Dendritic cell maturation in cancer

Abstract

Dendritic cells (DCs) are specialized antigen-presenting cells that are present at low abundance in the circulation and tissues; they serve as crucial immune sentinels by continually sampling their environment, migrating to secondary lymphoid organs and shaping adaptive immune responses through antigen presentation. Owing to their ability to orchestrate tolerogenic or immunogenic responses to a specific antigen, DCs have a pivotal role in antitumour immunity and the response to immune checkpoint blockade and other immunotherapeutic approaches. The multifaceted functions of DCs are acquired through a complex, multistage process called maturation. Although the role of inflammatory triggers in driving DC maturation was established decades ago, less is known about DC maturation in non-inflammatory contexts, such as during homeostasis and in cancer. The advent of single-cell technologies has enabled an unbiased, high-dimensional characterization of various DC states, including mature DCs. This approach has clarified the molecular programmes associated with DC maturation and also revealed how cancers exploit these pathways to subvert immune surveillance. In this Review, we discuss the mechanisms by which cancer disrupts DC maturation and highlight emerging therapeutic opportunities to modulate DC states. These insights could inform the development of DC-centric immunotherapies, expanding the arsenal of strategies to enhance antitumour immunity.

Introduction

Since their discovery by Ralph Steinman and Zanvil Cohn half century ago¹, dendritic cells (DCs) have been shown to be the critical link between innate immune responses in the periphery and the induction of adaptive immunity against presented antigens. The indispensable role of DCs in orchestrating antigen-specific immunity and immune tolerance is fulfilled through their unique life cycle and set of abilities: to effectively sample the surrounding environment through various means of uptake; preserve and present antigens derived from engulfed materials; migrate from the periphery to secondary lymphoid organs; and interact with adaptive immune cells and instruct them via cytokines and receptor–ligand interactions^2,3. In steady state and injury, DCs are recruited from the blood to tissue sites, where they capture antigens. They then migrate to the T zones of lymph nodes to engage with naive or memory T cells, either inducing tolerance to peripheral cell-associated antigens and maintaining the pre-existing regulatory T (T_reg) cell pool in steady state or priming T cells against exogenous antigens during injury. During inflammation and/or infection, several pathogen-associated molecular patterns (PAMPs) or damage-associated molecular patterns (DAMPs) have been shown to trigger DC maturation and migration to lymph nodes^4–8.

In light of their specialized functions, DCs are essential in mounting effective antitumour immunity and mediating the response to immune checkpoint blockade (ICB)^9–12. In addition to priming tumour-specific T cells in secondary lymphoid organs, DCs also have a critical role in the recruitment of tumour-specific T cells to the tumour site and their reactivation and differentiation into effector T cells upon exposure to ICB. However, DCs could also prime and activate T_reg cells, highlighting the delicate balance between immunogenic responses and tolerogenic responses calibrated by DCs in tumour lesions and their draining lymph nodes.

Through the complex process of maturation, DCs acquire the versatile functions that have such an important role in establishing effective antitumour immunity¹³. Studies of DC maturation have often reduced it to a phenotypic description, with increased surface expression of major histocompatibility complex (MHC) molecules and other co-stimulatory markers (CD40, CD80 and CD86) required to activate T cell differentiation into either effector T cells or T_reg cells. Yet, the maturation process in fact involves an elaborate interplay between DC ontogeny and environmental cues. Together, all of these factors dictate DC phenotype and function and ultimately the quality of DC-mediated immune responses^6,14. The process of maturation is directly consequential to the type of downstream immune response it elicits, as such, maturation is a vital node of immune regulation that is often perturbed and co-opted by cancer cells^15,16. Thus, gaining mechanistic understanding of DC maturation and of the specific molecular axes through which it is exploited in cancer would lay an important foundation towards the robust development of DC-centred immunotherapeutics.

In this Review, we consider the complexities underlying DC maturation in cancer by: delineating the diversity of DCs based on lineages and states of maturation found in tumour contexts; outlining the cell-intrinsic mechanisms of maturation; discussing various functions of mature DCs; and highlighting opportunities for therapeutic interventions.

DC diversity in cancer

In dissecting the heterogeneity of DCs found in tumoural contexts, we propose that DC diversity in cancer can be understood through an organizational paradigm revolving around two main factors: (1) DC ontogeny; and (2) DC states. In terms of ontogeny, the diversity of DCs is largely conserved across lymphoid and non-lymphoid tissues and consists of type 1 DCs (DC1) and DC2, which are conventional DC (cDC) subsets that arise from common DC precursors³ (Fig. 1). DC1s are considered to be an ontogenically singular subset, and their pivotal contribution to antitumour immunity and responsiveness to immunotherapy has been demonstrated in multiple studies^{10–12,17–20}. By contrast, DC2s are heterogenous, and the function of each DC2 subset is just beginning to be understood^21–26.

Fig. 1 |. Conceptual overview of DC diversity.

The diversity of dendritic cells (DCs) in cancer can be defined by both ontogeny and functional cell state. Considering the functions of different DC types in antitumour immunity, a potential therapeutic strategy is to target molecular pathways that modulate DC lineage commitment. Transcription factors known to regulate type 1 DC (DC1) and DC2 lineage specification from common DC progenitors (CDPs) to the differentiated DC1 or DC2 stage are shown, with red, yellow and blue indicating transcription factors implicated in specification of DC1, DC2 and both lineages, respectively. Plasmacytoid DC ontogeny seems to be of mixed myeloid and lymphoid origin, and therefore we have omitted them from this figure focused on myeloid-derived DCs. Mobilized mostly during inflammation, DC3s (also known as monocyte-derived DCs) derive from an ontogenic lineage (monocyte-DC progenitors; MDPs) that is distinct from CDP-derived conventional DCs (cDCs). In addition to ontogenic diversity, a few functional cell states have been identified, mainly through single-cell RNA sequencing of tumour-associated DCs: interferon-stimulated gene (ISG)⁺ DCs; CD207⁺ DCs (also known as langerin⁺ DCs); and mature DCs enriched in immunoregulatory molecules (mregDCs). Key markers for each state are indicated. So far, only DC2s and DC3s have been shown to engage a CD207⁺ DC state, whereas DC1s and DC2s have been shown to engage in both ISG⁺ DC and mregDC states. Whether DC3s are able to engage these other two states remains unclear. Although a couple of studies^120,178 have demonstrated that ISG⁺ DCs can be a transitional state before DCs engage in the mregDC state, the exact relationship between these two states remains unclear (indicated with a question mark) and deserves further investigation. FCGBP, IgGFc-binding protein; IFIT, interferon-induced protein with tetratricopeptide repeats; IRF, interferon regulatory factor; NFIL3, nuclear factor interleukin 3-regulated protein; SOCS, suppressor of cytokine signalling.

This Review focuses on common DC precursor-derived cDC subsets, but two other DC subsets, DC3s (also known as monocyte-derived DCs) and plasmacytoid DCs (pDCs), have been considered as part of the DC pedigree^2,3,27. DC3s share some of the phenotypic markers of cDCs (that is, MHC class II^high and CD11c⁺), yet they are mobilized mostly during injury and rapidly populate inflamed tissues²⁸. Although DC3s are capable of priming naive T cells in certain contexts^29–33, whether their priming capacity is truly commensurate with that of their cDC counterparts remains unclear³⁴, as does whether their physiological function is strictly redundant considering that they are predominantly found in inflamed tissue. Furthermore, in light of a 2023 study indicating that DC3s seem to arise from monocyte-DC progenitors (MDPs)³⁵ (Fig. 1), a careful functional characterization of their contribution to antitumour immunity is warranted. By contrast, pDCs are chiefly known for their ability to produce large amounts of type I and type III interferon in response to viral antigens^36,37; however, their contribution to antitumour immunity is decidedly mixed and context dependent (reviewed in ref. 27). The ontogeny of pDCs remains unclear, which has continued to fuel the debate on whether pDCs should be considered as another ‘true’ DC subset^38–40.

The cell-surface markers and transcriptional programme that distinguish the three predominantly myeloid-derived DC subsets have been extensively reviewed^2,27,41. Here we propose that the current knowledge of two main pillars of DC diversity — ontogeny and cell state — can be further exploited to enhance DC-mediated antitumour immunity. First, the generation of peripheral DCs requires haematopoietic activity in the bone marrow, where targeting molecular pathways responsible for DC lineage commitment could be therapeutic. Second, the cell states that these DCs acquire during the process of maturation can dictate DC control of antitumour immunity and might be amenable to favour immunogenic functions upon therapeutic targeting.

The generation of peripheral DCs by DC-poiesis is especially important to promote the presence of DCs in tumours, in which a paucity of DC1 is a well-described hindrance against effective antitumour immunity⁴². For example, FLT3L is a DC-poiesis-promoting cytokine that has been explored as an antitumour therapy to replenish intratumoural DC subsets (reviewed in ref. 43). Furthermore, enhancing DC bias during myelopoiesis via epigenetic modulation of pre-DC1 or pre-DC2 specification is another potential therapeutic approach to augment DC presence in tumours (Fig. 1). Considering the emerging evidence that tumour-derived signals, such as IL-6 and granulocyte–macrophage colony-stimulating factor (GM–CSF), can modulate DC-poiesis in the bone marrow^44,45, such therapeutic approaches could synergize with other immunotherapeutics that locally act on the tumour site. However, the epigenetic mechanisms that underlie DC lineage specification are beginning to be elucidated^46–49, and whether all modes of epigenetic regulation (that is, presence or absence of transcription factors, or chromatin accessibility) can be targeted in a DC progenitor-specific manner to modulate DC-poiesis remains to be tested in future proof-of-concept studies.

Potential therapeutic strategies involving DC lineage specification are still in their nascent stages, therefore, the diversity of cell states is the subsequent focus of this Review. The considerable accrual of single-cell transcriptomic studies on the DC compartment in mouse and human tumours has led to detailed characterizations of transcriptional signatures that could delineate conserved DC states during maturation^27,50–54 (Table 1). Although not exhaustive, several common DC molecular states have been identified across various human and mouse tumours. For example, ‘mature DCs enriched in immunoregulatory molecules’ (mregDCs), which are also known as migratory DCs, CCR7⁺ DCs or homeostatically mature DCs. Another state is characterized by an enrichment in interferon-stimulated genes (ISGs; ISG⁺ DCs). In addition, a distinct molecular state previously associated with Langerhans cells is referred to here as ‘CD207⁺ DCs’ (relevant studies are outlined in Table 1).

Table 1 |.

Diversity of DC states uncovered by scRNA-seq technology

States	Denomination used in study	DC1 or DC2	Context	Species and ref.	Shared transcriptional programme
CD207+ DCs	LC histiocytosis-like DCs	DC2	Lung cancer	Mouse and human52	CD1A, CD207, FCGBP, LTB, SPI1 and S100B
	LanghDCs	NA	Breast cancer	Human139
	LC-like DCs	DC2	Pan-cancer	Human301
	LC-like DCs	NA	Breast cancer	Human302
			HNSCC	Human303
	Langerhans cells	NA	Breast cancer	Human60
			Lung cancer	Human304
ISG+ DCs	Early mature DCs	DC1	Homeostasis	Mouse120	ISG15, ISG20, CXCL9, CXCL10, IFIT1, IFIT3, IFI6 and GBP2
	Inf-cDC2	DC2	Viral infection	Mouse34
	NA	DC1	Melanoma	Mouse178
	CCR7−MHC class II+	DC1	Melanoma	Mouse164
	IFN-DCs	Both	Subcutaneous models of melanoma and fibrosarcoma	Mouse preprint study165
	ISG+ DCs	DC2	Fibrosarcoma	Mouse163
	CXCL9+ DCs	DC2	Pan-cancer	Human51
	ISG15+ DCs	DC2	Pan-cancer	Human51
mregDCs	Late mature DCs	DC1	Homeostasis	Mouse120	CCR7, FSCN1, CD40, CD86, RELB, SOCS2, SOCS1, IL4I1, TNFRSF9, IDO1, BIRC2, BIRC3, FAS, MARCH1, CCL5, CCL17, CCL22, IL15, TAPBP and B2M
	CCL19+ DCs	Both	Breast cancer	Human302
	IDO1+ DCs	Both	HNSCC	Human305
	Tolerogenic DCs	Both	Myocardial infarction	Mouse306
	Activated DCs	Both	Adipose tissue	Mouse307
			Lung cancer	Human304
	NA	DC1	Melanoma	Mouse178
	CCR7+MHC class II+ DCs	DC1	Melanoma	Mouse164
	Migratory DCs	Both	TLR ligand-stimulated DCs in lung lymph notes	Mouse140
			Steady-state lymph nodes	Mouse308
			Healthy embryonic and adult skin and inflammatory skin disorders	Human309
			Fetal and adult small intestine	Human310
			Conjunctiva	Mouse311
			HNSCC	Human303
			Breast cancer	Mouse312 and human313
			Bladder cancer	Human314
			Glioblastoma	Mouse315 and human315,316
			Cutaneous squamous cell carcinoma	Human317
	mregDCs	Both	HDM-treated skin	Mouse141
			TLR ligand-stimulated DCs in lung lymph nodes	Mouse140
			Adult tissues in steady state and disease	Human142
			Neonatal lungs	Mouse318
			Subcutaneous tumour models	Mouse106,143–145
			Non-small-cell lung cancer	Mouse130,134,146 and human52,136–138
			Lewis lung carcinoma	Mouse319
			NASH and liver cancer	Mouse148 and human147
			Melanoma	Mouse149,150
			HNSCC	Human151–153
			Colorectal cancer	Mouse154
			Breast cancer	Mouse155 and human156
			Glioblastoma	Mouse156
			Ovarian cancer	Human157
			Bladder cancer and urine	Human159
			Pan-cancer	Human158
			COVID-19	Human160
			Inflammatory skin disorders	Human161
			Atherosclerosis	Mouse162
	DC3	Both	Non-small-cell lung cancer	Mouse and human53
			Melanoma	Mouse277
			Melanoma brain metastasis	Human320
			Nasopharyngeal cancer	Human321
			Ovarian cancer	Mouse322
			Pan-cancer	Human323
	LAMP3+ DCs	Both	Lung cancer	Human324
			Breast cancer	Human325
			Hepatocellular carcinoma	Human326
			Colorectal cancer and metastasis	Human327
			Bladder cancer	Human328
			Oesophageal squamous cell carcinoma	Human329
			HNSCC	Human152,330,331
			Pan-cancer	Human51,158

DC, dendritic cell; DC1, type 1 DC; DC2, type 2 DC; DC3, type 3 DC; HDM, house dust mite; HNSCC, head and neck squamous cell carcinoma; IDO1, indoleamine 2,3-dioxygenase 1; IFN, interferon; Inf-cDC2, inflammatory conventional DC2; ISG, interferon-stimulated gene; LC, Langerhans cell; MHC, major histocompatibility complex; mregDC, mature DC enriched in immunoregulatory molecules; NA, not applicable; NASH, non-alcoholic steatohepatitis; scRNA-seq, single-cell RNA sequencing; TLR, Toll-like receptor.

Defined by co-upregulation of both maturation (CD40, CD86, CCR7 and IL12B) and immunoregulatory (PDL1, PDCD1LG2 and ALDH1A2) genes and proteins, mregDC is a cell state of maturation induced in DC1 and DC2 upon cholesterol mobilization triggered by capture of cell debris (discussed in more detail later). ISG⁺ DCs have also been found in both steady state and inflammatory contexts. Despite the different nomenclatures in various studies, this DC state shares a core set of ISGs (that is, CXCL9, CXCL10, IRF7, ISG15 and IFITM3), suggesting a collective state of responsiveness to interferon signalling. Finally, CD207⁺ DCs are a separate DC cluster that express CD1a and CD207 (also known as langerin) and have been identified in several cancer types, including cancers of non-skin tissues (Table 1). Unlike bona fide Langerhans cells in the epidermis that also express CD207 and derive from embryonic precursors⁵⁵, these CD207⁺ DCs probably represent a subset of DC2s that are induced into this cell state by tissue cues such as transforming growth factor-β (TGFβ)^56,57, retinoic acid⁵⁸ and others previously implicated in induction of CD207 in DCs⁵⁹. Although high enrichment of CD207⁺ DCs in tumours were associated with a good prognosis in a subset of cancers (colorectal and breast)⁶⁰, further studies are needed to better understand the functional contribution of this state to antitumour immunity. Multiple DC subtypes have been shown to induce these signatures and have been associated with distinct antitumour mechanisms (discussed later); yet, how the functional capacities attributed to each ontogenic lineage shape the behaviour of DCs engaged in these transcriptional states remains to be investigated.

DC maturation

As transcriptional DC states become refined through multidimensional modalities that provide proteomic, epigenomic, metabolomic and spatiotemporal information at a single-cell level, the assembled molecular programmes of mature DC states should enable a better delineation of the molecular circuitry of DC maturation and function. Thus, we posit as the central premise of this Review that understanding the molecular underpinnings of DC maturation (triggers, signalling pathways and metabolic rewiring), and how they are exploited by cancer cells to suppress immunogenic functions of DCs, will help to chart the unexplored avenues for DC-targeted therapies that can synergize with the current T cell-centric arsenal of immunotherapeutics.

Triggers

Innate sensing.

As sentinels of the body, DCs are equipped with molecular machinery to sense foreign microorganisms and tissue injury via recognition of PAMPs and DAMPs. In tumours, DCs can sense DAMPs released by tumour cells, including high mobility group box 1 (ref. 61), heat shock proteins (HSPs)⁶² and various nucleic acids⁶³. These DAMPs are recognized by pattern recognition receptors (PRRs) such as Toll-like receptors (TLRs) and cGAS–STING pathway molecules present in DCs. PRRs can be spatially organized into intracellular (for example, TLR3, TLR7, TLR8, TLR9 and cGAS–STING) and extracellular (for example, TLR1, TLR2, TLR4, TLR5 and TLR6) compartments⁷.

Box 1 |. PRR engagers in the clinic.

Pattern recognition receptor (PRR) agonists, particularly Toll-like receptor (TLR) agonists, have garnered considerable attention in cancer therapy. Clinical trials of PRR agonists in patients with cancer have highlighted their potential in stimulating antitumour immunity, owing to their ability to induce pro-inflammatory cytokine production (including type I interferon) and activate various immune subsets including dendritic cells (DCs)⁷. Of note, the Bacillus Calmette–Guérin (BCG) vaccine has demonstrated efficacy in bladder cancer, leading to its approval by the US Food and Drug Administration³³³. Similarly, agonists targeting TLR3, TLR4, TLR5, TLR7–TLR8 and TLR9 have shown promise in various cancer types, including lung cancer, melanoma and glioblastoma^334–342. Combining TLR agonists with other anticancer therapies has emerged as a strategy to enhance efficacy. For instance, a strategy that combines TLR9 or TLR3 agonists with radiotherapy, chemotherapy or vaccination against tumour antigens has shown improved outcomes in patients with various blood and solid cancers^{335,343–347}. Moreover, TLR agonists have been shown to synergize with immune checkpoint blockade in both preclinical studies and early clinical trials, offering new avenues for cancer immunotherapy^348–350.

Despite the immense therapeutic potential of TLR agonists, their effective application as an immunotherapeutic requires several gaps in knowledge to be addressed. Although their potent ability to stimulate various immune subsets is beneficial in energizing targeted antitumour immune responses, it can also trigger several inflammatory adverse effects³⁵⁰. Furthermore, both preclinical and clinical studies of TLR agonists often show therapeutic efficacy without resolving cell-type-specific contributions to their mechanism of action. Considering that TLR agonism has been shown to induce pro-tumoural activities in preclinical studies³⁵¹, a deeper understanding of TLR signalling and its downstream effects within the complex and diverse ecosystems of tumours is essential in developing safe and productive TLR agonist-based therapies.

Beyond TLRs, STING agonists have emerged as potential cancer treatment agents due to their ability to induce persistent antigen-specific T cell immunity in multiple preclinical tumour models^126,127,352. The detailed mechanisms of action of STING agonists have been extensively reviewed elsewhere^353,354. In brief, a short and potent burst of STING agonism has been shown in preclinical studies to stimulate antitumour immune responses through strong induction of type I interferon and through promotion of a CD8⁺ T cell antitumour cytotoxic response that is mediated by activated DC1s^19,352,355. Despite the potent antitumour effect demonstrated in preclinical studies, STING agonists are not without their drawbacks. STING expression has been shown to decline in subsets of advanced human cancers (colon and melanoma)^356,357; in preclinical tumour models, excessive STING activation has been shown to not only create a pro-metastatic TME but also induce apoptosis in various myeloid and T cell subsets^358–361, thereby underscoring the need to refine the dose and the specificity of STING agonists. Despite such challenges, STING agonists have shown initially encouraging results in combination with immunotherapy in clinical trials^362,363. Incorporation of technical platforms to enhance cell-specific targeting, such as antibody–drug conjugates³⁶⁴ and lipid nanoparticles will help to optimize their antitumour efficacy.

As PRR engagement has shown some antitumour efficacy in both preclinical and clinical studies, a deeper understanding of how signalling dynamics from each PRR differentially contributes to DC maturation in cancer will help to inform therapeutic decisions on the types of PRR adjuvants and their optimal dosing scheme, for their effective incorporation into the arsenal of DC-targeting therapies.

The activation of PRRs via PAMPs and DAMPs or their pharmacological analogues is firmly established to drive DC maturation through the upregulation of co-stimulatory molecules (CD40, CD80 and CD86), induction of antigen processing machinery, metabolic rewiring and production of immunostimulatory cytokines such as IL-12 and type I interferon^4–8. The potent ability of PRR signalling to drive DC maturation has led to the use of numerous PRR engagers in the clinic to enhance antitumour immunity (Box 1). Although the intricate details of each PRR signalling modality and its functional nuance are beyond the scope of this Review (reviewed in refs. 8,64–68), consideration of such details should inform the choice of PRR agonists targeted towards intratumoural DCs. For instance, the ability of TLRs to induce type I interferon production is dependent on their localization to the endosomes, raising the importance of designing PRR agonists to be trafficked to the correct subcellular compartments⁸. Furthermore, DC maturation driven by TRIF signalling alone (for example, via TLR3 activation) is dependent on autocrine type I interferon, whereas MYD88-dependent signalling downstream of TLR activation can drive DC maturation independent of type I interferon, highlighting the specific signalling dependencies of each TLR pathway⁶⁹.

As part of their immune evasion strategy, tumours have been shown to dampen PRR-mediated signalling. For instance, various tumours upregulate ENPP1, a nucleotide pyrophosphate phosphodiesterase that can degrade extracellular 2′,3′-cGAMP, which is a potent agonist for the cGAS–STING pathway⁷⁰. Furthermore, increased expression of negative regulators of PRR signalling in DCs (for example, AIM2 (ref. 71)) correlates with poor clinical prognosis in various cancer types such as glioma, uveal melanoma and pancreatic and renal carcinomas^72,73.

Despite demonstrations of its potent ability to activate antitumour immunity, not all PRR engagement is beneficial. For instance, the specific blockade of TLR2 or MAVS (a signalling adaptor molecule for the RNA sensor RIG-I) have been shown to enhance the immunogenicity of intratumoural DCs^74,75. Moreover, although TLRs and STING are pivotal in initiating an efficient antitumour response, chronic or excessive signalling could also promote pathogenic inflammation and tumour growth^76–78.

Uptake of dead cells as a driver of homeostatic DC maturation.

For professional phagocytes, such as DCs and macrophages, the uptake of dead cells simultaneously fulfills two functions essential to maintaining tissue homeostasis and sculpting immune responses: the removal of dead cells to prevent triggering of an inflammatory response towards the homeostatic turnover of cells, and information capture of tissue state in the form of antigens that can be relayed to other immune cells^79,80. Unlike macrophages that act mainly in peripheral tissue, DCs are very well equipped to migrate to the secondary lymphoid organs and to prime or regulate the adaptive immune response about peripheral cues, highlighting the key role of DCs in peripheral immunosurveillance^2,3.

Upon uptake of dead cells, DCs undergo three sequential phases: ‘sensing’, ‘engulfing’ and ‘digesting’ dead cells in both steady-state conditions and the tumour microenvironment^79,80 (Fig. 2). The initial sensing phase involves recognizing ‘find-me signals’ emitted by dying cells — both non-tumoural and tumoural cells — including nucleotides (such as ATP and UTP)^63,81, chemokines (such as CX3CL1)⁸² and lipids (such as sphingosine-1-phosphate)^83–86. They guide DCs to regions requiring debris clearance but can also modulate DC phenotype and function. For instance, ATP binding to P2X₇ (a receptor for ATP) triggers IL-1β secretion in DCs, which is crucial for priming cytotoxic CD8⁺ T cells, whereas adenosine derived from ATP binding to its cognate receptors (such as adenosine receptor A2A (ADORA2A)) promotes a tolerogenic DC phenotype^87–90.

Fig. 2 |. Capture of cell debris by DCs leads to their maturation.

At each step of the uptake of dead cells, different sets of signals can skew the outcome of dendritic cell (DC) maturation towards either tolerogenic (red) or immunogenic (blue) responses. Brown indicates signals coming from dying cells or genes involved in digestion of dying cells. Here we focus on the genes that have a direct role in DCs or are expressed in DCs. a, During sensing, DCs detect ‘find-me’ signals released by dying tumour cells, including nucleotides (for example, ATP and UTP), chemokines (for example, CX3CL1) and lipids (for example, cholesterol, oxidized lipids and sphingosine-1-phosphate (S1P)), guiding them to sites requiring debris clearance and influencing DC phenotype and function. Activation of P2RX₇ (ref. 332), Toll-like receptor 3 (TLR3) and TLR4 triggers an immunostimulatory response, whereas activation of TLR2 and adenosine receptor A2A (ADORA2A)⁹⁰ can promote DC dysfunction. The asterisk (*) indicates that TLR2 stimulation can be dysfunctional in certain tumoural contexts⁷⁵. b, During engulfment, DCs identify ‘eat-me’ signals (for example, phosphatidylserine) exposed by dying cells. Depending on the phagocytic receptors engaged during engulfment, uptake can induce either immunosuppressive (for example, AXL and MERTK) of immunostimulatory (for example, CLEC9A and TIM4) programmes. Secreted (for example, gelsolin) or membrane-bound (CD31, CD47 and major histocompatibility complex (MHC) class I) ‘don’t eat-me’ signals on dead cells can interact with distinct receptors on DCs to block phagocytosis or to block recognition by receptors of eat-me signals. c, During digestion, DCs degrade the internalized cargo to extract dead cell-derived antigens (right) and to manage the influx of metabolites from cell debris (left). Signals received during the previous steps can influence how the cargo is processed; for example, stimulation of CLEC9A in DC1s during engulfment increases the probability of antigen-loaded phagosomes to rupture, leading to increased cross-presentation of dead cell-derived antigens. Agonizing SIRPα on DC2s via CD47 on tumour cells can enhance degradation of cargo-derived DNA, leading to diminished activation of the cGAS–STING pathway and decreased type I interferon production. Metabolites also influence function; for example, cholesterol influx from dead cells can stimulate de novo synthesis and induce increased cholesterol trafficking to the DC cell membrane, leading to increased antigen presentation and maturation. DCs can also induce liver X receptor-β (LXRβ)-mediated cholesterol efflux to maintain intracellular lipid homeostasis and fine-tune the degree of maturation. LILRB1, leukocyte immunoglobulin-like receptor B1.

The subsequent engulfing phase entails identification of ‘eat-me signals’ exposed by dying cells, where phosphatidylserine translocation from the inner to the outer leaflet of the cell membrane during apoptosis serves as a pivotal step⁹¹. Phagocytic receptors (such as TIM1–TIM4 (refs. 92,93) and stabilin 2 (ref. 94)) bind directly to phosphatidylserine, whereas receptor tyrosine kinases (including TYRO3, AXL and MERTK) require bridging molecules such as GAS6 and protein S to effectively recognize phosphatidylserine⁹⁵. Additional signals include oxidized low-density lipoproteins, calreticulin, annexin-1 and opsonizing antibodies that bind to complement and Fc receptors^79,96.

Recognition of eat-me signals by phagocytic receptors that facilitate engulfment is thought to be an ‘immunologically silent’ process. For instance, in addition to facilitating engulfment of dead cells, phosphatidylserine-recognizing receptors such as MERTK and AXL can induce immunosuppressive programmes by blocking NF-κB activation^97,98 or triggering the expression of SOCS1 and SOCS3 that inhibit the production of inflammatory cytokines^95,99. Accordingly, blocking the recognition of phosphatidylserine or blocking complementary phagocytic receptors in vivo consequently enhances the immunogenicity of dead cell antigens and, in the context of cancer, antitumour immunity (in preclinical models)^98–102. For example, inhibiting TIM3 on DCs can boost antitumour immune responses^103,104 by enhancing cGAS–STING pathway and activating the inflammasome in DCs^105,106. Broadly, the targeting of these immunosuppressive phagocytic receptors has recently emerged as a potential therapeutic strategy in cancer treatment, with several inhibitors currently in clinical trials (NCT02817633, NCT03425279, NCT05438420 and NCT04762199). However, whether blocking phosphatidylserine-recognizing receptors, such as TIM3, is sufficient to enhance DC antitumour function clinically remains to be shown^107–109.

Not all phagocytic receptors are immunosuppressive, as recognition and uptake of dead tumour cells through TIM4 and CLEC9A have been shown to be critical in initiating antitumour responses^110–114. In fact, secretion of gelsolin (an endogenous antagonist for CLEC9A) has been shown to impair DC1-mediated cross-presentation of tumour antigens and dampen antitumour immunity in various orthotopic tumour models¹¹⁵. Incidentally, several human cancers express gelsolin and its high expression has been correlated with poor prognosis in liver, head and neck and stomach cancers¹¹⁵. Enhanced recognition of dead tumour cells via CLEC9A upon blockade of the mevalonate pathway, which induces exposure of actin filaments (a ligand for CLEC9A) in dying tumour cells, led to potentiation of antitumour T cell responses in a B16 melanoma model¹¹⁶. Although the engagement of each phagocytic receptor induces a distinct immunostimulatory or immunosuppressive response, various phagocytic receptors are likely to be engaged simultaneously in tumour contexts.

The final digestion phase involves sensing and degradation of the internalized cargo. Engulfed PAMPs and DAMPs can be detected via endosomal TLRs, whereas cargo metabolites can be sensed within the lysosomal compartment^117–119. One notable challenge in this process is managing the doubling of intracellular contents, including carbohydrates, lipids, proteins and nucleotides from engulfed material. DCs must adeptly handle this metabolic burden, either by recycling to replenish cellular stores, or by releasing certain contents and reprogramming their metabolism^120–123. Cargo degradation influences metabolic processes and also triggers signalling pathways within DCs^{120,123–125}. For instance, incomplete DNA degradation within endosomes can activate the cGAS–STING pathway, leading to type I interferon production^126,127. Tumour cells interfere with degradation of cargo DNA via the CD47–SIRPα axis, which upon engagement of SIRPα on DCs, leads to phagosomal acidification and subsequent degradation of tumour-derived mitochondrial DNA and diminished type I interferon production by DCs¹²⁸. In summary, uptake of dead cells is a unique trigger, in that each step of uptake (sensing, engulfing and digesting) can engage several signalling pathways that can each dictate the immunogenicity of the presented antigen through the modulation of DC phenotype and function. It is tempting to speculate that from the initial sensing of a dead cell to the final metabolism of phagocytosed cargo, DCs are continually fine-tuning the immunogenicity of the antigen marked for presentation, by integrating signals that they receive throughout the uptake process. Therefore, how DCs gather and process the information contained with the phagocytosed cargo might dictate how such information will be presented during its interaction with their cognate T cell partners.

Molecular transformation induced upon capture of cell debris.

In steady states in which overt microbial PAMPs or tissue-derived DAMPs are absent, the molecular triggers that cause DCs to mature and migrate to lymph nodes after capturing cell debris from normal turnover remain unclear. Homeostatic maturation has been speculated to either be a stochastic process or be driven by migratory cues, subtle DAMPs or microbiota-derived PAMPs. Within the past decade, DC capture of dead, uninfected cells has been shown to be sufficient to trigger a state of maturation^120,129,130, in contrast to the prevalent notion of early-to-mid 2000s that capture of apoptotic cells is an immunologically inert process that maintains immaturity of DCs¹³¹. Similar to what was previously termed the homeostatic maturation^120,132 or migratory programme¹³³, this state of maturation, now referred to as the mregDC state^120,129, is elicited independently of CCR7-mediated migration and signalling pathways involving TLR, the inflammasome, IL-1R and type I interferons¹³⁰. The mregDC state consists of the induced expression of maturation markers (for example, CD40, CD80 and CD86), migration-related molecules (for example, CCR7, FSCN1, MARCKS, ICAM1, EXTL1 and MYO1G), immunoregulatory molecules (for example, PDL1 and PDL2, CD200, retinal dehydrogenase 2 (ALDH1A2), FAS, indoleamine 2,3-dioxygenase 1 (IDO1), IL4I1, SOCS1 and SOCS2), chemokines and cytokines to orchestrate optimal DC–T cell interactions (for example, CCL5, CXCL3, IL-27, CCL17, CCL22, IL-12B, IL-7 and IL-15) and downregulation of TLRs, phagocytic receptors and negative regulator of antigen presentation (for example, MARCH1)¹³⁰. DC co-expression of both maturation and regulatory molecules highlights a dual potential to prime either a tolerogenic or immunogenic response against the captured antigen.

The mregDC state represents DCs that are equipped with the necessary molecular machinery to perform various functions attributed to mature DCs (Fig. 3). First, PRRs and phagocytic receptors are downregulated to prevent further sensing of threats and focus on acquired cargo. Subsequently, antigen processing and migratory machinery are upregulated to enable antigen-carrying DCs to present antigens in the T zones of lymph nodes (for example, VAMP8, B2M and TAPBP). The interaction of antigen-carrying DC with its cognate T cell leads to activation of the non-canonical NF-κB (ncNF-κB) pathway in the DC, which promotes T cell activation (for example, RELB, NFKB2, MAP3K14, BIRC2 and BIRC3). In addition, enhanced expression of regulatory molecules in DCs prevents excessive T cell activation. Finally, upregulation of death molecules, such as FAS, promotes DC apoptosis after engaging with cognate T cells. Together, the mregDC programme enables DCs, upon acquisition of their antigen cargo, to leave the periphery and migrate towards the lymph nodes, where they inform naive T cells about peripheral cues.

Fig. 3 |. Molecular machineries encoded within the mregDC state.

Upon uptake of dead cell debris, both type 1 dendritic cells (DC1s) and DC2s can enter a cell state referred to as mature DCs enriched in immunoregulatory molecules (mregDC), also known as migratory DCs, CCR7⁺ DCs or homeostatically mature DCs. The mregDC state encompasses various transformations that are critical for mature DC function. After encountering an activating signal (a ‘threat’), DCs cease engulfing and sensing of external pathogen-associated molecular patterns and danger-associated molecular pathogens and, instead, focus on their engulfed cargo by notably dampening expression of Toll-like receptors (TLRs) and phagocytic receptors. The engulfed cargo is then processed into antigens that can be presented on major histocompatibility complex (MHC) molecules (1). While the antigen is being processed, DCs also upregulate their migratory machinery to facilitate their trafficking into secondary lymphoid organs (2). Upon their entry into secondary lymphoid organs, mregDCs produce many chemokines (for example, CCL17, CCL19 and CCL22) that facilitate their search for cognate T cells. Once they have found and interacted with their cognate T cells, the non-canonical NF-κB (ncNF-κB) pathway is activated in DCs (3), which can lead to production of additional signals critical for T cell priming (for example, co-stimulatory molecules and IL-12). In parallel, DCs also upregulate immunoregulatory molecules (for example, PDL1, PDL2, suppressor of cytokine signalling 1 (SOCS1), SOCS3, indoleamine 2,3-dioxygenase 1 (IDO1) and retinal dehydrogenase 2 (ALDH1A2)) to forestall hyperactivation of the primed adaptive arm (4). At some point during the DC–T cell interaction, DCs become susceptible to FAS-mediated apoptosis (5), leading to termination of priming of additional T cells against the presented antigen. CLR, C-type lectin receptor; FASL, FAS ligand; IFNγ, interferon-γ; PRR, pattern recognition receptor; RA, retinoic acid; TNFRSF, TNF receptor super family.

Although the uptake of dead cells can induce the mregDC state^120,129,130, the exact molecular pathways linking dead cell uptake to this transcriptional programme are still being elucidated. Emerging evidence has indicated that cholesterol movement, whether derived from engulfed apoptotic cargo, extracellular sources or de novo biosynthesis, can induce DC maturation, thereby highlighting that signals from dead cell uptake are sufficient triggers^120,134. Collectively, these studies suggest that the mregDC signature is a fundamental cell state triggered by the capture of dead cell-derived information (such as cholesterol) that transforms immature DCs into mature DCs capable of presenting cell-derived antigens and interacting with T cells. Importantly, the mregDC state has been identified both in normal tissues and during infection, autoimmunity and various cancers (Table 1). The prevalence of the mregDC state indicates that the capture of cell debris and subsequent migration to lymph node T zones is a crucial aspect of the DC life cycle in both homeostasis and inflammation.

In addition to uptake of dead cell debris, other triggers (such as PAMPs) might induce the mregDC state. Of note, the transcriptional state of DCs treated with poly:I:C (a synthetic PAMP) has been found to eventually converge with that of DCs undergoing homeostatic maturation (presumably induced by uptake of dead cells)¹³², which strongly suggests that the mregDC state serves as a fundamental mode for antigen presentation after capturing peripheral information, with or without the overt presence of PAMPs. Furthermore, data from a 2024 study have shown that PAMP-induced DC maturation was also reliant on cholesterol movement and synthesis, which suggests a shared molecular programme of maturation induced upon different triggers¹³⁴. The mregDC programme seems to represent the final convergent state of maturation across different contexts, yet different maturation triggers or cytokine milieus can further modulate the state of mature DCs. For example, PRR stimulation is crucial for promoting the secretion of inflammatory cytokines and enhancing the immunogenicity of the presented antigens; by contrast, in steady state, the absence of such triggers fosters regulatory responses towards self-antigens. In the context of cancer, immunosuppressive environments, such as the tumour microenvironment (TME), can dampen the production of inflammatory cytokines, thereby weakening antitumour immunity.

Considering the multifaceted role of mregDCs in modulating antigen-specific T cell responses in cancer, a better understanding of how signalling dynamics and metabolic rewiring contributes to the induction and modulation of the mregDC state could inform its therapeutic targeting to enhance DC-mediated antitumour immunity.

Signalling dynamics

In addition to engagement of PRRs by PAMPs, a wide-range of cell-to-cell interactions and soluble factors have been implicated in inducing and modulating DC maturation¹³⁵. Numerous studies have shown how various molecules contribute to DC maturation and function; however, they often lack holistic characterization of how each factor contributes to the molecular state of maturation, which complicates efforts to identify key DC maturation pathways, especially in the context of cancer. We propose that examining molecular programmes of maturation delineated by single-cell RNA sequencing studies of tumour-associated DCs from the past decade enables the identification of putative signalling networks that principally contribute to the establishment and maintenance of DC states. Although we acknowledge that other molecular states probably remain to be identified, we focus on two transcriptional programmes associated with mature DCs: the mregDC programme^{52,106,130,134,136–162} and ISG⁺ DCs^{34,51,163–165}. A pathway analysis on these transcriptional programmes revealed the expected association of ISG⁺ DCs with type I and II interferon-induced signalling, whereas canonical NF-κB (cNF-κB) and ncNF-κB pathways are enriched in mregDCs (Fig. 4a). Here we discuss the role of these pathways in DC maturation and function, outline how these pathways can be co-opted or manipulated in cancer, and lay out potential therapeutic opportunities to enhance DC-mediated antitumour immunity.

Fig. 4 |. Signalling dynamics underpinning DC states.

a, A simplified heatmap of various dendritic cell (DC) clusters or states found in single-cell RNA sequencing (scRNA-seq) datasets of tumoural DCs using a curated list of genes representative of each cell state. The scRNA-seq dataset of DCs collected from the MC57 fibrosarcoma model¹⁶³ was used to generate the heatmap. In brief, the mean number of mRNA transcript counts was computed for each gene across defined DC clusters identified by unsupervised clustering. Z-scores were computed using expression values across cell clusters for each gene. Red denotes high expression; blue indicates low expression. In addition to immature type 1 DC (DC1) or DC2 states, interferon stimulated gene (ISG)⁺ DCs and mature DCs enriched in immunoregulatory molecules (mregDCs) have been characterized in several scRNA-seq studies (outlined in Table 1). Pathway analysis of ISG⁺ DCs and mregDCs have shown that both type I and type II interferon signalling pathways are enriched in ISG⁺ DCs, whereas canonical NF-κB (cNF-κB) and non-canonical NF-κB (ncNF-κB) signalling pathways are enriched in mregDCs (pathways are highlighted in red). b, A working model of the signalling dynamics that potentially underpin various DC states and their transitions. Sensing pathogen-associated molecular patterns (PAMPs) derived from microbiota, plasmacytoid DCs (pDCs) secrete type I interferon, providing tonic type I interferon signalling that is pivotal for priming immature DCs to be responsive to additional activation signals (for example, PAMPs). In DCs with active tonic type I interferon signalling, DCs can respond to various triggers such as PAMPs or danger-associated molecular patterns (DAMPs), apoptotic cell debris and type I and II interferons that can induce DC maturation states; however, this response is mediated partly by an intact cNF-κB pathway. The exact relationship between ISG⁺ DCs and mregDCs remains to be elucidated (indicated by question mark), but the current evidence indicates that ISG⁺ DCs might represent a transitional state between immature DC states and mregDCs. The activated ncNF-κB pathway could potentially underlie the transition between ISG⁺ DC and mregDC states via interactions between the TNF super family–TNF receptor super family (TNFSF–TNFRSF) proteins, as activated ncNF-κB via RELB can inhibit production and autocrine signalling of type I interferon (shown in dashed box). LTβ, lymphotoxin-β; LTβR, lymphotoxin-β receptor; TLR, Toll-like receptor; TNF, tumour necrosis factor; TRAF, TNF receptor-associated factor.

Type I interferon.

The critical role of type I interferon (IFNα and IFNβ) in DC maturation and function has been appreciated since the late 1990s^166,167. DC-intrinsic type I interferon signalling enhances the expression of co-stimulatory molecules and CCR7, leads to a sustained increase in antigen processing and the continued synthesis and cycling of MHC class II molecules^168,169, and mediates the priming and differentiation of various T cell subsets (that is, T helper 1 (T_H1) cells, T_H2 cells and T follicular helper cells)^170–172. In addition, type I interferon seems to mediate DC responsiveness to various PAMPs^173,174. For example, tonic type I interferon production induced by the sensing of microbiota by pDCs helps to configure DCs in a particular epigenomic state that enables them to induce transcriptional and metabolic maturation-associated changes upon encountering PAMPs. Of note, the absence of the microbiota or type I interferon-mediated signalling does not affect the number and phenotype of mature DCs or their transcriptional state¹³². However, without type I interferon, DCs in vivo are unable to produce tumour necrosis factor (TNF) and IL-12 in response to PAMPs or CD40 agonizing antibody, or to prime CD8⁺ T cells to presented antigen — a defect that was not rescued by type II or type III interferon in steady state. Overall, type I interferon signalling puts DCs in a basal state that enables them to respond to maturation cues¹⁷⁴.

Type I interferon signalling in DCs is vital for a successful antitumour immune response, especially in the early stages of tumour development. For example, defective type I interferon signalling in DCs hampers cross-presentation of tumour antigens to CD8⁺ T cells and results in decreased DC1 accumulation within the tumour^175–177. Furthermore, a 2022 study has identified a distinct DC2-specific ISG⁺ DC transcriptional signature in a spontaneous regressing fibrosarcoma tumour model¹⁶³. These tumour-localized ISG⁺ DCs activate CD8⁺ T cells through ‘cross-dressing’ and cell-surface expression of tumour-derived peptide–MHC class I complexes. Moreover, single-cell RNA sequencing studies of DC1s in both steady-state¹²⁰ and tumoural contexts¹⁷⁸ have identified a distinct cluster of DC1s characterized by expression of ISGs, suggesting that both DC1s and DC2s can engage in a similar transcriptional programme. Interestingly, using trajectory inference analysis upon DC clusters, both studies found that DCs seem to transiently upregulate an ISG module before entering a fully mature state. Specifically, when DC maturation was driven by capture of apoptotic cell debris, downregulation of the ISG module marked a key transition point between early and late mature DC states¹²⁰. Of note, these data suggest that exiting an interferon-responsive state might be necessary to attaining full maturation (Fig. 4b). Although the transient upregulation of an ISG module might not be a prerequisite to DC maturation, further investigation on how the ISG module contributes to the immunogenic function of DCs remains to be established.

Many cancer types have developed mechanisms to dampen production of type I interferon within the TME by suppressing induction of PRR signalling (for example, cGAS–STING) or by impairment of intratumoural pDCs via immunosuppressive cytokines²⁷. Despite its antitumour effects, chronic exposure to type I interferon in cancer can skew DCs towards an immunosuppressive phenotype (increased expression of PDL1, IDO1 and IL-10) and induce various cancer-intrinsic mechanisms that enhance cancer cell survival and immune evasion¹⁷⁹, highlighting the effect of type I interferon temporal dynamics in the shaping of antitumour immune response.

Unfortunately, the clinical application of systemic type I interferon in cancer treatment has yielded mixed results, especially in solid tumours¹⁷⁹. Efforts are ongoing in both preclinical development and early-phase clinical trials to use targeted delivery of type I interferon-inducing agents or the interferon itself to DCs. Yet, critical knowledge gaps exist, including: how the transcriptional programme of ISGs elicited by type I interferon signalling contributes to the maturation and function of DCs, and how differential dose and duration of type I interferon exposure dictate the intricate balance between immune stimulation of DCs versus tumour-mediated DC immunosuppression.

Type II interferon.

The type I interferon and type II interferon (IFNγ) signalling pathways share an overlapping transcriptional signature and contribute jointly to DC maturation¹⁸⁰. However, IFNγ signalling holds a pivotal, non-redundant position in DC maturation and antitumour immunity. Similar to type I interferon, IFNγ triggers the expression of MHC class I and class II molecules, along with key components of the antigen processing machinery^181–184. Moreover, compared with constitutively expressed proteasome, IFNγ-induced immunoproteasome subunits enable presentation of a different and perhaps more immunogenic repertoire of antigens through their different catalytic activity¹⁸⁵. IFNγ signalling also orchestrates a transcriptional programme that supports essential functions of mature DCs, affecting antigen presentation, apoptosis signalling, migration and cytokine signalling, such as IL-12 and IL-15RA¹⁸⁶.

In bacterial and viral in vivo infection models, IFNγ coordinates a complex network involving natural killer cells, DCs and T cells, essential for mounting an effective immune response against foreign agents^187,188. This network is partly mediated by IFNγ, which through its amplification of IL-12 production can help to mount an effective T_H1 response against invading pathogens¹⁸⁹. Importantly, IFNγ signalling is also vital to DC-mediated antitumour immunity and the response to ICB therapies¹⁹⁰. For instance, recruitment of activated CXCR3⁺CD8⁺ T cells to the tumour site, which is critical for the response to ICB therapy, has been shown to be dependent on IFNγ-induced expression of CXCL9 and CXCL10 in DCs in a transplantable model of MC38, a mouse colon tumour line¹⁹¹. Once DC–T cell interactions are established in the tumour site, DC-intrinsic IFNγ signalling has been shown to induce a feedforward loop of IL-12–IFNγ production in tumoural DC–T cell crosstalk, which subsequently mediated the response to PD1 blockade in a MC38 model¹⁹². High correlations between ICB response and the IFNγ transcriptional signature in multiple studies of tumour biopsies further underscore the importance of IFNγ signalling in antitumour immunity^193–197.

Unfortunately, the efficacy of IFNγ as anticancer therapy in the clinic has been poor¹⁹⁸. Aside from its pharmacological instability and systemic toxicity¹⁹⁹, IFNγ can promote increased expression of immunosuppressive molecules in cancer and immune cells that further reduces the effectiveness of IFNγ therapy. In DCs, IFNγ signalling can enhance expression of PDL1 (ref. 200), SOCS2 (ref. 186) and IDO1 (refs. 201–203). Chronic IFNγ signalling in tumour cells can upregulate multiple T cell inhibitory receptors and their ligands (for example, TNFRSF14 or LGALS9), leading to PDL1-independent, cancer-intrinsic resistance mechanisms to ICB therapy across multiple transplantable cancer models²⁰⁴. Accordingly, suppression of IFNγ signalling in tumour cells led to impaired tumour growth in a KP lung carcinoma model²⁰⁵.

Given the context-dependent effects of IFNγ signalling, targeted delivery of molecules that induce IFNγ, inhibit its negative regulators or enhance IL-12 production to DCs and other antigen-presenting cells represent promising therapeutic approaches. Emerging technologies such as bispecific antibodies (BsAbs) and lipid nanoparticles (LNPs) offer selective delivery of IFNγ or enhancers of its signalling pathway to specific immune subsets, to leverage the antitumour potential of IFNγ.

cNF-κB pathway.

Discovered in 1986, the cNF-κB signalling pathway profoundly influences immune development, homeostasis and responses to inflammatory stimuli²⁰⁶. Serving as a model for inducible transcription factors, this pathway has exemplified how extracellular signals swiftly modify cellular transcriptional states²⁰⁶. In DCs, the pathway is involved in development, maturation and cell function. For instance, early studies identified that a distinct combination of cNF-κB signalling components regulate DC development and DC survival in response to an activating signal, CD40L^207,208. Moreover, activation of cNF-κB in DCs is required for expression of pro-inflammatory cytokines, not only in response to stimulatory ligands (such as lipopolysaccharide) but also in autoimmune contexts^209,210.

Activated cNF-κB is required for DCs to mount inflammatory responses to various stimuli, yet a functional cNF-κB pathway is equally vital in maintaining immune tolerance. For instance, in the absence of NF-κB-activating signals, p50 (encoded by Nfkb1) forms homodimers that act as transcriptional repressors²¹¹ of NF-κB-inducible genes and a subset of ISGs²¹². In an adoptive DC transfer model, unstimulated DCs derived from mice with a knockout of Nfkb1 produced higher amounts of IL-6 and TNF and showed enhanced priming of autoreactive T cells and diminished ability to induce T_reg cell differentiation²⁰⁹. Similarly to phenotypes seen in DCs with a knockout of Nfkb1, knockout of the gene encoding A20 (Tnfaip3), a negative regulator of cNF-κB, induced spontaneous maturation of DCs, endowing them with enhanced capacity to differentiate B cells into plasma cells and to activate and prime T_H1 cells in vivo — all of which contributed to the development of several autoimmune phenotypes^213,214. Reflective of preclinical phenotypes, several polymorphisms in human TNFAIP3 have been associated with various autoimmune diseases²¹⁵. Collectively, these data suggest that the cNF-κB pathway mediates a critical balance between an immunogenic and tolerogenic response of DCs to its environmental cues. Accordingly, inhibiting negative regulators of cNF-κB (for example, TNFAIP3) to unleash an autoimmune-like response within tumours represents a potential approach to enhance DC-mediated antitumour immunity.

In line with previous findings that implicated cNF-κB in DC maturation and regulation of adaptive immunity, under steady-state conditions, mature DCs demonstrate an enriched transcriptional profile regulated by NF-κB²¹⁶. DC-specific knockout of Ikbkb, which encodes IKKβ, in mice (a key signalling component of cNF-κB) showed decreased maturation of skin DCs, defective migration of mature DCs carrying tissue antigens to draining lymph nodes and decreased differentiation of T_reg cells, fostering spontaneous autoimmunity²¹⁶. In tumoural contexts, DC1-specific knockout of Ikbkb in an immunogenic melanoma model led to impaired DC maturation and decreased recruitment and activation of tumour-infiltrating CD8⁺ T cells, leading to increased tumour growth. Mechanistically, the cNF-κB pathway regulated an IFNγ-induced transcriptional programme, which mediated the dynamics of DC maturation and the antitumour function of intratumoral DC1s¹⁷⁸. Together, these findings demonstrate a pivotal role of the intact cNF-κB pathway in mediating several functions of mature DCs: to maintain tolerance towards self-antigens in steady state and induce an immunogenic response against tumour antigens.

Whether and how cancer modulates the cNF-κB pathway in DCs remains unclear. In light of the extensive contribution of cNF-κB in other immune cell types, drugging the cNF-κB pathway remains a fairly underdeveloped therapeutic strategy for cancer. However, with increased targeting enabled by emerging technologies (BsAbs²¹⁷ or LNPs²¹⁸) to DCs or other immune cells, an attractive approach for investigation could be to antagonize inhibitors of cNF-κB (such as A20) to help unleash antitumour immunity.

ncNF-κB pathway.

The cNF-κB and ncNF-κB pathways have interconnected signalling nodes, yet the unique function of ncNF-κB in immunity and inflammation became evident through studies on mice with loss-of-function mutations in NIK (the main signalling component of ncNF-κB), which have a severe alymphoplasia phenotype²¹⁹. These mice displayed similar phenotypes to TNF receptor superfamily member 3 (TNFRSF3)-deficient mice, highlighting TNFRSFs as main activators of the ncNF-κB pathway. TNFRSFs and their ligands have since been shown to coordinate interactions between innate and adaptive immune cells and stroma for development, organ homeostasis and host immunity, establishing ncNF-κB as one of the key communication modalities for immune cell interactions. In support of this notion, the kinetics of ncNF-κB activation, unlike the rapid and transient kinetics of cNF-κB (less than 30 minutes), are slow and persistent, suggesting that ncNF-κB activation can mediate lasting consequences post cell-to-cell interactions²²⁰.

The role of ncNF-κB on DC maturation and function has been most studied in the context of CD40, a key molecule in CD4⁺ T cell licensing of DCs for CD8⁺ T cell priming. Upon crosslinking of CD40 with CD40L from activated CD4⁺ T cells, CD40 on DCs, via activation of ncNF-κB, help to upregulate maturation and co-stimulatory molecules^221,222, increase expression of IL-12 (ref. 223) and have enhanced capacity to induce T cell proliferation and cytokine production²²⁴. Consequently, in immunogenic fibrosarcoma tumour models, lack of CD40 signalling in cDC1s leads to impaired survival and maturation of migratory DCs, causing defective priming of CD4⁺ and CD8⁺ T cell responses and failure of the antitumour immune response to reject the tumour^225,226. Furthermore, deficient CD40 signalling also resulted in mitochondrial dysfunction in DCs, suggesting a role in metabolic rewiring upon ncNF-κB activation²²⁶.

In addition to CD40, downstream components of the ncNF-κB pathway similarly affect DC maturation and function. Although knockout of the gene encoding NIK or IKKα in mice did not affect DC development or maturation markers under steady-state conditions, ncNF-κB was required for IDO1 and IL-12 expression upon CD40 engagement and for DC-mediated priming of antigen-specific T_H1 and T_H17 responses^227–230. Analogous to CD40 findings, the ncNF-κB pathway has been shown to mediate MST1 and MST2-dependent mitochondrial fitness in DC1s, which demonstrates that ncNF-κB mediates metabolic rewiring that is critical for DC-mediated T cell responses²³¹. Of note, several signalling genes associated with the ncNF-κB pathway (for example, BIRC2, BIRC3, RELB and NFKB2) are highly enriched in the mregDC state (Table 1). Considering that the mregDC state represents DCs that have captured the antigenic information to present to their T cell counterparts¹³⁰, upregulation of the ncNF-κB pathway might be one of the key signalling dynamics facilitating mature DC functions.

Interestingly, activation of the ncNF-κB pathway can inhibit production of type I interferon in DCs. Signalling components of ncNF-κB (RELB and p52) are recruited to promoters of Ifnb at the expense of its potent activator, RELA²³². Accordingly, loss of RELB in DCs resulted in heightened expression of ISGs, which was dependent on enhanced type I interferon signalling²³³. Considering that a transcriptional programme of ISGs has been observed to be enriched in early mature cDC1s and downregulated in late mature cDC1 subsets that express the mregDC transcriptional programme¹²⁰, it is tempting to speculate whether inhibition of type I interferon production by activated ncNF-κB might contribute to the switch in transcriptional programmes seen between ISG⁺ DC and mregDC clusters (Fig. 4b).

How cancer modulates the ncNF-κB pathway in DCs remains unclear. However, enhancement of the ncNF-κB pathway as an antitumour therapeutic strategy was most widely adopted in the form of CD40 agonists. Unfortunately, CD40 agonists have exhibited prohibitive toxicity levels in clinical trials^234,235, leading groups to develop strategies to target CD40 agonists to dendritic cells. Using BsAbs or DC-targeting LNPs could be promising approaches to harness the immunogenic potential of CD40 signalling^236,237. Beyond CD40 agonists, enhancing ncNF-κB signalling through modulation of its negative regulators (for example, via SMAC mimetics) has been shown to improve responsiveness to ICB treatment across multiple preclinical cancer models²³⁸ and further underscoring the therapeutic potential of targeting this pathway¹⁹². Importantly, activation of ncNF-κB could limit the antitumour effect of radiation therapy by suppressing type I interferon production, highlighting the context-dependent role of ncNF-κB in DC-mediated antitumour immunity²³⁹. Overall, the ncNF-κB pathway considerably affects the dynamics of DC maturation, survival and downstream T cell responses. Therapeutic strategies targeting this pathway show promise; their clinical efficacy and potential interactions in various treatment contexts need further exploration.

Although we have only discussed signalling networks that were explicitly nominated in the pathway analysis of mature DC states, others exist that contribute to DC maturation — some of which have been modulated as potential therapeutic approaches against cancer. As exemplars, we briefly highlight MAPK and mTOR pathways as their inhibitors have gained traction in the clinic as anticancer therapies (Supplementary Box 1).

Metabolic rewiring

It is increasingly recognized that DC maturation requires metabolic rewiring, which occurs in parallel with the transcriptional changes and signalling events that characterize maturation. Yet, the exact molecular circuitry linking distinct transcriptional and signalling modules with specific metabolic pathways remains to be elucidated. Using a clever combination of in vitro metabolomic studies with transcriptional and proteomic approaches and functional in vivo validations, researchers have begun to tease apart how key signalling pathways contribute to metabolic rewiring of DCs and how different metabolites regulate DC maturation and function. Furthermore, in light of an unexpected finding that myeloid cells outcompete T cells and cancer cells in the consumption of glucose in various tumour contexts²⁴⁰, understanding how DCs shape the tumour metabolic milieu and how, in turn, cancer cells exploit DC-intrinsic metabolic programmes to impair antitumour immunity is vital to establish immune–metabolic modulation as an antitumour therapeutic strategy.

In this section, we provide a non-exhaustive update of the emerging field of DC immunometabolism in cancer and highlight three major pathways most extensively investigated so far (Fig. 5). As it has become apparent that GM–CSF-driven bone marrow-derived DCs are metabolically distinct from cDC subsets found in vivo (reviewed in ref. 241), we have specifically limited our review to studies whose findings either are based on bona fide cDCs or have been validated using in vivo models.

Fig. 5 |. Metabolic rewiring during DC maturation.

a, Dendritic cell (DC)-intrinsic metabolic rewiring during DC maturation in steady state and cancer. The maturation of DCs involves considerable changes in their metabolic activities, notably in glucose, glutamine (left column) and lipid metabolism (right column). Mature DCs show increased glycogen and lipid content. In non-tumoural contexts, DCs that are stimulated by activating molecules (for example, LPS and dead cells) rapidly utilize glycogen to boost intracellular glucose levels to facilitate apoptotic cell uptake, migration and generation of tricarboxylic acid (TCA) metabolites for fatty acid synthesis. In cancer, tumour-derived factors (for example, lactate) negatively affect DC function by inhibiting antigen presentation and expression of pro-inflammatory cytokines upon pattern recognition receptor (PRR) stimulation. As glutamine can supplement the TCA cycle, it is highly sought after by cancer cells and myeloid cells within the tumour microenvironment (TME). Low intracellular levels of glutamine severely impair DC antigen presentation. For lipid metabolism in DCs, lipid bodies containing cholesterol and fatty acids are vital for antigen presentation. Upon maturation, DCs increase their intracellular cholesterol levels either from de novo synthesis, extracellular uptake via an apolipoprotein E–low-density lipoprotein receptor (APOE–LDLR) axis or digestion of dead cell debris. Intracellular cholesterol can be trafficked via NPC1 to the lipid nanodomains in the plasma membrane to enhance expression of maturation markers (for example, CD40, CD80, PDL1 and major histocompatibility complex (MHC) molecules), which is counterbalanced by liver X receptor-β (LXRβ)-mediated cholesterol efflux, to limit potential endoplasmic reticulum (ER) stress and modulate the mature DC phenotype. Tumour cells produce oxidized lipids and oxysterols, impairing DC function by inhibiting antigen presentation and decreasing maturation marker expression. Intracellular oxidized lipids are integrated into lipid bodies and can impair cross-presentation of dead cell antigens via heat shock protein 70 (HSP70). Sensing of oxidized lipid species in the ER triggers the unfolded protein response and induces an XBP1-mediated lipid biosynthetic programme that engenders further accumulation of oxidized lipid species. Oxysterols promote the LXRβ pathway and induce increased efflux of cholesterol, which can diminish expression of maturation markers. b, DCs can influence the TME through metabolic modulation. Mature DCs upregulate indoleamine 2,3-dioxygenase 1 (IDO1) and IL4I1, which degrade tryptophan into kynurenines. Kynurenines can promote angiogenesis, induce regulatory T cell differentiation and sustain immunosuppressive pathways in nearby DCs, collectively contributing to DC-mediated immunosuppression within the tumour. DAMPs, danger-associated molecular patterns; AHR, aryl-hydrocarbon receptor; TLR, Toll-like receptor; T_reg cell, regulatory T cell.

Glucose and glutamine metabolism.

Upon stimulation with various maturation signals (such as PAMPs), DCs rapidly increase glycolytic activity to support various aspects of maturation. Heightened glycolytic activity supports an array of phenotypic and functional changes that include: augmented co-stimulatory marker expression, secretion of cytokines and CCR7-dependent migration to the periphery²⁴¹. A rapid and early increase in glycolytic activity upon DC activation is reinforced by glycogen stores within the cell. Moreover, inhibition of glycogenolysis substantially attenuates early DC maturation driven by TLR signalling and DC priming of T cells²⁴². Furthermore, glycogenolysis contributes to phagocytic capacity of DCs²⁴³. Intriguingly, the main function of enhanced glycolytic activity is not necessarily to generate ATP but to increase the production of pyruvates that can be used to fuel the tricarboxylic acid (TCA) cycle and generate its key intermediates²⁴⁴. TCA cycle intermediates (for example, citrate) can be shunted to fuel de novo synthesis of fatty acids, which are used by DCs to expand the endoplasmic reticulum (ER) and Golgi apparatus, presumably to support the anticipated synthesis, transport and secretion of proteins involved in DC maturation. In support of such a mechanistic link between glucose and lipid metabolism, DCs have been reported to become ‘lacy’ in appearance due to increased intracellular storage of fat and glycogen during maturation²⁴⁵. In the context of cancer, lactate (a predominant glycolytic metabolite in the TME) has been shown to suppress the secretion of type I interferon and IL-12 in response to STING or TLR3 stimulation and lead to dysfunction of antigen presentation and T cell priming in a Lewis lung carcinoma model²⁴⁶. Furthermore, induction of SOCS3 expression in DCs by tumour-derived factors led to inhibition of pyruvate kinase type M2, a key enzyme in glycolysis, which led to dysfunctional antigen presentation²⁴⁷.

The TCA cycle is the central metabolic hub in all eukaryotic cells and its adequate fuelling is vital to the function and persistence of all cells. Therefore, when nutrients become constrained, cells go under strong selection pressures for the acquisition of factors that sustain their core metabolism. Both glucose and glutamine are major substrates for the TCA cycle and thus are often competed for by different cell populations within inflammatory contexts²⁴⁸. Inhibition of glutamine uptake in the TME across multiple cancer models in mice led to increased uptake of glucose by myeloid cells, which strongly suggests that these metabolites are limiting factors for function within the TME, driving metabolic competition between cells²⁴⁰. Cancer cells and DCs compete for glutamine through differential expression of SLC38A2 (a glutamine transporter), which is a prime example of how nutrient competition can serve as a critical checkpoint in the antitumour response²⁴⁹. Specifically, when outcompeted by cancer cells for glutamine uptake, DC1s enter a state of dysregulated endolysosomal homeostasis and defective antigen presentation. However, metabolite competition between DCs and other cells does not always result in immunosuppressive outcomes. For instance, T cell-mediated restriction of extracellular glucose enhances DC maturation and T cell priming under inflammatory contexts, suggesting that the overall outcome of metabolite competition between DCs and other cells might be highly context dependent²⁵⁰. This concept is supported by observations in mouse models of diabetes mellitus that hyperglycaemic conditions impair DC function during viral infection, which can be rescued by treatments that lower glucose abundance²⁵¹.

A firm understanding of the competitive or cooperative dynamics that underlie metabolic interactions between DCs and their partners in vivo will be critical to decipher how metabolism mediates DC maturation and function across various disease contexts.

Lipid metabolism.

Fuelled in part from augmented glycolytic activity, DCs stimulated with maturation signals (such as PAMPs) elevate the synthesis of fatty acids. Such metabolic rewiring is postulated to serve two main functions: (1) the expansion of the ER and Golgi apparatus to support an increased demand for protein synthesis and secretion, and (2) production of lipid droplets (also referred to as lipid bodies). Both of these functions have wide-ranging implications in DC maturation and function²⁵². In steady state, accumulation of lipids (fatty acids and cholesterol) within DCs has been associated with increased immunogenicity through increased expression of maturation markers²⁵³, increased MHC class II-dependent antigen presentation²⁵⁴ and increased expression of pro-inflammatory cytokines²⁵⁵. Moreover, lipid bodies produced by accumulation of fatty acids and cholesterol have been shown to enhance cross-presentation potential within DC1s²⁵⁶. Finally, regulation of intracellular cholesterol by cholesterol-trafficking machinery (that is, NPC1) and the liver X receptor (LXR) signalling pathway can control DC maturation. According to studies, inhibition of cholesterol trafficking to the membrane led to decreased DC maturation and impaired antitumour immunity¹³⁴, whereas heightened cholesterol levels in DCs with a knockout in the genes encoding LXR were associated with induction of ISGs characteristic of immunogenic maturation¹²⁰.

In the TME, which is replete with various lipid species²⁵⁷, intratumoural DCs readily accumulate lipids (including oxidized and acetylated lipids) through intake from their environment via a scavenger receptor (MSR1)²⁵⁸ and through intracellular generation of peroxidized lipid species²⁵⁹. Accrual of these altered lipid species leads to dysfunctional endolysosomal trafficking of peptide–MHC class I and impairs cross-presentation of dead cell antigens via interaction of oxidized lipid bodies with HSP70 (refs. 260,261). Moreover, sensing of the oxidized lipid species in the ER triggers the unfolded protein response and induces an XBP1-mediated lipid biosynthetic programme that results in further accumulation of lipids within the cell²⁶². Eventually, the high lipidic environment can lead to metabolic remodelling of DCs into a perturbed state with a dysfunctional TCA cycle that generates more mitochondrial reactive oxygen species (ROS). Elevated ROS can then lead to further production of oxidized lipid species and constitutive activation of XBP1, thereby exacerbating the cycle of lipid-driven dysfunction of DCs. Intracellular accumulation of altered lipid species is strongly associated with compromised DC function; however, increased efflux of intracellular cholesterol through tumour-mediated modulation of the LXR signalling pathway has been shown to downregulate CCR7 and prevent migration of tumour antigen-loaded DCs to the draining lymph nodes²⁶³.

Together, proper regulation of lipid homeostasis represents a critical metabolic checkpoint for DC maturation and function that is often exploited by cancer to dampen DC-mediated antitumour immunity.

Tryptophan metabolism.

The concept that DCs can shape their metabolic milieu to modulate downstream adaptive immune responses emerged from the discovery of IDO1 and its primary function in tolerance and immune suppression²⁶⁴. Induced by both immunostimulatory (IFNγ and ncNF-κB) and immunosuppressive (WNT–β-catenin and TGFβ) signalling pathways, IDO1 upregulation in DCs is a central feature of maturation along with co-expression of other co-stimulatory molecules (CD80, CD86 and CD40). IDO1 represents a DC-intrinsic feedback mechanism to restrain potential overactivation of T cells and subsequent autoimmune responses. Enzymatically, IDO1 metabolizes tryptophan to kynurenines²⁶⁵. Production of kynurenines can induce differentiation and activation of T_reg cells through the aryl-hydrocarbon receptor (AHR) pathway²⁰³, which is thought to be the main mechanism of immunosuppression through IDO1. In addition to suppression of T cells, kynurenines can maintain expression of IDO1 in DC1s through the AHR pathway even in the absence of its initial inductive signalling²⁰² and can educate DC2s (which do not express IDO1 upon PRR stimulation) to express the enzyme once it is expressed in the DC1 compartment²⁶⁶.

Despite the mounting evidence that IDO1 represents a critical mechanism of DC-mediated tolerogenesis, a phase III clinical trial of epacadostat, an inhibitor of IDO1, combined with PD1 blockade in patients with metastatic melanoma failed to show any additional clinical benefit of epacadostat²⁶⁷. Several reasons for this failure have been proposed²⁶⁸ ranging from incorrect dosage to potential compensatory mechanisms by other tryptophan-degrading enzymes (such as IDO2 and TDO) or enzymes within the tryptophan metabolism pathway (such as IL4I1)²⁶⁹. Interestingly, a 2023 study of patients with advanced ovarian cancer treated with epacadostat has shown that IDO1 inhibition induced activation of other immunosuppressive metabolic pathways: notably, the elevation of nicotinamide adenine dinucleotide (NAD⁺). Co-inhibition of NAD⁺ generation with IDO1 inhibition led to improved survival in mice models of ovarian cancer²⁷⁰, highlighting the need for more preclinical and translational studies that help to elucidate the main resistance mechanisms to IDO1 inhibition.

Mature DC functions

Migration and T cell engagement

Upon going through a cascade of molecular transformations that accompany the maturation process, mature DCs perform an array of functions that are carefully choreographed to ensure efficient priming of the adaptive immune response against the presented antigen. Upon capturing cell debris from the periphery in steady state and in the context of cancer, DCs can enter an mregDC state and respond via the upregulation of CCR7 in response to its ligands, CCL19 and CCL21, chemotactic signals that are expressed by cells within the lymphatics and secondary lymphoid organs^271,272. This interaction guides the mature DCs into micro-anatomical zones within these structures, where they can efficiently present antigens to T cells. Specifically, the expression of CCR7 by DCs enables them to localize within the T cell zone, whereas CXCR5 and GPR183 (a receptor for oxysterols) help to maintain DCs at the T–B cell border, preventing their deep infiltration into the B cell zone^273–275.

Once mature DCs position themselves in appropriate niches to interact with their T cell partners, they are equipped to prime naive T cells by providing three canonical signals vital for T cell activation and differentiation. First, DCs present antigens that have been captured and processed from the periphery on MHC class I and class II molecules for cognate T cell receptor recognition (‘signal 1’). In the context of cancer, mature DCs simultaneously present tumour-associated antigens on both classes of MHC molecules to convey help from CD4⁺ T cells towards CD8⁺ T cells in the lymph nodes, a critical step for efficient priming of cytotoxic CD8⁺ T cells against tumour antigens^147,177,225. Once MHC–T cell receptor interactions have been made between mature DCs and T cells, mature DCs deliver co-stimulatory and co-inhibitory signals (‘signal 2’) to T cells. Through the balanced expression of numerous co-stimulatory (for example, CD40, CD80 and CD86) and co-inhibitory (for example, PDL1, PDL2, PVR and CD200) molecules, DCs can regulate downstream T cell activation and prevent unwarranted autoimmunity or tolerance. Owing to their ability to fine-tune T cell responses, a considerable portion of currently available immunotherapeutics is built on agonizing or blocking these very co-stimulatory and co-inhibitory molecules^9,15.

In addition to fine-tuning T cell responses through co-stimulatory and co-inhibitory signals, mature DCs also modulate T cell differentiation and effector functions through the secretion of cytokines (‘signal 3’). In the context of cancer, IL-12 and IL-15 are central players. As a potent inducer of the T_H1 response, IL-12 induces production of IFNγ, which along with IFNβ and IL-15, helps to enhance survival and effective activation of CD8⁺ T cells, natural killer cells and natural killer T cells^189,276. In addition to IL-12, IL-15 has been reported as critical for the survival of effector-like TCF1⁻CD8⁺ T cells in the TME²⁷⁷. TCF expression is used in combination with other markers to discriminate T cell progenitor populations from their terminally differentiated counterparts²⁷⁸. Although DCs are critical for priming tumour-specific T_H1 cells and cytotoxic CD8⁺ T cells, they are also able to prime tumour-specific T_reg cells, thereby contributing to tolerance of tumour antigens. Owing to their potent ability to shape and magnify antitumour immunity, a notable portion of current immunotherapeutic approaches seeks to either exogenously deliver these immunostimulatory cytokines to tumours or amplify their intratumoural expression by relevant immune cells²⁷⁹.

Interestingly, data from the past decade suggest that in chronic inflammatory diseases such as cancer or chronic viral infections, DCs could also contribute to preventing the exhaustion of antigen-specific T cells. In a chronic lymphocytic choriomeningitis virus infection model, DC1s promote the maintenance of exhausted T cell progenitors by insulating them in a splenic niche via MHC class I-dependent interactions to prevent their activation and further exhaustion²⁸⁰. Furthermore, an increased number of activated DC1s in tumour-draining lymph nodes driven by FLT3L–CD40 therapy correlated with an increase of TCF1⁺CD8⁺ exhausted T cell progenitor frequencies and a decrease in tumour burden in an autochthonous model of lung adenocarcinoma²⁸¹.

Local reactivation of T cells in tumours

DCs are guided to subanatomical niches within secondary lymphoid organs to optimally interact with T cells for their subsequent priming, yet they also have a critical role in recruiting and subsequently activating T cells upon their arrival in peripheral tissues. In some cancer types, such as β-catenin⁺ tumours, an absence of DCs at the tumour site is thought to drive resistance to ICB, which is partly driven by decreased recruitment of effector T cells by DCs^19,282. In addition to recruitment, the quality of DC–T cell interactions at the tumour site is likely to be pivotal to dictating the effector response against tumour antigens. For example, a 2023 study reported that CD8⁺ T cells initially primed in lymph nodes must be reactivated by antigen-presenting cells at the tissue site to differentiate to the effector state²⁸³. This conceptual framework begs the question of what types of mature DC states enable optimal interactions with tumour-infiltrating T cells that unleash their antitumour functions.

DCs in the mregDC state were found to physically interact with exhausted T cell progenitors and CXCL13⁺ T_H cell subsets in lesions of patients with lung or liver cancer^136,147. Considering that the mregDC state is acquired upon capture of dead cell-associated antigen¹³⁶, the data strongly suggest that these interactions are driven by antigen presentation. Furthermore, these interactions were enriched in patients with hepatocellular carcinoma that responded to PD1 blockade compared with non-responders raising an intriguing hypothesis on whether terminal differentiation of exhausted T cell progenitors to their effector-like fates can be constrained by mregDCs via a PD1–PDL1 axis, which could be unleashed upon PD1 blockade¹⁴⁷. Multimodal approaches that provide insights on cellular crosstalk and spatial localization (for example, physically interacting cells sequencing²⁸⁴, universal labelling immune partnerships by SorTagging intracellular contacts (uLIPSTIC)²⁸⁵, multiplex epitope-based tissue imaging²⁸⁶ or spatial transcriptomics²⁸⁷) will shed light on the heterogeneity of DC–T cell interactions at the tissue site. However, the initial evidence suggests that DCs engaged in a particular state preferentially interact with specific T cell subsets — interactions that can regulate antitumour functions of these T cells and responsiveness to immunotherapies.

Interestingly, studies focused on understanding these critical intratumoural DC–T cell interactions indicate that they are often enriched in subtissular niches^{138,154,164,277,288,289}. Potential therapeutic roles of these subtissular niches have been underscored by accumulating observations that tertiary lymphoid structures (TLSs) in tumours correlate with better prognosis and clinical outcome upon immunotherapy²⁹⁰. Whether all the immune aggregates being reported in these studies represent true, mature TLS or immature versions in progression to mature structures (gradations between immature and mature TLS are reviewed elsewhere²⁹¹) remains unclear, as does whether the quality of these aggregates has differential prognostic potential. Yet, emerging evidence suggests that DCs in mature states are often enriched within these aggregates and can even orchestrate their interactions with particular T cell subsets via specific combinations of chemokines. For instance, in lung cancer lesions of patients who respond to PD1 blockade, DCs in the mregDC state aggregated with TCF1⁺PD1⁺CD8⁺ T cells and TCF1⁺CD4⁺ T cells, aggregates that differ in quality from TLSs^138,288. Moreover, mregDCs coordinated the recruitment of highly functional effector-like T cells that derive from TCF1⁺ stem-like cells via a CXCL16–CXCR6 axis into their perivascular niches and promoted the survival of these T cell subsets through trans-presentation of IL-15 in both melanoma and lung carcinoma mice models²⁷⁷.

Although these specialized niches can nurture critical DC–T cell interactions that support a robust antitumour T_H1 and cytotoxic CD8⁺ T cell response, they can also foster immunosuppressive DC–T_reg cell interactions. TLSs can be enriched with T_reg cells that constrain their expansion²⁸⁹ and DCs can recruit T_reg cells into tumours²⁹² — findings that are consistent with a recent study showing that DCs in the mregDC state were co-enriched with T_reg cells in peri-lymphatic niches in a genetically engineered mouse model of colorectal cancer and were the most likely cell type to recruit and activate T_reg cells within these niches¹⁵⁴. Collectively, these findings suggest that depending on their composition and maturation trajectory, these subtissular niches can have differential functional consequences on antitumour immunity.

In addition to modulating T cell states within the subtissular niches, mature DCs can also contribute to the construction or maintenance of such niches themselves, for optimal interactions with lymphocytes. For instance, DCs in their mregDC state express high levels of CCL19, which can recruit CCR7⁺ naive and central memory T cells^130,293, and CCL17 and CCL22, which can recruit CCR4⁺ CD4 T cells²⁹⁴. Furthermore, mature DCs also express CXCL9 and CXCL10, which are ligands for CXCR3⁺ cells that include activated T cells, natural killer or natural killer T cells^19,104,295 and pre-cDC1s²⁹⁶. In addition to recruiting essential lymphocytes, DCs in the mregDC state produce lymphotoxin-β (LTβ), a molecule vital for formation of secondary lymphoid structures. As LTβ can act on high endothelial venules to promote their maintenance and lymphocyte homing to lymph nodes in vivo²⁹⁷, LTβ-expressing mregDCs might be able to establish and maintain stromal components of TLS-like structures. Indeed, DCs produce LTβ in human lung²⁹⁸ and breast²⁹⁹ cancer lesions, and the density of mregDCs has been found to correlate with high endothelial venule density in breast cancer and melanoma lesions³⁰⁰. Together, DCs in the mregDC state might contribute to TLS formation and maintenance in cancer through various molecules that support both lymphocyte recruitment and stromal maintenance.

Conclusions

In the ensuing decades since the discovery of DCs by Cohn and Steinman, it became gradually appreciated that only DCs in a particular state, termed mature DCs, could effectively interact with naive T cells and prime them for subsequent immune responses. This insight has driven efforts to identify molecular regulators of DC maturation, aiming to modulate their role in adaptive immunity based on disease context. With their pivotal role in antitumour immunity and response to immunotherapies now recognized, strategies such as BsAbs and LNPs targeting mature DC states are beginning to hold tangible therapeutic potential in cancer.

Numerous additional questions remain unresolved regarding the biology of DC states during their maturation and their translational potential in cancer. For instance, whether immunogenic versus regulatory modules that are often co-upregulated in DC maturation programmes can be uncoupled is currently unclear. Future research should also investigate how DC1s can be specifically expanded in tissues, considering their pivotal role in antitumour immunity. Moreover, an in-depth spatiotemporal characterization of DC states will help to probe how these states relate to one another within the broader DC maturation dynamics and shed light on potential regulatory mechanisms that govern their transition. In addition, exploring how other parameters of cell states that are not assessed by transcriptomic methods, including their epigenome, metabolism, spatial positioning within tissue and interactions with other cell types, affect DC maturation and function could open up more nuanced therapeutic approaches. Finally, the effects of various physiological and pathological states (such as ageing, obesity and cardiovascular disease) on DC states and their function in cancer should be examined.

These unresolved questions represent an exciting frontier of DC immunology, exploration of which could have an important effect in the successful development and deployment of a DC-centric arsenal of immunotherapeutics against cancer.

Glossary

Bispecific antibodies

(BsAbs). Engineered antibodies that contain two binding sites directed at two different epitopes.

CCR7-mediated migration

A main mode of migration for dendritic cells to the secondary lymphoid organs upon maturation in response to CCL19 and CCL22, ligands for CCR7.

cGAS–STING pathway

A signalling pathway involving cyclic GMP–AMP synthase (cGAS) and stimulator of interferon genes (STING) that detects cytosolic DNA and activates the production of type I interferons in response to infection or cellular stress.

Damage-associated molecular patterns

(DAMPs). Molecules released by dying cells that trigger immune responses.

DC maturation

A process by which dendritic cells (DCs) gain the ability to interact with other haematopoietic cells (predominantly T cells) and shape the downstream immune response towards either a tolerogenic or immunogenic outcome.

Exhausted T cell progenitors

A distinct subpopulation of T cells found in chronic infection and cancer that can self-renew, proliferate and differentiate into either terminally exhausted T cells or cytolytic effector T cell fates.

Immunogenic responses

In dendritic cell-mediated immunity, immunogenic responses arise when a dendritic cell-presented antigen elicits T cell activation, proliferation and release of immunostimulatory cytokines.

Indoleamine 2,3-dioxygenase 1

(IDO1). An enzyme that degrades tryptophan to kynurenines, which can promote activation and differentiation of regulatory T cells.

Lipid nanoparticles

(LNPs). Nanoparticles composed of various mixtures of lipids optimized for nanoparticle stability, cell entry and endosomal escape; these entities are as a technical platform to deliver drugs and/or genetic material (for example, mRNA).

Metabolic rewiring

Changes in the metabolic state of a cell, often brought on to adapt to environmental perturbations or to facilitate various functions of a cell upon differentiation or transition to another cell state.

Ontogeny

The developmental lineage that gives rise to a specific cell type.

Pathogen-associated molecular patterns

(PAMPs). Highly conserved, molecular motifs that derive from microorganisms and are recognized by pattern recognition receptors.

Pattern recognition receptors

(PRRs). A diverse group of receptors expressed both on the cell surface and inside the cell that recognize pathogen-associated molecular patterns and damage-associated molecular patterns and initiate immune responses.

Phagosomal acidification

The phagosome is a cellular compartment containing engulfed material and its acidification facilitates the degradation of internalized cargo by acidic hydrolases.

SMAC mimetics

A class of small-molecule inhibitors that structurally resemble the amino-terminal inhibitor of apoptosis (IAP)-binding motif and can inhibit various IAP proteins.

SOCS1 and SOCS3

Proteins that regulate cytokine signalling by inhibiting the JAK–STAT pathway.

Tertiary lymphoid structures

(TLSs). Ectopic lymphoid structures found in inflammatory tissue sites that feature several organizational aspects akin to a lymph node including T cells, B cells, dendritic cells and high endothelial venules.

Tolerogenic responses

In dendritic cell-mediated immunity, tolerogenic responses arise when a dendritic cell-presented antigen induces a T cell regulatory response.

Transcriptional programme

A set of genes, of which changes in expression are highly correlated with one another; is differentially expressed by a distinct cluster of cells; and is linked with a putative function (or functions) of the cell cluster in question.