Deciphering Human-KSHV Protein Interactions Driving Kaposi Sarcoma with AI Models

About This Grant

Project Summary Kaposi’s sarcoma-associated herpesvirus (KSHV) is a gammaherpesvirus responsible for Kaposi’s sarcoma (KS) and primary effusion lymphoma (PEL), malignancies that predominantly affect immunocompromised individuals. The cyclic GMP–AMP synthase (cGAS)–stimulator of interferon genes (STING) pathway plays a crucial role in antiviral immunity by detecting viral DNA and inducing type I interferon responses. However, KSHV has evolved mechanisms to inhibit this pathway, enabling immune evasion, viral persistence, and tumorigenesis. Despite its importance, the precise KSHV proteins that interact with and suppress cGAS-STING signaling remain largely unidentified. Traditional experimental approaches for mapping host-virus protein-protein interactions (PPIs) are labor-intensive, low-throughput, and often fail to capture the full complexity of viral immune modulation. To address this challenge, we aim to develop an AI-driven computational framework that can systematically identify KSHV-human PPIs, with a focus on viral proteins that inhibit cGAS-STING signaling. Our central hypothesis is that these interactions can be accurately predicted using an AI-based framework and experimentally validated to reveal underlying mechanisms of KSHV-driven pathogenesis. This study aims to uncover novel viral immune evasion mechanisms, enhance our understanding of KSHV pathogenesis, and uncover new therapeutic targets for KSHV-associated cancers by (1) combining deep learning and protein language models to predict high- confidence viral-host interactions using large-scale protein sequences, (2) identifying cGAS/STING-based KSHV proteins with molecular validation, and (3) characterizing functional cGAS/STING inhibitions and evaluating candidate KSHV ORFs in latency establishment or lytic replication. The to-be-developed AI models and tools will be open-sourced and widely disseminated within the community. The success of this project will have broader implications for studying immune evasion mechanisms in other pathogenic viruses, paving the way for innovative antiviral and immunotherapeutic strategies.