Abstract

Immunoregulatory proteins expressed by SARS-CoV-2 interfere with host antiviral defences in infected cells and play critical roles in the pathogenesis and clinical manifestations of COVID-19. Here, we established a prediction algorithm by integrating a pretrained protein-language model and gene weights in immune-related pathways to quantify perturbations of SARS-CoV-2 proteins in host immunity. The results revealed that the canonical NF-κB pathway was dynamically regulated by SARS-CoV-2 infection and that nonstructural protein 1 (Nsp1) significantly suppressed the activation of the NF-κB pathway by other viral proteins and proinflammatory cytokines, such as IL-1β. Nsp1 binds to TAK1 at the TAB1-binding domain, promoting TRIM21-mediated K48-linked ubiquitination and subsequent proteasomal protein degradation, leading to the inactivation of the NF-κB signalling pathway. This work presents a novel framework to identify viral immunoregulators at the pathway level and provides mechanistic insights into immune evasion by SARS-CoV-2. SARS-CoV-2 infection causes a clinical spectrum ranging from asymptomatic infection to fatal illness, in which the dysregulated immune response plays a role. Unravelling the mechanisms of immunoregulation by viral proteins would aid in the development of new immunotherapies. In this study, we established a deep-learning algorithm to bridge the gap between ‘binary’ virus-host protein-protein interactions and complex pathway-level regulation. With the help of this predictor, GWPIS, we identified a key NF-κB suppressor, Nsp1. Mechanistically, Nsp1 binds to the TAB1-binding domain of TAK1, thereby inhibiting the formation of the TAK1-TAB1 complex and downstream signaling, including the NF-κB, MAPK and AP-1 pathways. Moreover, Nsp1 promotes TRIM21-dependent K48-linked polyubiquitination of TAK1, leading to proteasomal degradation and downregulation of TAK1, further suppressing downstream signaling. This study provides novel insights into how the