ABSTRACT: Why some individuals recover from COVID-19, while others do not, remains unknown. Mitochondrial dysfunction has been identified as a hallmark of Long COVID, however, multiple mitochondria haplotypes exist in the human population. Here, we used two mouse models that share the same nuclear genome but differ in mitochondrial genomes and compare their recovery rate after SARS-CoV-2 infection. While in one model, mice recovered, in the other model, male mice did not recover. RNAseq analysis, over a post-infection time course, revealed activation of the integrated stress response (ISR), which is known to be activated during acute viral infection through the PKR kinase. Therefore, and as expected, in recovered mice, activation of the ISR is transient and restricted to the acute phase of infection. However, in mice that did not recover, the ISR shows a sustained activation at time points where viral particles are no longer detectable. Rather, the activation of the ISR correlates with the up-regulation of endogenous mitochondria derived dsRNA, which were recently shown to be the preferred substrate of the PKR kinase. Sustained activation of the ISR was confirmed in 3 independent datasets of Long COVID patients in human. We show that treatment with the ISR inhibitor, ISRIB, correlates with recovery in mice affected by Long COVID. Since ISRIB is currently tested in clinical trials for other conditions and is reported to improve symptoms associated with Long COVID such as memory lost, gut inflammation and cardiomyopathy in preclinical models, our data opens the possibility to explore ISRIB as an intervention for the treatment of Long COVID.
