane-tethered mouse IL12 induced regression of untreated lesions in the absence of increases in circulating IL12, supporting the hypothesis that local IL12 could induce a systemic antitumor immune response against distal lesions. Finally, we report that in ex vivo patient tumor slice cultures, human IL12 mRNA (MEDI1191) induced dose-dependent IL12 production, IFNγ expression, and TH1 TME transformation.

mIL12 mRNA and MEDI1191 mRNA incorporate wild-type mouse IL12b and IL12a sequences (NM_001303244.1 and NM_001159424.2) and optimized human IL12A and IL12B mRNA sequences (human NM_002187.2 and NM_000882, in-house algorithm applied to optimize translation). For both MEDI1191 and mIL12 mRNA, a linker was added between the IL12B and IL12A sequences as published previously (21) to generate a linked monomeric IL12p70 (IL12B-IL12A). The IL12B signal peptide was retained for secretion, and the IL12A signal peptide was removed [first 22 amino acids for both mouse and human IL12A, according to Uniprot (mouse P43431 and human P29459)]. For tethered mIL12, the C terminus of the single-chain IL12B-IL12A construct was linked via a peptide linker (SG₃SG₄SG₄SG₄SG₃SLQ) to a PDGFRβ transmembrane domain, followed by a G₄S linker and V5 peptide tag at the intracellular end. mRNA constructs incorporated a miR122 binding site in the 3′ UTR (22), except where indicated in Fig. 1B. Control mRNA is nontranslating where the initiating AUG codon and any other potential initiating codons were modified. mRNA was synthesized and LNP formulated as described previously (23–26) and briefly detailed in the Supplementary Methods.

In vitro IL12p70 expression was quantified by DuoSet ELISA (R&D Systems) or Mesoscale Discovery assay (Mesoscale Discovery) in the supernatants and lysates of IL12 mRNA–treated cells. Human and mouse CD8⁺ T cells were cultured with species-specific CD3/CD28 Dynabeads and either HeLa supernatant containing mIL12 mRNA- or MEDI1191-derived IL12p70 or HeLa cells transfected with tethered mIL12 mRNA. IFNγ levels were measured by DuoSet ELISA (R&D Systems).

All in vivo experiments were carried out either in accordance with the Institutional Animal Care and Use Committee (IACUC) at Moderna Inc., MI Bioresearch and Crown Bioscience San Diego, Medimmune, or the UK Animals (Scientific Procedures) Act for studies at MedImmune. For syngeneic models, female C57BL/6 mice (The Jackson Laboratory or Charles River UK) or Balb/c mice (Charles River Laboratories) were implanted subcutaneously (SC) with MC38-S, MC38-R, B16F10-AP3, or A20 tumor cells. For PDX models NOD-SCID mice were implanted with subcutaneous melanoma (ME12057, ME12058) or head and neck (HN5111, HN5116) tumor cell slurries (Crown Bioscience). Tumors were measured with calipers, and volumes were calculated using the formula volume = (length × width²)/2. Tumor volumes at randomization (24 hours prior to intratumoral mRNA treatment or on the day of treatment) are listed in Supplementary Table S2. mRNAs were administered intratumoral in a fixed 25 μL volume to mice bearing established tumors. Antibodies were administered intraperitoneally as detailed in Supplementary Table S3.

Survival events were either recorded on the day when total tumor volume exceeded 1,500 or 2,000 mm³, or on the day when animals were removed from study due to specific negative clinical signs (e.g., tumor ulceration, weight loss > 20%), whichever event occurred first. CRs were defined as animals with no measurable tumor at all implant sites at study completion. CR rates and Kaplan–Meier survival plots are based on all animals enrolled in each group, except for the A20 tumor model where any animals that grew secondary masses are removed from the displayed data. IL12p70 and IFNγ were quantified with the ProcartaPlex Mix&Match Mouse 5-plex, Cytokine and Chemokine 36-Plex Mouse ProcartaPlex Panel 1A (Thermo Fisher Scientific), electrochemiluminescent assay for mouse IL12p70 or mouse IFNγ (V-plex, Meso Scale Discovery), or DuoSet ELISA (R&D Systems) as indicated.

Syngeneic tumors were dissociated to single cells, counted, normalized, and stained for viability (Fixable Viability Dye or Live/Dead Aqua, Thermo Fisher Scientific; Supplementary Table S4). Peripheral blood and spleen cells were stained with Zombie NIR viability dye or Live/Dead Aqua (Thermo Fisher Scientific). Tumor and spleen cells were blocked with TruStain FcX (anti-mouse CD16/32, BioLegend) prior to fluorescent antibody staining (antibodies detailed in Supplementary Table S4). Tumor samples were run on an LSRFortessa or FACSymphony (BD Biosciences) with 123count eBeads (Thermo Fisher Scientific). Peripheral blood and spleen samples were run on an Attune NXT flow cytometer or FACSymphony (BD Biosciences). All analyses were carried out using FlowJo V10 (TreeStar).

A total of 1 × 10⁵ splenocytes per well were seeded into precoated 96-well mouse IFNy ELISpot plates (MABTECH) in RPMI with 10% FBS, 1% penicillin and streptomycin, 50 nmol/L 2-Mercaptoethanol, 10 ng/mL IL2 (Roche). Splenocytes were stimulated with 1 or 10 μg/mL tumor antigenic peptides (H-2Kb MuLV P15E peptide KSPWFTTL and H-2K TRP-2 peptide SVYDFFVWL both from MBL). Plates were incubated at 37°C, 5% CO₂ for 48 hours before completing the IFNγ detection assay. Plates were counted on a CTL ImmunoSpot 6 Ultra-V analyzer.

All samples were obtained with written informed consent from patients. Studies were conducted in accordance with recognized ethical guidelines (Declaration of Helsinki) and were approved by the East of England—Cambridge East Research Ethics Committee (MedImmune Research Tissue Bank, RTB 16/EE/0334, HTA license number 12283). Fresh endometrial, colorectal adenocarcinoma, and colorectal liver metastasis tumor samples from surgical resections (Nottingham Health Service Biobank, Tissue Solutions and Scievita) were transferred to MedImmune Ltd in ice-cold Aqix RS-I solution (Life Science Group). Upon receipt (<24 hours postresection), samples were embedded in 4% low-melting point agarose (VWR) prepared in PBS, and vibratome sectioned (300-μm slices, VT-1200S, Leica). Fresh tumor slices were transferred to an organotypic insert (Millipore) in a 6-well plate containing RPMI1640 with 10% FBS and 1% penicillin/streptomycin. MEDI1191 or control RNA was added directly to the media at the desired concentration. Every 24 hours, an aliquot of culture media was collected for analysis, and replaced with the same volume of fresh complete media. Culture supernatants and tissue slices were harvested at endpoint after 3–5 days. IL12p70, CXCL10, and IFNγ were quantified in endpoint culture supernatants by MesoScale Discovery assay (MesoScale Discovery) and ELISA (R&D Systems).

mRNA was isolated from snap-frozen syngeneic tumors using an RNeasy Mini Kit (Qiagen) and quantified on the Affymetrix mouse 430 2.0 microarray. Interchip normalization was performed using the fRMA method (27). Technical outlier samples were identified by principle component analysis and removed from the analysis. Significantly differentially expressed genes (DEG) in mIL12 mRNA versus control mRNA-treated tumors were identified as those with fold change ≥ 2 with an P_adj < 0.05. ToppFun pathway analysis was performed on DEG lists (https://toppgene.cchmc.org/) with a false discovery–corrected P value cutoff of 0.05. Full DEG lists and pathway analysis results (Supplementary Table S1) underwent further manual curation to identify relevant groups of transcripts prior to the generation of heatmaps (TIBCO Spotfire). Microarray data have been deposited in the ArrayExpress database at EMBL-EBI (www.ebi.ac.uk/arrayexpress) under accession number E-MTAB-9316.

Patient tumor slices were placed in TRIzol (Invitrogen), disrupted in a Tissue Lyser (Qiagen) and total RNA was extracted with a Direct-zol RNA Miniprep Kit (Zymo Research). Fifty nanograms total RNA was evaluated using the 770-gene PanCancer IO 360 panel (NanoString Technologies). Gene counts were normalized to internal reference standards using the nSolver software. A further interchip normalization was performed with the ComBat algorithm (28) as implemented in R (R Foundation for Statistical Computing), and genes were identified, which were upregulated >1.5-fold versus matched tumor slices treated with control mRNA with an unadjusted P value of < 0.05. All significantly upregulated genes are displayed in Fig. 6H.

Syngeneic tumors collected at study endpoint and representative patient tumor slices collected immediately after vibratome sectioning were formalin fixed and paraffin embedded. Sectioning and immunostaining of all samples was performed as detailed in Supplementary Methods. Slides were digitally scanned (Aperio AT2 system, Leica Biosystems) and analyzed with Tissue Studio 4.4.2 (Definiens) or HALO 2.1 software (Indica Labs).

MC38-R and B16F10-AP3 tumor flow cytometry data statistical analysis was performed in R (R Foundation for Statistical Computing). Cell count and CD8/Treg ratio data were log₁₀ transformed, and percentage data were arc-sine square-root transformed prior to ANOVA analysis with p value correction according to the method of Edwards and Berry (29). All other statistical analysis was performed in GraphPad Prism. Comparisons of survival curves between groups were conducted using the log-rank test without any P value adjustment for multiple testing. Comparisons of plasma mIL12p70 or mIFNγ concentrations between groups were conducted by first performing a Log₁₀ transformation, then using one-way ANOVA statistical analysis with Tukey honest significance difference post hoc test. Statistics were not carried out on IL10 levels due to zero values in controls. Comparison of tumor slice culture supernatant cytokine levels was performed using a Friedman test with Dunn multiple comparison posttest. For all other datasets, comparisons of two groups were made using a Student t test, and comparisons between multiple groups were made by ANOVA with post hoc Tukey multiple comparison test. Data are presented as mean ± SD unless otherwise indicated (*, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001).

We first designed mouse (m) IL12 mRNA to enable us to investigate responses in immunocompetent syngeneic mouse models of cancer, as human IL12p70 protein is not bioactive in mice (30). mIL12 mRNA (Fig. 1A) encodes a subunit-linked, monomeric mIL12p70 protein as described previously (21). Lipid nanoparticle (LNP)-encapsulated mIL12 mRNA (Fig. 1A) induced dose-dependent release of mIL12p70 from colorectal MC38 tumor cells in vitro (Supplementary Fig. S1A), and mIL12 mRNA–derived mIL12p70 protein had comparable activity to heterodimeric recombinant mIL12p70 on mouse CD8⁺ T cells (Supplementary Fig. S1B). A miR122–3p (miR122) binding site was included in the mIL12 mRNA 3′UTR to inhibit IL12 protein production in hepatocytes without impacting levels in cancer cells (Fig. 1B; ref. 22).

The pharmacodynamic effects and antitumor activity of mIL12 mRNA were investigated in mice bearing established subcutaneous syngeneic tumors with differing sensitivities to ICB. The SC A20 lymphoma model is a well-established ICB-sensitive system (31). We recently reported that the MC38-resistant (MC38-R) variant of the MC38 colorectal cell line is less responsive to inhibition of PD-1 and PD-L1 versus MC38-sensitive (MC38-S) tumors after subcutaneous implantation in mice (23). mIL12p70 expression was investigated in MC38-R and A20 tumor-bearing mice following a single intratumoral dose of LNP-encapsulated mIL12 mRNA. In both models, mIL12 mRNA induced dose-dependent tumor expression of mIL12p70, which was also associated with increased circulating plasma mIL12p70 concentrations (Fig. 1C and D; Supplementary Fig. S1C and S1D). IL12 levels remained elevated out to 7 days postdose (Fig. 1C and D; Supplementary Fig. S1C and S1D). This contrasts with the 5- to 10-hour half-life of recombinant IL12 (32) and is consistent with previous reports that mRNA administration leads to continued protein production and so prolonged bioavailability versus recombinant protein administration (23, 33).

A single dose of mIL12 mRNA induced dose-dependent tumor growth delay, tumor regression, and in some cases CRs in mice bearing established A20, MC38-S, and MC38-R tumors (Fig. 1E–G; Supplementary Fig. S1F). In contrast to mIL12 mRNA, nontranslating control mRNA had no effect on tumor growth or overall survival versus untreated animals. A single dose of as little as 0.05 μg mIL12 mRNA significantly extended overall survival versus control mRNA-treated mice in all three models (P ≤ 0.0032, Fig. 1E–G). Animals treated with 0.5 μg and 5 μg mIL12 mRNA survived significantly longer than those treated with the lowest 0.05 μg dose of mIL12 mRNA (P ≤ 0.0472, Fig. 1E–G). Complete regression of these established tumors was also observed in a larger proportion of animals in the 0.5 μg and 5 μg mIL12 mRNA groups versus in the 0.05 μg mIL12 mRNA group in all three models (Supplementary Fig. S1F). mIL12 mRNA had greater antitumor activity in the A20 and MC38-S models versus the MC38-R model (Fig. 1E–G), consistent with the previously observed higher sensitivity of the A20 and MC38-S models to PD-1/PD-L1 blockade compared with MC38-R (23, 31). While weekly repeated dosing continued to induce IL12 as well as downstream cytokines (Supplementary Fig. S2), repeated dosing did not significantly enhance mIL12 mRNA antitumor activity versus a single dose in the MC38-S, A20, or MC38-R models (Supplementary Fig. S1E and S1G). mIL12 mRNA was well tolerated, particularly at the 0.05 and 0.5 μg dose levels, with less than 10% body weight loss detected (Supplementary Fig. S3A–S3D). The 5-μg mIL12 mRNA dose level was similarly well tolerated in three out of four studies and the one outlier study in MC38-R is shown in Supplementary Fig. S3D, where an average of 6.98% body weight loss was observed at day 17 postimplantation. Importantly, in nearly all animals with a CR to mIL12 mRNA, tumors did not grow at the rechallenge implantation site in all three models (Fig. 1H; Supplementary Fig. S1H). This infers that an immune memory response was established to A20, MC38-S, or MC38-R cells in most animals with a CR to mIL12 mRNA treatment.

Because mice with a CR to mIL12 mRNA showed evidence of the establishment of an immune memory response to tumor cells, we investigated the role of the adaptive immune system and specifically CD8⁺ T cells in mIL12 mRNA antitumor activity. In MC38-R tumor-bearing animals, mIL12 mRNA induced a dose-dependent increase in tumor-infiltrating CD8⁺ T cells 7 days postdose (Fig. 2A). mIL12 mRNA did not alter tumor FoxP3⁻ effector CD4⁺ cells or FoxP3⁺ Treg numbers (Fig. 2B and C). A dose-dependent increase in the ratio of CD8⁺ T cells to FoxP3⁺ Treg cells was, therefore, observed 7 days after a single dose of 0.05 μg or 0.5 μg mIL12 mRNA (Fig. 2D). To further investigate the impact of mIL12 mRNA on the MC38-R TME, we performed tumor transcriptomic evaluation followed by pathway analysis and manual review of differentially expressed genes. This analysis highlighted a significant increase in transcripts relating to T-cell lineage abundance 7 days postdose (> 2 fold change, false discovery rate < 0.05, Fig. 2E; Supplementary Table S1).

CD8⁺ T-cell depletion significantly reduced mIL12 mRNA antitumor activity in the MC38-R model compared with mice with an intact immune system (P < 0.0001; Fig. 2F). mIL12 mRNA still delayed tumor growth in CD8⁺ T-cell–depleted mice (P < 0.0001 vs. CD8⁺ T-cell–depleted mice treated with control mRNA), but mIL12 mRNA no longer induced any CRs in CD8⁺ T-cell–depleted animals. In contrast, CD4⁺ T-cell depletion had no impact on mIL12 mRNA antitumor activity in this model. Tumor and peripheral T-cell depletion was confirmed by flow cytometry (Supplementary Fig. S4). These data suggest that CD8⁺ T-cell–dependent adaptive immunity is required for optimal antitumor activity of mIL12 mRNA in the MC38-R tumor model.

Many of the known effects of IL12 are dependent on IFNγ release from activated NK and T cells (12). Therefore, we next investigated the role of T- and NK-cell activation and IFNγ release in mIL12 mRNA-induced antitumor immunity. Within 24 hours, mIL12 mRNA induced dose-dependent activation of tumor NK and NKT cells (increase in the activation marker CD69, by % positive cells; Fig. 3A; Supplementary Fig. S5A). In contrast, no early changes were observed in the activation of CD4⁺ or CD8⁺ T cells within MC38-R tumors (Supplementary Fig. S5B–S5D). NK-cell numbers in tumors were unchanged 7 days postdose (Supplementary Fig. S5E) and NKT-cell numbers in tumors were mildly but significantly increased 7 days postdose (Supplementary Fig. S5F). Tumor transcriptomic analysis revealed the upregulation of additional NK-cell activation markers (Supplementary Fig. S5G). Tumor NK activation within the first 24 hours postdose was followed by a dose dependent increase in tumor and plasma IFNγ levels in MC38-R tumor–bearing mice (Fig. 3B and C). A similar dose-dependent increase in IFNγ levels was also observed in A20 tumor–bearing mice (Supplementary Fig. S5H and S5I).

To determine the role of IFNγ in the antitumor immune response to mIL12 mRNA more directly, we treated MC38-R tumor–bearing mice with an IFNγ-blocking antibody prior to administration of a single intratumoral dose of mIL12 mRNA. Blockade of IFNγ completely abrogated the antitumor activity of mIL12 mRNA in this model (Fig. 3D and E; P < 0.0001 mIL12 mRNA plus isotype vs. mIL12 mRNA plus anti-IFNγ; P = 0.5878 mIL12 mRNA + anti-IFNγ vs. control mRNA + anti-IFNγ).

Because NK-cell activation has the potential to lead to both IFNγ release and tumor cell cytotoxicity, we investigated the impact of NK and NKT-cell depletion on mIL12 mRNA-induced IFNγ expression and antitumor activity in the MC38-R model. NK- and NKT-cell depletion with anti-NK1.1 led to a small but significant reduction in the peripheral IFNγ response to mIL12 mRNA in mice (Supplementary Fig. S6A and S6B). However, NK- and NKT-cell depletion had no statistically significant impact on mIL12 mRNA antitumor activity in this model (Fig. 3F; Supplementary Fig. S6C and S6D). NK- and NKT-cell depletion was confirmed by flow cytometry (Supplementary Fig. S6E).

By 7 days postdose in MC38-R tumor–bearing mice, mIL12 mRNA induced TH1 TME transformation (upregulation of TH1 immune response genes, Fig. 3G). mIL12 mRNA–driven TH1 TME transformation also included an early increase in the maturation of dendritic cells (DC) including CD103⁺ and CD8⁺ cross-presenting DC subsets (increased % cells positive for the maturation marker CD86; Fig. 3H; Supplementary Fig S7A and S7B), followed by a more general increase in the expression of genes relating to DC abundance and antigen presentation 7 days postdose (Fig. 3G). Importantly, nontranslating control mRNA did not induce nonspecific DC activation (Fig. 3H; Supplementary Fig S7A and S7B), consistent with the modifications incorporated into the synthetic mRNA to reduce nonspecific immune activation, and with our previous results (23, 24). Additional pathway analysis of the transcriptomic landscape of MC38-R tumors revealed the broad extent of TME transformation 7 days postdose with mIL12 mRNA. This included increases in expression of T- and NK-cell activation genes, cytokines and chemokines, complement factors, as well as genes relating to extracellular matrix (ECM) modulation and lipid metabolism (Supplementary Fig S7C–S7F; Supplementary Table S1). The broad extent of MC38-R tumor transcriptomic changes 7 days postdose correlated with significant inhibition of tumor growth induced by mIL12 mRNA at this timepoint (Supplementary Fig S7G).

IFNγ is a potent inducer of PD-L1 expression on both tumor and myeloid cells (1, 34). We therefore sought to determine whether intratumoral mIL12 mRNA induces PD-L1 expression in the tumor microenvironment of syngeneic mouse tumor models. Indeed, intratumoral mIL12 mRNA increased PD-L1 expression on immune infiltrating cells in MC38-R tumors 7 days postdose (Fig. 4A). Flow cytometry analysis revealed PD-L1 expression increased on tumor-infiltrating CD11b⁺ myeloid cells and on CD11b⁺ Ly6C^hi monocytes (Fig. 4B; Supplementary Fig. S8A). No significant change in PD-L1 expression on MC38-R tumor cells was detected at any timepoint (Supplementary Fig. S8B). On tumoral total CD11b⁺ myeloid cells, PD-L1 was only induced 24 hours postdose by the low dose and 72 hours postdose by the higher dose of mIL12 mRNA, and PD-L1 levels on these cells had returned to baseline 7 days postdose (Supplementary Fig. S8A). In contrast, both 0.05 and 0.5 μg mIL12 mRNA increased PD-L1 expression on Ly6C^hi monocytes in the tumor, and this increase was sustained from 24 hours through to 7 days postdose (Fig. 4B). The CD11b⁺ myeloid compartment includes neutrophils, monocytes, macrophages, MDSC, and some DC. The more sustained expression of PD-L1 on the Ly6C^hi monocyte/mMDSC subset versus other CD11b⁺ myeloid populations might for example reflect greater suppressive functions of mMDSCs. In the blood of MC38-R tumor–bearing mice, mIL12 mRNA also induced PD-L1 expression on Ly6C^hi monocytes 24 hours postdose, and this PD-L1 induction was significantly inhibited in the presence of an IFNγ-blocking antibody (Supplementary Fig. S8C).

Because PD-L1 interaction with PD-1 can suppress antitumor immunity, the effect of PD-L1 blockade (anti–PD-L1 antibody) on mIL12 mRNA antitumor activity was investigated in animals bearing established MC38-R tumors. As expected, anti-PD-L1 plus control mRNA had no impact on tumor growth or overall survival in the ICB-resistant MC38-R model (P = 0.1793 for anti-PD-L1 plus control mRNA compared with isotype antibody plus control mRNA; Fig. 4C). In contrast, the combination of anti–PD-L1 with a single dose of 0.5 or 5 μg mIL12 mRNA significantly improved overall survival and increased the number of animals with CRs, compared with the same doses of mIL12 mRNA alone (P ≤ 0.0250; Fig. 4C). However, the combination of anti–PD-L1 with the lowest dose of 0.05 μg mIL12 mRNA did not significantly improve overall survival versus 0.05 μg mIL12 mRNA alone (P = 0.9689; Fig. 4C). In addition, there was a nonsignificant trend toward improved survival with three weekly doses of mIL12 mRNA in combination with anti–PD-L1 versus a single dose of mIL12 mRNA in combination with anti–PD-L1 (P = 0.1225; Supplementary Fig. S8D). The combination of anti–PD-L1 with three doses of mIL12 mRNA also led to more CRs (12/15 CRs, 80%), versus anti–PD-L1 in combination with a single dose of mIL12 mRNA (8/15 CRs, 53%).

We next investigated the mechanism by which anti–PD-L1 enhanced the antitumor activity of mIL12 mRNA in MC38-R tumor–bearing mice. We quantified the frequency of splenic tumor peptide-reactive T cells by rechallenging splenocytes with the MC38 immunodominant retroviral antigenic peptide p15E (35). mIL12 mRNA alone expanded tumor-reactive T cells in the spleen 7 days postdose, but anti–PD-L1 combined with mIL12 mRNA did not further expand these cells versus mIL12 mRNA monotherapy–treated animals (Fig. 4D). However, the combination of mIL12 mRNA with anti–PD-L1 did significantly increase the number of CD8⁺ T cells infiltrating into tumors compared with treatment with either anti–PD-L1 alone (P < 0.0001) or mIL12 mRNA alone (P < 0.05; Fig. 4E and F). In particular, a dense CD8⁺ immune infiltrate was detected surrounding a completely necrotic tumor core 7 days postdose in 4 of 8 animals treated with 0.5 μg mIL12 mRNA plus anti–PD-L1 (Fig. 4G, these 4 animals indicated by star symbol in Fig. 4F). In most animals with a CR to mIL12 mRNA plus anti–PD-L1, no tumor growth was observed following rechallenge implantation of MC38-R cells on the opposite flank of the animal (Supplementary Fig. S8E and S8F). Together, these data infer that PD-L1 blockade enhances CD8⁺ T-cell recruitment to MC38-R tumors and CD8⁺ T-cell–mediated tumor lysis in response to mIL12 mRNA.

We also investigated mIL12 mRNA antitumor activity alone and in combination with anti–PD-L1 treatment in the B16F10-AP3 model, a PD-L1 blockade–resistant melanoma tumor model with minimal immune infiltrate (36). In this model, a single dose of mIL12 mRNA transformed the TME, activating NK cells and DCs (Supplementary Fig. S9A and S9B), increasing CD8⁺ T-cell infiltration and the CD8⁺ T-cell to Treg ratio (Supplementary Fig. S9C and S9D), inducing IFNγ release (Supplementary Fig. S9E) and additionally increasing PD-L1 expression on both tumor and myeloid and cells (Supplementary Fig. S9F–S9H). A single dose of 0.5 μg mIL12 mRNA significantly increased survival compared with control mRNA (P < 0.0001; Supplementary Fig. S9I and S9J). There was a nonsignificant trend toward improved survival and an increased number of CRs in animals treated with mIL12 mRNA plus anti–PD-L1 versus mIL12 mRNA (Supplementary Fig. S9I and S9J). However, in the B16F10-AP3 model, neither mIL12 mRNA alone nor mIL12 mRNA combination with αPD-L1 expanded splenic tumor peptide–reactive T cells (Trp2 or p15E peptides; Supplementary Fig. S9K), suggesting that mIL12 mRNA is less efficient at inducing antitumor immunity in B16F10-AP3 versus MC38-R tumor–bearing mice.

We next tested the hypothesis that intratumoral mIL12 mRNA alone and in combination with anti–PD-L1 would drive regression of distal, untreated tumors in mice bearing bilateral MC38-S tumors (Fig. 5A). A single injection of the lower 0.5 μg dose of mIL12 mRNA induced complete regression of both treated and untreated tumors in 3 of 20 animals (bilateral CRs, Fig. 5A) and significantly increased overall survival versus animals treated with isotype antibody plus 5 μg control mRNA (P < 0.0001; Fig. 5B). The higher 5 μg mIL12 mRNA dose induced significantly greater overall survival and a larger number CRs compared with 0.5 μg mIL12 mRNA (P < 0.0001, 16/20 CRs; Fig. 5A and B). Anti–PD-L1 antibody only had marginal antitumor activity in this model, providing a modest but statistically significant survival benefit versus animals treated with isotype antibody plus control mRNA without extending median overall survival or inducing any CRs (P = 0.0214; Fig. 5B). In contrast, the combination of αPD-L1 antibody with 0.5 or 5 μg mIL12 mRNA significantly improved overall survival compared with the same doses of mIL12 mRNA alone (P ≤ 0.0373) and compared with anti–PD-L1 alone (P < 0.0001; Fig. 5A and B). The antitumor activity of the combination of anti–PD-L1 and mIL12 mRNA was also dose-dependent in this model, with anti–PD-L1 plus 5 μg mIL12 mRNA providing a significantly greater survival benefit and larger number of CRs versus anti–PD-L1 + 0.5 μg mIL12 mRNA (P < 0.0001, 20/20 CRs and 8/20 CRs, respectively; Fig. 5B).

Intratumoral injection of mIL12 mRNA induces local tumoral mIL12p70 expression, with some increases in peripheral mIL12p70 exposure in syngeneic tumor-bearing mice (Fig. 1C and D). To investigate the ability of local IL12p70 alone to drive regression of uninjected lesions, we used mRNA designed to encode a membrane-tethered mIL12p70 to limit peripheral mIL12p70 exposure (Fig. 5C). While cell-associated mIL12p70 was detected in HeLa cells transfected with tethered mIL12 mRNA, mIL12p70 release into the cell culture supernatant was very low relative to the secreted IL12 variant (Fig. 5D). HeLa cells expressing tethered mIL12 mRNA enhanced IFNγ release from cocultured mouse CD8⁺ T cells, confirming the bioactivity of this membrane-associated mIL12p70 protein (Fig. 5E). Intratumoral delivery of tethered mIL12 mRNA in mice bearing bilateral MC38-S tumors led to complete regression of both treated and distal tumors in 4 of 20 mice (bilateral CRs, Fig. 5F). Minimal increases in circulating mIL12p70 were detected above controls in tethered mIL12 mRNA–treated animals, in contrast to animals treated with secreted mIL12 mRNA (Fig. 5G). Furthermore, plasma IFNγ induction was also greatly reduced in animals treated with tethered versus secreted mIL12 mRNA (Fig. 5H). These data with tethered mIL12 mRNA indicate that local tumor mIL12p70 expression can induce regression of untreated lesion in the absence of increases in circulating mIL12p70 levels.

Having observed promising antitumor activity of mIL12 mRNA in mouse syngeneic tumor models, we developed MEDI1191 as a potential new treatment for patients with solid tumors. MEDI1191 is an LNP-formulated mRNA encoding a linked monomeric bioactive human (h) IL12p70 protein. MEDI1191 induced dose-dependent release of hIL12p70 from human monocyte-derived macrophages and from a panel of 16 human tumor cell lines. hIL12p70 production varied significantly between tumor cell lines in vitro, whereas it was very consistent between monocyte-derived macrophages from different donors (Supplementary Fig. S10A and S10B). MEDI1191-derived linked monomeric hIL12p70 activity was comparable with that of recombinant heterodimeric hIL12p70 in a human CD8⁺ T-cell IFNγ release assay (Supplementary Fig. S10C).

We investigated the ability of MEDI1191 to drive IL12p70 production following intratumoral administration in 2 melanoma and 2 head and neck squamous cell carcinoma (HNSCC) PDX models. Intratumoral MEDI1191 induced dose-dependent increases in tumor and plasma hIL12p70 concentrations in the ME12057 melanoma and HN5111 HNSCC PDX models, the two in which multiple dose levels were assessed (Supplementary Fig. S10D). No significant difference in tumor or plasma hIL12p70 levels was observed between all four PDX models 6 hours postdose with 0.5 μg MEDI1191 (Supplementary Fig. S10E). Twenty-four hours postdose, hIL12p70 levels in tumor and plasma were significantly lower in the HN5116 HNSCC tumor-bearing mice versus the other models.

Finally, we investigated the ability of MEDI1191 to transform the microenvironment of patient colorectal and endometrial tumors in an ex vivo tumor slice culture system. Fresh samples of patient tumors collected following surgical resection were vibratome sectioned (37) and cultured in vitro at the air/media interface in the presence of MEDI1191 or control mRNA. Details on patient tumor samples are in Supplementary Fig. S11.

MEDI1191 induced significant, dose-dependent hIL12p70 release from all patient tumor slice cultures tested (Fig. 6A). MEDI1191 also induced dose-dependent IFNγ and CXCL10 release, whereas control mRNA had minimal effect on expression of these mediators (Fig. 6B and C). While MEDI1191 induced hIL12p70 release from every patient tumor tested, IFNγ and CXCL10 release was more variable between patient tumor slices (Fig. 6A–C). We therefore investigated whether MEDI1191-dependent induction of an early IFNγ response in tumor slice cultures might depend on the density of tumor-infiltrating NK and T cells at baseline. NK- and T-cell densities were quantified by IHC in patient tumor slices fixed immediately after vibratome sectioning (Fig. 6D and E). No correlation was observed between patient tumor slice IFNγ release and baseline T-cell infiltration in this acute model (Fig. 6F, IFNγ release measured after 3–5 days in culture). However, a positive correlation was observed between patient tumor slice acute IFNγ release and baseline NK-cell infiltration (Fig. 6G). Finally, because mIL12 mRNA induced TH1 TME transformation in the murine MC38-R model, we investigated whether MEDI1191 also induced similar TME transformation in patient tumor slices. Indeed, NanoString transcriptomic analysis of patient tumor slices treated with MEDI1191 revealed significant increases in expression of TH1 response genes compared with tumor slices treated with control mRNA (Fig. 6H). Notably, IFNG, STAT1, and GBP2 expression was upregulated in both MEDI1191-treated patient tumor slices and mIL12 mRNA–treated MC38-R tumors (Figs. 3D and 6H). The remaining TH1 genes induced by MEDI1191 in patient tumor slices all also came from the same gene families as TH1 genes upregulated in mIL12 mRNA–treated tumors (e.g., human CXCL10 vs. mouse Cxcl9).

We report here that in syngeneic mouse tumor models and patient tumor slice cultures IL12 mRNA drives TH1 transformation of the tumor microenvironment. In mouse models, this leads to IFNγ and cytotoxic T-cell–dependent antitumor immunity that is further enhanced by PD-L1 blockade.

A single intratumoral dose of mIL12 mRNA induced dose-dependent mIL12p70 expression in syngeneic tumor-bearing mice. Similarly, MEDI1191 induced robust, dose-dependent hIL12p70 protein production in all patient tumor-derived models tested, including 2 melanoma and 2 HNSCC PDX models and 7 colorectal and endometrial cancer patient tumor slice cultures. Our data indicate that IL12p70 expressed in mouse syngeneic models and PDX models is likely derived from both cancer cells and tumor-infiltrating myeloid cells transfected with mRNA. Both human macrophages and human tumor cell lines produced hIL12p70 in response to LNP-formulated MEDI1191 in vitro. It is difficult to directly determine which cells in the TME are responsible for IL12 mRNA-induced IL12p70 production in vivo, because IL12p70 is a secreted protein. However, we recently reported that intratumoral delivery of LNP-formulated mRNA encoding transmembrane OX40L in syngeneic tumors led to protein translation in cancer cells, tumor-infiltrating myeloid cells and tumor draining lymph node myeloid cells (23).

Repeated doses of IL12 mRNA continued to induce IL12 expression and IFNγ release, but did not significantly enhance the antitumor activity of mIL12 mRNA. This may reflect the limitations in using rapidly growing syngeneic models to address questions of dose and schedule. Alternatively, this may suggest that repeated intratumoral injection of IL12 mRNA as a monotherapy in the same tumor has limited benefit. Indeed, while we found that the antitumor response to a single dose of IL12 mRNA requires IFNγ, others have reported that chronic IFNγ can lead to upregulation of immune checkpoints, T-cell dysfunction, and impaired antitumor immunity (38).

mRNA-derived IL12p70 protein drove TH1 TME transformation, consistent with the expected activity of IL12p70 as a key driver of type I immune responses (2). In syngeneic mouse tumors, early tumor NK activation after a single dose of mIL12 mRNA was followed by dose-dependent increases in IFNγ expression, DC activation, and upregulation of a broad range of TH1 response genes. TH1 TME transformation in syngeneic mouse tumors led to dose-dependent tumor regression, CRs, and induction of antitumor immunity via a mechanism requiring both IFNγ and CD8⁺ T cells. The activity of mIL12 mRNA observed in multiple tumor models suggests IL12 mRNA therapy has the potential to benefit patients with varying TME composition.

MEDI1191 also drove TH1 gene expression in patient tumor slices, building confidence in the potential for translation of this therapeutic approach to a clinical setting. IFNγ expression in MEDI1191-treated slice cultures correlated positively with baseline NK infiltration, but not with baseline CD3⁺ T-cell infiltration. Detailed T-cell subset analysis might reveal the contribution of T cells to IFNγ release in this model (e.g., separating Treg, CD4⁺ effector, and CD8⁺ cytotoxic T cells). Analysis of a larger number of patient tumor samples might reveal a contribution of T cells to IFNγ production in a subset of tumors. In addition, this ex vivo system likely underestimates the contribution of T cells due to the short experimental time-course and the lack of modeling of immune cell migration between tumor, lymph node, and blood.

Importantly, a single dose of mIL12 mRNA also promoted regression of untreated lesions in a dual flank MC38-S model, consistent with the ability of mIL12 mRNA to promote systemic antitumor immunity. The hypothesis that intratumoral mIL12 mRNA drives true abscopal activity was supported by the discovery that mRNA encoding membrane-tethered IL12-induced complete regressions in uninjected distal tumors, albeit with reduced potency versus mRNA encoding secreted mIL12. It is, therefore, likely that the abscopal effects are due to the IL12 mRNA-induced expansion of tumor-specific T cells observed in the periphery of tumor-bearing mice, rather than due to induction of circulating IL12p70. Intratumoral delivery of an adenovirus-encoded, membrane-tethered IL12 also inhibited growth of distal CT26 mouse colorectal cancer lesions in a dual flank model (39). Abscopal effects of intratumoral IL12 plasmid plus electroporation have been observed in uninjected lesions in patients with metastatic melanoma in the absence of detectable elevations of circulating IL12 (20, 40). While a dual flank syngeneic model does not fully recapitulate the complexities of disseminated cancer in patients, these data highlight the potential for locally delivered MEDI1191 to drive immune-mediated regression of distal lesions in patients with metastatic disease.

Our preclinical data support the hypothesis that MEDI1191 might expand the utility of PD-1/PD-L1 blockade to some patients currently resistant to ICB (1). mIL12 mRNA monotherapy drove regression of ICB-resistant MC38-R mouse tumors, and combination with PD-L1 blockade enhanced antitumor immunity in this model. These results are in agreement with improved antitumor activity of both an IL12 immunocytokine and an IL12-expressing oncolytic virus in combination with anti-PD-1 or anti-PD-L1 in an otherwise PD-1/PD-L1 blockade–resistant MC38 model (15, 41).

Our data suggest that mIL12 mRNA may overcome PD-L1 resistance by bypassing the need for tumor DC–derived IL12 in MC38-R tumors, priming an antitumor immune response that can be enhanced through blockade of PD-1 signaling by de novo expressed PD-L1 in the TME. IL12 release from tumoral DC was reported recently to be required for effective anti-PD-1 immunotherapy in a PD-1 blockade–sensitive MC38 model (42). We reported previously that the PD-L1 blockade–resistant MC38-R model has a smaller DC infiltrate compared with the PD-L1 blockade–sensitive MC38-S model (23). Here we demonstrate that intratumoral mIL12 mRNA alone increased tumoral cytotoxic T-cell infiltration in the MC38-R model, expanded peripheral tumor-specific T cells, and increased PD-L1 expression in the TME. Combination with PD-L1 blockade further enhanced cytotoxic T-cell recruitment to MC38-R tumors and increased the incidence of tumor regression.

In the ICB-resistant B16F10-AP3 model which has a particularly small immune infiltrate (23, 36), mIL12 mRNA had significant monotherapy activity, but mIL12 mRNA plus PD-L1 blockade combination therapy only provided a modest, nonsignificant survival benefit versus mIL12 mRNA alone. Notably, mIL12 mRNA also did not drive peripheral expansion of tumor peptide–reactive T cells in B16F10-AP3 tumor-bearing mice. Thus, different, unknown mechanisms may be driving ICB resistance in the B16F10-AP3 model versus in MC38-R. In patients, a wide variety of mechanisms are emerging as drivers for primary and adaptive resistance to ICB (43), and combinatorial therapeutic approaches will likely be required to overcome resistance in these diverse patient populations.

The development of strategies for tumor-targeted IL12 delivery has been driven by extensive preclinical evidence that IL12 can promote antitumor immunity, combined with the poor tolerability and limited efficacy of systemic recIL12 in early clinical trials (11, 12). Tumor-targeted IL12 delivery may improve this therapeutic margin by driving local IL12 exposure (required for antitumor immunity) while limiting circulating IL12 exposure (likely reducing undesirable toxicity).

Various strategies have therefore promoted tumor-targeted IL12 delivery and they have all demonstrated promising activity in mouse tumor models (14–17, 41, 44, 45). Systemic administration of tumor-targeted IL12 would avoid the complexity of intratumoral injection in patients. However, systemically delivered IL12 immunocytokines lead to high peripheral versus tumor IL12 protein exposures, and a potentially still narrow therapeutic window in patients (46). Delivery of intratumoral IL12 plasmid plus electroporation (tavokinogene telseplasmid) in metastatic melanoma patients has reported clinical activity (20, 40). However, a separate electroporation device is required, limiting this therapeutic approach to patients with cutaneous or subcutaneous lesions (47). An LNP-formulated mRNA therapy can be delivered without electroporation, which expands the potential of intratumoral administration to patients with deep-seated lesions. Following intratumoral viral IL12 delivery, the challenge of maintaining tight control over IL12 expression has led to complex genetic switch technology where a second orally dosed small-molecule controls IL12 expression (17). In contrast, the limited half-life of LNP-formulated mRNA used here (25) leads to controlled and transient protein production, as demonstrated in multiple syngeneic and PDX models.

MEDI1191 was therefore developed as an optimized IL12 mRNA for intratumoral delivery. A research version of LNP-formulated IL12 mRNA, delivered systemically, inhibited tumor growth in a mouse model of hepatocellular carcinoma (44). However, higher protein expression was observed in normal liver tissue and spleen than in tumors, indicating that this approach was insufficient to target IL12 protein expression to tumors. Improving on this early research version, a miR122-binding site was incorporated in the MEDI1191 mRNA 3′UTR to minimize peripheral IL12 expression in miR122-positive normal hepatocytes (22). In addition, MEDI1191 incorporates aspects of a novel LNP formulation that demonstrates improved pharmacokinetics and better local mRNA endosomal escape and translation (25). These advances, in conjunction with direct intratumoral injection, aim to provide a clinically viable therapeutic window between efficacy and toxicity, and have enabled testing of MEDI1191 in patients.

In conclusion, the robust preclinical activity reported here of MEDI1191 (IL12 mRNA) and its mouse surrogate provide support for the development of MEDI1191 as a potential novel treatment for patients with both superficial and deep-seated solid tumors, alone and in combination with inhibitors of the PD-L1/PD-1 checkpoint. Intratumoral MEDI1191 is currently being evaluated in a phase I trial in patients with solid tumors in combination with durvalumab (anti–PD-L1, NCT03946800).

S.L. Hewitt reports other from Moderna (current employee who receives compensation and benefits from employer) during the conduct of the study and is listed as a coinventor on a patent (WO2017/201350) regarding Polynucleotides Encoding interleukin-12 (IL12) and Uses Thereof (e.g., in the treatment of cancer); Moderna shares worldwide commercial rights to MEDI1191 with AstraZeneca. D. Bailey reports personal fees from Moderna during the conduct of the study. J. Zielinski reports personal fees from Moderna (employee) during the conduct of the study. A. Apte reports personal fees from Moderna (employee) during the conduct of the study. F. Musenge reports personal fees from Moderna (employee) during the conduct of the study. R. Karp reports personal fees from Moderna (employee) during the conduct of the study. S. Burke is an employee of AstraZeneca, which is developing MEDI1191 as a potential anticancer therapy for patients with solid tumors suitable for direct intratumoral dosing. F. Garcon is an employee of AstraZeneca, which is developing MEDI1191 as a potential anticancer therapy for patients with solid tumors suitable for direct intratumoral dosing. A. Mishra reports personal fees from Moderna (employee) during the conduct of the study. S. Gurumurthy reports personal fees from Moderna (employee) during the conduct of the study. A. Watkins is an employee of AstraZeneca, which is developing MEDI1191 as a potential anticancer therapy. K. Arnold reports personal fees from Moderna (employee) during the conduct of the study. J. Moynihan is an employee of AstraZeneca, which is developing MEDI1191 as a potential anticancer therapy for patients with solid tumors suitable for direct intratumoral dosing. E. Clancy-Thompson is an employee of AstraZeneca, which is developing MEDI1191 as a potential anticancer therapy for patients with solid tumors suitable for direct intratumoral dosing. K. Mulgrew is an employee of AstraZeneca, which is developing MEDI1191 as a potential anticancer therapy for patients with solid tumors suitable for direct intratumoral dosing. G. Adjei is an employee of AstraZeneca, which is developing MEDI1191 as a potential anticancer therapy for patients with solid tumors suitable for direct intratumoral dosing. K. Deschler is an employee of AstraZeneca, which is developing MEDI1191 as a potential anticancer therapy for patients with solid tumors suitable for direct intratumoral dosing. D. Potz reports personal fees from Moderna (employee) during the conduct of the study. G. Moody is an employee and stockholder of AstraZeneca, which is developing MEDI1191 as a potential anticancer therapy for patients with solid tumors suitable for direct intratumoral dosing. D.A. Leinster is an employee of AstraZeneca, which is developing MEDI1191 as a potential anticancer therapy for patients with solid tumors suitable for direct intratumoral dosing. S. Novick is an employee of AstraZeneca, which is developing MEDI1191 as a potential anticancer therapy for patients with solid tumors suitable for direct intratumoral dosing. M. Sulikowski is an employee of AstraZeneca, which is developing MEDI1191 as a potential anticancer therapy for patients with solid tumors suitable for direct intratumoral dosing. C.J. Bagnall is an employee of AstraZeneca, which is developing MEDI1191 as a potential anticancer therapy for patients with solid tumors suitable for direct intratumoral dosing. P. Martin is an employee of AstraZeneca, which is developing MEDI1191 as a potential anticancer therapy. J. Lapointe is an employee of AstraZeneca, which is developing MEDI1191 as a potential anticancer therapy. H. Si is a former employee of AstraZeneca, which is developing MEDI1191 as a potential anticancer therapy. C. Morehouse is an employee of AstraZeneca, which is developing MEDI1191 as a potential anticancer therapy. M. Sedic reports personal fees from Moderna (employee) during the conduct of the study. R.W. Wilkinson is an employee of AstraZeneca, which is developing MEDI1191 as a potential anticancer therapy for patients with solid tumors suitable for direct intratumoral dosing. R. Herbst reports other from AstraZeneca (former employee and current stockholder) outside the submitted work. J.P. Frederick reports other from Moderna (current employee who receives compensation and benefits from employer) during the conduct of the study and is listed as a coinventor on a patent (WO2017/201350) regarding Polynucleotides Encoding interleukin-12 (IL12) and Uses Thereof (e.g., in the treatment of cancer); Moderna shares worldwide commercial rights to MEDI1191 with AstraZeneca. N. Luheshi is an employee of AstraZeneca, which is developing MEDI1191 as a potential anticancer therapy for patients with solid tumors suitable for direct intratumoral dosing.

S.L. Hewitt: Conceptualization, formal analysis, supervision, investigation, visualization, methodology, writing-original draft, project administration, writing-review and editing. D. Bailey: Formal analysis, investigation, visualization, methodology. J. Zielinski: Formal analysis, investigation, visualization, methodology. A. Apte: Investigation, visualization. F. Musenge: Investigation, visualization. R. Karp: Investigation, visualization, methodology. S. Burke: Formal analysis, investigation, visualization, methodology. F. Garcon: Formal analysis, investigation, visualization, methodology. A. Mishra: Formal analysis, investigation, visualization, methodology. S. Gurumurthy: Conceptualization, supervision, project administration. A. Watkins: Formal analysis, investigation, visualization, methodology. K. Arnold: Formal analysis, supervision, visualization, methodology, project administration. J. Moynihan: Formal analysis, investigation. E. Clancy-Thompson: Formal analysis, investigation, visualization. K. Mulgrew: Formal analysis, investigation. G. Adjei: Investigation. K. Deschler: Formal analysis, investigation, visualization. D. Potz: Investigation. G. Moody: Investigation. D.A. Leinster: Conceptualization, methodology. S. Novick: Formal analysis. M. Sulikowski: Formal analysis, investigation, methodology. C. Bagnall: Formal analysis, investigation, methodology. P. Martin: Formal analysis, investigation, visualization. J.-M. Lapointe: Formal analysis, investigation, visualization. H. Si: Formal analysis. C. Morehouse: Formal analysis. M. Sedic: Investigation. R.W. Wilkinson: Conceptualization, supervision, project administration, writing-review and editing. R. Herbst: Conceptualization, supervision, project administration, writing-review and editing. J.P. Frederick: Conceptualization, supervision, project administration, writing- review and editing. N. Luheshi: Conceptualization, formal analysis, supervision, visualization, methodology, writing-original draft, project administration, writing-review and editing.

Support for mRNA manufacture and LNP formulation was provided by Moderna Inc, in particular, Sarah Peterson, Mengfei Sun, and Weijia Wang. We acknowledge support from teams within Moderna including Toxicology and Pathology and Non-clinical Sciences, and from teams within AstraZeneca including IVS and Core Tissue Culture. We thank Jill Grenier, Philippe Garnier, and Lisa Johansen for program management.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.