Characterization of RIG-I-like receptor activation by <em>in vitro</em> transcribed mRNA vaccines and modulation by nucleoside modifications

Abstract

The therapeutic potential of mRNA has been studied for decades, but the COVID-19 pandemic accelerated its development and approval for human use. A major challenge is the potential of mRNA to activate innate RNA-sensing receptors, triggering inflammatory responses that affect vaccine dosage and efficacy. The underlying immune mechanisms are not well understood, although this knowledge would greatly aid in optimizing <em>in vitro</em> transcribed (IVT)-mRNA and modulating its immunostimulatory effects. <br/> In this study, the innate immune response to COVID-19 mRNA vaccines by BioNTech and Moderna, as well as to IVT-mRNA, was characterized <em>in vitro</em>. The cytosolic double-stranded RNA (dsRNA) sensing RIG-I-like receptors (RLRs), RIG-I and MDA5, were identified as the primary receptors responsible for mRNA-induced type I interferon responses. Mechanistically, RLRs were activated by unspecific dsRNA byproducts generated during <em>in vitro</em> transcription. LC/MS analysis revealed that 5′ triphosphate-dsRNA byproducts, capable of activating RIG-I, were not permissive to capping by Vaccinia Capping Enzyme. Various methods were explored to remove dsRNA byproducts and reduce RLR activation, such as oligo(dT)-purification and dephosphorylation. <br/> Additionally, the study characterized species-specific differences in innate mRNA sensing between human and murine cells. In human cells, IVT-mRNA was primarily sensed by RIG-I, while MDA5 was the dominant sensor in murine cells. <br/> Moreover, effects of nucleotide modifications, including Ψ, m1Ψ, 5moU, m5C, and m6A, on individual innate RNA receptors were investigated, focusing on RLRs. RLRs were differentially modulated by these modifications in human and murine cells. Reduced dsRNA content was the leading cause for lower RIG-I activation in response to modified IVT-mRNAs. However, 5moU directly impaired activation of specifically murine RIG-I, while human RIG-I was only slightly affected. Furthermore, Ψ and m1Ψ enhanced activation of murine but not human MDA5. Both human and murine MDA5 activation was completely abolished by 5moU, m5C, and m6A, even when pure dsRNA was used, indicating direct effects on MDA5. <br/> Furthermore, mRNA translation and its modulation by nucleoside modifications were investigated. Some modifications, such as m1Ψ and 5moU, are known to enhance mRNA translation, for which the lower innate immune activation is assumed to be responsible. While RLR activation and type I interferon signaling did indeed restrict mRNA translation in human cells, as expected, the enhanced translation by nucleoside modifications only partially depended on this. Thus, it was revealed that nucleoside modifications enhance translation by directly affecting the translation machinery via currently unknown mechanisms. <br/> In conclusion, the incorporation of either m1Ψ or 5moU has been identified as the most promising measure to reduce mRNA-induced innate immune responses while achieving maximal translation. Overall, this study provides valuable insights into immune responses to mRNA vaccines, presents methods for modulating these responses, and highlights differences between experimental systems relevant to preclinical study design. Altogether, this could be applied to design mRNA therapeutics that achieve an optimal balance of antigen expression and adjuvant activity on the one hand, or entirely non-immunostimulatory mRNA on the other hand, which could be used for protein replacement or combined with exogenous adjuvants tailored to specific pathogens.