Senescence and the tumor-immune landscape: Implications for cancer immunotherapy

Abstract

Cancer therapies, including conventional chemotherapy, radiation, and molecularly targeted agents, can lead to tumor eradication through a variety of mechanisms. In addition to their effects on tumor cell growth and survival, these regimens can also influence the surrounding tumor-immune microenvironment in ways that ultimately impact therapy responses. A unique biological outcome of cancer therapy is induction of cellular senescence. Senescence is a damage-induced stress program that leads to both the durable arrest of tumor cells and remodeling the tumor-immune microenvironment through activation of a collection pleiotropic cytokines, chemokines, growth factors, and proteinases known as the senescence-associated secretory phenotype (SASP). Depending on the cancer context and the mechanism of action of the therapy, the SASP produced following therapy-induced senescence (TIS) can promote anti-tumor immunity that enhances therapeutic efficacy, or alternatively chronic inflammation that leads to therapy failure and tumor relapse. Thus, a deeper understanding of the mechanisms regulating the SASP and components necessary for robust anti-immune surveillance in different cancer and therapy contexts are key to harnessing senescence for tumor control. Here we draw a roadmap to modulate TIS and its immune-stimulating features for cancer immunotherapy.

1. Introduction

Cytotoxic regimens that directly kill tumor cells while sparing normal cells have historically been the primary pursuit of cancer medicine and believed necessary to achieve cure. Through a better understanding of the mechanisms by which cytotoxic and other cancer drugs work, we have come to understand treatment does not need to directly kill tumor cells to achieve clinically-relevant tumor responses [1–3]. Indeed, there are numerous biological outcomes of cancer therapy. In addition to inducing non-immunogenic (apoptosis) and immunogenic (pyroptosis, necroptosis) cell death, cancer therapies can also block tumor cell growth through a transient (quiescence) or more durable (senescence) cell cycle arrest that impacts not only the tumor itself but also its surrounding tumor microenvironment (TME) in ways that dictate therapy responses [2, 4]. The TME includes a diverse array of cellular and non-cellular components, including stromal fibroblasts, vascular and lymphatic vessels, immune cells, and the extracellular matrix (ECM) that can exert both tumor suppressive and promoting effects [5]. Over the past decade the golden age of cancer immunotherapy has emerged, based on the principle that reactivating the immune system, and in particular cytotoxic T cells, can be harnessed as a means to indirectly kill tumor cells systemically throughout the body [6, 7]. Thus, understanding the biological outcomes of cancer therapy and the mechanisms by which they impact both cancer cells and the surrounding tumor-immune microenvironment is critical to engineering the next generation of cancer therapeutics and treatment paradigms.

Senescence is a damage-induced cellular response to cancer therapy. Cellular senescence was first described by Leonard Hayflick and colleagues, who observed that human diploid fibroblasts undergo a finite number of doublings before irreversibly arresting [8, 9]. As such, this process was named senescence (derived from the Latin root senex meaning “old”) in reference to its association with cellular aging. However, we now know senescence to be a conserved response to many different types of external and internal cellular stress, including not just telomere shortening associated with cell division (i.e. replicative senescence) but also oncogenic, genotoxic, metabolic, and oxidative damage, all of which can accumulate following cancer therapy [10–12]. Though initially thought a cell culture artefact, over the past two decades a plethora of tools and models to remove senescent cells in normal and disease contexts in in vivo settings, as well as observation of senescence-related markers in patients with cancer and other age-related chronic diseases, has revealed senescence to be a bona fide biological process. In fact, senescence has been shown to be a rather broad physiological response to tissue damage that plays a pleiotropic role in aging, embryonic development, wound healing, tissue regeneration, and, importantly, response to oncogenesis and cancer therapy [13–18].

Hallmarks of senescence include: (a) stable cell cycle arrest mediated by two tumor suppressor pathways, p53/p21^CIP1 and p16^INK4a/RB, that lead to transcriptional and epigenetic repression (visible as senescence-associated heterochromatin foci (SAHF)) of key cell cycle genes even after the senescence inducer is removed; (b) changes in cellular morphology (cell flattening, enlargement, multinucleation); (c) dissolution of the nuclear envelope (Lamin B1 loss); (d) enhanced lysosomal activity as marked by senescence-associated beta-galactosidase (SA-β-gal); (e) upregulation of anti-apoptotic and pro-survival pathways (e.g. BCL-2 family members) leading to resistance to cell death; (f) macromolecular damage (e.g. DNA, protein damage) and a persistent DNA damage response (DDR); and (g) altered metabolism (e.g. mitochondrial dysfunction, increased glycolysis, lipid accumulation) [19]. Enhanced transcription and protein synthesis in senescent cells converges to promote a unique aspect of senescence, the senescence-associated secretory phenotype (SASP), which allows senescent cells to communicate damage signals with neighboring cells, including immune cells, fibroblasts, endothelial cells, and adjacent non-tumor epithelial cells in their surrounding microenvironment in a non-cell autonomous manner [20, 21]. Enhancer regions of hundreds of SASP factors are made accessible, transcribed by NF-κB and other factors, and translated in an mTOR-dependent manner, resulting in the robust secretion of inflammatory cytokines and chemokines and angiogenic, growth, and ECM degrading signals [22]. The SASP is also associated with upregulation of cell surface molecules that modulate the interaction between senescent cells and the immune system [23, 24]. It is important to note that these hallmarks listed above are neither specific nor universal to all forms of cellular senescence, and that multiple markers are currently necessary to distinguish senescence from other biological outcomes of cancer therapy [10].

The dynamic nature of the SASP and its ability to modulate surrounding tissue microenvironments and immune responses in different ways is thought to contribute to the many contrasting physiological phenomenon associated with senescence. Indeed, senescence can be anti- or pro-tumorigenic depending on the senescence inducer, the duration of senescence, the SASP factors produced, and tissue and disease context [25, 26]. Markers of cellular senescence have been identified in pre-cancerous lesions in various solid organs in humans, including the lungs, prostate, pancreas, and skin, and are subsequently lost during neoplastic progression [18, 27]. As such it has been postulated that senescence may be tumor suppressive and act as a roadblock to tumor development by preventing the proliferation of potentially malignant cells. Supporting this hypothesis, oncogene-induced senescence following aberrant RAS activation leads to arrest of pre-malignant cells and secretion of pro-inflammatory SASP factors that promotes innate and adaptive immune clearance of incipient cancer cells and blocks tumor formation [28]. Similarly, following treatment of established KRAS mutant cancers with RAS pathway targeted therapies, we have shown that therapy-induced senescence (TIS) inhibits tumor growth and leads to influx of cytotoxic CD8⁺ T cells and Natural Killer (NK) cells that promote tumor regression [29, 30]. Conversely, there is a large body of evidence demonstrating that unresolved senescence, and in particular the SASP following chemotherapy, can be tumor promoting through secretion of immune suppressive factors and attraction of immune suppressive cell types, as well as production of angiogenic and other growth factors that lead to increased invasion and metastasis of adjacent non-senescent tumor cells [31–33]. Moreover, it is now appreciated that senescence is likely not a permanent or irreversible state, and that tumor cells that bypass senescence through acquisition of genomic instability (e.g. polyploidy) can gain enhanced stemness and tumorigenic potential that contributes to treatment relapse [34–37].

In order to effectively leverage senescence and its immune stimulating properties for cancer treatment, we need a better understanding of: (a) the biological markers of senescence in cancer and following therapy, (b) the context dependent regulation of the SASP and its impact on immunity in different cancer settings, and c) those senescence and SASP factors that are necessary for immune activation and tumor suppression. It is possible that senescence is a more common outcome of cancer therapy than initially appreciated, but may have different markers and immune and tumor outcomes in various settings. This review will focus on the known impact of senescence on the tumor-immune microenvironment, the mechanisms by which this is regulated and can be manipulated, and in the end propose a strategy for how to best harness senescence for cancer immunotherapy.

2. SASP remodeling of the tumor microenvironment: Lessons from genetic models of senescence.

In vivo mouse models offer the ability to tightly control senescence in specific tumor and non-tumor cell types within the context of an intact tissue microenvironment, and have been instrumental in uncovering the impact of the SASP on remodeling the TME in cancer [15]. Genetic induction of senescence in pre-malignant or tumor contexts in vivo has been achieved through activation of oncogenes (RAS, BRAF) or modulation of tumor suppressor genes (p53, p21, p27^KIP1, PTEN), leading to both cell cycle arrest and SASP production through accumulation of DNA damage and/or activation of p53 and RB pathways [28, 38–42]. Below we highlight current evidence from these models demonstrating how the SASP affects immune activity and stromal remodeling in a manner that ultimately impacts tumor onset and progression.

2.1. Immune surveillance

Incipient tumor cells, including those driven into senescence by oncogenic insults, undergo immune surveillance as a failsafe mechanism to block tumor development [43, 44]. This initially involves the innate immune system, including phagocytic macrophages that engulf pre-malignant cells, and NK cells that can secrete cytolytic granules to eliminate damaged target cells that express stress ligands [45–47]. The adaptive immune system is also activated following presentation of antigens on MHC Class I (MHC-I) molecules on tumor cells and MHC Class II (MHC-II) molecules on dendritic cells (DCs) to cytotoxic (CD8⁺) and helper (CD4⁺) T cells, respectively, which upon binding to matched T cell receptors (TCRs) can lead to T cell clonal expansion and killing of cells harboring tumor-specific antigens through similar mechanisms as NK cells [48]. Tumorigenic cells also secrete soluble factors in the form of cytokines (e.g. interleukins (ILs)) that can control immune cell activity, and chemokines (e.g. C-C motif chemokine ligands (CCLs) and C-X-C chemokine ligands (CXCLs)) that attract immune cells into tissues, leading to the initiation and maintenance of immune surveillance at the tumor site [49, 50].

Many of these immunomodulatory cell surface and secreted factors involved in cancer immune surveillance are expressed and produced by senescent cells as part of the SASP and contribute to its tumor suppressive functions. In mouse models of Nras^G12V-driven senescence in the liver, senescence not only inhibits the proliferation of incipient cancer cells, but also leads to their clearance by the innate (phagocytic macrophages) and adaptive (CD4⁺ helper T cells) immune system that is necessary to block liver tumor development [28]. Recent evidence using similar models of Kras^G12V-induced senescence suggests that immune-mediated clearance is dependent on p21 and its ability to induce the SASP chemokine CXCL14 that attracts macrophages to surveil senescent cells and activate cytotoxic T cell responses in the liver [40]. Upregulation of MHC-I and MHC-II on human melanocytes and melanoma cells through SASP factors such as IL-1β following mutant NRAS and BRAF overexpression also leads to increased CD8⁺ and CD4⁺ T cell proliferation and cytotoxicity in vitro and in vivo [51, 52].

Reactivation of senescence in established tumors can also have powerful anti-tumor effects through the SASP. Genetically restoring p53 tumor suppressor expression in Hras^G12V mutant hepatocellular carcinoma (HCC) produced senescence and SASP factors such as CCL2, IL-15, and CXCL1 that led to innate immune-mediated tumor regressions through macrophage, neutrophil, and NK cell mobilization [39]. A follow-up study demonstrated that anti-tumor NK cell immunity was mediated in part by (a) expression of stimulatory NK cell receptor ligands (e.g. RAE-1, ULBP1, H60a) on senescent tumor cells that bind to the activating NKG2D receptor on NK cells to trigger cytotoxicity, and (b) SASP-associated secretion of CCL2 that attracted NK cells into the liver TME [53]. Moreover, p53-induced senescence in hepatic stellate cells leads to induction of SASP factors such as IL-6 and interferon-γ (IFN-γ) that promote polarization of macrophages from an immune suppressive M2 to an immune stimulatory M1 state where they are capable of phagocytosing and clearing senescent cells [54]. Thus, the SASP can contribute to anti-tumor immunity through secreted and cell surface proteins that can activate both the innate and adaptive arms of the immune system.

2. Immune escape

Tumors undergo immune editing as they evolve over time, leading to downregulation of immune-stimulating receptors and ligands and upregulation of inhibitory checkpoints that block the ability of the innate and adaptive immune system to recognize and eliminate pre-malignant and malignant tumor cells [43, 44]. In addition, anti-inflammatory cytokines and other chemokines secreted by tumor cells can lead to an influx of suppressive immune cells, most notably regulatory T cells (T_regs) and myeloid-derived suppressor cells (MDSCs), that inhibit NK and CD8⁺ T cell cytotoxicity and sedate anti-tumor immunity [55].

Prolonged senescence and chronic SASP induction can also lead to similar mechanisms of immune escape. For example, BRAF^V600E mutant melanocytic nevi undergo senescence but are able to escape immune surveillance, and as such accumulate with age in humans [38]. Recent work suggests that senescent nevi do not express activating NKG2D receptor ligands, and that their secretion of SASP factors such as IL-6 even leads to upregulation of the inhibitory MHC molecule HLA-E that suppresses NK and T cell clearance of these pre-malignant lesions [56, 57]. Moreover, if pre-malignant hepatocytes are not effectively cleared by the immune system following Nras^G12V-induced senescence, the SASP through chemokine CCL2 can recruit immature suppressive myeloid cells (e.g. MDSCs) that inhibit NK cell function and promote progression to HCC [58]. In agreement, increased senescence markers in peritumoral tissue correlated with worsened prognosis in HCC patients [58]. These findings underscore the differential effects of acute and timely SASP versus a chronic and unresolved SASP on immune responses in cancer.

Certain SASP components can even regulate different immune responses depending on the cancer type. Whereas activation of COX-2 and production of the lipid SASP factor species prostaglandin E2 (PGE2) is necessary for macrophage maturation and surveillance of pre-malignant hepatocytes following Nras^G12V-induced senescence in the liver [59], Hras^G12V-mediated induction of PGE2 in senescent thyrocytes conversely promotes immune suppressive M2 macrophage polarization [60]. The cell type undergoing senescence, as well as the senescence-inducing stimuli, also impacts immune responses. Stromal senescence following p27 induction in fibroblasts can promote tumor progression through an IL-6-mediated influx of MDSCs and T_regs that inhibit anti-tumor cytotoxic T cell responses [41]. Similarly, PTEN tumor suppressor loss, which induces senescence in pre-malignant prostate lesions through activation of the oncogenic PI3K pathway and subsequent p53 stabilization [42], produces an immune suppressive SASP including factors such as CSF-1, IL-1β, IL-6, IL-10, and CXCL1 that attract MDSCs that inhibit cytotoxic NK and T cell responses and promote the transition to prostate adenocarcinoma [61, 62]. These findings suggest that the SASP can lead to differential anti-tumor (e.g. NK cells, helper and cytotoxic T cells, M1 macrophages) and pro-tumor (e.g. M2 macrophages, MDSCs, T_regs) immunity in different contexts.

2.3. Vascular remodeling

Besides direct immune modulation via release of cytokines and chemokines, the SASP may indirectly affect immune activity by impacting other stromal cells in the TME, in particular endothelial cells. Senescent cells produce high levels of pro-angiogenic SASP factors of the vascular endothelial factor (VEGF), platelet-derived growth factor (PDGF), and fibroblast growth factor (FGF) families that can mediate new blood vessel formation and vascular remodeling [31]. Early efforts to understand the impact of the SASP on the TME and how that ultimately influences tumor progression demonstrated that senescent fibroblasts secrete paracrine factors such as VEGF that promote new blood vessel formation (i.e. angiogenesis) and breast cancer tumorigenesis and progression in immunocompromised mice [63, 64]. In addition, lymphangiogenesis provoked by secretion of CXCL12 from BRAF^V600E mutant senescent thyroid cancer cells at the invasive front can also promote metastatic dissemination [65].

Though SASP-mediated induction of blood or lymphatic vessel formation can have tumor-propagating properties, it can also have immune stimulating properties by facilitating immune cell trafficking into tumors. We have recently shown that therapy-induced senescence in pancreatic cancer models leads to induction of the pro-inflammatory SASP factors CCL5, CXCL1, and IL-6 that mediate a phenomenon known as endothelial activation [66]. This led to the upregulation of cell adhesion molecules such as VCAM-1 on blood vessels that facilitated the extravasation of CD8⁺ T cells that express its ligand, VLA-4, into tumors [30]. Thus, depending on the tumor context, both the pro-angiogenic and pro-inflammatory SASP can lead to vascular remodeling that can promote tumor cell intravasation through blood vessels and dissemination to distant organs or conversely immune cell extravasation and infiltration into tumors. More research is needed to understand the role of the pro-angiogenic SASP and senescence-mediated vascular remodeling in immune responses in other cancer settings. Nonetheless, evidence from preclinical models that senescence can directly or indirectly modulate anti-tumor immune responses through the SASP, and in some instances effectively block tumor formation or even shrink existing established tumors, has paved the way toward exploring whether senescence-inducing cancer therapies can achieve similar outcomes.

3. Therapy-induced senescence and immune regulation in cancer

Many FDA-approved chemo-, radio-, and targeted therapies have now been shown to induce senescence in cancer cells in vitro and tumor models in vivo depending on the dose, cancer type, and genetics of the tumor (Fig. 1), as summarized extensively in other reviews [17, 67]. Studies in preclinical mouse models that allow functional dissection of tumor and immune responses in a physiologically-intact setting have informed us of the mechanisms by which TIS regulates of the immune system, highlighting roles for the SASP in mediating anti-tumor immune surveillance, or, opposingly, the generation of a tumor-permissive immune suppressive microenvironment. The off-target effects of systemic therapies on senescence induction in non-tumor cells, as well as prevalent loss of tumor suppressor pathways (p53/p21, p16/RB) associated with cellular senescence in late-stage tumors, likely contribute to these diverse tumor and immune responses to TIS. Here we detail the known effects of different classes of senescence-inducing therapies on immune regulation, stromal remodeling, and immunotherapy outcomes in preclinical cancer models and clinical cancer patient samples (summarized in Table 1).

Fig. 1. Overview of senescence regulation of tumor-immune interactions following cancer therapy and potential senotherapeutic interventions.

(1) Certain cancer therapies, including radiation, chemotherapy, and targeted therapies such as CDK4/6 and AURKA inhibitors, can cause cellular senescence through induction of DNA damage, a prolonged DDR orchestrated by ATM/ATR, and/or downstream activation of p53 and RB-regulated pathways, leading to cell cycle arrest. (2) RB and epigenetic modifiers lead to chromatin changes in senescent cells that repress accessibility at cell cycle genes but increase open chromatin and binding of BRD4 to enhancer regions at SASP gene loci. (3) RB, ATM/ATR, p38MAPK, GATA4, and NOTCH1, as well as the cGAS-STING pathway that is activated by binding cytosolic double-stranded DNA, also interact with and activate TFs such as NF-κB, C/EBPβ, and STATs that bind to SASP gene promoters and promote their transcription. (4) Depending on the mechanism of action of the therapy, the cancer and cell type it acts on, and the upstream epigenetic and transcriptional regulators activated, the composition of the SASP can be incredibly heterogeneous, comprising pro-inflammatory, anti-inflammatory, and angiogenic factors, as well as lipids, MMPs, and other growth factors. (5) Many of these SASP factors highlighted have been shown to modulate the immune system and the vasculature in both tumor suppressive and tumor-promoting ways. In addition, the upregulation of cell surface proteins on senescent cells as part of the SASP can allow them to be recognized by (MHC-I, NK ligands) or alternatively evade (HLA-E, PD-L1) the immune system. (6) The secretome and surfaceome of senescent tumor cells can be leveraged for immunotherapy, including ICB therapies to block PD-L1 and HLA-E, CAR-T cells and antibodies engineered to bind senescence-related surface proteins, and neutralizing antibodies to target immune suppressive SASP factors, such as IL-6, IL-8, and TGF-β, and their receptors. (7) Similarly, senomorphic agents targeting SASP transcriptional regulation, including inhibitors of BRD4 (JQ1), STAT3 (Ruxolitinib), NF-κB (BAY 11–7082, Metformin), and p38MAPK (SB203580) can be used to modulate specific SASP programs. (8) Alternatively, senolytic agents targeting pro-survival and anti-apoptotic pathways in senescent cells such as BCL-2 (Navitoclax), MDM2 (UBX101), and PI3K signaling (Dasatinib + Quercetin) can be used to kill deleterious senescent tumor cells. AURKA, Aurora Kinase A; CAR-T, chimeric antigen receptor (CAR) T cell; DDR, DNA damage response; ICB, immune checkpoint blockade; mAb, monoclonal antibodies; MMP, matrix metalloproteinases; NCID, NOTCH intracellular domain; Pol II, RNA polymerase II; SASP, senescence-associated secretory phenotype; TF, transcription factor.

Table 1.

Senescence-inducing therapies and their known effects on immune and immunotherapy responses in cancer.

Therapy Class	Therapy/Combination	Cancer Type (model)	Senescence Biomarkers	Immune Response	Tumor Response	Refs
Ionizing Radiation (IR)	IR	IR-induced osteosarcoma mouse model	SA-β-gal, p16, p21, SASP (IL-6, CCL2/3/4/5)	NKT ↑	↓	88
	IR	NSCLC human cell line xenografts	SA-β-gal, STING, NF-κB, L1, p21, SASP (IFN-β, IL1α, IL-6)	Mac ↑	↓ (distal tumor)	98
	IR + PARPi (ex vivo in tumor cells)	Melanoma and PDAC syngeneic transplant mouse models	SA-β-gal, p16, p21, SASP (CCL5, IFNβ, CXCL9/10/11)	DC ↑ CD8+T ↑ NK ↑	↓	103
Chemotherapy	Cyclophosphamide	B cell lymphoma syngeneic transplant mouse model	SA-β-gal, NF-κB, p15, SASP (IL-6, IL-8, ICAM-1, CXCL1)	NK ↑	↓	74
	Doxorubicin or melphalan	MM syngeneic transplant mouse model	SA-β-gal, p16, p53, NK ligands (RAE-1, MICA, MULT-1, PVR)	NK ↑	↓	75–77
	Cisplatin + irinotecan (ex vivo in tumor cells)	Ovarian cancer syngeneic transplant mouse models	SA-β-gal, STING, p16, yH2AX, SASP (IL-6, VEGFA, GM-CSF)	DC ↑ CD8+T ↑	↓	78
	Docetaxel	PCa GEMM	SA-β-gal, p16, p21, SASP (GM-CSF, CSF-1, IL-10, CCL2, CXCL1/2)	MDSC ↑ NK ↓ CD8+T ↓	↑	61
	Mitoxantrone, other agents	PCa human xenografts; PCa patient samples	SA-β-gal, p16, SASP (IL-6, IL-8, MMPs, AREG), PD-L1	CD8+T ↓	↑	72
Aurora kinase inhibitors	MLN8054/MLN8237 (AURKAi)	Melanoma human xenografts, PDXs, syngeneic transplant mouse models	SA-β-gal, NF-κB, SASP (IL-6, IL-8, CCL5, CXCL1/2)	Mac ↑ CD8+T ↑	↓	111, 113
	MLN8237 (AURKAi)	Melanoma patient samples	SASP (CCL5)	CD8+T ↑	N/A	113
	AZD1152 (AURKBi)	Melanoma and CRC syngeneic transplant mouse models	SA-β-gal, p21	CD8+T ↑	-	114
Cell Cycle inhibitors	Abemaciclib (CDK4/6i)	ER+ breast cancer GEMM and PDXs	SA-β-gal, MHC-I	CD8+T ↑ Treg ↓	↓	124
	Abemaciclib (CDK4/6i)	Melanoma syngeneic transplant mouse models	SA-β-gal, SASP (CCL20, CX3CL1)	↓ T cell suppression signature	-	126
	Palbociclib (CDK4/6i) + Trametinib (MEKi)	LUAD GEMM	SA-β-gal, NF-κB, p15, SASP (TNFα, ICAM-1, IL-15, NKG2D ligands)	NK ↑	↓	29
	Palbociclib (CDK4/6i) + Trametinib (MEKi)	PDAC GEMM	SA-β-gal, SASP (VEGFs, PDGFs, MMPs, IL-6, CXCL1, CCL5), MHC-I, PD-L1	CD8+T ↑	-	30
	Palbociclib (CDK4/6i) (ex vivo in fibroblasts)	Melanoma syngeneic transplant mouse models	SA-β-gal, NF-κB, p16, SASP (IL-6, MMP3, CCL6, CCL8, CCL11)	MDSC ↑	↑	127
	XL413 (CDC7i)	HCC GEMM and human xenografts	SA-β-gal, p16	Mac ↑ CD8+T ↑ CD4+T ↑	↓	128
Senescenceinducing + Immunotherapy Combinations	Cisplatin + irinotecan (chemotherapy) + α-PD-1 ICB	Ovarian cancer syngeneic transplant mouse models	SA-β-gal, STING, p16, yH2AX, SASP (IL-6, VEGFA, GM-CSF)	CD8+ T infiltration + activation ↑, DC ↑	↓↓	78
	Mitoxantrone (chemotherapy) + α-PD-1 ICB	PCa human xenografts	SA-β-gal, p16, SASP (IL-6, IL-8, MMPS, AREG), PD-L1	CD8+T infiltration + activation ↑	↓	72
	MLN8237 (AURKAi) + α-CD137 (T cell agonist)	Melanoma syngeneic transplant mouse models	SA-β-gal, NF-κB, SASP (IL-6, IL-8, CCL5, CXCL1/2)	CD8+T infiltration + activation ↑	↓↓	113
	AZD1152 (AURKBi) + α-CTLA-4 ICB	Melanoma and CRC syngeneic transplant models	SA-β-gal, p21	CD8+T ↑	↓	114
	Abemaciclib (CDK4/6i) + α-PD-1 ICB	ER+ breast cancer GEMM	SA-β-gal, MHC-I	CD8+T ↑ Treg ↓	↓↓	124
	Abemaciclib (CDK4/6i) + α-PD-1/CTLA-4 ICB	Melanoma syngeneic transplant mouse models	SA-β-gal, SASP (CCL20, CX3CL1)	↓ T cell suppression signature	↓	126
	Palbociclib (CDK4/6i) + Trametinib (MEKi) + α-PD-1 ICB	PDAC GEMM	SA-β-gal, NF-κB, SASP (VEGFs, MMPs, PDGFs, IL-6, CXCL1, CCL5), MHC-I, PD-L1	CD8+T numbers + activation ↑	↓	30

AURKA, Aurora Kinase A; AURKB, Aurora Kinase B; CRC, colorectal cancer; DC, dendritic cell; ER, estrogen receptor; GEMM, genetically engineered mouse model; HCC, hepatocellular carcinoma; i, inhibitor; ICB, immune checkpoint blockade; IR, ionizing radiation; L1, LINE-1; LUAD, lung adenocarcinoma; Mac, macrophage; MDSC, myeloid-derived suppressor cell; MM, Multiple Myeloma; NK, Natural Killer cell; NKT, Natural Killer T cell; NSCLC, non-small cell lung cancer; Pca, prostate cancer; PDAC, pancreatic ductal adenocarcinoma; PDX, patient-derived xenograft; SA-β-gal, senescence-associated beta-galactosidase; SASP, senescence-associated secretory phenotype. For tumor response: (↑) = increased tumor growth compared to control; (-) = no change in tumor growth compared to control; (↓) = decreased tumor growth compared to control; (↓↓) = further decrease in tumor growth compared to single agents alone. Ex vivo indicates that cells were treated with drugs in culture prior to their implantation into mouse models with existing tumors.

3.1. Chemotherapy

Chemotherapies can trigger cellular apoptosis or senescence depending on dose and mechanism of action of the chemotherapeutic agent [67, 68]. Low dose chemotherapy induces senescence over time through induction of DNA damage and a prolonged DDR mediated by ATM/ATR that converges to activate p53/21 and RB pathways and block cell division (Fig. 1). Classes of chemotherapeutic agents that promote senescence include: (a) DNA synthesis inhibitors (e.g. gemcitabine, 5-fluorouracil (5-FU), methotrexate) that lead to replication fork stalling and DNA damage; (b) topoisomerase inhibitors (e.g doxorubicin, etoposide, mitoxantrone) that prevent DNA re-ligation after supercoil unwinding, leading to single and double-strand DNA breaks (DSBs); (c) platinum-based and alkylating agents (e.g. cisplatin, cyclophosphamide, melphalan) that induce extensive genotoxic damage through DNA-crosslinking; and (d) microtubule inhibitors (e.g. docetaxel, paclitaxel) that interfere with microtubule spindles during the metaphase-to-anaphase transition to activate the mitotic checkpoint [17]. Markers of senescence, including SA-β-gal, p16, p21, and p53, as well as SASP factors IL-6, IL-8 (CXCL8), amphiregulin (AREG), VEGF, and CCL2 have been identified locally and in circulation in breast, lung, and prostate cancer patient samples following treatment with different classes of chemotherapeutic agents [20, 69–72]. In mice, senescence has been shown to be critical for effective chemotherapy responses in B cell lymphomas that have disabled apoptotic pathways through BCL-2 overexpression [73], suggesting that chemotherapy-induced senescence may contribute to tumor suppression in certain contexts.

Some of these tumor suppressive aspects of chemotherapy may depend on senescence-mediated anti-tumor immunity. Cyclophosphamide treatment drives an NF-κB-mediated SASP that activates NK cell immune surveillance in B cell lymphoma models [74]. In multiple myeloma, low doses of doxorubicin and melphalan induce expression of ligands on senescent tumor cells for the activating NK cell receptors NKG2D (RAE-1, MICA, MULT-1) and DNAM-1 (PVR) that lead to NK cell recognition and cytotoxicity toward tumor cells [75–77]. A recent study showed that intraperitoneal injection of chemotherapy-induced senescent cells resulted in their infiltration into established ovarian tumors in mice and an increased influx of DCs and CD8⁺ T cells that sensitized resistant tumors to immunotherapy targeting the PD-1 checkpoint on T cells [78]. This demonstrates the proof-of-concept that senescent tumors cells following chemotherapy treatment could serve as an immune adjuvant to promote anti-tumor immunity and immunotherapy efficacy.

Conversely, chemotherapy-induced senescence can also lead to immune suppression in tumors and inflammation and fibrosis in normal tissues. Chemotherapeutic induction of matrix metalloproteinases (MMPs) as part of the SASP leads to cleavage of NKG2D ligands on the surface of senescent tumor cells that allows them to evade NK cell surveillance and persist following cancer therapy [57]. Moreover, in a prostate cancer model, docetaxel treatment drove an influx of MDSCs through induction of anti-inflammatory SASP factors that led to NK and T cell suppression and chemoresistance [61]. Due to their systemic administration, chemotherapies can promote senescence in stromal cells and normal tissues that drives local and systemic inflammation leading to chemotoxicity, accelerated aging, and tumor relapse and metastasis [32, 79–82]. Bleomycin, which is used to treat Hodgkin’s lymphoma and testicular germ cell tumors, induces senescence in fibroblasts in the lungs of normal mice that leads to development of idiopathic pulmonary fibrosis (IPF) through secretion of fibrogenic SASP factors such as TGFβ, PAI1, IL-6, and MMPs, as well as a class of lipids known as leukotrienes [83, 84]. In addition, chemotherapy treatment in prostate cancer patients and xenograft models led to secretion of the SASP factor AREG from senescent stromal cells that promoted expression of the immune inhibitory checkpoint PD-L1 on tumor cells that contributed to T cell suppression and chemoresistance [72]. Thus, induction of senescence in bystander cells and tissues can limit the anti-tumor activity of systemic cytotoxic therapies. In the future it will be important to understand how the mechanisms of action of different classes of chemotherapy may dictate the degree and type of senescence induction and SASP-mediated immune and tumor responses in preclinical and clinical settings.

3.2. Radiation

Unlike the systemic effects of senescence-inducing chemotherapy, radiotherapy is administered locally to the tumor site, limiting damage to noncancerous tissues. Ionizing radiation (IR) therapy can trigger senescence through induction of DSBs and an ATM/ATR-dependent DDR, leading to activation of p53 and its target p21 (Fig. 1). IR induces senescence markers in vitro and in vivo, including SA-β-gal, p53, p21, and various SASP factors, and preclinical evidence suggests that senescence plays a major role in radiosensitivity in cancer [85–89].

Radiotherapy is known to promote anti-tumor immunity in cancer patients [90, 91]. Indeed, CD8⁺ T cells are required for IR-mediated tumor ablation in mice [92, 93]. Besides triggering the presentation of immunogenic tumor antigens [94], IR-induced DNA damage and accumulation of cytoplasmic double-stranded DNA fragments also activates nucleic acid sensors, such as cGAS-STING and the downstream NF-κB pathway, to stimulate production of SASP-associated cytokines and chemokines and expression of NK cell activating ligands on irradiated cells that promote innate and adaptive immune surveillance [95–102]. For instance, release of extracellular vesicles (EVs) containing cGAS-STING-activating DNA:RNA hybrids and retrotransposable elements (RTEs) and SASP factors IL-6, IL-1α, and IFN-β following tumor irradiation can mediate paracrine senescence, macrophage activation, and immune-mediated tumor regression in distal non-irradiated tumors through a rare phenomenon known as the abscopal effect [98]. IR has also been shown to induce RB-dependent senescence and IL-6 secretion that promotes immune surveillance by Natural Killer T (NKT) cells in osteosarcomas [88]. Combining IR with PARP inhibition to reinforce DNA repair suppression and exacerbate DNA damage in melanoma and pancreatic tumor models led to production of inflammatory SASP factors such as CCL5, IFN-β, and CXCL9/10/11 that promoted DC proliferation and enhanced both CD8⁺ T cell and NK cell anti-tumor immunity [103]. Interestingly, in an immune-deficient glioblastoma model, radiation-induced senescence accelerated tumor growth [87], underscoring the importance of immune responses in determining the therapeutic outcome of senescence-inducing radiotherapy.

As with chemotherapy, some of the immune stimulating aspects of radiation therapy are counterbalanced by off-target effects in normal cells. Radiation-induced immunosenescence can inhibit effective immune responses [104, 105], and clearance of senescent immune cells restores T cell proliferation and macrophage phagocytic activity [106]. Furthermore, persistent senescence following radiation in normal tissues contributes to chronic inflammation, fibrosis, and tissue dysfunction reminiscent of accelerated aging [107, 108]. Therefore, to improve therapy outcomes it will be important to develop therapies which target senescence more specifically to tumor cells.

3.3. Aurora kinase inhibitors

Aurora kinases are mitotic regulatory proteins responsible for proper segregation of chromosomes during mitosis and are commonly overexpressed in human cancers [109, 110]. Inhibition of Aurora kinases is known to induce senescence through blocking mitosis (Fig. 1), leading to genomic instability, polyploidy, and an ATM/CHK2-mediated DDR [111, 112]. In melanoma models, Aurora kinase A (AURKA) inhibition with MLN8054 or MLN8237 induced DNA damage, SA-β-gal expression, and an NF-κB-driven SASP that promoted macrophage recruitment, immune surveillance, and long-term immune control of tumors [111]. Follow-up studies revealed that AURKA inhibition could also induce SASP production of chemokine CCL5 to recruit T cells into melanoma lesions, and that combining AURKA inhibitors with a T cell agonist antibody targeting CD137 further enhanced anti-tumor responses [113]. This provided one of the first examples that senescence-inducing and immunotherapies could synergize through the ability of the SASP to recruit cytotoxic immune cells into the TME. Supporting these findings, melanoma patients that gained clinical benefit from treatment with AURKA inhibitors displayed increased CCL5 expression associated with greater infiltration of activated cytotoxic T cells [113]. Similarly, treatment with an Aurora kinase B (AURKB) inhibitor in melanoma models was shown to promote T cell-driven anti-tumor immunity and synergize with an antibody targeting the immune checkpoint CTLA-4 on T cells in a manner that was partially dependent on p21-mediated senescence in tumor cells [114]. As Aurora kinase inhibitors have also been shown to act directly on T cells and conversely inhibit their activation and proliferation [115–117], moving forward it will be important to consider the correct dose and schedule for combining senescence-inducing and immunotherapies to maximize immune activity and therapeutic benefit.

3.4. CDK inhibitors

Cyclin-dependent kinases (CDKs) regulate cell cycle checkpoints that control growth and arrest in both normal and malignant cells [118, 119]. As cyclins and CDKs are aberrantly hyperactivated, and their inhibitors, including p16, p21, and p27, are commonly suppressed in cancer cells, the use of pharmacological CDK inhibitors is a promising approach to induce cell cycle arrest in tumors that have lost cell cycle control [120]. The most clinically developed and well-studied CDK inhibitors are dual CDK4/6 inhibitors, which include the FDA-approved compounds palbociclib, ribociclib, and abemaciclib [121]. CDK4/6 inhibitors act in a way as a p16 mimetic to inhibit the CDK4 and CDK6 kinases that regulate the G1/S phase checkpoint by blocking their ability to phosphorylate RB, which can then sequester E2Fs to arrest cells in G1 (Fig. 1). As such, a functional RB protein is required for CDK4/6 inhibitors to have anti-neoplastic activity [122]. Strong RB activation following CDK4/6 inhibitor treatment alone can in certain cell types and contexts lead to senescence, for example in CDK4-amplified liposarcomas, estrogen receptor (ER)-positive breast cancers, and melanomas, as well as in normal fibroblasts [123]. While studies have begun to elucidate certain mechanisms necessary for CDK4/6 inhibitors to mediate senescence, including MDM2 downregulation, ATRX redistribution, HRAS suppression, and mTOR activation, the determinants governing whether CDK4/6 inhibitors induce transient quiescent or durable senescent states in different cancer settings are not completely understood [122].

It is now appreciated that CDK4/6 inhibitors can also directly (through their effects on immune cells) or indirectly (through induction of senescence) promote anti-tumor immunity in various cancer settings. Abemaciclib treatment in ER⁺ breast cancer models induced SA-β-gal activity, increased MHC-I levels, and triggered expression of endogenous retroviral elements and a type III interferon response that led to increased cytotoxic CD8⁺ T cell infiltration and synergy with anti-PD-1 immune checkpoint blockade (ICB) [124]. CDK4/6 inhibition also prevented the expansion of immune suppressive Tregs, though this response is likely a direct effect of CDK4/6 inhibitors on Treg function [124, 125]. Similarly, abemaciclib could reverse tumor intrinsic immune suppression associated with ICB resistance in human melanoma patients, induce SASP components such as CCL20 and CX3CL1, and improve combined anti-PD-1/CTLA-4 ICB efficacy in melanoma models [126]. It is important to note that CDK4/6 inhibitors can also have off-target effects on other stromal cells in the TME. Normal fibroblasts treated with palbociclib underwent senescence and produced an NF-κB-driven SASP that promoted MDSC infiltration and enhanced melanoma growth in vivo [127]. Targeting alternative cell cycle regulators may offer even more specificity towards activating senescence in cancer cells. For instance, inhibition of DNA replication kinase CDC7 with XL413 was recently shown to induce senescence specifically in tumors with p53 mutations and lead to SASP-mediated macrophage and CD4⁺ and CD8⁺ T cell immune surveillance in liver tumors [128].

Combining CDK4/6 inhibitors with other molecularly targeted therapies directed at cancer-specific oncogenic driver pathways also provides a means to increase selectivity and reinforce cellular senescence in tumor cells. We and others have shown that palbociclib in combination with the MEK inhibitor trametinib (T/P) can induce senescence and a robust SASP in KRAS-driven lung and pancreatic cancer models [29, 30, 129]. In Kras^G12D mutant; p53-null lung adenocarcinoma models, SASP-mediated TNFα secretion and ICAM-1 and NKG2D ligand expression on senescent tumor cells following T/P treatment led to an influx of NK cells, NK cell-mediated tumor clearance, and long-lasting tumor control [29]. In Kras^G12D mutant; p53-null pancreatic cancer models, a different SASP response and outcome on the surrounding TME was observed. T/P-induced senescence produced a SASP rich in both angiogenic factors, including VEGFs, FGFs, PDGFs, and MMPs that led to increased vascularization, as well as inflammatory factors such as IL-6, CXCL1, and CCL5 that simulated endothelial cell activation, culminating in increased vascular trafficking of cytotoxic CD8⁺ T cells into pancreatic tumors [30]. T cell influx, combined with increased MHC-I and PD-L1 expression on senescent tumors, made these immunologically “cold” pancreatic tumors “hot” and sensitized them to anti-PD-1 ICB [30].

Collectively, evidence from these different classes of cancer therapies suggests that senescence could be a means to enhance immunotherapy efficacy through SASP-mediated remodeling of the TME. Indeed, genomic correlates and transcriptional signatures related to senescence and the SASP are also enriched in cancer patients that respond well to PD-1/PD-L1 ICB [126, 130, 131], supporting the potential clinical relevance of senescence as an immunotherapy booster. Though more evidence connecting TIS and the mechanisms by which it achieves immune modulation is needed in vivo and especially in clinical samples overall, these results highlight the dynamic changes that senescence-inducing targeted therapies can have on the surrounding TME. A deeper knowledge of how the SASP is regulated in different therapy scenarios and which SASP factors mediate these divergent immune responses will allow us to strategize how to fully harness the SASP for cancer immunotherapy going forward.

4. Dynamic SASP regulation in cancer

The SASP is regulated on multiple levels, including epigenetically and transcriptionally, at the protein translation and secretion stage, and at the cellular and tissue level. Here we examine dynamic SASP regulation following senescence induction to gain insight into how the SASP output can be optimized for therapy in different cancer settings.

4.1. Epigenetic regulation

Senescent cells undergo a vast array of epigenetic changes, including altered chromatin occupancy, composition, and modifications [132], which in addition to repressing cell cycle genes can prime SASP gene expression (Fig. 1). RB forms a complex with histone deacetylases (HDACs) and the methyltransferase SUV39H1 to mediate H3K9 trimethylation (H3K9me3), often referred to as SAHF, at E2F target and cell cycle regulatory genes to repress their expression and maintain senescent cells in a growth arrested state [133–135]. Despite increased facultative chromatin at SAHF loci, in general senescent cells have reduced heterochromatin globally, correlating with loss of gene silencing and increased accessibility and expression of SASP genes and other senescence-related transcriptional targets [136]. Recent work demonstrated that the chromatin modifier HMGB2 is critical for limiting repressive heterochromatin on SASP gene loci during SAHF formation at cell cycle genes [137]. HDACs such as SIRT1 that deacetylate H3K27 and components of polycomb repressor complex 2 (PRC2) such as methyltransferase EZH2 that trimethylates H3K27 (H3K27me3) to condense chromatin and block enhancer occupancy at the p16 gene locus as well as SASP genes are silenced during senescence [138, 139]. Chromatin readers such as the bromo and extra terminal (BET) domain protein 4 (BRD4) can then be recruited and are free to bind open enhancer regions [140]. BRD4 activity is required for SASP gene expression during oncogene-induced senescence and is crucial for senescence-mediated NK cell immune surveillance and tumor suppression in the liver [140].

In addition, degradation of the methyltransferases G9a and GLP leads to further loss of H3K9 methylation and activates promoters at SASP genes including IL6 and IL8 [141]. Some chromatin modifiers, such as the disruptor of telomeric silencing 1-like (DOT1L), can lead to increased H3K79me2/3 occupancy at specific SASP genes, such as IL1A [142]. Finally, reorganization of histone variants including removal of macroH2A1 [143] and accumulation of H2A.J [144] at SASP loci reinforces SASP gene expression during senescence. This demonstrates that remodeling of chromatin landscapes, particularly in enhancer regions, is critical for SASP gene program regulation and may be necessary for immune clearance of senescent cells. A more comprehensive view of the epigenetic changes that influence expression of particular SASP gene loci in oncogene-induced senescence and other contexts, including TIS, will provide a deeper understanding of dynamic SASP transcriptional regulation. Further application of 3D and 4D DNA topological analyses (i.e. Hi-C, 3C) may help to provide additional granularity into how DNA structure regulates SASP expression [145–147].

4.2. Transcriptional regulation

Cell cycle and DDR pathways unite to activate the expression of SASP genes through downstream transcriptional regulators (Fig. 1). Apart from their canonical roles in mediating senescence-associated cell cycle arrest, p21 and p16 have very recently been shown to activate and maintain the SASP, possibly through RB-mediated transcriptional mechanisms [40, 148]. In addition to activating RB, it has been well-established that many senescence-inducing stimuli produce DNA damage that directly or indirectly, through mobilization of DDR proteins such as ATM/ATR and CHK1/CHK2, converges to activate a set of common transcription factors (TFs) that regulate SASP gene expression, including NF-κB, C/EBP-β, and GATA4 [21, 74, 149–151] (Fig. 1).

NF-κB and C/EBPβ can directly bind to DNA and synergistically activate the expression of mainly pro-inflammatory SASP genes such as IL1A, IL6, and IL8 [21, 74, 149]. Following DNA damage, ATM and ATR activate GATA4 upstream of NF-κB to regulate SASP expression [151]. The p38MAPK pathway is activated by oxidative stress and can also reinforce NF-κB transcriptional activity and stabilize SASP mRNA transcripts independent of the DDR [152–154]. Additionally, senescent cells harbor high levels of cytoplasmic DNA fragments due in part to loss of Lamin B1 and nuclear envelope dissolution [155, 156]. Cytosolic DNA is sensed by cGAS, which activates the downstream STING pathway resulting in the activation of interferon regulatory factor 3 (IRF3) and NF-κB and induction of interferon genes (e.g. IFN-β) and other pro-inflammatory SASP factors in senescent cells [100–102, 157]. Interestingly, p38MAPK has been shown to inhibit STING-mediated induction of interferon signaling while leaving NF-κB-driven SASP intact [102]. These inflammatory pathways downstream of the DDR also induce the expression of ligands for activating receptors NKG2D and DNAM-1 on NK cells that are important for their recognition and targeting of senescent cells [158, 159]. Underscoring their importance as SASP transcriptional regulators, NF-κB or cGAS-STING suppression alone is sufficient to ablate the expression of many pro-inflammatory SASP factors and subsequent immune surveillance of senescent cells. NF-κB signaling blockade inhibits key SASP factors (TNFα, CCL2) and cell adhesion molecules (ICAM-1) necessary for NK cell infiltration and targeting of senescent tumors [29, 74], and leads to recruitment of macrophages and suppressive MDSCs into tumors [160]. cGAS or STING genetic knockout in mice leads to reduced expression of SASP factors IL-1α, IL-1β, IL-6, IL-8, and CXCL10 in pre-malignant hepatocytes and inhibits their clearance by the innate and adaptive immune system [100, 102].

Other transcriptional regulators have been shown to generate a different SASP milieu beyond canonical IL-1α, IL-6, IL-8, and other pro-inflammatory SASP factors (Fig. 1). The JAK2/STAT3 pathway, which can be induced by autocrine IL-6 signaling, regulates the expression of immunosuppressive SASP factors (e.g., GM-CSF, M-CSF (CSF-1), IL-10, CXCL1, CXCL2) following Pten loss-induced senescence in prostate cancer that hinders immune surveillance [61]. JAK2 inhibitor treatment reprogrammed the immunosuppressive SASP, reversed MDSC-driven immune suppression, and promoted T cell-mediated anti-tumor immunity following docetaxel-induced senescence [61]. In oncogene-induced senescence, activation of NOTCH1 suppresses C/EBPβ activity to block the pro-inflammatory SASP and produces an alternative TGFβ-rich SASP, highlighting a pivotal role of NOTCH1 in governing functional composition of SASP [161]. Inhibition of activated NOTCH signaling promoted increased infiltration of T cells and clearance of pre-malignant Nras^G12V senescent hepatocytes [161], suggesting that targeting NOTCH in senescent tumors cells could be a potential strategy to enhance immune surveillance in tumors. Interestingly, recent work suggests that p21 can mediate a unique SASP (termed PASP) through facilitating RB transcriptional complexes with STAT and SMAD TFs, including STAT1/3 and SMAD2/3, which promote the secretion of pro-inflammatory SASP factors such as CXCL14 that potentiate immune surveillance in the liver [40]. P53 itself may also play dichotomous roles in SASP regulation. Whereas p53 restoration in established liver tumors can promote a pro-inflammatory SASP [39], other reports have demonstrated that p53 can actually suppress p38MAPK and NF-KB activity, and that p53 loss further enhances pro-inflammatory SASP production [20, 153, 162–164]. Thus, different TFs and TF complexes can regulate different arms of the SASP and contribute to its dynamic regulation of the immune system following TIS (Fig. 1).

4.3. Post-transcriptional regulation

mTOR regulates the translation of SASP proteins, including IL-1α, which in turn reinforces NF-κB activation to promote the expression of other pro-inflammatory SASP genes [165]. Indeed, IL-1α knockout was sufficient to suppress NF-κB-driven SASP production and macrophage recruitment into preinvasive pancreatic lesions following Kras^G12D induction [166]. Other SASP factors, including IL-6, IL-8, VEGF, CXCL1/2, CCL2, CCL20, TGFβ, and lipid species such as prostaglandins have been shown to reinforce autocrine senescence and SASP as well as promote paracrine senescence in neighboring cells through activation of downstream TFs such as NF-κB [21, 59, 149, 167–169]. mTOR also regulates the translation of MK2, which phosphorylates the RNA binding protein ZFP36L1 and inhibits its ability to degrade SASP transcripts [170]. As such, mTOR inhibition using rapamycin abolishes SASP induction [165, 170, 171]. In vivo, rapamycin treatment blocked recruitment of T cells, B cells, NK cells, and macrophages into the liver and subsequent immune clearance of senescent Nras^G12V mutant hepatocytes [170]. Enzymatic cleavage of SASP-associated proteins can also impact their maturation and regulation. IL-1 signaling in senescent cells is mediated in part by inflammasome components NLRP3 and TLR2, whose activation is required for caspase-1 cleavage of pro-IL-1α, -IL1β, and -IL-18 to their mature forms [168, 172]. Moreover, genetic loss of SASP factor TIMP1 in prostate cancer has recently been shown to lead to activation of MMPs, which in turn can regulate soluble factors present in the SASP through their proteinase activity [31, 173].

In addition, transcriptional derepression of RTEs such as LINE-1 (L1) that occurs during late senescence (following earlier SASP induction) leads to induction of type I interferon signaling that maintains the SASP [174]. PTBP1 has been shown to control SASP secretion by regulating the splicing of genes encoding proteins involved in intracellular trafficking. Paradoxically, SASP suppression following PTBP1 blockade limited liver tumorigenesis despite impairing immune surveillance [175]. Non-coding RNAs such as microRNAs (miRNAs) can also regulate SASP transcription, mRNA stability, and translation [176]. For example, miR-146a/b is induced by NF-κB but subsequently forms a negative feedback loop to suppress NF-κB activity and reduce IL-1α, IL-6, and IL-8 expression [177]. It is now come to be appreciated that SASP proteins, as well as DNA:RNA hybrids, miRNAs, and L1, can also be released from senescent cells in the form of EVs such as exosomes [98, 178–180] and signal in a paracrine manner. Still, whether soluble vs. vesicle-bound DNA, RNA, and protein species associated with the SASP may have differential effects on neighboring cells and immune responses is still unclear.

4.4. Tissue and cell type dependent SASP regulation

The SASP produced following senescence induction is highly variable and dependent on the senescence-inducing stimuli and cell and tissue type it acts on. For example, the same cells treated with different genotoxic therapies can have dramatically divergent SASPs with few overlapping genes [181]. The SASP composition following treatment with the same senescence-inducing agent can also be heterogenous across cancer cells with diverse genetic backgrounds, different cell types within the same tumors, and even among cells of the same type derived from different donors [12, 20, 182]. This may be partially dependent on the unique epigenetic footprints of cells from distinct origins [182]. As such, epithelial and stromal cells treated with the same senescence-inducing drugs have quantitatively and qualitatively different SASP profiles [20, 179, 181–184]. In addition, within a population of senescent cells, there can also be marked heterogeneity in SASP gene expression at the single cell level suggesting that different cells within a population may contribute to different SASP profiles [185]. A more comprehensive and high-throughput analysis of the SASP in distinct cell populations at the single cell level in different tumor and therapy contexts will be necessary to fully grasp the cellular mechanisms leading to differential SASP profiles and subsequent immune and tumor responses to senescence.

The resident tissue microenvironment that tumors reside in may also impact SASP expression and its outcomes on immune responses. RAS overexpression in pre-malignant hepatocytes or restoration of p53 in liver tumor cells or hepatic stellate cells leads to SASP-dependent innate and adaptive immune activation that effectively controls HCC inception and progression [28, 39, 54], suggesting the SASP promotes anti-tumor immunity in the liver. Similarly, in Kras^G12D mutant; p53-null lung adenocarcinoma, TIS following T/P treatment activated an NF-κB-driven pro-inflammatory SASP that triggered NK cell immunity and immune-mediated tumor regressions [29]. In contrast, T/P treatment in genetically similar Kras^G12D mutant; p53-null pancreatic cancer models triggered a SASP enriched in mainly pro-angiogenic factors that failed to activate NK cell-mediated tumor responses but rather promoted vascular remodeling that indirectly facilitated T cell infiltration [30]. This suggests that the same senescence-inducing stimuli can lead to different SASP outputs and immune and stromal outcomes based on the tissue. Our own preliminary experiments suggest that transplanting the same tumor cell line into different tissues in mice is sufficient to alter the SASP profile and subsequent immune and stromal responses in the TME following treatment with the same senescence-inducing therapies (unpublished). Therefore, it will be critical to characterize the types of SASP activated and their impact on the TME in each therapy and tumor context to (a) understand the specific SASP factors and regulators necessary for anti-tumor immunity and immunotherapy efficacy and (b) guide the development of strategies to harness or target the SASP to promote tumor responses following cancer therapy.

5. Modulating senescence with senotherapeutics for cancer immunotherapy

Combining senescence-inducing therapies with other agents to remove deleterious senescent cells or modulate SASP regulators or specific SASP factors in different cancer contexts may be necessary to optimize TIS for cancer therapy. The use of senotherapeutics to eliminate senescent cells (i.e. senolytics) or modulate the SASP (i.e. senomorphs) are promising strategies to block senescence-mediated immune suppression and/or potentiate SASP-mediated anti-tumor immunity (Fig. 1). Alternatively, overlaying senescence-inducing therapies with novel immunotherapy strategies directed toward the vulnerabilities that arise in senescent tumor contexts may allow us to further leverage the immune modulatory functions of senescence for tumor eradication (Fig. 1).

5.1. Senolytics

Senescence and the resulting SASP can in certain contexts suppress anti-tumor immune responses and promote cancer progression and relapse following cancer therapy [22, 31, 186]. Inducible ablation of p16-positive senescent cells in transgenic mice has been shown to mitigate age-related morbidities and inflammation, increase lifespan and healthspan, and reduce tumor initiation and relapse following therapy [32, 187–191]. This exciting proof-of-concept has led to the development and characterization of numerous senolytic agents to selectively eliminate senescent cells by targeting their vulnerabilities, most prominently anti-apoptotic pathways that allow them to survive and persist [192]. The most well-characterized senolytic agents include the multi-kinase inhibitor dasatinib, the BCL-2 family inhibitor ABT-263 (navitoclax), UBX0101 (MDM2 inhibitor) and FOXO4-DRI (FOXO4 blocker) that activate p53-mediated apoptosis, and natural products fisetin and quercetin that can inhibit pro-survival PI3K signaling [193] (Fig. 1). These agents have been shown to clear senescent cells in various tissues in old mice, reduce systemic SASP and inflammation, and alleviate age-related pathologies to extend health span and lifespan, thus blunting many of the deleterious effects of senescence during aging [190, 191, 194–198]. While ongoing clinical trials evaluating dasatinib and quercetin (D + Q) treatment have demonstrated select removal of senescent cells in humans and promising mitigation of physical dysfunction in IPF patients [199, 200], other trials testing UBX0101 in human osteoarthritis patients failed to show efficacy [201], highlighting that there is still much work to be done to effectively translate these senolytic approaches to chronic human diseases.

Senolytic strategies could also be used to potentiate or overcome resistance to cancer therapy and mitigate associated chemotoxicities. A promising tactic to leverage senolytics for cancer therapy is to combine a “one-two punch” of a senescence-inducing drug (1) followed by a senolytic agent (2) to induce synthetic lethality, as first described by Rene Bernards and colleagues [202]. Multiple studies have now shown that induction of tumor cell senescence with chemotherapy, radiation, and targeted Aurora kinase, PARP, or CDK4/6 inhibitors leads to induced susceptibility to senolytic BCL-2 family inhibitors and senescent cell apoptosis [203–209]. Another recent example is that CDC7 inhibitors induce senescence in HCC cells with p53 mutations that leads to synthetic lethality with sertraline, an antidepressant, or the mTOR inhibitor AZD8055 [128]. In addition, MDM2 inhibitor treatment in combination with senescence-inducing AURKA inhibition was shown to not only result in the death of senescent melanoma cells, but also lead to enhanced SASP expression of factors including CCL1, CCL5, and CXCL9 that further promoted the recruitment of both innate (macrophages, NK cells) and adaptive (T cells) immune cells and anti-tumor responses [210]. This suggests that senolytics could be used in certain settings to further enhance the immune stimulatory features of acute senescence while simultaneously removing detrimental senescent cells and chronic SASP expression. Navitoclax, FOXO4-DRI, and D + Q treatment has also been shown to eliminate senescent bystander cells following chemotherapy and radiation-induced senescence, leading to reduced treatment-related toxicities and morbidities and cancer recurrence [32, 84, 196, 211–213]. Thus, senolytics could be used as adjuvant therapy to maximize the effects of standard radiation and chemotherapy drugs that induce senescence. Indeed, numerous clinical trials combining navitoclax with chemotherapy or targeted therapy agents known to induce senescence have been carried out [214]; however, this approach has so far shown limited efficacy in various solid tumor settings.

Several roadblocks must be overcome to effectively apply senolytic agents to cancer therapy. First, the mechanisms by which many senolytic agents eliminate senescent cells need to be firmly defined. For instance, it is unclear how D + Q, or other natural products such as fisetin or sertraline selectively target senescent cells, as each of these drugs has multiple targets. Second, the activity of senolytic agents is not universal and is highly cell and tumor type specific. Response to navitoclax following TIS varies amongst cancer cell lines, which may be the result of genetic alterations in BCL-2 family or other pro-apoptotic genes [182]. Similarly, while D + Q treatment is ineffective in combination with senescence-inducing chemotherapy in liver cancer, this senolytic cocktail enhances the anti-tumor efficacy of senescence-inducing radiation in melanoma models [215, 216]. Biomarkers predictive of senolytic responses will thus need be identified. Finally, senolytic agents such as BCL-2 family inhibitors and activators of p53-mediated apoptosis (i.e. MDM2 inhibitors) are not specific for tumor or senescent cells and target cell death pathways common to all cells, leading to toxicities in patients [217, 218]. The recent development of galactose-modified nanoparticles and prodrugs that allow selective delivery of senolytic payloads to senescent cells through their high β-galactosidase activity may be an innovative means to overcome off-target toxicities [206, 219–221]. Still, additional on-target toxicities may arise from elimination of physiologically-relevant senescent cells, such as macrophages, megakaryocytes, and osteoblast progenitors [222–224], that could disrupt tissue or organismal homeostasis. It will be especially important to consider the impact of excessive senescent cell ablation in elderly individuals, where senescent cells may represent a large percentage of cells in a tissue and their removal could potentially threaten tissue integrity and function [225, 226]. Given the off-target effects of current senolytic drugs and potentially determinantal consequences of removing senescent cells throughout the body, a safer and more elegant approach may be to remodel SASP phenotypes and interactions between senescent cells and the tumor-immune microenvironment through senomorphic strategies.

5.2. Senomorphs

The use of senomorphic agents to abolish or remodel the SASP to limit tumor-propagating inflammation or promote anti-tumor immunity following TIS also holds therapeutic potential in cancer. The majority of senomorphs target key SASP regulators that can block all or parts of the SASP while leaving senescent cells and other tumor suppressive aspects of senescence in place [227]. Compounds targeting SASP epigenetic, transcriptional and post-transcriptional regulators, including JAK2/STAT3 (Ruxolitinib, NVP-BSK805) [61, 228, 229], NF-κB (metformin, BAY 11–7082) [74, 230–232], p38MAPK (SB203580, UR-13756) [153, 156, 233], mTOR (Rapamycin, RAD001) [165, 170, 232], L1 (lamivudine) [174], STING (H-151) [234, 235], and BRD4 (JQ1, iBET762) [140, 236, 237] have been shown to decrease SASP production (Fig. 1). However, the impact of SASP suppression on immune responses and pro- and anti-tumorigenic outcomes depends on the context. JAK2/STAT3 inhibition has been shown to block immune suppressive SASP programs to potentiate anti-tumor immune responses in Pten-null prostate cancer [61]. Likewise, the p38MAPK inhibitor SB203580 can reduce expression of the immune checkpoint HLA-E on senescent cells that blocks NK and T cell immune surveillance [56]. In contrast, mTOR, cGAS-STING, or BRD4 signaling blockade has been shown to inhibit anti-tumor immune surveillance that could contribute to tumor escape [102, 140, 170]. Thus, the impact of senomorphic agents on the TME is context-dependent and requires further evaluation in preclinical and clinical settings. Of note, these approaches are exquisitely dose dependent, with some drugs (e.g. BET, BRD4 inhibitors) mediating SASP suppression at low doses but senolysis at high doses [140, 236, 238].

An alternative senomorphic approach is to directly target specific SASP molecules that lead to immune suppression and tumor promotion. Monoclonal antibodies and other agents targeting cytokine signaling have been widely used in other cancer and inflammatory disease settings and are far along in clinical development [239]. These include drugs targeting IL-6/IL-6R (Tocilizumab, Siltuximab) [240, 241], IL-1α/IL-1β/IL1R (Anakinra, Canakinumab, Rilonacept) [242–244], and TNFα (Etanercept, Infliximab) [245, 246]. Though IL-6/IL-6R targeting alone or in combination with chemotherapy has largely failed to demonstrate therapeutic benefit in clinical trials for solid tumor malignancies [247–249], the IL-6R monoclonal antibody tocilizumab administered in combination with chemotherapeutic agents carboplatin and doxorubicin did lead to immune stimulation, including M1 macrophage polarization and activation of T cell effector functions, in patients with recurrent epithelial ovarian cancer [250]. Additionally, inhibition of IL-8 expression using liposome-encapsulated siRNAs reduced the growth of ovarian cancer xenografts [251]. Clinical efforts to target IL-8 or its receptors CXCR1/2 are underway, including the use of Reparixin in combination with paclitaxel in breast cancer patients that has already shown early promising results [252]. These studies push forward the paradigm that neutralizing immune suppressive SASP factors could be a strategy to enhance immune surveillance and anti-tumor responses following chemotherapy-induced senescence. As IL-8 has also been implicated in blunting the effects of ICB [253, 254], use of IL-8 signaling antagonists could be yet another strategy to improve immunotherapy outcomes following treatment with senescence-inducing therapies.

In the future it will be of interest to test inhibitors of other regulators of SASP-mediated immune suppression, including NOTCH and TGFβ antagonists that are under clinical development [255, 256], as well as activators of the pro-inflammatory SASP, such as STING agonists that are now being widely tested in the setting of cancer immunotherapy [257], for their ability to boost anti-tumor immunity following TIS. Though implementation of senomorphs may be safer than senolytics due to senescent and other normal cells remaining intact, senomorphic approaches still suffer from a lack of specificity and universality, and more work is needed to understand how their use in different cancer contexts may impact the SASP milieu in ways that ultimately drives tumor suppression or promotion.

5.3. Senescence-directed immunotherapy strategies

Alterations in both the secretome and surfaceome of senescent cells provide promising targets for existing and new immunotherapy strategies (Fig. 1). Senescent cells express altered levels of immunomodulatory cell surface molecules that prime them for recognition by cytotoxic lymphocytes, such as NKG2D receptor ligands that have been shown to be required for NK cell killing of senescent cells [258], and MHC-I and B2M [30, 51, 259] required for antigen presentation to CD8⁺ T cells. However, senescent cells also upregulate immune checkpoint proteins, such as PD-L1 and HLA-E, that block effective NK and T cell surveillance necessary for tumor eradication. Use of FDA-approved ICB immunotherapies targeting the PD-1/PD-L1 axis has been one strategy to boost the anti-tumor activity of senescence-inducing agents [260]. For example, we and others have shown that treatment with CDK4/6 inhibitors that induce senescence alone or in combination with other targeted therapies activate the inhibitory checkpoint PD-L1, and as such synergize with anti-PD-1/PD-L1 blockade to promote CD8⁺ T cell-mediated tumor clearance in melanoma, breast, lung, pancreatic, and colon cancers [30, 124–126, 261]. In addition, senescent tumor cells that evade NK cell immunity through shedding of NKG2D ligands [57] or upregulation of HLA-E molecules that inhibit NK cell activation through the receptor NKG2A [56] can be targeted with antibodies that are currently in clinical development to block the NKG2A checkpoint [262] or the shedding of the NKG2D ligands MICA/B [263].

Senescent cells also express unique cell surface proteins that can be used to design and engineer antibody and chimeric antigen receptor (CAR) approaches to immunologically target senescent cells following cancer therapy (Fig. 1). Elucidation of the surfaceome in senescent cells has identified potential targets for these approaches, such as urokinase-type plasminogen activator receptor (uPAR) [264], NOTCH1 [161], DPP4 [265], and membrane-bound Vimentin [266]. One strategy has been to design antibodies targeting these molecules (e.g. DPP4) to promote antibody-dependent cell-mediated cytotoxicity (ADCC) through innate immune cells such as macrophages, neutrophils, and NK cells that express Fc receptors that can bind to the constant region of these antibodies and kill antibody-positive senescent cells [265]. More recently, Amor and colleagues engineered CAR-T cells to target uPAR expressed on senescent cells, which they demonstrated could effectively eliminate senescent liver cells following fibrosis and lung tumor cells following TIS [264]. These methods highlight innovative ways to more specifically and effectively engineer the immune system to target senescent cells following cancer therapy.

As the SASP can also promote the infiltration and immune suppressive characteristics of MDSCs and other myeloid cells known to inhibit NK and T cell functionality, targeting these populations and the signals that attract them is another approach to further augment anti-tumor immunity following TIS. Antibodies can be utilized to neutralize immune suppressive SASP factors such as CSF-1, IL-6, PGE2, IL-1β, and TGFβ that are known to attract and activate MDSCs [267]. Orthogonally, clinically developed CSF-1R, CCR2, and CXCR2 inhibitors, the latter of which have been shown to inhibit MDSC infiltration following chemotherapy-induced senescence in prostate cancer models [268], can be utilized to block recruitment and/or reeducate suppressive myeloid cell populations that infiltrate following senescence induction [269, 270]. In all, these examples underscore that by understanding how the SASP impacts immune surveillance in cancer, it is possible to take advantage of immunomodulatory effects of the SASP to create rational combinations of senescence-inducing and immunotherapies to further enhance anti-tumor responses.

6. Future Directions and concluding remarks

6.1. Defining better markers of cellular senescence

Recent advances in genetic engineering and in vivo model systems allowing manipulation of senescence in tumor contexts with intact TMEs has shed light on the broad role of senescence in regulating tumor-immune microenvironments in preclinical settings. Still, there is a great degree of heterogeneity and context dependency that dictates the impact of senescence on immune and tumor responses in cancer, including but not limited to: (a) the senescence inducer (and its dose), (b) duration and time after insult, (c) the cell type it acts on, (d) genetic and epigenetic alterations found within a tumor, and (e) the composition of the surrounding TME. A clear limitation to developing a consensus framework on the role of senescence in cancer biology and tumor immunology has been the lack of specific and reproducible markers of cellular senescence, especially in vivo, to differentiation senescent cells from other cell states associated with cell cycle withdrawal, including quiescence, post-mitotic terminal differentiation, and dormancy [10, 16, 271]. While SA-β-gal, p16, and p21 have been used widely as senescence markers, these factors are not universal [179, 182, 183, 185, 272, 273] and can be present in other cell types, particularly macrophages [224, 274]. Certain types of senescence do not even produce a SASP [275, 276]. Comparison of transcriptional and proteomic signatures across different cell types following senescence induction has begun to pinpoint common genes and pathways associated with senescence [179, 182, 183, 272]. It will be critical to link these senescence genes with their transcriptional regulators and chromatin changes across the genome to comprehensively map senescence transcriptional regulatory networks. Multi-parametric approaches and machine learning tools currently being implemented should allow systematic discovery of not only core senescence genes but also markers and regulators of different senescence subtypes or SASP programs [182, 277]. Ongoing single cell and spatial transcriptomic efforts [278] will supply further granularity within the context of intratumoral heterogeneity to better separate SASP-secreting senescent cells from other tumor and stromal cell populations within tissues. Importantly, such analysis will help us to differentiate “good” senescence populations that contribute to immune stimulation and tumor constraint from “bad” senescent cells that promote immune suppression and disease relapse following therapy.

6.2. Characterizing immune-stimulatory senescence-inducing therapies and SASP regulators

It will be critical to define the SASP factors and regulators necessary for immune-mediated tumor control systematically across cancer settings. This will require detailed analysis of both the SASP output and immune responses in human cancer patients and physiologically intact murine tumor models following TIS, and manipulation of specific SASP components to understand how they functionally modulate tumor-immune microenvironments (Fig. 2). As a first step, we must identify cancer therapies that can robustly induce senescence and anti-tumor immunity. Several groups have developed screening approaches using single senescence markers or gene reporters to identify pro-senescence therapies [128, 203, 279]; however, such systems do not readout the SASP and cannot predict the impact of candidate target inhibition on immune responses and remodeling of the TME. Creative genetic and pharmacological screening platforms to (a) uncover new targets and drugs that can induce multiple senescence and SASP markers specifically in genetically-defined tumor cells in vitro and (b) rapidly evaluate their impact on tumor-immune interactions in immunocompetent mouse models with genetically engineered tumors [280] or in 3D co-culture systems that incorporate TME stromal components [281, 282] will be necessary to characterize those senescence-inducing therapies that produce robust anti-tumor immunity and mediate tumor control (Fig. 2). Use of patient-derived xenografts (PDX) in mice with a humanized immune system or patient-derived explants (PDEs) will be also be important to confirm findings made in murine systems in the human setting [283, 284]. Functional experiments to block, modulate, or induce specific SASP regulators following therapy-induced senescence are then required to identify candidate SASP factors that are essential for anti-tumor immunity across cancer types and could serve as potential biomarkers of immune and tumor responses following cancer therapy in the clinic (Fig. 2). Importantly, as the majority of work on senescence-mediated TME remodeling has focused on a handful of immune cells, most prominently NK cells, macrophages, MDSCs, and T cells, utilization of more high-throughput and unbiased technologies will help us understand how other immune (B cells, γδ T cells, granulocytes, innate lymphoid cells (ILCs)) and stromal (endothelial cells, pericytes, fibroblasts, nerves, microbiota) populations are impacted by and respond to SASP cues and contribute to tumor control.

Fig. 2. Pipeline to identify immune-stimulatory senescence-inducing cancer therapies and SASP biomarkers for the clinic.

(1) In vitro genetic and pharmacological screens can be performed on genetically-defined cancer cell lines engineered with fluorescent senescence and SASP reporters to identify new senescence-inducing therapies for defined cancer contexts. These candidate senescence-inducing therapies can then be tested in immunocompetent mouse models harboring transplanted or autochthonous tumors to evaluate their anti-tumor efficacy. (2) The impact of therapy-induced senescence on tumor-immune interactions can be assessed using in vitro co-culture systems and in vivo tumor mouse models with intact TMEs. (3) SASP factors and regulators that functionally impact tumor-immune interactions can then be identified by inducing or suppressing their expression genetically or pharmacologically through the pipeline in (1). (4) The immunomodulatory effects of senescence-inducing therapies and SASP biomarkers of immune and tumor responses can be validated in tissue and liquid biopsies from cancer patients treated in the clinic. mAb, monoclonal antibodies; TME, tumor microenvironment.

This pipeline will unveil new senescence-inducing therapies for different tumor and tumor genetic contexts, as well as elucidate the mechanisms by which the SASP can remodel tumor-immune microenvironments for tumor control. Still, a caveat to studying the immune modulatory mechanisms of TIS or pursuing it as a therapeutic strategy for cancer treatment is that non-tumor cells, in particular stromal fibroblasts, can also undergo senescence following treatment and may impact immune responses and tumor outcomes in negative ways [186]. Thus, innovative approaches are still needed to design better therapies that specifically induce senescence only in tumor cells, or to engineer methods to deliver payloads of pro-senescence molecules specifically to the tumor. As senescence-inducing therapies alone are unlikely to mediate complete tumor remission, understanding the new dependencies that arise through senescence-mediating remodeling of the tumor-immune landscape will allow for development of rational combinatorial therapies with senolytic and senomorphic agents and immunotherapies to fully harness the capabilities of senescence for cancer therapy.

6.3. Translating senescence to cancer therapy in the clinic

Finally, how do we apply what we have learned from preclinical models to inform the clinical application of TIS (Fig. 2)? Given the lack of generalizable senescence markers and difficulty clearly discerning senescence phenotypes in patient samples, our knowledge on the frequency and impact of senescence responses following chemo-, radio-, and targeted therapies in the clinic is limited. Still, some studies have begun to assess senescence biomarkers is tissue biopsies and even circulating T cells and correlate expression with chemotherapy outcomes and toxicities [69, 285]. As bulk and even single cell transcriptomics are becoming more commonplace with longitudinal tracking of tumor responses to therapy [286], transcriptional profiles from patient tumors may be the most straightforward path to defining TIS [58, 287]. In addition, other studies have begun to look at blood biomarkers (i.e. liquid biopsies) as a readout of senescence in animal models and human patients [72, 179, 264, 288, 289]. A number of senoprobes are now being tested for in vivo detection of senescent cells [219, 290], including a β-galactosidase PET probe for radiographical imaging that is in human trials in liver cancer patients [291]. Though promising, these approaches rely on use of one phenotype associated with senescence and cannot indicate which cell types are senescent and secreting SASP.

Correlating SASP factor expression with (a) tumor responses and progression-free and overall survival and (b) immune responses locally (by biopsy) and peripherally in the blood has yet to be achieved clinically. Such analysis will be paramount to understanding which aspects of senescence and SASP are necessary for different immune effects, and whether senescence is associated with a positive or negative impact on therapeutic efficacy (Fig. 2). Indeed, each senescence-inducing therapy (and its dose) in a given cancer context may have a different effect on the SASP, immune surveillance, and tumor responses, and broad cross-comparison between treatments and cancer types will elucidate key SASP components that contribute to “good” senescence mediating tumor control and “bad” senescence promoting tumor relapse. While informative and hypothesis-driving, it is important to note that correlative analysis of independent markers of senescent, tumor, and immune cells does not prove cause and effect but rather an association that can further be followed up and substantiated at the bench. As reproducible biomarkers of senescence and SASP and the processes that govern their effects on tumor-immune microenvironments are unveiled, we envision that cancer therapies that induce senescence will be more readily applied in different settings to promote tumor control, create new therapeutic vulnerabilities and rational combinatorial therapy regimens, and be used to potentiate immunotherapy to fully harness the immune system to fight cancer.

ABBREVIATIONS

3C

chromosome conformation capture

5-FU

5- fluorouracil

ADCC

antibody-dependent cellular cytotoxicity

AREG

amphiregulin

ATM

ataxia telangiectasia mutated

ATR

ataxia telangiectasia and rad3-related

ATRX

alpha thalassemia/mental retardation syndrome X-linked

AURKA

Aurora kinase A

AURKB

Aurora kinase B

B2M

Beta-2 microglobulin

BCL-2

B-cell lymphoma 2

BET

Bromodomain and extra terminal

BRD4

bromo and extra terminal domain protein 4

C/EBP-β

CCAAT/enhancer binding protein beta

CAR

chimeric antigen receptor

CCL

C-C motif chemokine ligand

CCR

C-C motif chemokine receptor

CDC7

cell division cycle 7

CDK

cyclin-dependent kinase

cGAS

cyclic GMP-AMP synthase

CHK1

checkpoint kinase 1

CHK2

checkpoint kinase 2

CIP

CDK interacting protein

COX-2

cyclooxygenase-2

CSF-1

colony-stimulating factor 1

CTLA-4

cytotoxic T-lymphocyte-associated protein 4

CXCL

C-X-C motif chemokine ligand

CXCR

C-X-C motif chemokine receptor

D + Q

Dasatinib and Quercetin

DC

dendritic cell

DDR

DNA damage response

DNAM-1

DNAX accessory molecule-1

DOT1L

disruptor of telomeric silencing 1-like

DPP4

dipeptidyl peptidase 4

DSB

double-strand break

ECM

extracellular matrix

ER

estrogen receptor

EV

extracellular vesicle

EZH2

enhancer of zeste homolog 2

Fc

fragment crystallizable

FDA

U.S. Food and Drug Administration

FGF

fibroblast growth factor

FOXO4

forkhead box O4

GATA4

GATA binding protein 4

GEMM

genetically engineered mouse model

GLP

G9a-like protein

GM-CSF

granulocyte-macrophage colony-stimulating factor

HCC

hepatocellular carcinoma

HDAC

histone deacetylase

Hi-C

high-throughput chromosome conformation capture

HLA-E

major histocompatibility antigen, Class I, E

HMGB2

high mobility group box 2

ICAM-1

intercellular adhesion molecule-1

ICB

immune checkpoint blockade

IFN

interferon

IL

interleukin

ILC

innate lymphoid cells

IPF

idiopathic pulmonary fibrosis

IR

ionizing radiation

IRF

interferon regulatory factor

JAK2

janus kinase 2

KIP

kinase inhibitory protein

L1

LINE-1

MDM2

murine double minute 2

MDSC

Myeloid-derived suppressor cell

MEK

mitogen-activated protein kinase (MAPK) kinase

MHC

major histocompatibility complex

MICA

MHC class I polypeptide–related sequence A

MICB

MHC class I polypeptide–related sequence B

miRNA

microRNA

MK2

mitogen-activated protein kinase (MAPK)-activated protein kinase

MMP

matrix metalloproteinase

mTOR

mammalian target of rapamycin

MULT-1

murine UL16 binding protein-like transcript-1

NF-κB

nuclear factor kappa B

NK cells

Natural Killer cells

NKG2A

NK group 2 member A

NKG2D

natural killer group 2 member D

NKT cell

Natural Killer T cell

NLRP3

NLR family pyrin domain containing 3

p38MAPK

p38 mitogen-activated protein kinase

PAI-1

plasminogen activator inhibitor-1

PARP

Poly (ADP-ribose) polymerase

PASP

p21-activated secretory phenotype

PET

positron emission tomography

PD-1

programmed cell death protein-1

PD-L1

programmed death ligand-1

PDE

patient-derived explant

PDX

patient-derived xenograft

PGE2

prostaglandin E2

PI3K

phosphoinositide 3-kinase

PRC2

polycomb repressive complex 2

PTBP1

polypyrimidine tract binding protein 1

PTEN

phosphatase and tensin homolog

RAE-1

retinoic acid early inducible gene 1

RB

retinoblastoma

RTE

retrotransposable element

SA-β-gal

senescence-associated beta-galactosidase

SAHF

senescence-associated heterochromatin foci

SASP

senescence-associated secretory phenotype

SIRT1

Sirtuin 1

STAT

signal transducer and activator of transcription

STING

stimulator of interferon genes

SUV39h

suppressor of variegation 3–9 homolog

TCR

T cell receptor

TF

transcription factor

TGFβ

transforming growth factor beta

TIMP1

TIMP metallopeptidase inhibitor 1

TIS

therapy-induced senescence

TLR2

Toll-like receptor 2

TME

tumor microenvironment

TNFα

tumor necrosis factor alpha

T_reg

regulatory T cell

ULBP1

UL16 binding protein 1

uPAR

urokinase-type plasminogen activator receptor

VCAM-1

vascular cell adhesion molecule-1

VEGF

vascular endothelial growth factor

VLA-4

very late antigen-4

γδ T

gamma-delta T cell