Resistance to oncogene-targeted therapies involves discrete drug-tolerant persister cells, originally discovered through in vitro assays. Whether a similar phenomenon limits efficacy of programmed cell death 1 (PD-1) blockade is poorly understood. Here, we performed dynamic single-cell RNA-Seq of murine organotypic tumor spheroids undergoing PD-1 blockade, identifying a discrete subpopulation of immunotherapy persister cells (IPCs) that resisted CD8+ T cell–mediated killing. These cells expressed Snai1 and stem cell antigen 1 (Sca-1) and exhibited hybrid epithelial-mesenchymal features characteristic of a stem cell–like state. IPCs were expanded by IL-6 but were vulnerable to TNF-α–induced cytotoxicity, relying on baculoviral IAP repeat-containing protein 2 (Birc2) and Birc3 as survival factors. Combining PD-1 blockade with Birc2/3 antagonism in mice reduced IPCs and enhanced tumor cell killing in vivo, resulting in durable responsiveness that matched TNF cytotoxicity thresholds in vitro. Together, these data demonstrate the power of high-resolution functional ex vivo profiling to uncover fundamental mechanisms of immune escape from durable anti–PD-1 responses, while identifying IPCs as a cancer cell subpopulation targetable by specific therapeutic combinations.
Kartik Sehgal, Andrew Portell, Elena V. Ivanova, Patrick H. Lizotte, Navin R. Mahadevan, Jonathan R. Greene, Amir Vajdi, Carino Gurjao, Tyler Teceno, Luke J. Taus, Tran C. Thai, Shunsuke Kitajima, Derek Liu, Tetsuo Tani, Moataz Noureddine, Christie J. Lau, Paul T. Kirschmeier, David Liu, Marios Giannakis, Russell W. Jenkins, Prafulla C. Gokhale, Silvia Goldoni, Maria Pinzon-Ortiz, William D. Hastings, Peter S. Hammerman, Juan J. Miret, Cloud P. Paweletz, David A. Barbie
Usage data is cumulative from April 2024 through April 2025.
Usage | JCI | PMC |
---|---|---|
Text version | 1,976 | 590 |
196 | 112 | |
Figure | 596 | 6 |
Table | 47 | 0 |
Supplemental data | 378 | 22 |
Citation downloads | 68 | 0 |
Totals | 3,261 | 730 |
Total Views | 3,991 |
Usage information is collected from two different sources: this site (JCI) and Pubmed Central (PMC). JCI information (compiled daily) shows human readership based on methods we employ to screen out robotic usage. PMC information (aggregated monthly) is also similarly screened of robotic usage.
Various methods are used to distinguish robotic usage. For example, Google automatically scans articles to add to its search index and identifies itself as robotic; other services might not clearly identify themselves as robotic, or they are new or unknown as robotic. Because this activity can be misinterpreted as human readership, data may be re-processed periodically to reflect an improved understanding of robotic activity. Because of these factors, readers should consider usage information illustrative but subject to change.