The question

For adjuvanted vaccines, the induction of an innate immune response is essential to generate a protective, long-lasting adaptive immune response. Lipid-formulated RNA vaccines against cancer or COVID-19 hold intrinsic adjuvant activity, as evidenced by the impressive induction of antibody and T cell responses after vaccination in humans.

However, the innate immunostimulatory activity of lipid-formulated RNA vaccines also induces systemic pro-inflammatory cytokine responses and dose-dependent, transient systemic reactions — such as fever and chills — that were not predicted from preclinical studies. The exact pathway by which such innate immune responses are generated is not known, so understanding species-specific differences of tolerability after lipid-formulated RNA vaccination in mice and humans could provide insight into potential mechanisms.

The discovery

We set out to understand the mechanism of inflammatory, species-specific toxicities associated with lipid-formulated RNA vaccines. We studied the RNA-lipoplex (RNA-LPX) cancer vaccine that uses liposomes to contain unmodified single-stranded RNA and can be sensed by Toll-like receptors 7 and 8 (TLR7 and TLR8, respectively). In addition, an alternative version of this LPX vaccine was formulated using N1-methyl-pseudouridine-modified RNA (modRNA-LPX) (Fig. 1a), with modRNA used to greatly reduce the innate immunostimulatory activity of the vaccine1. Moreover, we tested the modRNA-LNP vaccine using lipid nanoparticles (LNPs) —containing the ionizable lipid SM-102 — to mimic the Moderna COVID-19 vaccine formulation2. We used human peripheral blood mononuclear cells, mouse primary leukocytes or different knockout mouse models and measured the levels of different pro- and anti-inflammatory cytokines after vaccination to identify species-specific factors that contribute to vaccine-induced inflammatory responses.

Fig. 1: RNA vaccines induce IL-1 and other pro-inflammatory cytokines in human immune cells in vitro.
figure 1

a, Schematic of the different lipid-formulated RNA vaccines. b, Cytokine release from human peripheral blood mononuclear cells treated with unmodified RNA-LPX, N1-methyl-pseudouridine-modified RNA-LPX (modRNA-LPX) or modRNA-LNP formulated with SM-102 lipids. CCL, C–C motif chemokine ligand; CXCL, C-X-C motif chemokine ligand; EGF, epidermal growth factor; G-CSF, granulocyte colony-stimulating factor; GM-CSF, granulocyte–macrophage colony-stimulating factor; IFN, interferon; IL, interleukin; TNF, tumor necrosis factor; LTA, lymphotoxin-α; VEGF, vascular endothelial growth factor. © 2022, Tahtinen, S. et al.

Full size image

We discovered that the RNA-LPX vaccine induces the release of the cytokine IL-1. IL-1 initiates an innate immune cascade that results in systemic cytokine release and the adverse events that limit vaccine dosing in humans. Furthermore, we identified IL-1 receptor antagonist (IL-1RA) as an endogenous, inducible suppressor of this systemic inflammation in mice, explaining why mice can tolerate more than 1,000-fold higher doses of RNA-LPX than humans.

Surprisingly, our RNA-LPX data indicated that both the TLR7 or TLR8 agonistic function of the RNA and the presence of liposomes were required for the inflammasome-mediated production of IL-1, as IL-1 release was abolished when the modRNA-LPX vaccine was used, whereas modRNA-LNP vaccine particles were potent activators of the IL-1 pathway (Fig. 1b). These results suggest that the reactogenicity and immunogenicity of modRNA is context-dependent — modRNA can either be non-immunostimulatory when formulated in LPX or initiate a potent innate response when formulated in LNPs.

The implications

Knowing how RNA vaccines induce innate immunity and how those responses are differentially modulated in mice and humans not only provides valuable information about how to improve adjuvanted vaccine efficacy and safety but also addresses some basic principles of immunobiology. Moreover, these findings provide insights for oligonucleotide delivery beyond adjuvanted vaccines, especially in instances where the induction of an innate immune cascade mediated by IL-1 would be less desirable (such as LNP-based therapeutic agents for autoimmune and autoinflammatory diseases3).

Our data revealed that the use of modified RNA is not necessarily an essential feature of RNA vaccines. Whereas the use of modified RNA may have advantages with respect to protein translation, the elimination of the inherent adjuvanticity of RNA as a TLR7 or TLR8 agonist necessitates the addition of another innate stimulus. In the case of COVID-19 vaccines, this is provided by ionizable lipids4, which we found induce IL-1 and associated pro-inflammatory responses similar to unmodified RNA. However, our work did not address the mechanism by which lipid structures can induce the inflammasome-mediated production of IL-1 in human cells or whether such inflammatory responses would be exacerbated in IL-1RA-deficient mice in vivo.

Future work will determine whether ionizable lipids in LNPs can indirectly (via induction of cell damage) or directly (activating of caspases5 or other innate sensors) induce the activation of IL-1, and whether these physiochemical and possibly lytic properties could be modified to influence the overall immunostimulatory profile of LNP-based particles to design safer and more effective RNA-based vaccines and therapeutics.

Siri Tahtinen and Ira Mellman, Genentech, South San Francisco, CA, USA.

Expert opinion

“This paper reveals several ‘dirty little secrets’ of lipid-formulated RNAs. The authors identified that RNA vaccines trigger inflammasome activity and unleash a complex interplay between IL-1 cytokines and their antagonists. The downstream net effect of the immune response to the composite vaccine, which is largely different in preclinical species and humans, needs to be considered when assessing the toxicities and efficacy of this emerging therapeutic technology.” Eicke Latz, Institute of Innate Immunity, Bonn, Germany

Behind the paper

This project brought together scientists across different departments and areas of expertise at Genentech (USA) and BioNTech (Germany). It was sparked by observations in our joint phase 1 clinical trial in 2017, where patients with cancer receiving RNA-LPX vaccines exhibited transient flu-like symptoms at dose levels (50 µg) that were well-tolerated in mice; a difference that still seems striking to us considering the obvious size difference between the species.

We made considerable progress by characterizing the role of IL-1 in unmodified RNA-LPX-induced inflammatory responses. In 2021, after the success of modified RNA-LNP vaccines in the prophylaxis of COVID-19, we began studying what drives the adjuvancy of these LNP vaccines. We were surprised to see that IL-1 was strongly induced by modRNA-LNPs, creating an additional layer of complexity to the mechanistic underpinnings of IL-1-mediated inflammatory response and the adjuvancy triggered by different lipid-formulated RNA vaccines. S.T. & I.M.

From the editor

“Lipid-formulated RNA vaccines have been studied for many years and received a huge boost in interest following the COVID-19 pandemic. However, their dosing can be limited by their reactogenicity. The authors demonstrate that lipid-formulated RNA vaccines can trigger IL-1 release and ensuing inflammation in humans, whereas in mice this process is dampened by production of the endogenous IL-1 antagonist IL-1RA. These findings recommend caution when using certain preclinical models for studying reactogenicity.” Zoltan Fehervari, Senior Editor, Nature Immunology.