Single-cell analysis of human triple-negative breast cancer revealed heterogeneous macrophage populations with opposing phenotypes — proinflammatory and proresolution of inflammation. Paradoxically, both subsets accumulated in therapy-refractory residual tumors but showed inverse correlations across patients, suggesting mutually exclusive resistance mechanisms. Inflammatory macrophages localized preferentially to epithelial-like tumors, whereas proresolution macrophages were enriched in mesenchymal-like tumors. Mouse models faithfully recapitulated these patterns. After chemoimmunotherapy, mesenchymal-like tumors expanded proresolution macrophages through phagocytosis/efferocytosis, ω-3 fatty acid uptake, and resolvin production. Macrophage-secreted C1q emerged as a principal antagonist of T cell function by targeting mitochondria and inducing metabolic dysfunction. By contrast, epithelial-like tumors accumulated inflammatory macrophages and neutrophils that produced prostaglandins via ω-6 fatty acid pathways. Knocking down ELOVL5 — an elongase involved in ω-3 and ω-6 metabolism — mitigated both neutrophil- and macrophage-mediated immunosuppression. These distinct axes, driven by dysregulated inflammation and resolution programs, converged to undermine therapy-induced immunosurveillance; however, targeting their shared upstream regulators may overcome these resistance mechanisms.
Liqun Yu, Charlotte Rivas, Fengshuo Liu, Yichao Shen, Ling Wu, Zhan Xu, Yunfeng Ding, Xiaoxin Hao, Weijie Zhang, Hilda L. Chan, Jun Liu, Bo Wei, Yang Gao, Luis Becerra-Dominguez, Yi-Hsuan Wu, Siyue Wang, Tobie D. Lee, Xuan Li, Xiang Chen, David G. Edwards, Xiang H.-F. Zhang
Usage data is cumulative from February 2026 through June 2026.
| Usage | JCI | PMC |
|---|---|---|
| Text version | 2,002 | 0 |
| 624 | 0 | |
| Figure | 462 | 0 |
| Supplemental data | 316 | 0 |
| Citation downloads | 149 | 0 |
| Totals | 3,553 | 0 |
| Total Views | 3,553 | |
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.