Senescent Cell-Derived Extracellular Vesicles Inhibit Cancer Recurrence by Coordinating Immune Surveillance

Abstract

Senescence is a nonproliferative survival state that cancer cells can enter to escape therapy. In addition to soluble factors, senescence cells secrete extracellular vesicles (EV), which are important mediators of intercellular communication. To explore the role of senescent cell (SC)-derived EVs (senEV) in inflammatory responses to senescence, we developed an engraftment-based senescence model in wild-type mice and genetically blocked senEV release in vivo, without significantly affecting soluble mediators. SenEVs were both necessary and sufficient to trigger immune-mediated clearance of SCs, thereby suppressing tumor growth. In the absence of senEVs, the recruitment of MHC-II+ antigen-presenting cells (APC) to the senescence microenvironment was markedly impaired. Blocking senEV release redirected the primary target of SC signaling from APCs to neutrophils. Comprehensive transcriptional and proteomic analyses identified six ligands specific to senEVs, highlighting their role in promoting APC-T cell adhesion and synapse formation. APCs activated CCR2+CD4+ TH17 cells, which seemed to inhibit B-cell activation, and CD4+ T cells were essential for preventing tumor recurrence. These findings suggest that senEVs complement the activity of secreted inflammatory mediators by recruiting and activating distinct immune cell subsets, thereby enhancing the efficient clearance of SCs. These conclusions may have implications not only for tumor recurrence but also for understanding senescence during de novo carcinogenesis. Consequently, this work could inform the development of early detection strategies for cancer based on the biology of cellular senescence. Significance: Chemotherapy-treated senescent tumor cells release extracellular vesicles that trigger an immune response and suppress tumor recurrence. See related commentary by Almeida and Melo, p. 833.

Figure 1.. A novel engraftment-based senescence model enables study of the senescence microenvironment.

(A) Optimization of in vitro senescence induction in MOC2 squamous carcinoma cells through 48-hour treatment with specified concentrations of paclitaxel and cisplatin. Ideal conditions ensured elevated senescence levels (quantified by increased SA-β-Gal staining and reduced BrdU incorporation), minimized apoptosis/cell death (measured by annexin V and 7-AAD staining, respectively), and maximization of cell recovery (quantified by cell counting) (n=4). The first column on the left of each heatmap has 0nM paclitaxel and therefore represents cisplatin only. The bottom row of each heatmap has 0uM cisplatin and therefore represents paclitaxel only. The rest of the heatmap shows combination of the two chemotherapeutic agents. (B) Differential expression of senescence (Cdkn1a [p21], Cdkn2a [p16], Glb1 [SA-B-Gal]) and proliferation (Ki67) markers in paclitaxel-treated (Pac, 600 nM) vs. untreated (UT) MOC2 cells in vitro analyzed by bulk RNA sequencing (n=3). (C) Inhibition of senescent cell (MOC2) population expansion by continuous 600 nM paclitaxel treatment for 5 days with fresh media replenishment every 48 hours (n=3, error bars=SD). Statistical analysis conducted using repeated-measure one-way ANOVA with multiple testing correction. (D) Inhibitory effect of paclitaxel treatment on MOC2 proliferation over a 24-hour period, resulting in 15% BrdU incorporation (~7-fold reduction from 100% maximum BrdU incorporation; n=3; error bars=SD). (E) Effect of paclitaxel treatment on BrdU incorporation rate in MOC2 cells, demonstrating an 8-fold reduction following a 2-hour BrdU pulse (n=3). (F) Persistence of the senescent phenotype in vitro, as evidenced by unaltered SA-B-Gal activity for 7 days after paclitaxel removal (n=3). (G) Delay in tumor formation by paclitaxel treated senescent MOC2 cells implanted orthotopically in the dermis of mice (1M per mouse, n=10). (H) Flow cytometric analysis of immune cell subsets infiltrating tumor or senescent MOC2 cell implants on day 3 post-challenge with 5E5 cells orthotopically (n=5). Statistical analysis conducted using non-parametric one-way ANOVA (Kruskal-Wallis test) with multiple testing correction.

Figure 2.. Senescent cell-derived EVs differ from tumor-derived EVs in morphology and protein composition.

(A) Increased extracellular vesicle (EV) protein content per million senescent MOC2 cells (600 nM paclitaxel for 48 hours), determined to be 6-fold higher compared to non-senescent cells (control), as assessed by NanoDrop (n=3). EV isolation conducted from cell supernatant utilizing ultracentrifugation. Data representative of both MOC2 and mEER+ cell lines. (B) Nanoparticle tracking analysis (NTA) of EVs from senescent MOC2 cells: A 16% increase in median diameter corresponding to a 56% increase in volume compared to EVs from proliferating cells (n=5). No differences in EV numbers between proliferating (Ctrl) and senescent (Pac) cell-derived EVs (rightmost graph). (C) Electron microscopy (EM) analysis of senescent EVs from both MOC2 and mEER+ cell lines confirms that senEVs are 16% larger than tumor-derived EVs. Data from >10 images per condition. (D) Immunoblot for the pan-EV marker Flotillin-1. (E) Differential proteomic analysis of EVs isolated from senescent and proliferating squamous carcinoma cells reveals that 150 and 120 proteins exhibit significantly altered expression levels in EVs derived from senescent cells (MOC2 and mEER+, respectively), compared to EVs from proliferating cells (tEVs). (F) Pathway analysis (DAVID) of enriched proteins in senescent EVs from senescent MOC2 revealed significant associations with core cellular functions such as ribosomes, spliceosomes, DNA replication, and type-1 immunity (padj<0.01). (G) Comparison of proteins enriched in senescent cell-derived EVs (as compared to EVs from proliferating cells) shows a 30% overlap (31 out of 109) between two distinct senescent cell types. Statistical tests were conducted utilizing Mann-Whitney or One-Way ANOVA. Proteomic analysis employed multiple comparisons correction using the Benjamin, Krieger, and Yekkutieli method.

Figure 3.. Expression of Rab35-DN does not inhibit the senescence-associated secretory phenotype

(A) Effective and targeted reduction of EV release detected in both proliferating and senescent MOC2 cells expressing Rab35-DN using immunoblot assay for the pan-EV marker Flotillin-1. EV preparations were normalized by cell number. UT: untransduced, WT: wild type, DN: dominant negative. (B) Detection of reduced extracellular vesicle (EV) release in both proliferating and senescent MOC2 cells expressing Rab35-DN utilizing nanoparticle tracking analysis (NTA), which assesses EV numbers and size via light scatter (left) and fluorescence (right) measurements. (C) Senescence-associated secretory phenotype (SASP) factors are not decreased by Rab35-DN expression, as evidenced by quantification of 40 secreted factors using Mouse Cytokine Array on senescent MOC2 cell culture supernatant. (D) Bulk RNA sequencing of Rab35-WT and Rab35-DN senescent MOC2 cells (left) and of proliferating and senescent MOC2 cells (right) confirms that Rab35-DN expression has minimal impact on the global SASP (red), defined as genes containing a signal peptide and lacking a transmembrane domain (left); and that changes in gene profile after Rab35-DN expression are minor compared to changes induced by senescence stimuli (right). Significance level set at adjusted p-value < 0.01. Statistical tests by two-way ANOVA with multiple testing correction.

Figure 4.. Senescent cell-derived EVs recruit antigen-presenting cells and are necessary and sufficient to inhibit cancer recurrence

(A) Rab35-DN expression in senescent cells promotes tumor recurrence in C57BL/6J mice challenged orthotopically (dermis or tongue) with 5E5 senescent cells (three independent experiments, two different senescent cell types, n=10). (B-C) Rab35-DN expression in senescent cells promotes accumulation of senescent MOC2 cells 3 days after engraftment, as measured by flow cytometry (B; n=5) and by single-cell sequencing (C; n=3), utilizing the transgenic marker dLNGFR. Data representative of two independent experiments. (D) Quantification of immune cell infiltrates by flow cytometry 3 days after implantation of Rab35-WT or Rab35-DN senescent MOC2 cells (data representative of two independent experiments, see Figure S4D). (E) Administration of senEVs to mice challenged with Rab35-DN senescent MOC2 cells reverts tumor recurrence (N=10). Statistical tests by one-way ANOVA, two-way ANOVA or Mann-Whitney. Multiple testing correction applied (Sidak).

Figure 5.. Antigen presenting cells recruited by senEVs require T_H17 cells to inhibit tumor recurrence

(A) Identification of three major cellular lineages within the senescence microenvironment through single-cell sequencing: immune, epithelial, and stromal cells, identified based on differentially expressed genes in each lineage (merged data of two independent experiments). (B) Sub-clustering of cells belonging to the immune lineage distinguishes between different immune cell subsets. (C-D) Identification of engrafted senescent cell subset among epithelial lineage cells. Proliferating MOC2 cells, stem-like senescent MOC2 cells and senescent MOC2 cells were defined based on markers of proliferation (Mki67), stemness (Sox9, Hmga2, Ly6a/Sca1, and Krt19), and senescence (Cdkn1a and Cdkn2a). (E) Visualizing cell-cell communications between senescent cells and immune cells mediated by membrane-bound ligand-receptor interactions using chord diagrams. Each sector in the chord diagrams is an arrow depicting the direction of the ligand-receptor signaling. (F) Identification of membrane ligands specifically present on senescent cell-derived EVs. Communications emanating from senescent cells that were enriched in presence of senEVs (differential membrane signaling between Rab35-WT versus Rab35-DN senescent MOC2 cells, left) were intersected with the senEV proteome (right). This analysis was restricted to membrane-bound ligands. The resulting short-list (right) represents putative ligands specifically present on senEVs and not on senescent cell surface. The bubble plot (right) identifies the cellular targets within the senescent microenvironment. (G) Visualizing cell-cell communications between antigen presenting cells (APC) and T_H17 cells mediated by membrane-bound ligand-receptor interactions using chord diagrams. Each sector in the chord diagrams is an arrow depicting the direction of the ligand-receptor signaling. (H) Tumor recurrence by Rab35-WT senescent MOC2 cells is increased in absence of CD4 T cells, to levels like those observed in Rab35-DN senescent MOC2 cells. (I) Schematic of the proposed mechanism of EVs in senescence surveillance. Created with BioRender.com (FP; https://app.biorender.com/illustrations/66c3d10ff323a174357c3c4e?slideId=3a8de68c-e682-4f15-87f2-1ea5986d7a89))