In contrast to TH2-biased approaches, TH1 COVID-19 vaccines protect Syrian hamsters from severe disease in the absence of dexamethasone-treatable vaccine-associated enhanced respiratory pathology

Since December 2019, the novel human coronavirus SARS-CoV-2 has spread globally, causing millions of deaths. Unprecedented efforts have enabled development and authorization of a range of vaccines, which reduce transmission rates and confer protection against the associated disease COVID-19. These vaccines are conceptually diverse, including e.g. classical adjuvanted whole-inactivated virus, viral vectors, and mRNA vaccines. We have analysed two prototypic model vaccines, the strongly TH1-biased measles vaccine-derived candidate MeVvac2-SARS2-S(H) and a TH2-biased Alum-adjuvanted, non-stabilized Spike (S) protein side-by-side, for their ability to protect Syrian hamsters upon challenge with a low-passage SARS-CoV-2 patient isolate. As expected, the MeVvac2-SARS2-S(H) vaccine protected the hamsters safely from severe disease. In contrast, the protein vaccine induced vaccine-associated enhanced respiratory disease (VAERD) with massive infiltration of eosinophils into the lungs. Global RNA-Seq analysis of hamster lungs revealed reduced viral RNA and less host dysregulation in MeVvac2-SARS2-S(H) vaccinated animals, while S protein vaccination triggered enhanced host gene dysregulation compared to unvaccinated control animals. Of note, mRNAs encoding the major eosinophil attractant CCL-11, the TH2 response-driving cytokine IL-19, as well as TH2-cytokines IL-4, IL-5, and IL-13 were exclusively up-regulated in the lungs of S protein vaccinated animals, consistent with previously described VAERD induced by RSV vaccine candidates. IL-4, IL-5, and IL-13 were also up-regulated in S-specific splenocytes after protein vaccination. Using scRNA-Seq, T cells and innate lymphoid cells were identified as the source of these cytokines, while Ccl11 and Il19 mRNAs were expressed in lung macrophages displaying an activated phenotype. Interestingly, the amount of viral reads in this macrophage population correlated with the abundance of Fc-receptor reads. These findings suggest that VAERD is triggered by induction of TH2-type helper cells secreting IL-4, IL-5, and IL-13, together with stimulation of macrophage subsets dependent on non-neutralizing antibodies. Via this mechanism, uncontrolled eosinophil recruitment to the infected tissue occurs, a hallmark of VAERD immunopathogenesis. These effects could effectively be treated using dexamethasone and were not observed in animals vaccinated with MeVvac2-SARS2-S(H). Taken together, our data validate the potential of TH2-biased COVID-19 vaccines and identify the transcriptional mediators that underlie VAERD, but confirm safety of TH1-biased vaccine concepts such as vector-based or mRNA vaccines. Dexamethasone, which is already in use for treatment of severe COVID-19, may alleviate such VAERD, but in-depth scrutiny of any next-generation protein-based vaccine candidates is required, prior and after their regulatory approval.