Turning cold tumors into hot tumors to ignite immunotherapy

Abstract

The revolution in cancer immunotherapy, particularly through immune checkpoint inhibitors (ICIs), underscores the significant role of the tumor microenvironment (TME) in determining therapeutic outcomes. At the heart of this is the classification of tumors into “cold” and “hot”, which significantly influences the efficacy of immunotherapy. “Cold” tumors are characterized by scant immune cell infiltration and an immunosuppressive TME, which effectively evades immune detection and resists ICIs. In contrast, “hot” tumors, characterized by abundant immune cells and a proinflammatory environment, are more receptive to immunotherapeutic approaches. This review comprehensively examines the molecular and cellular foundations of the “cold” tumor phenotype, delving into the mechanisms of camouflage (impeding immune priming and infiltration), coercion (suppressing immune functions)​​, and ​​cytoprotection (resisting inflammatory death)​​ that contribute to maintaining immune silence. Furthermore, it critically evaluates emerging strategies for converting “cold” tumors to “hot”, immune-reactive entities, including the role of biomaterials in remodeling the TME to increase the effectiveness of immunotherapy. Through an in-depth exploration of these foundational mechanisms and therapeutic advancements, this review seeks to shed light on the way forward in cancer treatment, framing the transformation of “cold” tumors to “hot” tumors as a crucial approach to enhancing the reach of immunotherapy to a broader array of cancer types.

Introduction

The field of cancer therapeutics has undergone significant transformation with the introduction of immunotherapy, especially through the use of immune checkpoint inhibitors (ICIs) that target the programmed cell death-1 (PD-1)/programmed cell death Ligand 1 (PD-L1) And cytotoxic T-lymphocyte associated protein 4 (CTLA-4) axes [1, 2]. These treatments offer remarkable capacity for producing durable, and sometimes curative, outcomes in a variety of cancers[3, 4]. However, the effectiveness of these therapies varies significantly, with a variety closely linked to the characteristics of the tumor microenvironment (TME) [5–8]. The divide between immunologically “cold” and “hot” tumors is crucial for this variability, critically impacting therapy outcomes and standing as a key area of focus in oncology research [9–11].

Immunologically “cold” tumors are inherently resistant to immunotherapy, a phenomenon that is systematically classified by the immunoscore system into distinct phenotypes [12, 13](Fig.1). Immune-desert tumors lack T-cell presence both in the tumor’s core and at its invasive margins. Altered-excluded tumors show lymphocytes restricted to stromal margins, with physical barriers hindering penetration into the tumor core. Altered-immunosuppressed tumors are characterized by a uniformly low T-cell density, with functional impairment arising from soluble inhibitors and suppressive cellular infiltrates [14]. These phenotypes share fundamental biological attributes: a low tumor mutational burden, compromised antigen presentation machinery, and dense stromal barriers coupled with aberrant vasculature, which together contribute to inherent resistance to immunotherapy [15]. In contrast, “hot” tumors are distinguished by robust infiltration of cytotoxic T cells, the presence of tertiary lymphoid structures (TLSs) fostering sustained immune activation, and a proinflammatory microenvironment conducive to effective antitumor immunity [16, 17]. This reactive state underlies the markedly improved clinical outcomes observed with ICIs in “hot” tumors, in stark contrast to the considerable challenges presented by the inherent resistance mechanisms of “cold” tumors [14, 18]. Consequently, converting “cold” tumors into “hot” variants is pivotal for enhancing immunotherapy, as it revives antitumor immune responses and surpasses the obstacles limiting ICI effectiveness [10, 17]. This transformation, achieved by remodeling the TME to improve T-cell priming, recruitment, and functionality, renders otherwise resistant tumors vulnerable to immune-mediated eradication, thereby broadening the scope and efficacy of cancer immunotherapy [13].

Fig. 1.

Tumor immune phenotypes. The immunoscore system divides immunologically “cold” tumors into distinct phenotypes. Immune-desert tumors are characterized by the absence of T-cell infiltration in both the tumor core and invasive margins, resulting in minimal immune surveillance. Altered-excluded tumors display T cells localized to stromal regions but are physically obstructed from infiltrating the tumor core owing to dense stromal barriers, creating an immunologically excluded state. In altered-immunosuppressed tumors, there is a widespread reduction in T-cell density coupled with functional impairments, including diminished cytotoxic activity (evidenced by decreased interferon-γ and granzyme B production) and an exhausted phenotype. These impairments are attributed to the presence of soluble inhibitors, such as immunosuppressive cytokines (TGF-β, IL-10), and cellular infiltrates, including regulatory T cells (Tregs) and myeloid-derived suppressor cells (MDSCs). Common characteristics among these phenotypes include a low tumor mutational burden, impaired antigen presentation machinery, dense stromal barriers, and aberrant vasculature, contributing to natural resistance to immunotherapeutic approaches. In contrast, “hot” tumors demonstrate robust infiltration of cytotoxic T lymphocytes, the presence of tertiary lymphoid structures that sustain immune activation, and a proinflammatory tumor microenvironment conducive to effective antitumor immunity and enhanced responsiveness to immune checkpoint inhibitors

Essential to effective antitumor immunity, the tumor-immunity cycle constitutes a dynamic sequence that orchestrates systemic defense against malignancies [13] (Fig. 2). This cycle begins with the release of tumor antigens and their capture by dendritic cells (DCs), proceeds through the maturation and migration of these cells to lymph nodes, and culminates in the priming of naïve CD8⁺ T cells [19]. These primed lymphocytes then navigate to the tumor site via chemokine gradients and endothelial adhesion signals to permeate the TME and engage in antigen-specific interactions via major histocompatibility complex class I (MHC-I) complexes, leading to the elimination of target cells [19]. Previous studies have established a foundational “three Cs” model—camouflage, coercion, and cytoprotection—to describe the mechanisms of tumor immune evasion [1, 20]. Expanding on this model, our review provides an in-depth examination of the molecular and cellular foundations underlying the immunologically “cold” tumor phenotype, focusing on the aforementioned mechanisms. These include camouflage, which blocks immune priming and infiltration; coercion, which suppresses immune function; and cytoprotection, which prevents inflammatory cell death [1, 21](Table1)(Fig.3). Whereas previous reviews have categorized tumor immune landscapes and sketched out broad therapeutic approaches, our examination enriches the “three Cs” conceptual model by dissecting the nuanced pathways that perpetuate the immunologically “cold” condition [1, 10, 11]. Moreover, our review integrates emerging and underexplored areas such as neural–immune crosstalk, circadian regulation, endosomal sorting complex required for transport (ESCRT)-mediated membrane repair, the gut microbiome, tertiary lymphoid structures, and innovative biomaterial-based strategies, thus providing a comprehensive update on “cold” tumors. This holistic analysis not only advances our understanding of “cold” tumors but also links fundamental mechanisms to potential therapeutic interventions, identifying new opportunities for overcoming resistance to immunotherapy in challenging cancer types.

Fig.2.

The tumor-immunity cycle. The tumor-immunity cycle represents a dynamic, multi-step process essential for effective antitumor immunity. This cycle begins with the release of tumor antigens and their capture by dendritic cells (DCs), proceeds through the maturation and migration of these cells to the lymph nodes. Within lymphoid tissues, DCs prime naïve CD8⁺ T cells, activating antigen-specific cytotoxic T lymphocytes (CTLs). These primed CTLs subsequently traffic to the tumor site, guided by chemokine gradients and endothelial adhesion signals, where they infiltrate the tumor microenvironment. Upon infiltration, CTLs engage tumor cells through antigen-specific recognition via major histocompatibility complex class I (MHC-I) molecules, culminating in targeted tumor cell elimination

Table 1.

Molecular and cellular mechanisms of “cold” tumor phenotypes

Mechanisms	​​Core mechanistic elements​​	Examples	Refs
Camouflage	Defective antigen processing and presentation	B2M mutations result in the failure of MHC-I assembly	[23]
		​​IRGQ targets MHC-I for degradation	[31]
		SUSD6-mediated lysosomal degradation of MHC-I	[33]
	Impaired DC-T cell crosstalk	Calreticulin sequestered in mitochondria with STC1	[41]
		CRL5-SPSB3 ubiquitin ligase targets nuclear cGAS for degradation	[364]
	Impaired immune cell recruitment	β-catenin represses CCL4	[45]
	Physical barrier establishment	ETBR regulates NO to inhibit ICAM-1 expression	[61]
		ECM cross-linking by LOX increases matrix stiffness	[65, 66]
	Epigenetic alterations	HDAC8 deacetylates H3K27, suppressing CCL4	[76]
		KDM5D recruits Sin3-HDAC complexes, suppressing MHC-I	[80]
Coercion	Immunosuppressive cellular networks	Anemia-induced erythropoiesis	[135]
		TAN-driven NET formation	[128]
	Metabolic reprogramming	Tumor-cDC1 competes for glutamine via SLC38A2	[170]
		Cytidine deaminase-mediated UDP production	[180]
	Neural–immune axis modulation	Sympathetic norepinephrine suppresses CD8+ T cells through β-AR signaling	[202, 203]
		Nociceptors in tumors secrete CGRP	[209, 210]
	Circadian rhythm modulation	Modulation of circadian endothelial/leukocyte clocks	[226]
Cytoprotection	Immune synapse defects	Low ICAM-1 expression in tumors impairs immune synapse formation via UHRF1/DNMT1 methylation	[233]
	Cell death signaling dysregulation	Methylation of GSDMD/GSDME promoters prevents gasdermin cleavage	[262, 263]
	ESCRT-mediated membrane repair	ESCRT-III repairs gasdermin pores, preventing pyroptosis	[273, 277]

Fig.3.

Mechanisms underlying the “cold” tumor phenotype. This schematic outlines the “cold” tumor phenotype mechanisms within a comprehensive framework encompassing camouflage, coercion, and cytoprotection. Camouflage involves compromised antigen processing and presentation mechanisms—such as silencing major histocompatibility complex class I (MHC-I), impaired dendritic cell (DC) functionality, dysregulated chemokine signaling leading to ineffective immune cell recruitment, and alterations in tumor-associated vasculature and extracellular matrix alongside epigenetic changes—all of which collectively impair immune detection and infiltration. Coercion is characterized by the formation of immunosuppressive networks including M2 tumor-associated macrophages (TAMs), regulatory T cells (Tregs), myeloid-derived suppressor cells (MDSCs), erythroid progenitor cells (EPCs), and tumor-associated neutrophils (TANs) releasing inhibitory substances, alongside the metabolic reshaping of cancer cells which exhausts nutrients and produces immunosuppressive metabolites. This also involves neural–immune interface alteration leading to T-cell impairment and circadian rhythm adjustments disrupting immune cell movements and efficacy. Cytoprotection confers resistance against inflammatory death by impairing immune synapse formation, altering death signaling pathways, and leveraging the endosomal sorting complex required for transport (ESCRT) for membrane repair in response to cytotoxic threats. Together, these integrated mechanisms promote immunological inactivity, thereby diminishing the effectiveness of immunotherapy in “cold” tumors

Molecular mechanisms underlying “cold” tumors: camouflage, coercion, cytoprotection

Camouflage: impeding immune priming and infiltration

Defective antigen processing and presentation

Effective antitumor immunity is predicated on the accurate processing and display of tumor antigens through MHC-I molecules to cytotoxic T lymphocytes (CTLs) [1]. However, in “cold” tumors, cancer cells deploy a range of tactics to interfere with this essential process, leading to impaired antigen presentation—a key factor contributing to their ability to evade immune detection [1]. These evasion strategies involve a comprehensive suite of molecular mechanisms that impair the antigen presentation system, encompassing genetic alterations, epigenetic modifications, and posttranslational adjustments.

MHC-I silencing mechanisms

Genetic alterations are fundamental to the failure of antigen presentation. Mutations in the genes responsible for the structural elements of MHC-I, such as the heavy chains or beta-2-microglobulin (B2M), or those affecting the ancillary components within the peptide-loading complex, have a pronounced impact on peptide loading and the subsequent cell surface expression of MHC-I heterodimers [22]. Notably, common mutations or deletions in B2M, particularly those observed in cancers such as lung adenocarcinoma, lead to a failure in MHC-I assembly, thus rendering tumor cells invisible to surveillance and elimination by CTLs [23]. Additionally, mutations within peptide-loading complex elements, including proteasome subunits, antigen-processing transporters, and chaperones, disrupt peptide production, transportation, or MHC-I loading, thereby significantly reducing the immunogenicity of tumor cells [24].

Epigenetic dysregulation presents a formidable alternative mechanism for silencing antigen presentation. Repressive histone modifications such as histone H3 lysine 27 trimethylation​ (H3K27me3), which are induced by polycomb repressive complex 2 (PRC2), and H3K9me3 actively suppress the expression of genes encoding MHC-I components and vital cofactors across various malignancies [24–26]. Specifically, the PRC2 complex, via its catalytic subunit ​​enhancer of zeste homolog 2​​ (EZH2) in lymphomas, contributes to the direct silencing of MHC-I transcription—a pathway exploited by tumors to circumvent immune detection [25]. Moreover, DNA hypermethylation in the vicinity of MHC-I gene promoters further inhibits their expression, solidifying a stable barrier against antigen presentation [26, 27].

Furthermore, posttranslational modifications that inhibit antigen presentation through protein degradation or inhibitory signaling are complex. In pancreatic ductal adenocarcinoma (PDAC) and non-small cell lung cancer (NSCLC), the autophagy apparatus plays a critical role in the lysosomal degradation of MHC-I molecules [28–30]. The immunity-related GTPase family Q protein (IRGQ) emerges as a selective receptor for directing misfolded MHC-I molecules for lysosomal degradation, a route leveraged by hepatocellular carcinoma cells [31]. This exploitation, characterized by reduced MHC-I surface expression, leads to evasion from CD8⁺ T cell-mediated eradication, with profound implications for patient survival [31]. Additionally, the proprotein convertase subtilisin/kexin type 9​​ (PCSK9)-induced lysosomal degradation of MHC-I, alongside the SUSD6-TMEM127-WWP2 E3 ubiquitin ligase complex-driven lysosomal degradation, delineates the multifaceted attack on antigen presentation [32, 33]. Furthermore, the overexpression of the nonclassical MHC molecule human leukocyte antigen G​​ (HLA-G) in a variety of cancers suppresses T-cell function and promotes the expansion of immunosuppressive cells, including regulatory T cells (Tregs) and myeloid-derived suppressor cells (MDSCs), intensifying immunosuppression [34].

Impaired DC-T cell crosstalk

Effective activation of Antitumor CTLs necessitates recruitment, activation, And effective antigen capture by conventional type 1 dendritic cells (cDC1s). However, in cold tumors, the chemokine gradients crucial for cDC1 recruitment, such as C–C motif chemokine Ligand 5​​ (CCL5) and X-C motif chemokine Ligand 1 (XCL1), are inhibited by tumor-derived factors, notably transforming growth factor-beta (TGF-β) and lactate [1, 35, 36]. The cyclic GMP–AMP synthase (cGAS)-stimulator of interferon genes (STING)-type I interferon (IFN-I) axis frequently hampers cDC1 activation, with epigenetic mechanisms that silence STING or the enzymatic breakdown of its ligand, cyclic GMP-AMP (cGAMP), by ectonucleotide pyrophosphatase/phosphodiesterase 1 (ENPP1), leading to impaired interferon production and, consequently, DC maturation and the ability to prime naïve T cells [37–39]. Furthermore, even when they infiltrate tumors, DCs are often functionally impaired. Tumor cells elevate “don’t eat me” signals, notably CD47, which interacts with signal regulatory protein alpha (SIRPα) on DCs, negating the prophagocytic action of calreticulin (CALR) [40]. Moreover, stanniocalcin 1 (STC1), which is secreted by tumor cells, confines CALR within mitochondria, inhibiting its surface presentation and further restricting phagocytic functionality [41]. Tumor-released extracellular gelsolin Impedes binding to the C-type lectin domain family 9 member A (CLEC9A) receptor on cDC1s, thus blocking the recognition of dead cell antigens and severely reducing the cross-presentation of tumor antigens to CD8⁺ T cells [42].

In summary, a combination of genetic, epigenetic, posttranslational, and environmental factors effectively disrupts the antigen processing and presentation pathway. This strategic dismantling prevents the visibility of tumor cells to CTLs and undermines DC-mediated T-cell activation, solidifying the immunologically “cold” state that eludes both spontaneous and induced immune responses.

Impaired immune cell recruitment

Chemokine gradient dysregulation

The orchestrated recruitment of immune cells into the TME is critically dependent on chemokine guidance, a process that is frequently subverted in cold tumors [43]. Chemokines, a family of chemotactic cytokines, play pivotal roles in directing myeloid and lymphoid cell extravasation from the bloodstream, positioning within specialized niches, and fostering functional interactions essential for antitumor immunity [44]. In cold tumors, cancer cells actively disrupt this system through multiple mechanisms to evade immune surveillance.

Cancer cells Suppress the expression of chemokines associated with type 1 immunity, which are pivotal for recruiting CTLs and natural killer (NK) cells. For example, in melanoma, oncogenic β-catenin signaling notably represses the secretion of CCL4, impairing the recruitment of conventional cDC1 populations crucial for cross-presenting tumor antigens and activating CD8⁺ T cells. This suppression consequently hampers T-cell priming and infiltration, facilitating immune evasion [45, 46]. Tumors further disrupt the chemotactic gradients required for precise immune cell navigation. Colorectal cancer cells, for example, release soluble C-X3-C motif chemokine Ligand 1​​(CX3CL1), removing the chemotactic gradients necessary for CX3CR1⁺ CTL trafficking into tumors—a striking example of how chemokine dysregulation can spatially limit antitumor immunity [47]. Chemokine redundancy and biased receptor usage further complicate Immune cell trafficking. C-X-C motif chemokine receptor 3 (CXCR3), which is integral to antitumor immunity and is expressed on both CTLs and immunosuppressive Tregs, is an example. Although its Ligands C-X-C motif chemokine Ligand 9 (CXCL9)/10/11 are intended to promote CTL infiltration, their effect can be negated by CXCR3⁺ Tregs, which attenuate effector responses, thus fostering a permissive environment for tumor progression [48, 49].

Moreover, tumors actively coopt chemokine networks to recruit immunosuppressive myeloid populations, enhancing the immunosuppressive milieu of the TME [50]. A landmark study demonstrated how mutant p53 (p53^R172H) in PDAC exploits distal enhancers to upregulate CXCL1 expression in a nuclear factor kappa-B (NF-κB)-dependent manner, driving substantial neutrophil infiltration that undermines the efficacy of immune checkpoint inhibitors—a direct correlation between transcriptional dysregulation and the failure of immunotherapies [50]. Similarly, epithelial cancer cells overexpress chemokines (CXCL1/2/8) to attract neutrophils. These neutrophils deploy neutrophil extracellular traps (NETs), which not only shield tumor cells from CTL- and NK-mediated destruction but also secrete CCL17 to attract CCR4⁺ Tregs, augmenting immunosuppression [51–53]. The redirection of monocyte-recruiting chemokines such as CCL2, CCL5, and CX3CL1 to attract tumor-associated macrophages (TAMs) further supports this effect [54]. TAMs secrete CXCL12, sequestering T cells in stromal niches remote from tumor cells and enabling metastatic spread [55].

In summary, these mechanisms—encompassing the suppression of immunostimulatory chemokines, enlistment of suppressive cell types, and obliteration of chemotactic gradients—collaborate to create a “camouflage” strategy. This strategy effectively deprives tumors of effector immune cells, cementing the immunologically “cold” nature of the TME and presenting significant hurdles for successful immunotherapy [56].

Physical barrier establishment

The architecture of the TME plays a pivotal role in the immune evasion strategies of “cold” tumors, reinforcing their capacity to elude immune surveillance and sustain an immunologically excluded phenotype [9]. Central to this evasion is the establishment of physical barriers, consisting of an irregular tumor-associated vasculature (TAV) and a pathologically remodeled extracellular matrix (ECM). These elements collaborate to restrict the infiltration and functionality of effector immune cells, thus cementing the resistance of tumors to immune-mediated attack.

Abnormal tumor vasculature

Abnormalities in the tumor vasculature are characterized by a tortuous vessel structure, irregular branching patterns, and disorganized endothelial cell layers, leading to widened intercellular junctions [57]. Such architectural anomalies induce vascular hyperpermeability, elevated interstitial fluid pressure, and disrupted blood flow, culminating in an environment that hinders oxygen and nutrient distribution. This results in a hypoxic, acidic TME that significantly curtails T-cell survival, activation, and effector functions. Furthermore, the compromised integrity of pericytes and basement membranes destabilizes the TAV, consequently impeding the extravasation of immune cells into the tumor’s parenchymal tissue [57].

In addition to these structural impediments, tumor endothelial cells exhibit a marked anergy, characterized by reduced responsiveness to proinflammatory signals such as interferon-gamma (IFN-γ) and tumor necrosis factor-alpha (TNF-α) [58, 59]. This anergic state is reflected in the diminished expression of key adhesion molecules and chemokines crucial for T-cell adhesion and migration [58, 59]. The molecular landscape further complicates this scenario: vascular endothelial growth factor (VEGF)-triggered Nitric oxide production disrupts intercellular adhesion molecule 1 (ICAM1) organization at endothelial junctions, fibroblast growth factor 2 (FGF2)-mediated actions inhibit adhesion molecule gene transcription, and endothelin B receptor (ETBR) activation compromises ICAM1 presentation on the cell surface under inflammatory stimuli [60, 61]. The aberrant expression of immune checkpoint ligands and Fas ligands by tumor endothelial cells promotes apoptosis in effector T cells while selectively sparing Tregs because of their inherent expression of cellular flice-like inhibitory protein (cFLIP) [62, 63].

Extracellular matrix remodeling

In parallel, the ECM undergoes significant remodeling into a fibrotic barrier that physically hampers the migration of cytotoxic T cells and NK cells toward the tumor core [64–66]. Collagen fibers, which are primarily secreted by cancer-associated fibroblasts (CAFs), undergo extensive cross-linking by enzymes such as lysyl oxidase (LOX), transglutaminase (TGase), and prolyl hydroxylase, resulting in a stiffened, dense matrix architecture [65, 66]. This structural remodeling reduces pore sizes (typically 50–200 nm) below the diameter of T-cell nuclei, physically obstructing CTL migration toward tumor cores [67]. The aligned collagen fibers further spatially confine T-cell trafficking, while elevated interstitial fluid pressure from the aberrant vasculature collapses intratumoral blood vessels, exacerbating perfusion deficits and hindering immune cell entry [68, 69].

Mechanistically, collagen deposition activates biomechanical signaling pathways that reinforce immune exclusion. Stiffened ECM engages β1-integrin on tumor cells, triggering ILK/PI3K/AKT signaling to promote stem-like properties and therapy resistance [70, 71]. Additionally, collagen binding to leukocyte-associated Immunoglobulin-like receptor 1 (LAIR1) on CD8⁺ T cells recruits tyrosine-protein phosphatase non-receptor type 6 (PTPN6), dampening activating signals and inducing exhaustion, thereby compromising antitumor immunity [72]. These physical and molecular barriers synergistically create an immune-excluded niche, facilitating tumor immune evasion [64].

Epigenetic alterations

Cancer cells utilize intrinsic epigenetic reprogramming as a principal mechanism for immune evasion, creating a “camouflaged” phenotype that profoundly disrupts antigen recognition and immune cell recruitment [73]. In contrast to permanent genetic mutations, epigenetic changes offer tumor cells a dynamically reversible mechanism, allowing them to adaptively downregulate immunogenic pathways that are pivotal for T-cell activation [74]. A central tactic in this strategy is the epigenetic Suppression of Antigen presentation mechanisms. The SET domain bifurcated histone lysine methyltransferase 1 (SETDB1), for example, promotes H3K9 trimethylation across clusters of immune-related genes, including IFN genes and MHC-I components. This activity induces a repressive chromatin configuration that mitigates tumor immunogenicity and diminishes CD8⁺ T-cell recognition [75]. Similarly, in Hepatocellular carcinoma, the ability of histone deacetylase 8 (HDAC8) to deacetylate H3K27 at the CCL4 promoter inhibits chemokine-driven T-cell infiltration [76]. These alterations contribute to the development of an immunologically “cold” phenotype characterized by compromised MHC-I expression and attenuated chemotactic signaling [77, 78].

In addition to traditional histone modifications, tumors exploit additional epigenetic mechanisms to deepen Immune camouflage. For example, the lysine demethylase 5B (KDM5B) collaborates with SETDB1 to repress the expression of endogenous retroelements, hindering double-stranded RNA (dsRNA) sensing and the expression of interferon-stimulated genes, which are crucial for immune cell activation [79]. Additionally, sex chromosome-associated epigenetic factors such as KDM5D engage with Sin3-HDAC complexes to specifically suppress MHC-I antigen processing genes, demonstrating a unique layer of immune evasion sex bias [80]. These epigenetic modifications frequently target areas essential for the immune synapse, with DNA hypermethylation and RNA N6-methyladenosine (m⁶A) modifications undermining immunogenic transcripts and modulating chemokine networks that control lymphoid cell recruitment [81, 82].

This immunosuppressive coordination extends to cytokine Signaling, which is modulated epigenetically by tumor entities. Methyltransferase Like 3 (METTL3), an m⁶A writer, enhances the stability of basic helix-loop-helix family member e41​​ (BHLHE41) or c-Myc mRNAs, increasing the production of CXCL1/CCL20 to attract MDSCs and neutrophils, which obstruct T-cell infiltration [83–85]. Furthermore, super-enhancer development in renal carcinoma cells increases CXCL8 expression, assisting in neutrophil-driven metastasis and the exclusion of cytotoxic lymphocytes [86]. Chromatin remodelers Such as AT-rich interactive domain-containing protein 1A (ARID1A) and chromodomain Helicase DNA binding protein 1 (CHD1), in conjunction with NF-κB, increase the levels of interleukin-6 (IL-6) and CXCL2, directing myeloid cells toward immunosuppressive phenotypes that protect the tumor from immune attack [87–89]. These complex epigenetic modifications collectively transform the tumor microenvironment into an arena where reduced antigen presentation, altered chemokine landscapes, and the incorporation of myeloid barriers effectuate comprehensive cloaking against the adaptive immunity of the host.

Coercion: suppressing immune functions

Immunosuppressive cellular networks

The TME of immunologically “cold” tumors presents a formidable challenge to effective antitumor immunity, largely due to the complex network of immunosuppressive cell types. These cells—which include TAMs, Tregs, MDSCs, immunosuppressive erythroid progenitor cells (EPCs), and tumor-associated neutrophils (TANs)—collaboratively create a landscape conducive to cytotoxic effector cell activity and infiltration [90–94].

Tumor-associated macrophages and regulatory T cells

Under the influence of tumor-derived signals, including hypoxia, exosomal microRNAs, and metabolites such as lactate and succinate, TAMs transition to an M2-like phenotype that suppresses the cytotoxic activity of CD8⁺ T cells and NK cells [95–100]. The secretion of IL-10 and VEGF further complicates immune elimination, while their involvement in “don’t eat me” signaling pathways, such as the CD47-SIRPα, CD24-sialic acid-binding Ig-like lectin 10 (Siglec-10), and STC1 axes, supports tumor evasion from immune surveillance [41, 101–104]. Tregs are pivotal in ensuring immune tolerance within the TME [91]. They are attracted to the tumor site by chemokines such as CCL22 and CCL28, and create a hostile metabolic environment that favors their survival and suppressive function [105–107]. These cells inhibit CD8⁺ T-cell responses through multiple mechanisms, including CTLA-4-mediated actions, adenosine production, and immunosuppressive cytokine release, thereby hampering effector functions [108–110].

Myeloid-derived suppressor cells

MDSCs, which are abundant in response to cancer-related hypoxia, nutrient scarcity, endoplasmic reticulum stress, and chronic inflammation, engage a sophisticated immunosuppressive arsenal [111–114]. They not only inhibit T-cell functionality through the upregulation of PD-L1, which is facilitated by tumor-secreted factors, but also exhaust key metabolites essential for T-cell activation, impairing T-cell receptor signaling and fostering Treg differentiation [111–116]. In addition, MDSC-derived reactive oxygen species (ROS) and NO disrupt T-cell receptor ζ-chain expression and induce apoptosis, further inhibiting leukocyte trafficking [117–121]. MDSCs intensify the immunosuppressive network by engaging Tregs, transforming B cells into regulatory types, and swaying TAMs toward an M2 phenotype, underscoring their critical role in maintaining tumor “coldness” [122–124].

Erythroid progenitor cells

Aberrant hematopoiesis, elicited by tumor-induced stressors such as anemia, hypoxia, and chronic inflammation, plays a critical role in crafting a potent immunosuppressive network that underlies the “cold” tumor phenotype [93]. This phenomenon primarily occurs through extramedullary hematopoiesis (EMH), with hematopoiesis shifting from the bone marrow to peripheral sites such as the spleen and liver. A major result of EMH is the significant expansion of EPCs, especially the immunosuppressive CD45⁺ subset, which in advanced cancer stages, frequently overtakes traditional immunosuppressive cell populations in both the splenic and intratumoral environments [125–127]. These cells secrete a range of mediators, arginase-1 (ARG1), ARG2, ROS, TGF-β, and IL-10, which impair T-cell function, reduce L-arginine availability, promote T-cell dysfunction or apoptosis, and promote Treg differentiation [126, 128–132]. Moreover, CD45⁺ EPCs also directly inhibit T-cell proliferation and cytotoxicity by engaging in coinhibitory signaling through PD-L1/PD-L2 and V-domain Ig-containing suppressor of T cell activation (VISTA), which act via receptors such as PD-1 [133, 134]. Pivotal TME adaptation includes the granulocyte‒macrophage colony stimulating factor (GM-CSF)-driven transformation of CD45⁺ EPCs into erythroid-differentiated myeloid cells (EDMCs), which display enhanced immunosuppressive abilities by upregulating PD-L1, inducible nitric oxide synthase (iNOS), and ARG1, thereby reducing CD8⁺ T-cell function and undermining the immune checkpoint blockade (ICB) response [135]. This EPC-mediated immunosuppression fosters a self-perpetuating cycle of T-cell dysfunction and tumor growth prompted by tumor-induced anemia and hypoxia [93].

Tumor-associated neutrophils

TANs further compound this immunosuppression by directly disrupting cytotoxic T-cell and NK-cell operations [94]. Under conditions of nutrient stress, TANs preferentially undergo a metabolic shift toward fatty acid oxidation, generating excess ROS that compromise CD4⁺ T-cell integrity and increase lactate production and glucose uptake [136, 137]. This metabolic imbalance triggers ferroptosis in polymorphonuclear MDSCs (PMN-MDSCs), releasing immunosuppressive oxidized lipids such as prostaglandin E₂ (PGE₂), which inhibit T-cell activity and facilitate neutrophil-to-PMN-MDSC transformation [138]. Additionally, TANs deplete crucial extracellular arginine reserves through the annexin A2/TLR2/MyD86 pathways, thereby diminishing T-cell viability and NK-cell activation [139–141]. In conjunction with CD39/CD73-expressing TANs converting extracellular adenosine triphosphate (ATP) into adenosine and secreting Treg and TAM-recruiting chemokines, these mechanisms generate an immunosuppressive milieu [142–145]. TAN-driven immunosuppression is further exacerbated by circadian disruptions, leading to advanced neutrophil aging and NET formation, aligning with tumor-derived stimuli to enhance signal Transducer And activator of Transcription 3 (STAT3) activation and bolster the presence of PMN-MDSCs, consolidating the tumor’s immune-evading capabilities [146, 147].

These intricate and interdependent strategies employed by various immunosuppressive cell populations highlight the complexity of the TME in “cold” tumors. Understanding these mechanisms provides critical insights into the hurdles that must be overcome to transform “cold” tumors into “hot” tumors, a transition essential for enhancing the responsiveness of these tumors to immunotherapy.

Metabolic reprogramming

Cancer cell metabolic reprogramming is a pivotal determinant in crafting the immunosuppressive TME typifying immunologically “cold” tumors. Through the commandeering of crucial metabolic pathways, including glucose utilization, lipid metabolism, amino acid turnover, and nucleotide processing, cancer cells effectively circumvent both innate and adaptive immune surveillance by diverse methods [148, 149]. This manipulation results in a TME paradoxically abundant in immunosuppressive metabolites but depleted of nutrients vital for effector immune cells, thereby exacerbating T-cell dysfunction and fostering the growth of immunosuppressive networks.

Glucose metabolism

The strategic exploitation of glucose competition by cancer cells severely compromises CTLs. The upregulation of glucose Transporters, notably glucose Transporter 1 (GLUT1), exhausts the supply of intratumoral glucose, hampering CTL infiltration and functionality in types of cancer resistant to therapy, such as melanoma and NSCLC [36, 150]. This glycolytic supremacy diminishes CXCL10 secretion, undermines CTL recruitment, triggers Hexokinase 2 (HK2)-dependent NF-κB signaling, enhances immune checkpoint expression, and promotes M-CSF/GM-CSF secretion, thereby augmenting the population of MDSCs [36, 151, 152]. Furthermore, lactate, a byproduct of glycolysis, promotes immunosuppression by inhibiting CTL proliferation and cytotoxic capabilities, amplifying Treg functionality, and orienting TAMs toward a protumoral M2-like phenotype through mechanisms including hypoxia-inducible factor 1-alpha (HIF-1α) stabilization and ARG1 induction [153–159]. Emerging evidence also indicates that lactate promotes the stabilization of PD-L1 on tumor cells via lactylation modifications, hence fueling a feedback mechanism that expedites immune evasion [160].

Lipid metabolism

Dysregulated lipid metabolism in cancer cells leads to the generation of immunosuppressive lipids, significantly impairing immune cell function. Increased fatty acid oxidation elevates CD47 expression, thus preventing phagocytosis [161]. Meanwhile, accumulation of lipid droplets promotes PGE₂ synthesis, adversely affecting DC maturation, suppressing NK cell activity, and elevating PD-1 expression on T cells and myeloid cells [35, 162–167]. Additionally, the expression of the fatty acid transporter CD36 in cancer cells enables the uptake of oxidized lipids, triggering ferroptosis and impairing the functionality of CD8⁺ T cells [168, 169].

Amino acid metabolism

Cancer cells also interfere with amino acid availability, creating an environment conducive to immune evasion. The dependence on glutamine hinders the ability of cDC1s to prime T cells by competitively inhibiting their uptake through solute carrier family 38 member 2 (SLC38A2) [170]. In the context of osteosarcoma, L-type amino acid transporter-2 (LAT2)-mediated glutamine uptake increases CD47 expression, further inhibiting phagocytosis [171]. The sequestration of methionine via LAT4 reduces available extracellular pools, leading to decreased H3K79me2 methylation and STAT5A signaling in T cells, thus dampening their effector functions [172]. Additionally, the catabolism of tryptophan into kynurenine by indoleamine 2,3-dioxygenase 1 (IDO1)/tryptophan 2,3-dioxygenase (TDO) fosters Treg differentiation and promotes an M2-like phenotype in TAMs, with the process being intensified in glioblastoma by quinolinic acid production and lysine catabolism reprogramming, which reduces the efficacy of dsRNA/dsDNA-induced type I interferon responses and CTL infiltration [173–175]. Notably, dietary interventions targeting amino acid metabolism have shown mixed outcomes. While serine/glycine deprivation in colorectal cancer may inhibit tumor growth and boost CD8⁺ T-cell infiltration through the induction of immunogenic cell death (ICD) and chemokine secretion, it paradoxically promotes PD-L1 lactylation in tumor cells, hastening the upregulation of immune checkpoints and leading to T-cell dysfunction [160]. Moreover, breast cancer cells exporting arginine to support polyamine biosynthesis in TAMs establishes a protumor environment through spermine-mediated, thymine DNA glycosylase (TDG)-dependent DNA demethylation, and peroxisome proliferator activated receptor gamma (PPARG) upregulation, creating a perpetuating immunosuppressive loop that is resistant to nutrient depletion [176].

Nucleotide metabolism

Dysregulation of nucleotide metabolism plays a critical role in creating potent immunosuppressive signals within the tumor microenvironment. The conversion of extracellular ATP into adenosine, which then binds to A2A receptors on immune cells, results in the suppression of T-cell activity, augmentation of Treg functionality, and inhibition of DC maturation [177–179]. Recent research has identified new nucleotide-mediated immunosuppressive pathways: the overexpression of cytidine deaminase in PDAC and melanoma cells results in the production of uridine diphosphate (UDP), which activates P2Y6 receptors on TAMs and fosters immunosuppression and resistance to anti-PD-1 therapies. Inhibiting cytidine deaminase disrupts this TAM-mediated CTL exclusion, rendering tumors more susceptible to immunotherapy [180]. Additionally, adenylosuccinate lyase (ADSL) activity in breast cancer cells under hypoxic conditions generates Fumarate, which binds to And inhibits STING, thereby preventing cGAMP-dependent interferon regulatory factor 3 (IRF3) activation and IFN-I production, ultimately impairing innate immune detection and CD8⁺ T-cell activation [181].

Immunometabolic perspective

Immunometabolism has emerged as a fundamental regulator governing the immune-permissive versus immune-excluded phenotypes of tumors, acting as a critical molecular “switch” in the cold-to-hot transition [11, 182–184]. “Cold” tumors are characterized by a metabolically hostile microenvironment wherein cancer cells, through oncogene-driven metabolic rewiring, consume vast quantities of glucose, amino acids, and oxygen, thereby starving infiltrating immune cells of essential nutrients [185]. This nutrient competition is exacerbated by the accumulation of immunosuppressive metabolites. High lactate levels resulting from aerobic glycolysis impair CD8⁺ T cell and NK cell function by inhibiting nuclear factor of activated T cells (NFAT) signaling and cytokine production [186, 187]. Concurrently, lactate drives the polarization of TAMs toward an immunosuppressive M2-like phenotype in a HIF-2α-dependent manner, further reinforcing the immune-suppressive niche [188].

Similarly, tryptophan catabolism mediated by IDO1 and TDO generates kynurenine, which activates the aryl hydrocarbon receptor (AHR) to facilitate Treg differentiation and directly inhibit effector T cell function [189, 190]. Arginine depletion via ARG1, highly expressed in MDSCs and some tumor cells, leads to loss of T-cell receptor ζ-chain expression and cell cycle arrest, inducing a state of T-cell dysfunction and exhaustion [191–194]. Lipid metabolic reprogramming also contributes significantly to immune suppression. Tumor-infiltrating T cells can upregulate the fatty acid transporter CD36, leading to the uptake of oxidized lipids from the TME. This influx triggers lipid peroxidation and ferroptosis, a form of iron-dependent cell death, which drives CD8⁺ T cell exhaustion and functional impairment [168]. Conversely, Tregs leverage this same pathway; their immunosuppressive capacity and survival within the TME are bolstered by CD36-mediated fatty acid uptake, which supports their mitochondrial oxidative metabolism [195]. This metabolic adaptation allows Tregs to thrive in the harsh tumor conditions where effector T cells falter.

Targeting these immunometabolic pathways provides a compelling strategy for reversing the “cold” tumor phenotype [182]. Inhibiting IDO1 or TDO restores tryptophan levels and blocks kynurenine-mediated immunosuppression, thereby reinvigorating antitumor T-cell responses and enhancing the efficacy of immune checkpoint blockade [189, 196]. Similarly, arginase inhibitors such as CB-1158 ameliorate arginine scarcity, reverse T-cell dysfunction, and reduce tumor growth in preclinical models [197]. Disrupting lactate handling through monocarboxylate Transporter 1 (MCT1) inhibition compromises Treg adaptability and mitigates acidosis-induced immunosuppression [157]. Furthermore, glutamine antagonists such as JHU083 impair tumor cell metabolism while promoting oxidative phosphorylation and sustained activation in CD8⁺ T cells, effectively remodeling the TME toward an immunostimulatory state [198]. These approaches, especially when combined with adoptive cell therapy or checkpoint inhibition, offer promising avenues for metabolic reprogramming and ignition of robust antitumor immunity.

In summary, metabolic reprogramming of cancer cells confers two benefits for tumor growth: it secures essential biosynthetic resources for proliferation and establishes metabolic barriers that suppress immune cell activity. This strategic alteration complements defects in antigen presentation and immune cell migration, solidifying an immunosuppressive milieu characteristic of “cold” tumors [148].

Neural–immune axis modulation

The complex interaction between the peripheral nervous system and the immune system within the TME plays a significant role in suppressing antitumor immunity, thereby maintaining the immunologically “cold” phenotype of tumors [199, 200]. Peripheral nerves, encompassing sensory, sympathetic, and parasympathetic fibers, permeate numerous solid tumors, secreting a diverse array of neurotransmitters and neuropeptides (e.g., noradrenaline, acetylcholine, calcitonin gene-related peptide, and substance P). These molecules modulate immune cell function and phenotype, fostering an immunosuppressive environment [199, 200].

Sympathetic nervous signaling

Sympathetic nervous signaling plays a central role in this mechanism. It suppresses cytotoxic CD8⁺ T cell activity, drives T cell exhaustion, and amplifies the immunosuppressive functions of MDSCs through the release of norepinephrine and the activation of beta-adrenergic receptors (β-ARs, predominantly β₂-AR) on immune cells [201, 202]. This adrenergic signaling pathway critically undermines antitumor immunity, sustaining the “cold” tumor state. Studies on melanoma have demonstrated that exposure to catecholamines increases adrenergic receptor beta 1 (ADRB1) expression on CD8⁺ T cells, leading to their exhaustion, reduced cytotoxic functionality, and decreased interferon-gamma production [202]. Moreover, the activation of β₂-AR further impairs T cell receptor signaling pathways, exacerbating T cell dysfunction and significantly compromising antitumor immunity [203]. Interestingly, blocking β-adrenergic receptors with pharmacological agents such as propranolol can counteract this suppression, restoring T-cell functions and promoting their infiltration into the tumor, thereby aiding the transformation of “cold” tumors into the more treatable “hot” state [202, 203]. Clinically, adrenergic innervation of tumors is correlated with higher rates of lymph node metastasis and lower survival in cancers such as head and neck squamous cell carcinoma (HNSCC) and prostate cancer, underscoring the profound impact of sympathetic signaling on immune suppression [204, 205]. The utilization of β-blockers has proven beneficial in patients undergoing immunotherapy, indicating the critical potential of inhibiting this pathway to enhance the efficacy of treatments for “cold” tumors [206, 207].

Sensory neuron signaling

Sensory nociceptor neurons, which are critical for pain sensation, play a significant role in the neural–immune crosstalk. These neurons, when associated with tumors, respond to cancer-related stimuli (e.g., nerve growth factor and cytokines) by releasing calcitonin gene-related peptide (CGRP) and substance P [208, 209]. Specifically, CGRP acts via the receptor activity-modifying protein 1 (RAMP1)/calcitonin receptor-like receptor (CALCRL) complex on CD8⁺ T cells, leading to their exhaustion [209]. This phenomenon has been validated through single-cell RNA sequencing of human melanoma, which revealed that CD8⁺ T cells with RAMP1 expression presented increased signs of exhaustion and a reduced response to ICB [209]. The genetic deletion of Calca, the gene encoding CGRP, in mice led to a marked decrease in the growth of HNSCC, coupled with an increase in the infiltration of CD4⁺ T cells, cytotoxic CD8⁺ T cells, and NK cells, providing a direct link between sensory neuron-derived CGRP and compromised antitumor immunity [210]. Additionally, CGRP influences the immunological composition of the tumor microenvironment by regulating the recruitment of immune cells [211].

Substance P, which primarily acts via the neurokinin-1 receptor (NK1R), induces T-cell exhaustion and is expressed at increased levels in several cancers, including melanoma, breast cancer, and glioma [212–215]. In a variety of immune cells, substance P-NK1R signaling plays a significant role in creating an immunosuppressive milieu. The effectiveness of targeting substance P-NK1R to counteract this immunosuppression significantly depends on the TME’s inflammatory condition [216].

The neural niche, comprising neurons, Schwann cells, immune cells, and cancer cells, engages in continuous signal exchange [200]. Neurotransmitters and neuropeptides serve as messenger molecules, binding to receptors on immune cells and facilitating neuron-immune cell interactions. Conversely, immune-derived cytokines such as IL-1β, IL-6, and TNF increase neuronal sensitivity, leading to increased release of neuropeptides [217, 218]. Such bidirectional communication fosters circuits that not only perpetuate but also intensify the immunosuppressive state of the TME, driving immune cells toward dysfunction and exhaustion. Moreover, the compromised integrity of the blood‒nerve barrier in tumors enhances the interaction between neural and immune elements, further deteriorating the TME and exacerbating the traits of cold tumors, namely diminished T-cell functionality, impaired infiltration, and reduced durability [219, 220]. Therefore, modulation of the neural–immune axis has emerged as a promising systemic strategy to disrupt this immunosuppressive network, reverse T-cell exhaustion, and promote a favorable antitumor immune response [221, 222]. Targeting this axis holds significant potential for converting “cold” tumors into “hot” tumors, ultimately enhancing the efficacy of contemporary immunotherapies.

Circadian rhythm modulation

Circadian rhythms play crucial roles in regulating immune cell trafficking and functionality [223–225]. DCs exhibit peak expression of costimulatory molecules (e.g., CD80) at zeitgeber time 7 (ZT7) under Brain And muscle ARNT-like 1 (BMAL1) control, a rhythm essential for T-cell priming. Recent studies have demonstrated that the abundance and function of tumor-infiltrating CD8⁺ T cells oscillate diurnally, peaking at ZT13 in murine melanoma models. This oscillation is governed by endothelial circadian clocks through rhythmic ICAM1 expression, which regulates leukocyte migration into the TME. Disruption of this endothelial clock (e.g., via Bmal1 deletion) eliminates time-of-day differences in T-cell infiltration, thereby impairing immunosurveillance [226]. From a therapeutic perspective, aligning immunotherapy with peak immune activity significantly enhances efficacy. For example, chimeric antigen receptor T (CAR-T) cell administration at ZT13 improved tumor homing and control in lymphoma models. Similarly, anti-PD-1 delivery at ZT13, when PD-1 expression is at its lowest level on CD8⁺ T cells, increases IFN-γ production and degranulation, augmenting responses in melanoma and NSCLC [223–226].

Circadian rhythm disruption, whether genetic (e.g., epithelial Bmal1 knockout) or environmental (e.g., chronic jet lag), promotes an immunosuppressive TME characteristic of “cold” tumors. This is achieved through the recruitment of MDSCs and the rewiring of cytokine networks [227, 228]. In intestinal and breast cancer models, such disruption amplifies proinflammatory signaling, notably Wnt-dependent CXCL5 upregulation, which drives neutrophil recruitment and MDSC differentiation. These MDSCs exhibit elevated PD-L1 expression, increased ROS production, and enhanced T-cell suppressive capacity, directly inhibiting cytotoxic CD8⁺ T-cell function and promoting fibrosis [147, 229].Consequently, the CXCL5-CXCR2 axis is hyperactivated in breast cancer under circadian misalignment, facilitating MDSC and Treg infiltration while dampening CD8⁺ T-cell and NK cell cytotoxicity. This establishes an immunosuppressive niche that enables immune evasion by cancer stem cells and disseminated tumor cells [147, 229].

Therefore, circadian modulation represents a pivotal axis for converting “cold” tumors to immunogenic “hot” states. Targeting this pathway through chronotherapy, such as timed anti-PD-L1 administration at the early active phase (ZT16) when PD-L1⁺ MDSCs peak, reduces stromal immunosuppression and enhances CD8⁺ T-cell infiltration across colorectal carcinoma, lung cancer, and melanoma models [147, 226, 230]. Synchronizing immunotherapy with circadian immune activity disrupts immunosuppressive niches, offering a strategic approach to reinvigorate antitumor immunity and overcome treatment resistance.

Cytoprotection: resisting inflammatory death

Immune synapse defects

Immune synapses (ISs) represent specialized junctions between cytotoxic lymphocytes (e.g., CD8⁺ T cells, NK cells) and tumor cells and are central to the targeted destruction of malignant cells through the delivery of cytolytic agents such as perforin and granzymes [231]. In “cold” tumors, disturbances in the formation or function of ISs are key mechanisms by which cancer cells resist immune clearance. A pivotal disruption is the downregulation of ICAM1, which is essential for the activation of lymphocyte function-associated antigen-1 (LFA-1) signaling in cytotoxic lymphocytes [232]. The repression of ICAM1, often through ubiquitin Like with PHD And ring finger domains 1 (UHRF1)/DNA methyltransferase 1 (DNMT1)-driven DNA methylation, impairs the killing efficacy of CTLs and NK cells in various cancers [233]. Therapeutic interventions aimed at reinstating ICAM1/LFA-1 interactions can enhance T-cell-mediated cytotoxicity, bolster immune cell infiltration, augment the effects of ICB, and facilitate a shift toward an effector CD8⁺ T-cell phenotype within the TME [233].

The TME actively undermines IS functionality through physicochemical changes. Acidification of the synaptic cleft, driven by tumor metabolism, impedes the release and polymerization of perforin from NK cells, thus diminishing their cytotoxic capacity [234]. Additionally, metabolic disturbances associated with the TME, such as hypoxia, lead to the downregulation of crucial activating receptors on NK cells (e.g., NKG2D, NKp30, and CD16) [235–238]. This reduction hampers the ability of NK cells to recognize tumor cells and establish functional synapses. Furthermore, lipid scarcity, notably a deficit in sphingomyelin, alongside disrupted serine metabolism, compromises the stability of membrane microdomains vital for immune synapse organization. These alterations weaken the physical linkage between immune effector cells and tumor cells [234].

Tumor cells not only passively modify the microenvironment but also actively recruit or express ligands that bind to inhibitory receptors on immune cells, thereby undermining the stability and functionality of immune synapses. For example, the interaction of tumor-expressed human endogenous retrovirus-H Long terminal repeat-associating 2 (HHLA2) with the inhibitory receptor killer cell Immunoglobulin-like receptor 3DL3 (KIR3DL3) on immune cells dampens the cytotoxic responses of both CD8⁺ T cells and NK cells, facilitating immune evasion [239, 240]. Similarly, the interaction between tumor-expressed poliovirus receptor (PVR) and KIR2DL5 on NK cells triggers the formation of inhibitory synapses, thus impeding NK cell activation [241].

The functional implications of IS deficiencies are significant. Inadequate remodeling of the actin cytoskeleton not only obstructs the targeted delivery of lytic granules to the IS but also impedes the effective termination of the synapse following target cell destruction. Consequently, this impediment restricts the ability of cytotoxic lymphocytes to undergo repeated targeting, diminishing their serial killing efficacy [242]. Tumor cells may leverage these limitations, such as insufficient actin contraction or compromised T-cell receptor (TCR) adaptability, to evade cytotoxic actions [1]. Additionally, the inefficient removal of signaling receptors such as the TCR through ectocytosis at the IS extends unneeded activation signals or hinders T-cell withdrawal, further compromising immune responsiveness [243].

Cell death signaling dysregulation

Dysregulation of cell death signaling in “cold” tumors forms a critical barrier against immune-mediated elimination, allowing cancer cells to evade destruction [1]. Under normal conditions, controlled cell death pathways engage the immune system by releasing damage-associated molecular patterns (DAMPs) and proinflammatory cytokines, which are crucial for summoning and activating immune cells [244]. However, in “cold” tumors, such pathways often malfunction, leading to diminished immunogenic death and reduced immune cell infiltration.

Autophagy

Autophagy significantly influences stress responses within tumor cells, markedly impacting antitumor immunity and the efficacy of ICIs [245–247]. This mechanism plays a dual role in cancer immunotherapy, either promoting immune escape or improving treatment outcomes [248]. One key mechanism by which autophagy facilitates immune evasion is through the promotion of lysosomal degradation of MHC-I molecules, leading to diminished antigen presentation and, subsequently, reduced T-cell-mediated cytotoxicity [28, 249]. In particular, the autophagy receptor, neighbor of BRCA1 gene 1 (NBR1), has been shown to guide MHC-I toward lysosomal degradation, especially in PDAC models [28]. Inhibiting autophagy, NBR1, or lysosomal functions has been demonstrated to restore MHC-I expression and enhance immune recognition [28]. Additionally, autophagy critically affects immune checkpoint expression and, consequently, T-cell activity [247, 250, 251]. The activation of autophagy, for example, results in PD-L1 degradation via specific receptors Such as Huntingtin-interacting protein 1-related protein (HIP1R), potentially facilitating increased immune detection [250]. Autophagy also regulates the functionality of immune cells, including DCs and macrophages, by influencing antigen processing, cytokine release, and cell polarization [252, 253]. Inhibiting autophagy can reprogram tumor-associated macrophages from an immunosuppressive M2 phenotype to a proinflammatory M1 phenotype, enhance DC maturation, and bolster CD8⁺ T-cell priming [254]. Additionally, blocking autophagy increases the secretion of chemokines such as CCL5 and CXCL10, fostering a more immunogenic tumor microenvironment [30, 255]. These mechanisms underscore the role of autophagy as a crucial promoter of immune resistance in immunologically “cold” tumors, characterized by low T-cell infiltration and effective immune evasion.

Preclinical studies underscore the potential of autophagy inhibition in overcoming resistance to ICIs [256]. Combining chloroquine, an autophagy inhibitor, with anti-PD-1/PD-L1 and anti-CTLA-4 antibodies has been shown to increase MHC-I expression and enhance CD8⁺ T-cell penetration, collectively reducing tumor growth in resistant models [28]. Moreover, targeting upstream regulators of autophagy, Such as phosphatidylinositol 3-kinase catalytic Subunit type 3 (PIK3C3) or palmitoyl-protein thioesterase 1 (PPT1), increases the efficacy of ICIs by stimulating chemokine-driven T-cell recruitment and counteracting myeloid cell-induced immunosuppression [30, 257, 258]. These insights highlight the potential of autophagy modulation in activating antitumor immunity, establishing autophagy as a promising therapeutic target to mobilize immune responses against “cold” tumors alongside ICIs.

Pyroptosis and necroptosis

In “cold” tumors, pathways leading to proinflammatory cell death, Such as necroptosis And pyroptosis, are often Suppressed. Necroptosis, which involves the proteins receptor-interacting protein kinase 1​​ (RIPK1), RIPK3, and mixed lineage kinase domain-like protein​​ (MLKL), is typically hindered by the epigenetic silencing of the RIPK3 gene through promoter hypermethylation across various cancer types [259, 260]. This mechanism disrupts the assembly of the necrosome complex and the subsequent release of proinflammatory cytokines, thereby diminishing the recruitment of immune cells [261]. Similarly, the disruption of pyroptosis occurs through the hypermethylation of promoters of gasdermin (GSDM) proteins, such as GSDMD and GSDME, in gastric cancer and other tumors. This alteration leads to decreased expression of these proteins, preventing the formation of membrane pores and the subsequent release of IL-1β and IL-18, both of which are essential for eliciting an adaptive immune response [262, 263].

Ferroptosis

Ferroptosis, an iron-dependent cell death process characterized by lipid peroxidation, is similarly disrupted in “cold” tumors. The upregulation of solute carrier family 7 member 11 (SLC7A11) boosts glutathione synthesis, thereby activating glutathione peroxidase 4 (GPX4) to decrease lipid peroxides and prevent ferroptosis [264, 265]. Moreover, the epigenetic silencing of Runt-related Transcription factor 3​​(RUNX3) via DNA methylation increases SLC7A11 expression, enhancing resistance to ferroptosis in various cancers, including gallbladder carcinoma [266]. Furthermore, the overexpression of ferroptosis inhibitory protein 1 (FSP1) in certain tumors offers an alternative cytoprotective strategy by counteracting lipid peroxidation [267]. The suppression of ferroptosis diminishes the release of lipid peroxidation byproducts, which serve as potent DAMPs, thereby undermining the induction of antitumor immunity.

In essence, the orchestrated dysregulation of cell death pathways in “cold” tumors creates an environment conducive to immune evasion, reinforcing the overall immunosuppressive characteristics of the TME and facilitating tumor persistence.

ESCRT-mediated membrane repair

The ESCRT apparatus plays a critical role in protecting tumor cells from immune attack by mediating membrane repair and suppressing immunogenic cell death pathways [268]. A key element in this mechanism is the influx of calcium ions (Ca²⁺) through membrane disruption, which instigates the assembly of ESCRT components, including ESCRT-I, ESCRT-III, apoptosis-linked gene-2 interacting protein X (ALIX), And vacuolar protein sorting-associated protein 4 (VPS4), at the sites of damage [269–271]. This process effectively removes damaged portions of the membrane as ectosomes, thereby reducing the effectiveness of cytotoxic T lymphocytes and promoting immune evasion [268–271].

Furthermore, the ESCRT machinery is instrumental in augmenting tumor cell survival by neutralizing the principal pathways that instigate immunogenic death. For instance, during pyroptosis, the protein GSDMD forms pores in the plasma membrane, facilitating the release of proinflammatory cytokines such as IL-1β [272]. ESCRT-III is responsible for the removal of GSDMD pores through membrane shedding, and ESCRT-I assists in the encapsulation of both GSDMD and IL-1β into exosomes for expulsion from the cell [273–275]. This sequence of actions mitigates proinflammatory signaling and curtails the dispersion of DAMPs, thus limiting DC activation and perpetuating a “cold” TME conducive to immune evasion [274–277]. Interfering with the activation of ESCRT-III, which depends on Ca²⁺, disrupts effective pore removal, prolonging pyroptotic signaling and increasing the immune response [273, 277]. Recent findings have demonstrated that pharmacological inhibition of ESCRT-mediated repair, combined with ICB, can enhance gasdermin-induced pyroptosis, thereby increasing T-cell infiltration and leading to tumor shrinkage in preclinical studies [277, 278]. Additionally, ESCRT-III repairs membrane damage induced by lipid oxidation during ferroptosis, impeding the escape of DAMPs and cytokines that could otherwise activate a stronger antitumor immune reaction [279, 280]. In the event of necroptosis, ESCRT-I encapsulates phosphorylated MLKL, a critical pore-forming enzyme, within exosomes for removal from the cell, whereas ESCRT-III facilitates the extraction of compromised membrane regions [281].

The ESCRT complex also plays a critical role in maintaining the “cold” tumor phenotype through non-repair mechanisms. Specifically, phosphorylation of the ESCRT-0 subunit, hepatocyte growth factor-regulated tyrosine kinase substrate (HRS), by extracellular signal-regulated kinase (ERK) encourages the loading of PD-L1 into exosomes, subsequently inhibiting CD8⁺ T cell infiltration into tumors [282]. Moreover, the ESCRT-III component charged multivesicular body protein 2 A (CHMP2A) bolsters resistance to NK cells by regulating the expulsion of extracellular vesicles containing apoptosis-inducing factors such as MHC class I chain-related gene A (MICA), MICB, and TNF-related apoptosis-inducing ligand (TRAIL) [283].

Genetic modifications in the ESCRT machinery exacerbate its cytoprotective roles. Lung cancer cells with hemizygous deletions in the PTPN23 gene, which encodes an ALIX-related protein, increase B-cell lymphoma survival through integrin-mediated pathways [284]. Similarly, loss of either VPS4A or VPS4B can lead to synthetic lethality in cells lacking both paralogs, while inducing NF-κB-mediated inflammatory responses and ICD in tumors with single paralog deficiency, thereby paradoxically encouraging antitumor macrophage activity [285–287]. Furthermore, the loss of VPS37A interferes with the completion of autophagosome formation and increases endosomal signaling turnover, resulting in the accumulation of ubiquitinated receptors that promote cell survival [287–289]. Therefore, the ESCRT system establishes a multifaceted barrier against immune attack through membrane repair, immunomodulatory vesicle secretion, and genetic adaptations.

Therapeutic approaches for converting “cold” to “hot” tumors

Therapeutic approaches to​ counteracting camouflage

Chemokine axis rewiring

Therapeutic strategies to target camouflage mechanisms in immunologically “cold” tumors focus on reprogramming immune cell trafficking to overcome barriers to immune recognition and infiltration (Fig. 4A) (Table 2). Chemokine axis rewiring represents a core strategy and contributes fundamentally to converting “cold” tumors into “hot” tumors by reprogramming immune cell trafficking.

Fig. 4.

Therapeutic approaches targeting camouflage mechanisms in “cold” tumors. Immunologically “cold” tumors are characterized by diminished immune priming and infiltration, which is facilitated by the chemokine dysregulation, atypical vasculature, a fibrotic extracellular matrix (ECM) barrier, and hindered antigen presentation due to major histocompatibility complex class I (MHC-I) downregulation. To address these challenges, therapeutic strategies have been developed to bridge this transition. A These strategies include the rewiring of the chemokine axis to bolster the recruitment of effector T cells and dendritic cells (DCs). B Additionally, they involve disrupting physical barriers through vascular normalization, ECM remodeling, and targeting cancer-associated fibroblasts (CAFs). C-D Strategies also comprise activating the innate immune system with stimulator of interferon genes (STING) agonists to rejuvenate type I interferon responses and employing epigenetic reprogramming of both tumor and immune cells to revive antigen presentation and negate immunosuppressive pathways. Together, these integrated approaches specifically target the tumor’s camouflage mechanisms to facilitate T-cell infiltration and rejuvenate antitumor immunity

Table 2.

Therapeutic approaches for converting “cold” to “hot” tumors

​​Targeted mechanism​	Therapeutic approaches	Examples	Development stage	Trial Number	Refs
Targeting camouflage	Chemokine axis rewiring	BL-8040 (CXCR4 antagonist)	Phase II trial	NCT02907099	[303]
		Mogamulizumab (anti-CCR4)​	Phase II trial	NCT02281409	[295]
	Disrupting physical barriers	Bevacizumab (anti-VEGF)​	FDA approved	–	[306]
		​​PEGPH20 (PEGylated hyaluronidase)​	Phase II trial	NCT03193190	[328]
	Innate immune agonists	​ADU-S100 (STING agonist)​	Phase I trial	NCT02675439	[371]
		MK-1454 (STING agonist)​	Phase I trial	NCT03010176	[372]
		Bacillus Calmette-Guérin (TLR4 agonist))	FDA-approved	–	[380]
	Epigenetic reprogramming	​​Decitabine (DNMT inhibitor)​	FDA-approved	–	[389]
		GSK2879552 (LSD1 inhibitor)​	Phase I trial	NCT02034123	[400]
Targeting coercion	Targeting immunosuppressive cellular networks	SAR-439459 (TGF-β inhibitor)	Phase I trial	NCT04729725	[431]
		DS-8273a (TRAILR2 agonist)	Phase I trial	NCT02076451	[412]
		All-trans retinoic acid (Target ​​retinoic acid receptor)	Phase I/II trial	NCT03200847	[416]
		INCB001158 (ARG1 inhibitor)	Phase I trial	NCT02903914	[432]
	Tertiary lymphoid structures induction	Cytokines (LT family) or agonistic antibodies	Preclinical evidence	–	[487]
		CCL21 gene-modified DCs	Phase I trial	NCT01574222	[477]
	Metabolic interventions	PFK-158 (PFKFB3 inhibitor)	Phase I trial	NCT02044861	[491]
		​TVB-2640 (FASN inhibitor)​	Phase II trial	NCT03032484	[494]
		​Telaglenastat (Glutaminase inhibitor)	Phase II trial	NCT03163667	[497]
		Oleclumab (anti-CD73)	Phase II trial	NCT03822351	[505]
	Neural–immune axis interception	​​β-blockers	Phase III trial	NCT02362594	[206]
		Clonidine (α2-adrenergic agonists)​	Preclinical evidence	–	[510]
		ANT308 (VIP antagonist)​	Preclinical evidence	–	[512]
	Microbiome engineering	Salmonella-IL2	Phase I trial	NCT01099631	[518]
		​​Fecal microbiota transplantation	Phase I trial	NCT03772899	[528]
		Microbial ecosystem therapeutic 4	Phase II trial	NCT05743777	[529]
		CBM588 (Bifidogenic live bacterial product)​	Phase I trial	NCT03829111	[530]
	Circadian rhythm modulation	Timed anti-PD-1 administration	Preclinical evidence	–	[226]
Targeting cytoprotection	Promotion of inflammatory cell death	T-VEC (Oncolytic virus)	FDA approved	–	[539]
		Photodynamic therapy	Phase I trial	NCT00470496	[541]
		LTX-315 (Oncolytic peptides)	Phase I trial	NCT01986426	[544]
		Elesclomol (Cu ionophore)​	Phase II trial	NCT00888615	[559]
	Targeting the ESCRT machinery	VPS4 paralog inhibitors​	Preclinical evidence	–	[286]
		Calcium chelator	Preclinical evidence	–	[277]

Localized delivery techniques that target chemokines to recruit CXCR3⁺ T cells and DCs have demonstrated effectiveness in preclinical studies. For example, antibody-chemokine fusion proteins, such as the CXCL10-EGFRvIII single-chain variable fragment (scFv), bind chemokines directly to tumor cells, thus promoting T-cell infiltration in glioblastoma and increasing survival in mouse models [290]. Likewise, the combination of CCL4 with the von Willebrand factor (vWF) collagen-binding domain enhances chemokine retention in tumors, improving the recruitment of CCR5⁺ cDC1, and working in tandem with ICB in resistant melanoma [291]. Furthermore, the use of oncolytic viruses modified to express CXCL9 or CXCL11 has been investigated, although their impact may be curtailed by the natural chemokine response to viral infection [292–294].

Strategies to deplete Tregs by blocking their chemokine receptors impair their ability to migrate to tumors. The anti-CCR4 antibody mogamulizumab, which targets CCR4^high intratumoral Treg cells, has shown promise in stabilizing solid tumors and inducing partial responses in advanced esophageal cancer patients [295, 296]. Similarly, antibodies targeting CCR8 can eliminate Treg subsets within the TME, bolstering antitumor immunity in experimental models. Agents that neutralize myeloid-attracting chemokines such as CCL2 (carlumab) or CXCL8 (HuMax-IL-8) aim to reduce protumor monocytes and neutrophils, although patient outcomes have yet to significantly improve [297–299].

Adoptive cell therapies are being enhanced by the introduction of chemokine receptors that align with specific ligands within the TME. CAR-T cells modified to express CXCR1 or CXCR2 show increased affinity for CXCL8-abundant glioblastomas, markedly decreasing tumor volume [300]. CAR-T cells equipped with CXCR6 leverage CXCL16 gradients from tumor-associated DCs and macrophages, increasing their proliferation and effectiveness in models of pancreatic cancer [301]. In addition, to circumvent suppression in immunosuppressive zones, CAR-T cells coexpressing CCR8 and a dominant-negative TGF-β receptor coordinate for superior tumor suppression [302].

Innovations in chemokine modulation are increasingly being incorporated into combination treatments. The CXCR4 antagonist BL-8040, for example, augments anti-PD-1 therapy in pancreatic cancer by securing stem-like T cells within tumors and obstructing the exit of myeloid suppressors [303]. DNA vaccines that encode chemokine‒antigen combinations target CCR5⁺ DCs to activate tumor-specific T cells, showing promise in human papillomavirus (HPV)-related cancers [304, 305]. These approaches collectively reconstruct chemokine landscapes, disrupting the deceptive tranquility of “cold” tumors to facilitate robust immune engagement and destruction.

Disrupting physical barriers

Dysfunctional TAVs and dense ECMs form form physical barriers in immunologically “cold” tumors, severely restricting immune cell infiltration and promoting immunosuppression [9]. Overcoming these structural impediments is essential for converting immune-excluded tumors into T-cell-inflamed microenvironments. Therapeutic strategies focus on normalizing the aberrant vasculature and remodeling the fibrotic stroma (Fig. 4B).

Vascular normalization transforms aberrant TAVs into functional conduits, improving immune cell infiltration and synergizing with immunotherapy [58, 306]. The optimal administration of low-dose antiangiogenic agents, such as bevacizumab or DC101, reduces vascular leakage, improves pericyte coverage, and restores blood flow. This enables T-cell infiltration and avoids the hypoxia that high-dose regimens may exacerbate [307–310]. Multiple kinase inhibitors, such as lenvatinib, which targets VEGFR1-3, normalize the TAV in melanoma and hepatocellular carcinoma by reprogramming macrophages and diminishing MDSCs, increasing the efficacy of PD-1 blockade [311–314]. Dual targeting of VEGF And Angiopoietin 2 (ANG2) via bispecific antibodies remodels vessels and boosts CD8⁺ T-cell presence, while sensitizing tumors to PD-1/PD-L1 inhibitors by promoting a proinflammatory macrophage environment and reducing Treg presence [315–317]. Stroma-directed approaches, such as TGF-β inhibition with tranilast, decrease the ECM density, dilate vessels, and improve T-cell presence, working in synergy with ICIs [318, 319]. Additionally, low-dose radiation and metronomic chemotherapy remodel the TME by increasing the expression of adhesion molecules Such as vascular cell adhesion molecule 1 (VCAM1) and elevating thrombospondin-1 (TSP1), respectively, thereby normalizing perfusion and supporting CTL activity [320–323]. Innovations have extended to the metabolic targeting of tumor endothelial cells to improve vessel functionality and drug access, as well as CAR-T cells targeting endothelial cell antigens such as VEGFR2, to refine the vasculature and lymphocyte attraction, contingent on precise tumor-specific antigen selection for safety [324, 325].

Concurrently, strategies targeting the ECM aim to overcome its role as a barrier [326]. PEGylated recombinant hyaluronidase (PEGPH20) diminishes hyaluronic acid levels in PDAC, thereby easing stromal tension and enhancing drug accessibility [327, 328]. Although phase II trials have indicated only modest survival improvements, combining PEGPH20 with chemotherapy or ICIs has been shown to modify the dynamics of the TME [327, 328]. The application of collagenase to disrupt the ECM significantly promotes T-cell penetration into tumor interiors and amplifies responses to ICIs in preliminary studies [329, 330]. Furthermore, targeting enzymes that crosslink the ECM reduces stromal rigidity, thus facilitating immune cell movement [329, 330]. Inhibitors targeting LOX, such as simtuzumab, and repurposed medications, such as losartan, alleviate collagen crosslinking, decrease interstitial fluid pressure, and improve blood flow [64, 331, 332]. Innovative compounds such as the pan-LOX inhibitor PXS-5505 are currently undergoing clinical evaluation. Targeting collagen receptors also disrupts ECM-driven Immune exclusion. For example, discoidin domain receptor 1 (DDR1)-blocking antibodies reversed collagen alignment-induced T-cell exclusion in triple-negative breast cancer models [333, 334]. Furthermore, disrupting the interaction between the ECM and cells through anti-α5β1 integrin antibodies, such as volociximab combined with paclitaxel, has improved progression-free survival in ovarian cancer patients by reducing ECM stiffness and MDSC recruitment [64]. Effective strategies also include targeting the production of ECM by CAFs. Depleting CAFs or inhibiting profibrotic signaling (e.g., TGF-β with galunisertib) suppresses collagen deposition and synergizes with ICIs in colorectal and pancreatic cancers [335, 336].

Collectively, these approaches overcome physical barriers in “cold” tumors by normalizing the dysfunctional vasculature and degrading or remodeling the fibrotic ECM. This facilitates robust immune cell infiltration and effector function, transforming immune-excluded tumors into T-cell-inflamed microenvironments to potentiate immunotherapy responses.

Utilization of innate immune agonists

Strategies that leverage innate immune agonists, including STING, Toll-like receptors (TLRs), RIG-I-like receptors (RLRs), and NOD-like receptors (NLRs), offer considerable potential in activating antigen-presenting cells, attracting effector T cells, and reprogramming the tumor microenvironment to foster immunostimulatory conditions [337].

The cGAS-STING axis, which is critical for cytosolic DNA sensing, plays a pivotal role in the transition of tumors from an immunologically “cold” to a “hot” phenotype by detecting aberrant DNA and initiating a robust innate immune response [338, 339]. Upon encountering anomalous DNA from sources such as micronuclei, mitochondrial DNA, or therapy-induced damage, cGAS catalyzes the formation of cGAMP, which activates STING [338, 340–347]. This activation triggers a cascade, including the recruitment of TanK-binding kinase 1 (TBK1), phosphorylation of STING, and activation of IRF3, leading to the secretion of IFN-I and inflammatory chemokines [348, 349]. IFN-I is instrumental in inducing a “hot” tumor microenvironment by activating DCs, enhancing antigen presentation, and promoting CD8⁺ T-cell responses and tumor antigen cross-presentation [350]. Simultaneously, STING activation increases NF-κB signaling, leading to the production of chemokines that attract CD103⁺ DCs and CTLs to the TME [348, 351]. Notably, the intercellular transfer of tumor-derived cGAMP through specific transporters acts as an “immunotransmitter”, enabling STING activation in DCs and macrophages to circumvent potential cGAS defects in tumor cells [352–357]. This process promotes DC maturation, the upregulation of costimulatory molecules, and lymph node migration, priming tumor-specific T cells [358, 359]. Moreover, metabolic disturbances such as mitochondrial dysfunction further intensify this immune-stimulating pathway, linking cellular stress mechanisms to inflammation driven by interferon [360–362].

In “cold” tumors, the cGAS-STING axis is often inhibited by several mechanisms, including the extracellular breakdown of cGAMP by ENPP1/3, nuclear sequestration and proteasomal destruction of cGAS, and disturbances in STING trafficking due to mutations in coatomer subunit alpha (COPA) or ​​ADP-ribosylation factor 1​​ (ARF1) [38, 363–367]. These issues result in abnormal signaling and desensitization of the pathway. As such, reactivating the cGAS-STING- IFN-I axis is crucial for connecting innate DNA sensing within the cytosol to the mobilization of adaptive immune responses and efficient antigen presentation. This reactivation is essential for the development of T-cell-inflamed tumors that are receptive to immunotherapy treatments [368, 369].

Targeting the cGAS-STING axis represents a groundbreaking approach for transforming “cold” tumors into immunologically active sites [339, 370] (Fig. 4C). Targeting the cGAS-STING axis represents a groundbreaking approach for transforming “cold” tumors into immunologically active sites. Precision delivery of agonists such as synthetic cyclic dinucleotides (CDNs) such as ADU-S100 (MIW815) and MK-1454 directly stimulates STING, initiating IFN-I production and mobilizing DCs and CTLs [371, 372]. Early clinical trials (NCT02675439 and NCT03010176) have shown that intratumoral injections of these agonists not only augment CD8⁺ T-cell infiltration but also enhance the effectiveness of PD-1 inhibitors, despite the limited effectiveness of monotherapy due to adaptive resistance mechanisms [371, 372]. Engineered bacterial vectors, such as SYNB1891, and innovative nonnucleotide agonists, including diamidobenzimidazole (diABZI) and oral MSA-2, demonstrate superior pharmacokinetics and specificity to the TME [373, 374]. Notably, diABZI effectively induces significant tumor regression in colon cancer models by rapidly promoting IFN-β and CXCL10 production, thus altering chemokine gradients to favor T-cell trafficking [373]. Efforts to increase the activity of endogenous cGAMP involve ENPP1 inhibitors, such as RBS-2418, which have bolstered T-cell infiltration in clinical trials [38, 363]. Additionally, the use of Mn²⁺-loaded nanoparticles and nucleic acid-based technologies has refined delivery methods, enhancing the sensitivity of cGAS or shielding stimulatory DNA [375–377]. In the clinic, the integration of STING agonists with immunotherapies, including PD-1 inhibitors and CAR-T cell therapy, has shown promising outcomes, particularly in modulating T cell-responses and nurturing CAR-T cell functions within the TME [378, 379]. These combinatorial strategies illuminate the capacity of cGAS-STING agonists to activate innate immunity and remodel cold TMEs into hot, immune-responsive environments.

TLR agonists exploit microbial detection axes to counteract immunosuppression within the TME. Bacillus Calmette-Guérin (BCG), the first FDA-approved innate immune therapy for non-muscle-invasive bladder cancer, utilizes TLR4 to induce inflammation and macrophage polarization [380]. Similarly, the TLR7 agonist Imiquimod remolded the microenvironment in cutaneous T-cell lymphoma, achieving lesion remission in 75% of patients through DC activation (NCT01676831) [381]. Innovations such as SBT6050, which combines TLR8 agonists with HER2-targeting antibodies, increase the specificity of immune activation in tumor cells, minimizing side effects (NCT04460456). However, TLR engagement in stromal cells risks promoting metastasis through IL-6 and VEGF, highlighting the importance of targeted approaches [382, 383].

Commensal bacteria serve as natural sources of TLR ligands, exerting systemic effects on antitumor immunity. The oral administration of Lactobacillus rhamnosus GG (LGG) has been shown to enhance intratumoral IFN-β production in colorectal cancer models. This effect is mediated through activation of the cGAS-STING axis in DCs, leading to improved CD8⁺ T-cell cross-priming and synergistic effects with PD-1 blockade therapies [384]. Furthermore, fecal microbiota transplantation from ICI responders has been shown to increase the gut level of Akkermansia muciniphila, enhancing TLR4-dependent anti-PD-1 effectiveness in treatment-resistant patients [385]. In contrast, the depletion of gram-negative bacteria through antibiotics has been associated with reduced hepatic metastasis, which is attributed to decreased TLR4-mediated recruitment of PMN-MDSCs [383]. These results highlight the potential of microbiome engineering as a strategic means to amplify innate immune activation endogenously.

Epigenetic reprogramming

Epigenetic reprogramming represents a promising strategy for modulating the immunosuppressive TME in immunologically “cold” tumors, thereby increasing their responsiveness to immunotherapy [73]. Unlike irreversible genetic alterations, epigenetic modifications (e.g., DNA methylation, histone modifications, and RNA methylation) are dynamically reversible, offering tractable therapeutic targets to reshape immunosuppressive mechanisms [386] (Fig. 4D).

Aberrant epigenetic states in cancer cells promote “cold” TME characteristics by suppressing antigen presentation, inhibiting T-cell infiltration, and recruiting immunosuppressive myeloid cells [73]. A principal mechanism behind this is the epigenetic silencing of immunostimulatory genes, such as CXCL9 and CXCL10, essential chemokines for the recruitment of CD8⁺ T cells and NK cells to the tumor location [46, 387]. This is achieved through hypermethylation of promoter regions or the addition of repressive histone marks such as H3K27me3, orchestrated by DNMTs, EZH2, or other chromatin modifying enzymes, which suppress chemokine expression, thus impairing T-cell movement and endorsing an immunologically excluded phenotype [78, 388]. Similarly, the silencing of MHC class I and II gene components through DNA methylation or histone deacetylation diminishes tumor cell visibility to T cells [85]. The employment of DNMT and HDAC inhibitors, such as the FDA-approved decitabine (DAC), has shown promise in reversing these immunosuppressive mechanisms. Decitabine, for example, has been shown to be effective at reactivating silenced immune-related genes and enhancing T-cell penetration in patients with PDAC when paired with anti-PD-1 or anti-VISTA therapies [389]. Moreover, azacitidine, another DNMT inhibitor, in conjunction with anti-PD-1/PD-L1 agents, is under investigation for various solid tumors, including melanoma (NCT02816021) [85]. Selective inhibition of HDAC8 (e.g., PCI-34051) has been shown to restore chemokine expression and bolster CD8⁺ T-cell recruitment, enhancing anti-PD-L1 efficacy in conditions such as hepatocellular carcinoma [76]. Ongoing clinical trials are exploring HDAC inhibitors, such as entinostat in combination with pembrolizumab, which target melanoma, bladder cancer, and other malignancies to mitigate T-cell dysfunction and amplify ICI responses (NCT03765229, NCT03978624). Furthermore, EZH2 inhibitors (e.g., DZNep/EPZ) transform “cold” tumors into immunologically active states by activating the dsRNA/STING/IFN axis, which upregulates the CXCL10 and CXCL11 chemokines to recruit effector T cells while suppressing protumor neutrophils [390, 391]. Tazemetostat, an EZH2 inhibitor, has received FDA approval for treating epithelioid sarcoma and follicular lymphoma. Ongoing clinical trials are investigating its use in combination with ICIs for treating various solid tumors, including NSCLC and urothelial carcinoma (NCT03854474, NCT04705818) [392]. Similarly, lysine-specific demethylase 1 (LSD1) inhibitors such as GSK2879552 have shown potential in reversing suppressive epigenetic programs and enhancing the response to immunotherapy [393, 394]. Additional targets Such as additional sex comb-like 1​​ (ASXL1), SET And MYND domain containing 3 (SMYD3), and mixed Lineage leukemia 4​​ (MLL4) demonstrate that their inhibition primes T-cell activation and synergizes with ICIs in melanoma and other cancers [395–397].

Direct epigenetic modulation of immune cells counteracts T-cell exhaustion and myeloid-mediated immunosuppression. In CD8⁺ T cells, hypoxia-induced exhaustion involves HIF-1α/HDAC1/PRC2 complexes that suppress cytotoxic genes (e.g., IFNG). HDAC1 inhibition (e.g., entinostat) restores T-cell function and synergizes with anti-PD-1 therapy in breast cancer [80]. Targeting ​​suppressor of variegation 3–9 homolog 1 (SUV39H1) (with ETP-69) or LSD1 (with GSK2879552) reverses epigenetic silencing of IFNG and GZMB, enhancing T-cell persistence and ICI efficacy [398–400]. In myeloid cells, disruption of the METTL3-YTH domain family 2 (YTHDF2) axis in TAMs or MDSCs interferes with immunosuppressive polarization. YTHDF2 blockade (e.g., DC-Y13-27) impairs NF-κB-driven IL-10 and ARG1 expression in MDSCs, enhancing responses to radiotherapy and anti-PD-1 [89]. In neutrophils, β-glucan-induced trained Immunity remodels chromatin accessibility at the neutrophil cytosolic factor 1 (NCF1) and NCF2 loci, promoting ROS-dependent antitumor activity [401].

However, context-dependent effects necessitate biomarker-driven patient stratification. For example, RNA demethylase AlkB homolog 5 (ALKBH5)-mediated MDSC infiltration has opposing effects on colorectal versus cholangiocarcinoma, underscoring the need for cancer-type-specific epigenetic targeting [402, 403]. Ongoing efforts in clinical epigenetics are increasingly incorporating biomarker assessments, such as DNA methylation signatures or histone modification profiles, to identify patients most likely to benefit from combination epigenetic and immunotherapeutic regimens [404, 405]. Future efforts should integrate single-cell epigenomic profiling with CRISPR screening to identify novel context-specific regulators [406, 407].

Therapeutic approaches to combatting coercion

Targeting immunosuppressive cellular networks

Therapeutic strategies to combat coercion target immunosuppressive cellular networks, metabolic dysregulation, and neural–immune signaling that actively suppress effector T-cell function and infiltration within the immunologically “cold” TME. Immunosuppressive cellular networks, comprising TAMs, Tregs, MDSCs, EPCs, and TANs, play a central role in maintaining the immunosuppressive TME of “cold” tumors, thereby impeding effective antitumor immunity. Therapeutic interventions targeting these cells focus on reprogramming, depleting, or disrupting their functions to dismantle immunosuppressive hubs, enhance T-cell infiltration and activity, and ultimately convert “cold” tumors to immunologically “hot” states [90–94] (Fig. 5A).

Fig. 5.

Therapeutic approaches targeting immunosuppressive coercion in “cold” tumors. Immunologically “cold” tumors evade immune destruction through multifaceted coercion mechanisms, including immunosuppressive cellular networks, metabolic dysregulation, neural–immune axis modulation, and microbiome effects. Therapeutic strategies address these challenges through a multifaceted approach designed to enhance T-cell infiltration and antitumor immunity. A Specifically, interventions target the repolarization, depletion, or functional disruption of immunosuppressive cell types within the tumor microenvironment. B Moreover, metabolic interventions seek to rectify the altered metabolic landscape of cancer cells, targeting the dysregulated metabolism of glucose, lipids, amino acids, and nucleotides to reverse the immunosuppressive milieu. C Neural–immune axis modulation through use of β-blockers aims to inhibit sympathetic nerve-mediated T-cell suppression, whereas interventions targeting the calcitonin gene-related peptide (CGRP)/receptor activity-modifying protein 1 (RAMP1) pathway aim to mitigate sensory nerve-driven immunosuppression. D Additionally, advanced microbiome engineering techniques, including the strategic deployment of bacterial consortia, fecal microbiota transplantation (FMT), and dietary modifications, are leveraged to alter the gut-tumor immune axis, enhancing the efficacy of immune checkpoint inhibitors in traditionally unresponsive tumor contexts. Collectively, these approaches synergize to disrupt the coercive immunosuppressive networks within “cold” tumors, thereby rejuvenating effector T-cell functionality and fostering the emergence of tertiary lymphoid structures, critical for a robust and sustained antitumor immune response

Repolarizing TAMs from the immunosuppressive M2 phenotype to the proinflammatory M1 phenotype is crucial. Metformin, a clinically approved antidiabetic drug, successfully repolarizes M2 TAMs toward the M1 phenotype, enhancing anti-PD-1 therapy outcomes by increasing CD8⁺ T-cell infiltration and diminishing Tregs and MDSCs, thereby inhibiting tumor progression and metastasis [408]. Additionally, chimeric antigen receptor macrophages (CAR-Ms) have been engineered for increased phagocytosis and TME modification. Anti-HER2 CAR-Ms disrupt the extracellular matrix to improve T-cell penetration, and targeting CD147 with CAR-Ms lowers collagen levels, further facilitating T-cell access [409, 410]. Early trials, including those with HER2-targeted CAR-Ms (e.g., CT-0508), signal their potential for solid tumors, particularly when combined with checkpoint inhibitors [411].

MDSCs impose immunosuppression by disrupting metabolic pathways, expressing checkpoint molecules, and attracting regulatory immune cells, highlighting the need for diverse therapeutic strategies. Selective depletion via DS-8273a, a TNF-related apoptosis-inducing Ligand receptor 2​​ (TRAILR2) agonistic antibody, targets MDSCs in advanced tumors [412]. Bispecific T-cell engagers, notably AMV564, reroute T cells to target CD33⁺ MDSCs, enhancing the effectiveness of anti-PD-1 therapy in preclinical studies [413]. Chemokine receptor antagonists, such as PF-04136309 (CCR2 inhibitor) and motivafortide (CXCR4 antagonist), reduce MDSC infiltration and increase CD8⁺ T-cell abundance in pancreatic and colorectal cancers [303, 414]. Strategies for functional reprogramming include the use of STAT3 inhibitors (danvatirsen) to disrupt immunosuppressive signaling and all-trans retinoic acid (ATRA) to promote immunostimulatory differentiation, improving antigen presentation [415, 416]. Similarly, low-dose chemotherapeutic agents and tyrosine kinase inhibitors reduce the number and activity of MDSCs, restoring T-cell function in cancers such as ovarian cancer and melanoma [417, 418]. These combined approaches intended to improve ICB efficacy, yet the selection of patients on the basis of biomarkers is vital owing to MDSC diversity [419, 420].

EPC, which proliferate through tumor-stimulated extramedullary hematopoiesis, emerge as a key immunosuppressive force within advanced cancers [93]. The synergy of radiotherapy and ICB diminishes such tumor-elicited EPCs by activating CD8⁺ T-cell functionality and the IFN-γ signaling pathway [421]. Inhibiting extramedullary hematopoiesis, specifically through preventing the recruitment of hematopoietic stem and progenitor cells to the spleen via blockade of the CCL2/CCR2 or ​​CXCL12​​/CXCR4 axis, effectively reduces EPC amplification [422, 423]. Additionally, treating cancer-induced anemia with agents such as epoetin alfa or modulating the IL-33/suppression of tumorigenicity 2 (ST2) axis curtails the immunosuppressive influence of EPCs [424, 425]. Encouraging their maturation into functional erythrocytes via TGF-β inhibitors or GATA binding protein 1 (GATA1) activation further diminishes their suppressive effect [128, 426]. Integrating these methodologies with ICBs disrupts detrimental EPC–immunity interactions, revitalizes cytotoxic T cells, and cultivates a proinflammatory TME, thereby facilitating the transition from “cold” to “hot” tumors suitable for immunotherapy [135, 421, 427].

TANs support a “cold” tumor phenotype by interacting with Tregs, TAMs, and MDSCs, necessitating their reprogramming or elimination. Inhibitors of CXCR1/2, including AZD5069 and SX-682, obstruct the recruitment of neutrophils that tumors stimulate through specific chemokines. This intervention limits the infiltration of immunosuppressive N2-like TANs and disrupts their collaborative suppression of immune responses [428, 429]. When AZD5069 is used in conjunction with anti-PD-1 therapy, it shifts intratumoral TANs toward a proinflammatory, antitumor role, thereby increasing CD8⁺ T-cell infiltration and counteracting resistance in models such as non-alcoholic steatohepatitis-induced hepatocellular carcinoma [428]. Furthermore, TGF-β inhibitors, exemplified by SAR-439459, facilitate the conversion of TANs from a protumor N2 phenotype to an antitumor N1 phenotype in diseases, including colorectal cancer and NSCLC [430, 431]. With respect to metabolic pathways, the inhibition of ARG1 via CB-1158 cancels the TAN-induced suppression of T cells [197, 432]. Simultaneously, triggering the triggering receptor expressed on myeloid cells 1​​ (TREM1) pathway with agonists such as PY159 enhances IFN-γ production, contributing to the transformation of the TME toward a state favorable for immune activation [433].

Induction of tertiary lymphoid structures

TLSs are ectopic lymphoid aggregates that emerge within nonlymphoid tissues under chronic inflammatory conditions such as cancer, autoimmune diseases, and infections [434]. These structures, which architecturally mimic secondary lymphoid organs, evolve from synergistic interactions between lymphoid tissue inducer cells and lymphoid tissue organizer cells, guided by complex cascades of cytokines and chemokines. The development of TLSs signifies significant reorganization within the TME, establishing TLSs as critical architects in crafting immune-permissive niches within “hot” tumors [16, 435].

Composed of a rich assortment of immune and stromal components, TLSs are instrumental in enhancing adaptive antitumor immunity. The stromal framework within TLSs is established by fibroblasts, follicular dendritic cells (FDCs), and high endothelial venules (HEVs) [436]. Fibroblasts initiate the TLS formation process by secreting chemokines and cytokines that recruit lymphocytes [437–439]. FDCs, denoting TLS maturity, maintain antigens and facilitate germinal center reactions to assist in B-cell maturation [439–442]. HEVs, distinguished by peripheral node addressing (PNAd) expression, allow for lymphocyte entry into TLSs, ensuring a steady flow of immune cells [442–445]. CD103⁺ tissue-resident memory T (TRM) cells, through TGF-β-induced CXCL13 production, and follicular helper T (Tfh) cells, via IL-21 secretion and inducible T-cell co-stimulator (ICOS) signaling, further increase TLS functionality and support their expansion [446–451].

TLS progression is characterized by three distinct histological stages, each denoting varying degrees of structural and functional maturity. The initial stage features dense lymphocyte clusters without clear organization, succeeded by delineated T/B-cell zones lacking germinal centers, culminating in fully developed TLSs, complete with germinal centers, FDC networks, PNAd⁺ HEVs, and plasma cell areas [452]. The maturation process is propelled by stromal cell priming and reciprocal interactions between lymphocytes and stromal cells, amplifying chemokine production and HEV development, thus facilitating TLS maturation [437, 453–459].

In oncology, TLSs are strongly associated with improved clinical outcomes in diverse cancers such as NSCLC, HNSCC, ovarian carcinoma, and colorectal cancer, especially when they are internal and fully developed [443, 460–462]. Mechanistically, TLSs underpin both humoral and cellular immune responses, supporting B-cell somatic hypermutation and antibody production [463–466]. Additionally, CD21⁺ CD86⁺ B cells within TLSs act as antigen-presenting cells, stimulating CD8⁺ T-cell activity. [467–470]. Transcription factor 7-like 1 (TCF1) ⁺ PD-1⁺ CD8⁺ T cells within TLSs exhibit stem-like durability, resistance to exhaustion, and heightened responsiveness to ICB therapy [471–473]. As such, TLS density serves as a prognostic indicator for ICB therapy efficacy, with increased maturation correlating with improved disease-free survival rates [474, 475].

Efforts to initiate TLS formation are targeted toward the administration of lymphoid-organizing factors. DCs engineered to present specific chemokines have been proven to successfully drive TLS development. For example, DCs transduced with CCL21 (Ad-CCL21-DC) significantly enhance CD8⁺ T-cell penetration in lung adenocarcinoma models and work synergistically with ICB therapies [476, 477]. Similarly, DCs modified to overexpress IL-36γ or T-box Transcription factor 21 (TBX21) promote rapid lymphocyte accumulation and TLS development in sarcoma patients, with IL-36γ-driven mechanisms leading to substantial CXCL13⁺ T-cell attraction [478, 479]. The transplantation of stromal cells, particularly PDPN⁺ CD31⁻ fibroblastic reticular cells (FRCs) from lymph nodes, leads to the establishment of functional TLSs upon implementation, which are characterized by CD69⁺ PD-1⁺ T-cell proliferation and IFN-γ-facilitated tumor control [480]. Innovative 3D scaffolds that carry both stromal vascular fraction cells and DCs offer a framework for the accelerated construction of lymphoid tissues [481].

Local administration of TLS-promoting chemokines transforms the immune landscape. The intraperitoneal injection of CXCL13 in ovarian cancer models increases the intratumoral TLS density and optimizes the positioning of CD8⁺ T-cells around TLSs, leading to prolonged survival [461]. The combination of CXCL13 and CCL21 with gemcitabine in PDAC significantly enhances TLS development and decreases tumor size [482]. Additionally, the tumor necrosis factor Superfamily member 14 (TNFSF14, or LIGHT) is instrumental in normalizing the tumor vasculature and promoting the formation of HEVs, which are essential for lymphocyte movement. Fusion proteins, such as CGKRK-LIGHT, and CAR-T cells engineered to express LIGHT facilitate TLS-like formations in glioblastoma and melanoma, thereby improving the efficacy of ICB therapies [458, 483, 484].

Furthermore, agonists that target crucial receptors involved in lymphoid tissue genesis hold considerable promise for therapeutic application. TLR agonists alter the chemokine environment: activation of TLR4 triggers SRY-box Transcription factor 4 (SOX4)-mediated CXCL13 secretion from CD4⁺ helper T cells, initiating TLS formation [485]. The intratumoral application of G100, a TLR4 agonist, has been shown to increase T-cell infiltration in the clinic [486]. Agonists for the lymphotoxin beta receptor (LTβR), such as the ACH6 monoclonal antibody, have been shown to increase the influx of T and B cells in colorectal cancer models [487]. STING agonists, such as ADU-S100, activate endothelial cells to secrete LTα, CCL19, and CXCL10, supporting the development of nonclassical TLSs and augmenting the response to anti-PD-1 therapy [488].

Metabolic interventions

Metabolic reprogramming in cancer cells establishes an immunosuppressive TME, a hallmark of immunologically “cold” tumors. Targeting cancer metabolism represents a strategic approach to reverse immunosuppression and potentiate immunotherapy, with key therapeutic strategies focusing on disrupting specific metabolic pathways to reinstate antitumor immunity [149] (Fig. 5B).

The pharmacological blockade of glucose absorption through GLUT1 inhibitors, such as BAY-876, significantly enhances anti-PD-1 therapy [489]. This synergy is evident in preclinical models of pancreatic and lung adenocarcinoma, where it effectively counters glucose-driven immunosuppression. This immunosuppression is characterized by decreased recruitment of CTLs due to lower CXCL10 levels, upregulation of PD-L1 mediated by NF-κB, and the proliferation of MDSCs triggered by M-CSF/GM-CSF secretion [489]. Similarly, targeting the glycolytic enzyme PFKFB3 with PKF-015 enhances the effectiveness of PD-1 blockade in melanoma and colorectal cancer by modifying PD-L1 expression [490]. This strategy’s viability is supported by early-phase clinical trials (e.g., NCT02044861) [491]. Furthermore, to counter lactate-mediated immunosuppression, which compromises CTL and NK cell cytotoxic capacity while encouraging Treg activity, MCT4 inhibitors such as MSC-4381 have shown potential in preclinical investigations [492]. These inhibitors decrease the accumulation of extracellular lactate, acting in concert with ICIs. In addition, the genetic or pharmacological inactivation of lactate dehydrogenase A (LDHA) has been proven to reverse PD-1 resistance in glycolytic melanomas by destabilizing Treg cells within the tumor, indicating a novel therapeutic path to mitigate immune suppression [156, 493].

In addition, interventions in lipid metabolism have gained attention for their potential in cancer therapy. Denifanstat (TVB-2640), which targets fatty acid synthase (FASN), is currently being assessed in phase II trials for its efficacy against astrocytoma and NSCLC [494]. Furthermore, the combination of the carnitine palmitoyltransferase 1 (CPT1) inhibitor etomoxir with CD47-blocking antibodies has been shown to be effective in reestablishing phagocytosis and enhancing radiosensitivity in glioblastoma models [161]. CD36 blockade has also emerged as a synergistic partner of ICIs in preclinical studies involving melanoma and pancreatic cancer models, suggesting new avenues for therapeutic intervention [195, 495].

Interventions targeting glutamine and amino acid metabolism also show promise. The glutaminase inhibitor telaglenastat (CB-839) synergizes with anti-CTLA-4 and anti-PD-1 therapies in melanoma models by dampening cancer cell metabolism and augmenting oxidative phosphorylation (OXPHOS)-dependent CTL activation [496]. Although safety profiles have been confirmed in renal cell carcinoma trials (NCT03163667, NCT03428277), therapeutic outcomes remain modest [497–499]. Pharmacological inhibition of the glutamine Transporters alanine-serine-cysteine Transporter 2​​ (ASCT2) (via V-9302) or LAT2 (via BCH) enhances phagocytosis and T-cell-mediated suppression in triple-negative breast cancer and osteosarcoma [171, 500]. Notably, serine/glycine restriction diets have emerged as a novel strategy. Deprivation of exogenous serine and glycine induces ICD, enhances CD8⁺ T-cell infiltration, and increases antigen presentation in colorectal cancer models [160]. However, this diet simultaneously promotes PD-L1 lactylation—a novel posttranslational modification that stabilizes PD-L1 by delaying lysosomal degradation, enabling immune escape. Combining serine/glycine restriction with PD-1 blockade reverses this evasion and significantly inhibits tumor growth [419]. Similarly, targeting acid ceramidase in colorectal cancer triggers ICD, mitochondrial stress, and glutathione depletion, leading to ROS-dependent T-cell activation. Acid ceramidase inhibition synergizes with anti-PD-1 therapy to reduce the number of MDSCs and Tregs while enhancing M1 macrophage polarization and cytotoxic T-cell function [501]. Targeting tryptophan catabolism with IDO1 inhibitors (e.g., epacadostat) initially showed promise alongside PD-1 blockade but failed phase III melanoma trials (ECHO-301/KEYNOTE-252), underscoring the need for innovative approaches Such as ubiquitin-specific peptidase 14 (USP14) inhibition to overcome resistance [502, 503].

Recent research underscores the pivotal role of nucleotide metabolism in regulating immunosuppression within the tumor microenvironment. In breast cancer, hypoxia-induced overexpression of ADSL leads to its phosphorylation by inhibitor of kappa B kinase beta​ (IKKβ), facilitating its translocation to the endoplasmic reticulum and subsequent fumarate production. Fumarate impedes STING activation by occupying its cyclic dinucleotide-binding domain, thus thwarting cGAMP-induced signaling pathways and bolstering immune evasion mechanisms. Interfering with the ADSL-STING interaction enhances STING oligomerization and IRF3-dependent cytokine production, culminating in increased CD8⁺ T-cell numbers. This mechanism works in concert with anti-PD-1 therapy to arrest tumor proliferation in preclinical settings [181]. Moreover, in tumors resistant to immunotherapy, such as PDAC and melanoma, the upregulation of cytidine deaminase results in the accumulation of UDP. This surplus extracellular UDP coopts immunosuppressive TAMs through P2Y6 receptor activation, obstructing CTL recruitment. Interrupting the cytidine deaminase or P2Y6 axis via genetic or pharmacological means, for instance, with cedazuridine, revokes this immunosuppressive feedback, thus restoring tumor sensitivity to anti-PD-1 therapy by promoting a more immunostimulatory TAM profile and augmenting CTL infiltration [180].

In another vein, the adenosine pathway, principally through the action of the enzymes CD39 and CD73, transforms proinflammatory ATP into the immunosuppressive adenosine, facilitating immune escape. Advances in the clinical sphere include the combination of oleclumab (anti-CD73) with durvalumab (anti-PD-L1), which has shown efficacy in phase II trials for NSCLC [504, 505]. Additionally, adenosine receptor inhibitors, such as AZD4635 and taminadenant, have potential in the treatment of solid tumors, heralding new avenues for overcoming immunosuppression in the cancer landscape [506, 507].

Despite challenges such as tumor metabolic heterogeneity and the necessity for cancer cell-selective delivery systems, metabolic modulators offer a compelling avenue to overcome the immunosuppressive TME and stimulate antitumor immunity [489, 508]. Their integration into the therapeutic arsenal against “cold” tumors holds promise for enhancing immunotherapy outcomes.

Neural–immune axis interception

The neural–immune axis is pivotal in addressing the immunosuppression observed in “cold” tumors. The infiltration of peripheral nerves into the TME results in functional interactions between neural elements, cancer cells, and immune cells [199, 200]. These interactions, which are mediated by the release of neurotransmitters and neuropeptides, play significant roles in modulating immune responses and contributing to resistance against immunotherapy. Strategies aimed at disrupting these neural–immune interactions present a promising avenue to invigorate antitumor immunity [199, 200] (Fig. 5C).

Sympathetic nervous system activity, characterized by the release of norepinephrine, leads to CD8⁺ T-cell exhaustion through the stimulation of β-ARs. Preclinical research has shown that β-blockers, such as propranolol, can synergize with ICB therapies by counteracting catecholamine-induced T-cell dysfunction [202, 203]. In models of melanoma and pancreatic cancer, β-blockers have been shown to restore the effector function of CD8⁺ T cells and induce the formation of tissue-resident memory T cells, thereby rendering previously ICB-resistant tumors susceptible [202, 203]. Clinical trials have corroborated these findings, demonstrating that perioperative β-blocker administration can enhance the effectiveness of ICB treatments in patients with melanoma and NSCLC [206, 207]. Nevertheless, retrospective analyses emphasize that the outcomes are highly context-dependent, underscoring the importance of biomarker-driven patient selection [509]. Additionally, α₂-adrenergic agonists, such as clonidine, have been shown to enhance macrophage-mediated T-cell activation and are currently being evaluated in combination with ICB therapies [510].

Sensory nociceptors contribute to immunosuppression via CGRP and substance P. The interaction of CGRP with RAMP1 receptors on CD8⁺ T cells induces exhaustion, thereby diminishing the efficacy of ICB therapies [209]. The use of FDA-approved migraine medications, such as rimegepant, to inhibit the CGRP-RAMP1 axis has been shown to reverse T-cell exhaustion, inhibit autophagy in nutrient-deficient cancer cells, and increase survival in preclinical studies [511]. Similarly, blocking substance P through neurokinin-1 receptor antagonists, such as aprepitant, has been effective in reducing metastasis in models of inflammatory breast cancer, although the outcomes vary depending on the initial state of inflammation in the TME [216].

The cholinergic and vasoactive intestinal peptide (VIP) pathways significantly influence immune responses within the tumor microenvironment. The use of VIP receptor antagonists, such as ANT308, in combination with anti-PD-1 therapy has been demonstrated to enhance T-cell activity in PDAC [512]. Moreover, novel biomarkers, including autonomic nerve density and perineural invasion, hold promise for refining patient selection for therapies targeting neural mechanisms [513, 514].

Microbiome engineering

The gut microbiome plays a crucial role in regulating antitumor immunity, transforming immunologically “cold” tumors into “hot” tumors that are more amenable to ICIs [515–519] (Fig. 5D). This transformation is linked to specific microbial compositions marked by an increase in beneficial bacteria such as Akkermansia muciniphila,Faecalibacterium prausnitzii, and Ruminococcaceae, and a decrease in harmful bacteria such as Enterocloster spp. [520–525]. These changes are associated with increased ICI effectiveness across a variety of cancers. Factors such as antibiotic use or dietary imbalances leading to dysbiosis can adversely affect this balance, impeding the infiltration of intratumoral CD8⁺ T cells and encouraging the movement of immunosuppressive T cells [525]. Strategies in microbiome engineering aim to correct these imbalances via targeted approaches.

Fecal microbiota transplantation (FMT) has offered clinical evidence of its capacity to counteract ICI resistance. Early trials in patients with melanoma have shown that FMT from individuals responsive to ICIs, or from healthy donors, can restore anti-PD-1 efficacy in 20–30% of previously unresponsive patients. This outcome aligns with an increase in advantageous Ruminococcus spp. and a decrease in Enterocloster spp. [526–528]. Moreover, the TACITO trial (NCT04758507) highlighted how FMT, when combined with pembrolizumab/axitinib, enhances 1-year progression-free Survival in patients with advanced Renal cell carcinoma. At the mechanistic level, FMT promotes Gut epithelial integrity by upregulating mucosal vascular addressin cell adhesion molecule 1 (MAdCAM1), thus impeding the migration of immunosuppressive T cells into tumors [525].

Live biotherapeutic products (LBPs) and probiotics are crucial for targeted immunomodulation. Specific bacterial consortia, Such as microbial ecosystem therapeutic 4 (MET4), and single-strain probiotics, such as Clostridium butyricum CBM588, have been shown to enhance ICI responses by generating immunomodulatory metabolites [529, 530]. Notably, in metastatic renal cell carcinoma, the combination of CBM588 with nivolumab and ipilimumab markedly increased progression-free survival, a benefit associated with the enrichment of Bifidobacterium longum in responders [530]. Similarly, supplementation with Akkermansia muciniphila supports CD8⁺ T-cell recruitment through the activation of TLR2/TLR1 and inosine-mediated adenosine A2A receptor signaling [531, 532].

Diet influences the microbiome composition and metabolite production, significantly affecting immunotherapy outcomes. Diets rich in fiber increase the levels of Ruminococcaceae and short-chain fatty acids (e.g., butyrate), triggering the activation of G-protein-coupled receptor 43 (GPR43) on T cells and enhancing the infiltration of IFN-γ-producing CD8⁺ T cells [522, 533, 534]. The use of prebiotics such as inulin intensifies this mechanism by increasing Akkermansia muciniphila populations and promoting the development of memory T-cell responses [534]. Moreover, polyphenols, exemplified by castalagin derived from camu-camu, increase Akkermansia muciniphila and Ruminococcaceae levels, producing urolithin A. This metabolite is instrumental in fostering T-cell stemness and enhancing T-cell infiltration into tumors [535, 536].

Therapeutic approaches to overcoming cytoprotection

Promotion of inflammatory cell death

Immunologically “cold” tumors employ cytoprotective strategies to shield themselves from immune detection and destruction, posing a significant challenge to the efficacy of immunotherapies [244]. These tumors enhance membrane repair capabilities, disrupt cell death signaling pathways, and impair the formation or function of immune synapses to form a robust defense mechanism against immune attacks [244]. Strategies to induce inflammatory cell death, such as pyroptosis, necroptosis, ferroptosis, and cuproptosis, transform these immunologically silent tumors into active producers of danger signals and antigens [537]. Various approaches have been developed to effectively trigger inflammatory cell death, including oncolytic virotherapy (e.g., T-VEC), photodynamic therapy (PDT), microwave thermal ablation, selected chemotherapeutics (e.g., oxaliplatin), and oncolytic peptides (e.g., LTX-315) [538–544]. This approach initiates a self-reinforcing cycle of both innate and adaptive immune responses within the TME, effectively targeting and breaking down tumor mechanisms for cytoprotection and evasion from the immune system [244, 537] (Fig. 6A).

Fig. 6.

Therapeutic approaches targeting cytoprotection mechanisms in “cold” tumors. Therapeutic approaches aim to overcome cytoprotection in “cold” tumors, which are characterized by immune synapse defects, dysregulated cell death signaling, and resistance to inflammatory cell death, leading to low immunogenicity. A The introduction of novel inflammatory cell death inducers, such as pyroptosis activators that cleave gasdermin proteins to form membrane pores, challenges this evasive phenotype. Conversely, “hot” tumors exhibit inherent immunogenicity due to tumor antigens, damage-associated molecular patterns (DAMPs), and inflammatory factors, promoting dendritic cell (DC) maturation and T-cell activation, despite the survival of cytoprotective mechanisms via the endosomal sorting complex required for transport (ESCRT) activity. B The application of ESCRT inhibitors enhances pyroptosis and apoptosis by blocking membrane repair during gasdermin pore formation. This dual approach leverages inherent immunogenicity and impedes ESCRT-mediated cytoprotection, driving caspase-1/3-mediated activation of gasdermins, thus promoting extensive lytic cell death. The consequent release of DAMPs further stimulates antitumor immunity, undermining residual cytoprotection and enhancing DC and T-cell interaction. These targeted strategies exploit tumor subtype differences to eliminate cytoprotective barriers, fostering a robust inflammatory cell death transition throughout the immunological spectrum

Pyroptosis disrupts tumor cytoprotection through the action of GSDMs, which form pores in the plasma membrane upon activation by caspases or granzymes [272, 545]. This event allows the release of proinflammatory mediators such as ​​high mobility group box 1 (HMGB1), ATP, and IL-1β, stimulating DCs to facilitate tumor antigen presentation to CD8⁺ T cells [272, 545]. Interestingly, pyroptosis in as few as 15% of tumor cells can elicit a robust antitumor immune response, effectively eliminating both primary and metastatic tumors and enhancing the efficacy of ICIs such as anti-PD-1 [546]. The regulation of pyroptosis involves complex posttranslational modifications that affect the function and interaction of key molecules, including GSDMs and inflammasome components [547]. For example, the oncolytic parapoxvirus ORFV triggers caspase-3-mediated pyroptosis by stabilizing GSDME in tumor cells with low GSDME expression, working in concert with anti-PD-1 therapy to transform “cold” tumors into “hot” tumors through increased CD8⁺ T-cell infiltration and IFN-γ production [548].

Ferroptosis is a form of iron-dependent regulated cell death that specifically undermines cytoprotective metabolic pathways [549, 550]. This process is initiated by the deactivation of GPX4, leading to the accumulation of lipid hydroperoxides and resulting in the inflammatory death of cells. This type of cell death has a complex impact on the TME [549]. Early-stage ferroptosis releases DAMPs, such as oxidized phospholipids and mitochondrial DNA. These DAMPs then activate DCs, enhancing antigen presentation and priming CD8⁺ T cells, thereby increasing the immune system’s ability to combat tumors [551]. Targeting specific enzymes such as CPT1A has been shown to initiate ferroptosis in lung cancer cells, working in synergy with ICB therapies to increase T-cell activity and promote tumor regression [552]. Conversely, excessive or sustained ferroptosis may activate feedback loops that dampen immune responses [138, 549, 553]. Furthermore, CD8⁺ T cells are Susceptible to ferroptosis, with deficiencies in phospholipid phosphatase 1 (PLPP1) leading to an accumulation of polyunsaturated fatty acids and increased ferroptosis risk [554]. This sensitivity is compounded by PD-1 signaling, which downregulates PLPP1 expression, perpetuating a cycle of T-cell impairment and death through ferroptosis [554]. These findings emphasize the importance of carefully targeting ferroptosis to exploit its tumor-suppressing capabilities while avoiding the strengthening of immune-suppressive pathways in cancer treatment.

Cuproptosis, a novel form of regulated cell death, utilizes copper-induced proteotoxicity to circumvent metal ion chelation defenses [555, 556]. This process occurs via the aggregation of lipoylated proteins in mitochondria, such as dihydrolipoamide S-acetyltransferase (DLAT), leading to proteotoxic stress [555].​​ This process is regulated by ferredoxin 1 (FDX1), which catalyzes the reduction of Cu²⁺ to Cu⁺, enhancing the lipoylation of mitochondrial enzymes and inducing lipid peroxidation [555]. This disruption of the tricarboxylic acid (TCA) cycle and destabilization of iron‒sulfur clusters cause the release of DAMPs such as mitochondrial DNA and HMGB1, activating immune signaling pathways in DCs [557]. Copper ionophores, such as elesclomol, trigger significant cuproptosis, promoting DC maturation and CD8⁺ T-cell recruitment, thereby increasing the effectiveness of anti-PD-1 therapies through increased IFN-γ signaling [558, 559]. However, susceptibility to cuproptosis varies among cancer types and is dependent on the expression levels of copper transporters and lipoylation enzymes. Combining copper ionophores with ATPase copper transporting alpha​​ (ATP7A)-targeted small interfering RNA (siRNA) enhances cuproptosis in breast cancer, activating adaptive immunity via DC and T-cell engagement [560]. Furthermore, cancer cells can evade immune detection by epigenetically silencing FDX1. DNMT inhibitors, Such as 5-azacytidine, can reverse this silencing, increasing the responsiveness of tumors to cuproptosis inducers and immunotherapy [561].​​

PANoptosis, a unified cell death pathway that combines pyroptosis, necroptosis, and apoptosis, disrupts complex cytoprotective mechanisms in cancer cells [562, 563].​​ Initiated by factors such as oncolytic viruses, IFN-γ/TNF-α cytokines, or chemotherapeutic agents, it triggers Z-DNA binding protein 1 (ZBP1) and caspase-8 activation [564]. This leads to the formation of GSDMD pores, MLKL phosphorylation, and mitochondrial apoptosis. The process releases high levels of DAMPs, including HMGB1 and ATP, inducing a potent “cytokine storm” that activates DCs and CTLs [564]. This not only targets tumor-initiating cells but also promotes long-term immune memory by enhancing antigen presentation by DCs [564]. The use of oncolytic viruses, such as HSV-1, has shown promising results in inducing PANoptosis and suppressing tumor growth, especially when combined with Fusobacterium nucleatum outer membrane vesicles [565]. Nevertheless, the risk of excessive tissue damage and autoimmunity calls for careful control and targeted delivery strategies.

The strategic induction of inflammatory cell death represents a significant advancement in targeting cytoprotective mechanisms that preserve immunologically “cold” tumors [244]. This approach counters the evolution of resistance in tumors, especially the epigenetic repression of death pathways during immunoediting. Copper-based nanoinducers, for example, simultaneously initiate disulfidptosis and pyroptosis, undermining membrane and metabolic stability to bolster immune infiltration and circumvent cytoprotective measures [566, 567]. Importantly, combining these nanoinducers with epigenetic modulators, such as DNMT inhibitors, effectively reinstates the expression of crucial death effectors. Additionally, the integration of checkpoint inhibitors enhances the eradication of compromised tumors by T cells.

Targeting the ESCRT machinery

Cancer cells circumvent immunogenic cell death by harnessing the ESCRT-mediated membrane repair mechanism to withstand the perforin-induced pore formation initiated by CD8⁺ T-cell attacks [268]. Targeting and inhibiting this repair pathway allows for the sensitization of tumors to T-cell-mediated elimination, attributable to ongoing membrane damage and eventual cell death [268, 278, 568] (Fig. 6B). A pivotal approach involves disrupting the Ca²⁺-dependent ESCRT-III assembly, critical for repairing perforated membranes. Pharmacological blockade of ESCRT-III, thereby preventing gasdermin pore removal, prolongs pyroptosis and boosts inflammatory responses. Preclinical evidence suggests that combining ESCRT inhibitors with ICB significantly enhances T-cell penetration and induces tumor shrinkage [277, 278]. Another tactic leverages synthetic lethality within the ESCRT pathway. Frequently, cancer cells exhibit incidental losses of either VPS4A (located on chromosome 16q) or VPS4B (on chromosome 18q). Although the loss of a single paralog is manageable due to inherent redundancy, targeting the remaining VPS4 paralog can provoke synthetic lethality, leading to catastrophic sterile inflammation driven by NF-κB, ICD, and macrophage-led tumor eradication. Thus, the discovery and application of paralog-selective inhibitors for VPS4A or VPS4B offer a pathway to selectively eradicate tumor cells while preserving healthy tissues [285, 286].

ESCRT-dependent immunosuppressive vesicle release offers additional therapeutic opportunities. The phosphorylation of ESCRT-0’s HRS component by ERK facilitates PD-L1 enrichment in exosomes, thoroughly inhibiting CD8⁺ T-cell infiltration [282]. Similarly, the absence of VPS37A, which is crucial for autophagosome closure, leads to increased PD-L1 levels through STAT3 activation [569]. Conversely, ALIX expression is negatively correlated with the presence of PD-L1 on the surface of EGFR-predominant cancers [570]. Interrupting these specific ESCRT-driven processes might effectively counteract the “cytoprotection” phenotype. Furthermore, ESCRT-III’s CHMP2A protein significantly increases the release of immunosuppressive vesicles containing MICA/MICB ligands, triggering apoptosis in natural killer cells. These findings suggest that suggests CHMP2A is a viable candidate for reviving innate immune monitoring [283].

Overall, emerging strategies targeting ESCRT function, such as adjusting local calcium signaling or temporarily focusing on ESCRT-III/VPS4, show promise in making “cold” tumors more susceptible to immunotherapy [273, 277]. Moreover, the identification of ESCRT-associated biomarkers, such as increased CHMP2A expression or unique PD-L1 exosomal profiles, could enhance patient selection for these therapies [282, 283]. Ultimately, pharmacologically inducing pyroptosis or impairing ESCRT-dependent membrane repair represents a transformative avenue for potentiating antitumor immunity and converting “cold” tumors into “hot”, immune-responsive microenvironments.

Biomaterials in immunoengineering: innovations for modulating the “cold” tumor microenvironment

Biomaterials are revolutionizing the remodeling of the TME via precision engineering. These approaches, which facilitate the precise delivery of immunomodulators and alter the immunosuppressive milieu, overcome the fundamental challenges posed by conventional therapies in addressing immunologically “cold” tumors [571, 572] (Table 3). Passive targeting capitalizes on the enhanced permeability and retention (EPR) effect, allowing nanomaterials (50–150 nm) to accumulate selectively in tumor sites characterized by leaky vasculature and deficient lymphatic drainage [573]. Active targeting involves the use of ligands, such as antibodies, peptides, and folate, conjugated to biomaterials to specifically bind to receptors (e.g., CD44, EGFR, and PD-L1) that are overexpressed on tumor or immune cells [574]. This enhances both cellular uptake and targeting specificity. Stimuli-responsive targeting employs materials that are activated by signals within the TME (e.g., pH, ROS, enzymes) or by external stimuli (e.g., light, ultrasound), facilitating the controlled release of immunomodulators or initiating catalytic reactions that mitigate immunosuppression [575]. Biomimetic targeting adopts natural mechanisms of homing and immune evasion by cloaking nanomaterials with cell membranes (e.g., red blood cells, macrophages) or by using engineered microbes, thereby increasing tumor targeting efficiency [576]. Strategies leveraging polymeric platforms, inorganic nanomaterials, biomimetic technologies, and carrier-free nanodrugs enhance T-cell infiltration, thus transforming “cold” tumors into “hot” tumors that are more amenable to therapy. The following sections delve into the cutting-edge materials driving this transformative shift (Fig. 7).

Table 3.

Examples of biomaterials for modulating the “cold” tumor microenvironment

Classification	biomaterials	Therapy	Mechanisms	​​Advantages	Cancer type	Refs
Nanogels	CFe/βCP + Cis hydrogel	CDT	Induces ferroptosis and macrophage phenotype reprogramming, enhances CD8+ T-cell infiltration	Long-term local retention, injectable and formulation stability, synergistic efficacy	Colorectal cancer	[579]
	RPNPs/Ce6-CAT/PEGDA	PDT	Persistent hypoxia reduction, antigen-adjuvant synergy, enhanced CD8+ T-cell infiltration	​​Single-administration multi-round therapy, spatiotemporal control, oxygen-economizing design	Breast cancer, Colorectal cancer	[580]
	mPEG-b-PELG	–	Induces a pro-immunogenic microenvironment, activates and proliferates CD8+ T cells	​​Sustained and controlled delivery, potential combination with anti-PD-L1, chirality-dependent customization	Melanoma	[581]
	αCD47/Ce6@PPG	PDT	ROS-mediated immune-stimulation, immune memory induction, amplified phagocytosis and antigen presentation	​​ROS protection, controlled release, Raman traceability, long-term efficacy	​​Breast cancer	[583]
	PNP	–	Prolonged cytokine (IL-15) retention, controlled CAR-T motility, T-cell phenotype reprogramming	Dose-sparing, simplified manufacturing, biocompatibility	Medulloblastoma​​	[584]
Polymer prodrugs	NP/OXA-ASP2	–	Induces immunogenic cell death, inhibits glycolysis, reduces MDSC and Tregs	GSH/ascorbate dual reduction-responsiveness, synergistic drug ratio	Colorectal cancer	[586]
	LINC	NIR	Induces immunogenic cell death, inhibits IDO-1 activity, enhances CD8+ T-cell infiltration	Laser-triggered dePEGylation, high drug loading, synergistic immunotherapy	Breast cancer	[587]
	BCPN	–	Induces immunogenic cell death, inhibits IDO-1, enhances CD8+ T-cell-to-Treg ratio	​​Acid-triggered charge reversal, enhanced tumor penetration and uptake, high specificity	Breast cancer, Colorectal cancer	[588]
	PtIV-functionalized NPs	NIR	Induces immunogenic cell death, enhances CD8+ T-cell infiltration, DNA damage, ROS-induced cell damage	Deep tissue penetration, tumor and nuclear targeting, stimuli-responsive activation	Breast cancer	[589]
Mesoporous material	PMSN	SDT	Induces pyroptosis, synergizes PD-L1 blockade, recruits CD8+ T cells, NK cells, DCs, and M1-polarized macrophages	Integrin αvβ3-specific tumor targeting, ultrasound penetration	Liver cancer	[591]
	CMZM	CDT/PDT	Disrupts ROS homeostasis, polarizes M2-to-M1 macrophages, activates DCs	Self-supplies H2O2/O2, depletes GSH, comprehensive ROS generation	Breast cancer	[594]
	TCN-PPDA-COF	PTT/PDT	Induces immunogenic cell death, inhibits VEGFR2, enhances CD4+ and CD8+ T-cell infiltration	​​Enhanced drug delivery and specificity, angiogenesis-targeted resistance reversal	Breast cancer, Melanoma	[598]
Metal-Based nanomaterials	Mn3O4 NPs	–	Activates the STING pathway, innate immune activation, enhances mRNA translation, antigen presentation	Enhances adaptive immunity via antigen expression and robustly activates innate immunity via STING	HPV-positive oral cancer	[601]
	ZCPO@HA NPs	–	Induces ferroptosis and cuproptosis, activates the cGAS-STING pathway	Dual cell death induction and immune activation, acid-triggered decomposition	Breast cancer	[608]
Quantum dots	QD-Cat-RGD nanoprobes	NIR/Radiotherapy	Induces immunogenic cell death, alleviates hypoxia, causes DNA damage	Combines real-time NIR-IIb imaging with therapy, RGD mediates tumor-specific accumulation	Breast cancer, Melanoma	[610]
	FAQD	CDT/SDT	Alleviates hypoxia, amplifies ROS, enhances CD4+ and CD8+ T-cell infiltration	Controlled catalytic activity, switchable Fe(III)/Fe(II) cycling, peptide-driven self-assembly, multimodal synergy	Cervical cancer, Breast cancer, Melanoma	[612]
Cell membrane-camouflaged nanocarriers	mEHGZ	–	Induces immunogenic cell death, synergy with checkpoint inhibition	Tumor cell membrane coating enables homologous targeting	Breast cancer	[620]
	M@BTO NPs	–	Induces immunogenic cell death, alleviates hypoxia, enhances CD4+ and CD8+ T-cell infiltration	MMP2-responsive αPD-L1 masking prevents toxicity, prolonged circulation and binding enhance tumor accumulation	​Melanoma	[622]
Extracellular vesicle mimetics	gETL NPs	–	Repolarizes macrophages to M1 phenotype, enhances phagocytosis via CD47-SIRPα interaction	Enhanced drug delivery and targeting, temperature-triggered release	Peritoneal cancer, Colorectal cancer	[629]
	miR497/TP-HE NPs	–	Inhibits PI3K/AKT/mTOR pathway, potentiates ROS, repolarizes macrophages	Homologous and active cRGD-integrin targeting enhances tumor accumulation	Ovarian cancer	[631]
Carrier-free nanodrugs	FD/FM@siTOX NPs	–	Induces immunogenic cell death, reduces T-cell exhaustion, enhances CD8+ T-cell infiltration	Eliminates carrier-related toxicity, high drug loading, fluorination strategy	Liver cancer	[636]
	2’F-CpG/Obsl1 NPs	–	Activates TLR9-mediated innate immunity, enhances antigen uptake and cross-presentation	Avoids carrier toxicity, fluorination strategy	Melanoma​	[633]

Fig.7.

Emerging biomaterial strategies for converting immunologically “cold” tumors into immunotherapy-responsive “hot” tumors. Engineered biomaterial platforms, which include nanogels, polymer prodrugs, mesoporous materials, metal-based nanomaterials, bioinspired systems, quantum dots, and carrier-free nanodrugs, have been meticulously designed to achieve precise tumor targeting. These platforms utilize various mechanisms for targeting, including passive accumulation through the enhanced permeability and retention (EPR) effect, bioactive ligand‒receptor interactions for more specific targeting, stimuli-responsive targeting that responds to changes in the tumor microenvironment, and biomimetic recognition that mimics natural biological processes. Through these mechanisms, platforms are capable of remodeling the immunosuppressive tumor microenvironment (TME). This is achieved by inducing immunogenic cell death (ICD), promoting the presentation of tumor antigens, enhancing the infiltration of T cells, inducing tertiary lymphoid structure (TLS) formation, and working synergistically with existing immunotherapies. Together, these actions transform immunologically “cold” tumors into “hot” microenvironments characterized by T-cell inflammation, increasing their susceptibility to checkpoint blockade therapies

Polymeric platforms

Nanogels

Nanogels synergize the hydrating properties of hydrogels with the pharmacokinetic benefits of nanoparticles via crosslinked polymeric networks [577]. This design ensures the precise delivery of immunomodulatory agents, with the aim of altering the immunosuppressive TME of “cold” tumors. Stimuli-responsive systems, such as pH, ROS, or enzymes, allow for targeted release within the TME, reducing systemic effects and enhancing localized action [578].

A pivotal approach involves the cotransportation of therapeutics that induce ICD and DC activation. For example, hydrogels that integrate β-cyclodextrin-grafted polymer brushes with carbon nanozymes and cisplatin achieve extended retention within the tumor, yielding combined antitumor effects. Here, cisplatin increases cytotoxicity, whereas carbon nanozymes trigger ROS generation through Fenton-like reactions, initiating ferroptosis. Concurrently, the hydrogel formulation reorients TAMs toward a proinflammatory M1 phenotype and enhances CD8⁺ T-cell penetration, effectively transforming the TME [579]. Additionally, light-activated nanogels co-delivering catalase and chlorin e6 (Ce6) target hypoxia by converting H₂O₂ into oxygen, thereby intensifying PDT effectiveness and facilitating DC maturation [580].

Other methods directly reshape the immune landscape. For example, chiral polypeptide hydrogels loaded with antigens and adjuvants adjust the immune landscape, decreasing the number of PD-L1⁺ DCs and exhausted T cells and increasing antigen-specific T-cell priming and memory responses [581]. Hybrid hydrogels incorporating mesoporous bioactive glass nanoparticles counteract acidity in the TME, bolstering NK cell activity and preventing hepatocellular carcinoma recurrence [582].

To address adaptive immune resistance, ROS-sensitive hydrogels strategically deliver ICIs to the TME, combining therapies such as Ce6 and anti-CD47 antibodies to increase tumor cell phagocytosis and synergize with PDT, preventing metastasis in aggressive cancer models [583]. Moreover, thermosensitive hydrogels designed for sequential drug release further sustain key therapies, such as IL-15, to support CAR-T cell efficacy while reducing the risk of cytokine release syndrome [584].

Polymer prodrugs

Polymer prodrugs harness the distinct pathological characteristics of the TME, including altered redox balance, acidic pH, and elevated enzymatic activity, to accurately control drug activation [585]. By attaching therapeutic agents to stimuli-responsive polymeric carriers, these prodrugs reduce nontargeted toxicity and enhance drug availability within tumors, thereby disturbing immunosuppressive pathways and initiating antitumor immune responses [586]. A significant strategy employs ROS-responsive polymers to trigger ICD. Innovations include a light-activated nanocarrier system, utilizing a thioketal-based polymer prodrug coloaded with β-lapachone (a ROS enhancer) and an IDO-1 inhibitor [587]. Exposure to near-infrared (NIR) light generates exogenous ROS, facilitating autonomous drug deployment. This reaction effectively eradicates tumor cells while also encouraging calreticulin display and HMGB1 release, leading to DC maturation and increased CD8⁺ T-cell presence in triple-negative breast cancer models [587].

Enhanced specificity is achieved through dual-stimuli-responsive systems. Advances include pH/GSH-responsive nanoparticles formulated from a polymer conjugate delivering oxaliplatin and the IDO-1 inhibitor NLG919 [588]. In the acidic TME, aconitate linkers are hydrolyzed, revealing cationic surfaces that improve tumor penetration, whereas intracellular GSH converts Pt(IV) prodrugs into lethal Pt(II) forms and releases NLG919. This dual mechanism effectively counters immunosuppression by eliminating Tregs and activating effector T cells, drastically reducing tumor growth and lung metastasis in breast and colorectal cancer models [588].

Furthermore, photodynamic immunotherapy is effectively integrated with chemotherapy in polymer prodrug formulations. Polyplatin nanoparticles, which are composed of aggregation-induced emission (AIE) photosensitizers and Pt(IV)-conjugated polymers, initiate NIR-induced photoreduction to Pt(II), simultaneously generating singlet oxygen [589]. This synergistic effect enhances ICD and DNA damage, promoting DC maturation and activating cytotoxic T lymphocytes.

Inorganic nanomaterials

Mesoporous materials

Mesoporous materials, distinguished by their expansive surface area, adjustable pore dimensions, and easy surface modification, facilitate efficient encapsulation and directed transport of immunomodulatory agents to tumor locations [590]. Mesoporous silica nanoparticles (MSNs) exploit the EPR effect for passive tumor targeting, while surface modifications such as polyethylene glycol (PEG) enhancement extend circulation times and bolster tumor penetration. For example, PEG-coated MSNs attached to phthalocyanine amplify the effects of sonodynamic therapy (SDT), promoting pyroptosis and widespread antitumor immunity, especially when used in conjunction with PD-L1 inhibitors [591]. Furthermore, actively targeted MSNs, functionalized with iRGD peptides, ensure deep tumor penetration and precise delivery of catalytic ions, such as Mn²⁺, thereby activating the cGAS-STING axis, promoting DC maturation, and facilitating T-cell infiltration [592].

Metal–organic frameworks (MOFs), combining catalytic metal nodes with organic linkers, support diverse therapies. Porphyrin-based MOFs, acting as photosensitizers for PDT, produce singlet oxygen upon light exposure, initiating ICD and antigen release [593]. The porous architecture of these materials prevents photosensitizer self-quenching, increasing ROS production even under low oxygen levels [594]. Additionally, MOFs containing high atomic number elements such as Hf and Au serve as radiosensitizers for radiodynamic therapy (RDT), transforming X-rays into ROS to heighten oxidative stress and antitumor immune responses [595, 596]. Bimetallic Au@Mn-MOF heterojunctions enhance radiocatalytic immunotherapy by neutralizing radicals and boosting immune cell-mediated destruction [597].

Covalent organic frameworks (COFs) reveal superior photodynamic capabilities due to their broad π-conjugated systems. Porphyrin-infused COFs trigger ROS production and ICD upon light exposure [598]. When administered through microneedles, these COF-based photosensitizers permit accurate tumor targeting, increasing the success of phototherapy and immunotherapy against persistent tumors [598]. COFs respond to hypoxia, integrate azobenzene linkers, discharge immunoadjuvants in low-oxygen areas, and cooperate with PDT to induce pyroptosis and a potent systemic antitumor response [599].

Metal-based nanomaterials

Metal-based nanomaterials leverage their distinct physicochemical properties to breach the immunosuppressive fortifications of “cold” tumors [600]. When encapsulated within lipid nanoparticles (MnLNPs), manganese oxide nanoparticles (Mn₃O₄ NPs) increase the effectiveness of mRNA vaccines by producing oxygen in oxygen-poor microenvironments, thus activating the STING axis in DCs and enhancing antigen presentation [601]. Iron oxide nanoparticles, such as ferumoxytol, transform TAMs from an immunosuppressive M2 phenotype into a proinflammatory M1 phenotype, increasing T-cell infiltration and circumventing immune evasion [602, 603]. Glycocondensation-modified gold nanoparticles (AuNPs) increase tumor immunogenicity and the effectiveness of CTL responses in lung cancer models [604, 605]. Hafnium oxide nanoparticles (NBTXR3) serve as radioenhancers, intensifying radiotherapy-induced DNA damage and, in conjunction with anti-PD-1 therapy, addressing ICB resistance in clinical settings [606]. Copper-based nanomaterials utilize cuproptosis and ferroptosis processes under NIR-II irradiation, initiating ICD and broad antitumor immunity in breast cancer models [607, 608]. These approaches jointly transform the TME, metamorphosing immunologically “cold” tumors into “hot” targets susceptible to immunotherapy.

Quantum dots

Quantum dots (QDs), semiconductor nanomaterials known for their adjustable optoelectronic attributes, enable precisely guided tumor targeting through real-time imaging, thereby facilitating targeted interventions within the immunosuppressive TME [609, 610]. Beyond their diagnostic utility, QDs are instrumental in transforming immunologically “cold” tumors into “hot” tumors by initiating pathways that lead to ICD. Notably, CdSe/ZnS QDs are capable of producing cytotoxic ROS And causing mitochondrial damage, which in turn activates the NLR family pyrin domain-containing protein 3 (NLRP3) inflammasome through mitochondrial ROS and Ca²⁺ influx, thereby inducing pyroptosis [611]. This process results in the release of DAMPs, attracting DCs and T lymphocytes, and thus, promoting antitumor immunity [610].

Recent developments have led to multifunctional QD platforms that merge catalytic, sonodynamic, and immunomodulatory functions. Iron-silver-modified ZnSe QDs, featuring atomically dispersed Fe on their surface (FAQDs), utilize ultrasound to control valence switching: Fe³⁺ catalyzes the decomposition of H₂O₂ to mitigate hypoxia, whereas ultrasound-converted Fe²⁺ stimulates Fenton reactions, generating hydroxyl radicals [612]. The incorporation of silver enhances the production of singlet oxygen by hindering exciton recombination. When combined with matrix metalloproteinase (MMP)-responsive self-assembling peptides for improved tumor retention and selenium-induced immune activation, FAQD promotes a synergistic effect of primary tumor removal and systemic T-cell activation against metastases [612]. Similarly, nucleus-targeted graphene QDs, modified with TAT peptides, cause selective DNA damage through π–π stacking, leveraging dysregulated DNA repair mechanisms to counteract immune evasion [613].

Bioinspired systems

Cell membrane-camouflaged nanocarriers

Cell membrane-camouflaged nanocarriers (CMNPs) are at the forefront of biomimetic immunoengineering, transforming immunologically “cold” tumors into “hot” microenvironments by mimicking natural membrane components [614]. These nanoplatforms evade immune detection by displaying “self” identifiers, such as CD47 from red blood cell (RBC) membranes, which bind to SIRPα on phagocytes. This interaction suppresses “eat me” signals, enabling extended circulation and enhanced tumor targeting through the EPR effect [615, 616]. For instance, RBC membrane-coated nanoparticles deliver immunomodulatory agents such as low-molecular-weight curdlan (LCUR) to shift TAMs from the immunosuppressive M2 phenotype toward the antitumor M1 phenotype. This transformation promotes the infiltration and activity of CTLs, reversing T-cell exclusion in “cold” tumors [617].

Platelet membrane-coated nanoparticles utilize adhesive proteins, such as P-selectin, to target tumor cells and remodel the immunosuppressive TME. The delivery of compounds such as sulfasalazine or TLR agonists triggers cancer cell ferroptosis and DC maturation. This process results in increased CD8⁺ T-cell infiltration and reduces the number of Tregs [618]. Cancer cell membrane-coated nanoparticles (CCNPs) employ homotypic targeting to bring tumor-associated antigens and adjuvants directly to tumors, serving as in situ vaccines [619]. When combined with PDT, these nanoparticles induce ICD, DC activation, and cross-presentation to T cells. The outcome is increased systemic antitumor immunity and the conversion of immune-deserted tumors into immunogenic sites [620].

Further enhancing the conversion from “cold” to “hot” tumors, genetically engineered CMNPs with SIRPα-variant-expressing membranes inhibit CD47 “don’t eat me” signals, improving macrophage phagocytosis and T-cell activation [621]. MMP2-responsive membranes displaying anti-PD-L1 antibodies target PD-L1 within the TME, countering T-cell fatigue and, when paired with ROS-generating cores, increasing the infiltration of granzyme B⁺ CD8⁺ T cells. This multifaceted approach ignites immune responses in “cold” tumors [622]. The incorporation of hybrid membranes, such as those used for cancer cell/macrophage fusion, adds another layer of targeting and immunomodulation, establishing CMNPs as adaptable instruments for TME remodeling [623].

Extracellular vesicle mimetics

Extracellular vesicle (EV) mimetics have emerged as a cutting-edge category of bioinspired nanoplatforms, crafted to surpass the constraints of natural exosomes while harnessing their inherent biocompatibility and targeting potential for immunotherapy in “cold” tumors [624]. These platforms are broadly divided into modified natural vesicles (either chemically engineered or biologically engineered exosomes) and synthetic exosome analogs (such as liposomes or hybrid vesicles), all aimed at bolstering tumor-directed immunomodulation [625]. The key to their functionality is surface engineering to avoid immune detection and enhance the affinity for immunosuppressive regions within cold tumors. Techniques include attaching CD47 to exosomes through carbodiimide binding or click chemistry to prevent macrophage-mediated phagocytosis, thus extending their circulation and enabling tumor localization [626, 627]. Furthermore, equipping EVs with specific targeting moieties (such as RGD peptides or folate) facilitates their direct delivery to TAMs or DCs, thereby interrupting immunosuppressive circuits [628].

Innovative hybrid configurations, including the fusion of exosomes with liposomes or the integration of inorganic nanoparticles with EVs, significantly enhance the potency of immunotherapy. For example, gene-augmented exosome‒liposome hybrids deliver docetaxel and GM-CSF concurrently ​​to​​ activate T cells ​​and induce ICD​​ in metastatic peritoneal carcinoma models [629]. Similarly, liposomes mimicking macrophage-derived exosomes carrying paclitaxel and adorning sigma receptor-specific ligands mitigated lung metastases by reorienting TAMs and increasing T-cell infiltration [630]. Additionally, aptamer-decorated EV mimetics encoding siRNA (such as siS100A4) or miRNA-497 have been leveraged to mute immunosuppressive genes in cancer cells, overturning chemoresistance and igniting antitumor immune responses [631, 632].

Carrier-free nanodrugs

Carrier-free nanodrugs, consisting exclusively or largely of active pharmaceutical ingredients (APIs), embody a revolutionary approach to transforming immunologically “cold” tumors into “hot”, immune-responsive territories [633]. Owing to their exceedingly high drug loading capacity, elimination of exogenous carrier-associated toxicity, and streamlined manufacturing, these nanodrugs significantly benefit from breaching the “cold” TME’s defenses. By utilizing supramolecular self-assembly driven by noncovalent forces such as hydrophobic interactions, π–π stacking, and hydrogen bonding, these nanodrugs effectively amalgamate various therapeutic agents indispensable for triggering immune responses [634].

At the heart of this methodology is the simultaneous delivery of ICD inducers alongside immunomodulators. For example, carrier-free nanoassemblies that combine chemotherapeutic compounds such as doxorubicin (DOX) or cytolytic peptides, such as melittin, are renowned for their ability to induce ICD and liberate tumor neoantigens, with molecules that directly neutralize immunosuppression [635, 636]. Notably, recent advancements have yielded a carrier-free formulation (FD/FM@siTOX NPs) that administers DOX, melittin, and anti-TOX siRNA (siTOX). In combination, DOX and melittin significantly promote ICD and bolster CD8⁺ T-cell penetration into tumors, whereas siTOX diminishes TOX expression, thus alleviating CD8⁺ T-cell fatigue and generating a formidable antitumor immune assault that effectively reconditions “cold” tumors [636]. Likewise, carrier-free nanodrugs integrating TLR agonists (such as R848 and R837) or inhibitors of immunosuppressive circuits can directly incite innate immunity or obstruct critical immunosuppressive pathways within the TME, offering a substantial increase in stability and delivery efficacy over solitary molecules [637, 638].

Structural refinement further enhances their therapeutic potential for treating “cold” tumors. Adorning carrier-free nanodrugs with targeting entities such as antibodies or aptamers or employing biomimetic coatings such as cell membranes may increase their delivery precision and retention at tumor locations [639]. For example, nanovaccines formulated purely from melanoma neoantigen peptide (ObsII) And a 2’-fluorinated CpG adjuvant (2’F-CpG) secured exceptional antigen payloads and spurred potent, enduring antigen-specific T-cell responses in melanoma prophylaxis and therapy, highlighting the immense potential of carrier-free platforms for targeted immune activation [640].

Perspectives

The transformation of immunologically “cold” tumors into “hot” environments represents a pivotal challenge in the field of cancer immunotherapy. The emergence of spatial multi-omics and single-cell technologies has profoundly expanded our grasp of the TME, offering insights into its complex dynamics with unparalleled resolution (Fig. 8). Through the application of spatial transcriptomics, multiplexed imaging, and TCR/B-cell receptor (BCR) repertoire analysis, researchers can now concurrently map the cellular makeup, gene activity, protein distribution, and diversity of immune receptors within undisturbed tissues [641–645]. These methodologies reveal critical determinants of immunotherapy resistance, encompassing intricate cell interactions, immunosuppressive zones, and the operational conditions of TILs. Spatial omics is adept at identifying prognostic elements such as immune deserts or TLSs, whereas single-cell TCR sequencing provides a window into T-cell clonal dynamics and fatigue [459, 646–648]. The synthesis of these comprehensive datasets through artificial intelligence (AI) and machine learning is essential for dissecting TME heterogeneity. Predictive modeling pinpoints vital areas of stromal–immune dialog or metabolic hindrances [649, 650]. Furthermore, patient-derived organoids (PDOs) integrated with patient immune cells bridge the gap between discovery and clinical validation, facilitating the evaluation of neoantigen-specific TCRs or combination therapies [651, 652].

Fig. 8.

Perspectives and challenges in transforming immunologically “cold” tumors. Recent advancements in spatial multi-omics and single-cell technologies have enabled unprecedented resolution in dissecting the immunosuppressive tumor microenvironment (TME), allowing for detailed mapping of cellular interactions, metabolic barriers, and immune exclusion zones. Artificial intelligence plays a crucial role in integrating these multidimensional datasets, facilitating the identification of critical stromal–immune crosstalk nodes and predicting therapeutic vulnerabilities. Patient-derived organoids, coupled with autologous immune cells, serve as valuable ex vivo platforms for validating combination strategies. Beyond the TME, systemic host factors profoundly influence antitumor immunity. Innovative biomaterial-based approaches now enable precise spatiotemporal remodeling of metabolic, neural, and microbial ecosystems. The integration of these diverse approaches has accelerated the development of spatially targeted interventions aimed at dismantling immunosuppressive networks and reinvigorating antitumor immunity in immunotherapy-resistant cancers. However, successful clinical translation of these strategies necessitates biomarker-driven patient stratification. This comprehensive approach promises to advance our understanding and treatment of “cold” tumors, potentially transforming the landscape of cancer immunotherapy. Nevertheless, challenges remain in translating these complex, multifaceted strategies into clinically viable treatments, underscoring the need for continued research and interdisciplinary collaboration

Beyond the TME, systemic host factors fundamentally shape antitumor immunity. Obesity can lead to chronic inflammation and support immunosuppressive niches, whereas caloric restriction might enhance T-cell efficacy and checkpoint blockade effectiveness [653–656]. Evidence increasingly indicates that the neural–immune axis acts as a systemic modulator, potentially enhancing immunotherapy outcomes through β-adrenergic blockade [206, 207]. The gut–tumor axis is important, as specific microbiota can either strengthen immune responses or promote immunosuppression and angiogenesis [516, 657]. Intratumoral microbes can produce metabolites that directly inhibit T-cell function [658, 659]. Endocrine dysregulation further complicates immunotherapy, affecting T-cell infiltration and cytotoxicity [660, 661].

The rise of biomaterial-based strategies has led to a paradigm shift in the transformation of “cold” tumors [572, 662, 663]. These engineered materials modify the TME via targeted metabolic changes, microbiota reprogramming, and precise gene editing techniques. For example, bacteria-nanodrug biohybrids leverage microbial tropism to deplete immunosuppressive metabolites (e.g., cyst(e)ine), disrupt redox balance, and enhance T-cell infiltration [664]. Similarly, nanoparticle-embedded hydrogels can selectively modify the intratumoral microbiota to enrich immunostimulatory species and enhance immunotherapy [665]. Beyond the realm of microbial manipulation, oral CRISPR‒Cas9 delivery systems offer a revolutionary approach for targeting critical genes, such as the mitochondrial chaperone TRAP1. These methods not only facilitate ICD but also alter immunosuppressive myeloid cell profiles, underscoring the viability of noninvasive strategies for continuous immune system modulation [666].

Despite these advancements, considerable challenges in clinical application remain. The deployment of complex multi-omics platforms, including spatial transcriptomics and AI-driven TME analysis, encounters significant barriers in scalability, affordability, and standardization across healthcare facilities. Innovations in biomaterials, although promising, confront issues related to in vivo biodistribution, potential immunogenic responses, and consistency in production. Moreover, strategies addressing systemic factors such as circadian rhythms and neural–immune interactions risk unintended consequences or could aggravate existing conditions in susceptible patient cohorts. The inherent variability of “cold” tumors further complicates the universal applicability of these strategies, emphasizing the critical need for refined biomarker-driven patient stratification to increase therapeutic precision and outcomes. Additionally, the long-term safety and efficacy of novel interventions, including neural–immune axis interception and PANoptosis activation, demand thorough investigation beyond the scope of preclinical studies.

Moving forward, the integration of spatial multi-omics, AI-driven data analytics, PDO-based assays, and innovative biomaterials promises to accelerate the creation of spatially precise interventions. Overcoming translational hurdles, including the expansion of spatial techniques and the standardization of organoid–immune cell cocultures, is imperative. The implementation of biomarker-driven patient stratification will play a pivotal role in tailoring these sophisticated platforms to the unique immunosuppressive landscapes of individual tumors. Ultimately, this multifaceted approach aims to elicit sustained antitumor responses in “cold” tumors that are historically resistant to treatment.

Conclusion

In conclusion, transforming immunologically “cold” tumors into “hot” counterparts represents a pivotal advancement in fully realizing the capabilities of immunotherapy for cancer treatment. This innovative strategy, aimed at addressing the challenges posed by camouflage, coercion, and cytoprotection, requires a comprehensive approach that combines advancements in molecular insights, biomaterials, epigenetic modifications, metabolic strategies, neural–immune interactions, and microbiome engineering. The strategic integration of these therapeutic modalities offers the potential to rejuvenate the field of immunotherapy, broadening the spectrum of durable antitumor responses and paving the way for precision oncology in treating previously refractory cancers.

Abbreviations

ADSL

Adenylosuccinate lyase

ALIX

Apoptosis-linked gene-2 interacting protein X

ARG1

Arginase-1

ASXL1

Additional sex comb-like 1

ATP

Adenosine triphosphate

B2M

Beta-2-microglobulin

BAF1

Barrier-to-autointegration factor 1

BATF3

Basic leucine zipper ATF-like transcription factor 3

BMAL1

Brain and muscle ARNT-like 1

β-AR

Beta-adrenergic receptor

CAF

Cancer-associated fibroblast

CAR

Chimeric antigen receptor

CCL5

C-C motif chemokine ligand 5

cDC1

Conventional type 1 dendritic cell

CDT

Chemodynamical therapy

cGAMP

Cyclic GMP-AMP

cGAS

Cyclic GMP-AMP synthase

CGRP

Calcitonin gene-related peptide

CHMP2

Charged multivesicular body protein 2

COF

Covalent organic framework

CPT1

Carnitine palmitoyltransferase 1

CRL5

​​Cullin-RING ubiquitin ligase 5

CTLA-4

Cytotoxic T-lymphocytes associated protein 4

CTL

Cytotoxic T lymphocyte

CX3CL1

C-X3-C motif chemokine ligand 1

CXCL

C-X-C motif chemokine ligand

CXCR

C-X-C motif chemokine receptor

DAC

Decitabine

DAMP

Damage-associated molecular pattern

DC

Dendritic cell

DNMT1

DNA methyltransferase 1

DOX

Doxorubicin

dsRNA

Double-stranded RNA

ECM

Extracellular matrix

EMH

Extramedullary hematopoiesis

ENPP1

Ectonucleotide pyrophosphatase/phosphodiesterase 1

EPC

Erythroid progenitor cell

EPR

Enhanced permeability and retention

ERK

Extracellular signal-regulated kinase

ESCRT

Endosomal sorting complex required for transport

ETBR

Endothelin B receptor

EV

Extracellular vesicle

EZH2

Enhancer of zeste homolog 2

FASN

Fatty acid synthase

FDC

Follicular dendritic cell

FDX1

Ferredoxin 1

FMT

Fecal microbiota transplantation

FRC

Fibroblastic reticular cell

FSP1

Ferroptosis inhibitory protein 1

GLUT1

Glucose transporter 1

GM-CSF

Granulocyte-macrophage colony-stimulating factor

GPX4

Glutathione peroxidase 4

GSDM

Gasdermin

H3K27me3

Histone H3 lysine 27 trimethylation

HDAC8

Histone deacetylase 8

HEV

High endothelial venule

HIF-1α

Hypoxia-inducible factor 1-alpha

HLA

Human leukocyte antigen

HMGB1

High mobility group box 1

HNSCC

Head and neck squamous cell carcinoma

HPV

Human papillomavirus

HRS

Hepatocyte growth factor-regulated tyrosine kinase substrate

ICAM1

Intercellular adhesion molecule 1

ICB

Immune checkpoint blockade

ICD

Immunogenic cell death

ICI

Immune checkpoint inhibitor

IDO1

Indoleamine 2,3-dioxygenase 1

IFN-γ

Interferon-gamma

IFN-I

Type I interferon

IL-6

Interleukin-6

iNOS

Inducible nitric oxide synthase

IRF3

Interferon regulatory factor 3

IRGQ

Immunity-related GTPase family Q protein

IS

Immune synapse

KDM5B

Lysine demethylase 5B

LAT2

L-type amino acid transporter-2

LFA-1

Lymphocyte function-associated antigen-1

LOX

Lysyl oxidase

LSD1

Lysine-specific demethylase 1

LTβR

Lymphotoxin beta receptor

m⁶A

N6-methyladenosine

MCT1

Monocarboxylate transporter 1

MDSC

Myeloid-derived suppressor cell

METTL3

Methyltransferase like 3

MHC-I

Major histocompatibility complex class I

MICB

MHC class I chain-related gene B

MLKL

Mixed lineage kinase domain-like protein

MMP

Matrix metalloproteinase

MOF

Metal-organic framework

MSN

Mesoporous silica nanoparticle

NBR1

neighbor of BRCA1 gene 1

NCF1

Neutrophil cytosolic factor 1

NET

Neutrophil extracellular trap

NF-κb

Nuclear factor kappa-B

NIR

Near-infrared ray

NK

Natural killer

NK1R

Neurokinin-1 receptor

NLR

NOD-like receptor

NP

Nanoparticle

NSCLC

Non-small cell lung cancer

PCSK9

Proprotein convertase subtilisin/kexin type 9

PDO

Patient-derived organoid

PD-1

Programmed cell death-1

PD-L1

Programmed cell death ligand 1

PDAC

Pancreatic ductal adenocarcinoma

PDT

Photodynamic therapy

PEG

Polyethylene glycol

PGE₂

Prostaglandin E₂

PLPP1

Phospholipid phosphatase 1

PMN-MDSC

Polymorphonuclear-MDSC

PNAd

Peripheral node addressin

PRC2

Polycomb repressive complex 2

PTT

Photothermal therapy

PVR

Poliovirus receptor

QD

Quantum dot

RAMP1

Receptor activity-modifying protein 1

RBC

Red blood cell

RLR

RIG-I-like receptor

RIPK1

Receptor-interacting protein kinase 1

RORA

RAR-related orphan receptor alpha

ROS

Reactive oxygen species

SDT

Sonodynamic therapy

SETDB1

SET domain bifurcated histone lysine methyltransferase 1

siRNA

Small interfering RNA

SIRPα

Signal regulatory protein alpha

SLC38A2

Solute carrier family 38 member 2

SLC7A11

Solute carrier family 7 member 11

SMYD3

SET and MYND domain containing 3

STAT3

Signal transducer and activator of transcription 3

STC1

Stanniocalcin 1

STING

Stimulator of interferon genes

SUSD6

Sushi domain containing 6

TAM

Tumor-associated macrophage

TAN

Tumor-associated neutrophil

TAV

Tumor-associated vasculature

TBK1

TANK-binding kinase 1

TCA

Tricarboxylic acid

TCR

T-cell receptor

TDO

Tryptophan 2,3-dioxygenase

TGF-β

Transforming growth factor-beta

TLR

Toll-like receptor

TLS

Tertiary lymphoid structure

TME

Tumor microenvironment

TNFSF14

Tumor necrosis factor superfamily member 14

TNF-α

Tumor necrosis factor-alpha

Treg

Regulatory T cell

UDP

Uridine diphosphate

UHRF1

Ubiquitin like with PHD and ring finger domains 1

VCAM1

Vascular cell adhesion molecule 1

VEGF

Vascular endothelial growth factor

VIP

Vasoactive intestinal peptide;

VPS4

Vacuolar protein sorting-associated protein 4

VISTA

V-domain Ig-containing suppressor of T cell activation

XCL

X-C motif chemokine ligand

YTHDF2

YTH domain family 2

ZBP1

Z-DNA binding protein 1

ZT

Zeitgeber time

Authors’ contributions

Wrote and revised the paper: Yuan-Tong Liu, Yun-Long Wang, Shuo Wang, Jia-Jun Li, Wei He, Xin-Juan Fan, Xiang-Bo Wan. Figure preparation: Yuan-Tong Liu, Yun-Long Wang. Supervision: Wei He, Xin-Juan Fan, Xiang-Bo Wan. All authors read and approved the final manuscript.

Funding

This study was supported by the National Science Fund for Distinguished Young Scholars [82225040], National Natural Science Foundation of China [82504345, 82373513], National Key Research and Development Program of China [2022YFA1105300, 2022YFC2503700], and China Postdoctoral Science Foundation [2025M771439].

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.