Non-cytolytic viral vaccine vectors induce durable effector-memory CD8 T cell immunity by triggering type I interferon [scRNA-Seq]

Replication-deficient viral vector systems hold promise for CD8 T cell-based vaccination, but the molecular mechanisms accounting for platform-specific differences in immunogenicity remain ill-defined. Here we compared lymphocytic choriomeningitis virus- (rLCMV) and vesicular stomatitis virus-based (rVSV) single-cycle vectors to explore vector-specific features determining long-lived effector-memory CD8 T cell induction. rLCMV, a non-cytolytic vector, elicited more durable CD8 T cell responses and a higher proportion of effector-memory CD8 T cells than cytolytic rVSV. When rVSV was re-engineered to be non-cytolytic (rVSVMq), also this vector induced durable and effector-differentiated CD8 T cell memory, reminiscent of rLCMV-induced responses, and it afforded superior protection against Listeria challenge than rVSV. Improved CD8 T cell responses of non-cytolytic rVSVMq depended on an elevated type I interferon (IFN-I) response and its direct sensing by vaccination-induced CD8 T cells. Many rVSVMq-infected cells in the splenic marginal zone that were in contact with antigen-specific CD8 T cells expressed not only vectorized antigen but also IFN-I, suggesting that non-cytolytic viral vectors promote the integration of peptide-MHC and IFN-I signals during T cell activation. These mechanistic insights should help to refine vaccines aimed at eliciting durable and protective effector-memory CD8 T cell immunity. Overall design: Single-cell RNA sequencing was performed on antigen-specific CD8 T cells isolated from C57BL/6 mice vaccinated with replication-deficient viral vectors. Mice received either rVSV-S1, rVSVMq-S1, or rLCMV-S1 vectors, and SARS-CoV-2 spike S1 epitope-specific CD8 T cells were sorted using MHC class I tetramers at day 7 and day 14 post-immunization. Each condition included 3 biological replicates which were labelled with hashtag oligos (HTOs) before being pooled prior to GEM generation. Experimental variables include vector type (rVSV, rVSVMq, rLCMV) and time point (day 7 vs. day 14).