Abstract
Immune checkpoint inhibitors are successful immunotherapy modalities that enhance CD8+ T-cell responses. Although T cells are initially primed in draining lymph nodes, the mechanisms that underlie their reactivation inside the tumor microenvironment are less clear. Recent studies have found that not only is the cross-priming of conventional type 1 dendritic cells (cDC1) required to initiate CD8+ T-cell responses during tumor progression, but it also plays a central role in immunotherapy-mediated reactivation of tumor-specific CD8+ T cells for tumor regression. Moreover, many cancer treatment modalities trigger type I IFN responses, which play critical roles in boosting cDC1 cross-priming and CD8+ T-cell reactivation. Inducing type I IFNs within tumors can overcome innate immune resistance and activate antitumor adaptive immunity. Here, we review recent studies on how type I IFN-cDC1 cross-priming reactivates CD8+ T cells and contributes to tumor control by cancer immunotherapy.
Introduction
Cancer immunotherapy modulates the host's immunity to fight cancer, with immune checkpoint inhibitors (ICI) as successful examples. ICIs use antibodies to block immune inhibitory signals in T cells, such as cytotoxic T-lymphocyte–associated protein 4 (CTLA-4) and the programmed cell death protein 1 (PD-1) pathway, to reinvigorate antitumor T-cell responses (1). T cells, particularly CD8+ cytotoxic T cells, are important targets of cancer immunotherapy because they discriminate tumor cells from normal cells and, thus, kill tumor cells by recognizing tumor-derived mutant neoantigens (2). CD4+ T cells can help not only CD8+ T-cell activation, but also direct tumor killing in some models (3, 4). However, only a few patients completely respond to immunotherapy, which indicates either that directly modulating surface immune inhibitory signals alone does not sufficiently reactivate T cells within tumors, or that tumor-specific T cells are not being generated in the first place. Research is elucidating mechanisms through which tumor-specific CD8+ T-cell responses are reactivated after treatment as a crucial step toward developing new strategies to boost responses to immunotherapy.
Dendritic cells (DC) are specialized antigen-presenting cells with a key role in priming T cells. Specifically, conventional type 1 DCs (cDC1) excel in cross-presenting antigens to initiate CD8+ T-cell responses, while cDC2s preferentially prime CD4+ T-cell responses (5). DCs' recognition of foreign antigens by pattern recognition receptors (PRR) often triggers the secretion of type I IFNs, which are required for activating DCs to prime T cells (6). cDC1s that cross-prime CD8+ T-cell responses are involved in spontaneously rejecting tumor immunity (7). However, after a tumor has been treated, whether the reactivation of dysfunctional CD8+ T cells requires cDC1 cross-priming and how this occurs are still unclear. cDC1s lie within both peripheral lymphoid and tumor tissues; CD8+ T-cell reactivation can occur within both sites (8, 9). We have summarized the studies that reported immunotherapy's antitumor effects, focusing on cDC1 cross-priming that fully reactivates CD8+ T cells. Interestingly, type I IFNs triggered by anticancer treatments are required to stimulate cDC1s (Fig. 1). Combining immunotherapy with traditional therapy modalities that trigger type I IFN signaling and DC cross-priming may enhance CD8+ T-cell responses and synergistically improve treatment efficacy.
CD8+ T-Cell Responses Activated by cDC1s Are Critical in Antitumor Immunity
Conventional or classical DCs (cDC) are subsets of DCs arising from a common progenitor in the bone marrow that are especially adept at presenting exogenous and endogenous antigens to activate T cells (10). cDCs comprise two main distinct subsets, cDC1 and cDC2 (10, 11). Mouse and human cDC1s are well known for their ability to cross-present antigens and activate CD8+ cytotoxic T cells, although recently, the early priming of CD4+ T cells by cDC1s was found in the setting of cell-associated antigens (12). cDC1s express CD11c, major histocompatibility class II (MHC II), CD8α, XCR1, CLEC9a, CD24, and CD103. Human cDC1s also express CD141. The transcription factors, IRF8, ID2 and Batf3 are required for the development of cDC1. cDC1s include both lymphoid tissue–resident CD8α+ CD11b− cDCs and nonlymphoid tissue–resident CD103+CD11b− cDCs; the latter can migrate to draining lymph nodes after maturation upon activation (5). cDC2s correspond to the CD11b+ cDCs and preferentially initiate CD4+ T-cell responses in vivo (13). However, both mouse and human cDC2s can also cross-present soluble antigens to CD8+ T cells (11, 12). In addition to CD11c, MHC II, and CD11b, cDC2s also express Sirpα and CD301b (4). CD1c is a marker of cDC2 in humans, but not in mice. The development of cDC2 requires the transcription factors IRF4, RBPJ, KLF4, and RELB (4).
During cross-priming, exogenous antigens are taken up by antigen-presenting cells (APC), processed, and presented on MHC I, which eventually activates CD8+ T-cell responses (14). cDC1s and cross-priming play critical roles in the initiation of antitumor CD8+ T-cell responses (15). The lack of cDC1s in Batf3−/− mice prevents the priming of tumor-specific CD8+ T cells or tumor control (16). The unique role of cDC1s in cross-presenting tumor antigens might be related to the reduced antigen degradation in the endosomal compartment and the unique processing of cell-associated antigens (17, 18). The molecule WDFY4 was identified as being involved in cDC1s' cross-presentation of cell-associated antigens, perhaps through the regulation of vesicular trafficking pathways, but finding the molecules involved in this process will require much more research (18).
In addition to priming and activating, cDC1s also contribute to antitumor CD8+ T-cell responses through recruiting and positioning CD8+ T cells within tumor tissues. In a melanoma tumor model, CD103+ cDC1s within tumor tissues secreted the chemokines CXCL9 and CXCL10, which recruited adoptive transferred CXCR3+ effector CD8+ T cells for tumor clearance (19). Meanwhile, tumor cDC1s express higher amounts of IL12 than other DC subsets or macrophages and may help enhance the cytotoxic function of CD8+ effector T cells within the tumor microenvironment (TME; ref. 20).
Type I IFN Signaling to Regulate cDC1s' Cross-priming of Tumor Antigen–Specific CD8+ T Cells
Human and mouse type I IFNs include class II α-helical cytokines, such as IFNα (12 subtypes in humans and 14 subtypes in mice), IFNβ, IFNϵ, IFNκ, and IFNω (21). The IFNAR receptor, shared by all type I IFNs, is a heterodimer composed of two subunits and expressed in most cells (21). In cancer, type I IFNs, predominantly IFNα and IFNβ, play critical roles in preventing tumorigenesis and promoting antitumor immunity by acting on multiple cells (22, 23). Type I IFNs can directly induce tumor cells' apoptosis, inhibit their proliferation, and upregulate the presentation of cancer antigens. Type I IFNs promote cross-presentation of cDC1s to activate tumor-specific CD8+ T cells (23). Furthermore, type I IFNs are one of three signals that directly promote T-cell activation, expansion, and differentiation. Type I IFNs' antitumor effects are also complemented by their functions of recruiting CD8+ T cells through chemokine CXCL10, increasing natural killer–cell cytotoxicity, and suppressing the activity of protumorigenic immune cells, like regulatory T cells (Treg) or myeloid-derived suppressor cells (23, 24).
Among these functions, type I IFNs are well-known for promoting the activation of and cross-presentation by DCs (especially cDC1s), which are crucial for initiating antitumor CD8+ T-cell responses. DCs are particularly sensitive to and closely regulated by type I IFNs. Type I IFNs can reduce the rate of endosomal-lysosomal acidification to promote cell-associated antigen persistence in RAB5+ and RAB11+ compartments (6). Upon activation by type I IFNs, CCR7 expression drives the migration of antigen-bearing DCs to the lymph nodes to interact with and prime T cells. Type I IFNs can increase the survival of antigen-bearing cDC1s by upregulating the antiapoptotic genes bcl-2 and bcl-xL. Most importantly, type I IFNs enhance cDC1s' ability to activate CD8+ T cells through upregulating costimulatory molecules, including MHC I, MHC II, CD40, CD80, and CD86 (6).
Using spontaneously rejected tumor graft models, endogenous type I IFNs have been shown to play a central role in cancer immune editing (25). After tumor implantation, type I IFNs were increased within draining lymph nodes, and they enhanced cDC1s' cross-priming ability. Activating IFNAR signaling in cDC1s promoted the priming of tumor antigen–specific CD8+ T cells and resulted in the spontaneous rejection of immunogenic tumors (26, 27). After tumors are established, CD8+ T cells are usually tolerated and dysfunctional. Interestingly, studies have found that dysfunctional CD8+ T cells are reactivated after efficient tumor treatment, which also requires cDC1s activated by type I IFNs. In the next part, we will summarize these studies with a focus on how type I IFNs are induced and how cDC1s' cross-priming is activated to fully reactivate CD8+ T cells during cancer treatment.
The Efficacy of Tumor-Targeting Antibody Therapy Depends on Type I IFN and cDC1 Cross-priming
The first-generation of FDA-approved tumor-targeting antibodies, including trastuzumab, cetuximab, and rituximab, were designed to target surface oncogenic receptors and were found to inhibit tumor cells by interrupting oncogenic signals, by antibody-dependent cytotoxicity (ADCC), or by complement-dependent cytotoxicity (CDC; refs. 28–31). Studies also found that CD8+ T-cell responses play critical roles during antibody-mediated tumor regression. Trastuzumab binds to HER2, which is overexpressed in a large number of breast cancers. HER2 receptor signaling is connected to higher malignancy, tumor relapse, and mortality (32). The antitumor effect of anti-HER2/neu antibodies was found to be significantly impaired in Rag−/− mice or CD8+ T-cell–depleted wild-type mice. This suggests that anti-HER2/neu antibodies efficiently control tumor growth, depending on tumor-specific CD8+ T cells (33). Moreover, anti-HER2/neu–treated tumor cells released HMGB1 danger signals and activated the MyD88 pathway, both of which are critical for establishing antibody-mediated adaptive immunity. Interestingly, anti-HER2/neu therapy depends on the release of both type I and type II IFNs, suggesting that anti-HER2/neu–treated tumors release HMGB1 to trigger MyD88-dependent Toll-like receptor (TLR) signaling and stimulate type I IFNs, which then activate DCs to prime CD8+ T-cell responses (34). Similar results were found with lymphoma therapy when delivering anti-CD20. Early studies found that anti-CD20 could control tumors by direct killing or via ADCC/CDC. Recently, CD8+ T cells were found to contribute to effective anti-mouse CD20 therapy in a syngeneic B-cell lymphoma mouse model (30, 35, 36). Mechanistically, anti-CD20 treatment promoted DC-mediated cross-presentation, during which macrophage-secreted type IFN was involved (35). Overall, these studies show the essential contribution of DCs and CD8+ T-cell responses for tumor control by oncogenic receptor–targeting antibodies.
CD47 was found to be expressed in a broad range of malignant cells, including acute myeloid leukemia, non-Hodgkin lymphoma, bladder cancer, and breast cancer. CD47 is involved in tumor progression by interacting with receptor signal regulatory protein α (SIRPα) on phagocytes to prevent phagocytosis of tumor cells. Previous studies that used xenograft tumor models showed that antibodies that block human CD47 clear tumors by phagocytosis (37, 38). However, studies in immunocompetent mice revealed that anti-CD47 inhibits tumor progression by enhancing tumor-specific CD8+ T cells (39). Despite being major phagocytes regulated by anti-CD47 within tumors, macrophages are not involved in CD8+ T-cell cross-priming (39). Recent studies have shown that anti-CD47 boosts DCs' cross-priming for CD8+ T-cell activation. Anti-CD47 treatment induced DCs to express higher IFNα levels; tumor control depends on IFNAR signaling in DCs. Through type I IFN, CD47 blockade may enable the activation of NADPH oxidase NOX2 in DCs, thus inhibiting phagosomal acidification and reducing the degradation of tumor mitochondrial DNA (mtDNA) in DCs (40). Interestingly, the induction of type I IFNs in DCs depends on the cyclic GMP-AMP synthase (cGAS)–stimulator of IFN genes (STING) pathway, but not on the conventional MyD88 pathway. Anti-CD47 induced the release of mtDNA, which was recognized by cGAS in DCs' cytosol, thereby inducing type I IFN production (40). Overall, these studies suggest that tumor death induced by tumor-targeting antibody treatment can actually induce type I IFNs, which boost cDC1 cross-priming and CD8+ T-cell activation for complete tumor clearance.
cDC1 Cross-priming Contributes to ICI Therapy–Mediated CD8+ T-Cell Reactivation
ICIs may release immune inhibitory signaling to activate T cells, but recent studies have shown that CD8+ T cells reactivated by cDC1s are also involved. Preclinical studies showed that the therapeutic effect of anti-CTLA-4 (2) or anti-PD-L1 (8) was abrogated in cDC1-deficient Batf3−/− mice. The antitumor effect of combined anti-PD-1 and anti-CTLA-4, or anti-PD-1 and anti-CD137 also depends on cDC1 (16, 41). There is a concern that the impaired ICI therapy in Batf3−/− mice was due to a deficiency in CD8+ T-cell priming during tumor establishment or in exhausted CD8+ T-cell reactivation after antibody treatment. Recent studies on anti-PD-L1/PD-1 antibody treatment mechanisms have revealed the following: (i) PD-1 blockade induced exhausted CD8+ T-cell proliferation that depends on costimulatory CD28 signaling, which indicates the involvement of APC priming in reactivating exhausted CD8+ T cells (42, 43). (ii) Tumor-infiltrating myeloid cells express much higher PD-L1 than tumor cells and are critical for anti-PD-L1–mediated tumor control (44, 45). Another study showed that both PD-L1 and B7.1 were expressed in peripheral and tumor-associated DCs. Blocking PD-L1 on DCs relieves B7.1 sequestration in cis by PD-L1, which allows the B7.1/CD28 interaction to increase T-cell priming (46). (iii) cDC1s upregulate PD-L1's expression upon antigen uptake and are essential in PD-L1 blockade (47). T cells are primed in draining lymph nodes or reactivated inside tumor tissues, but which are essential for ICIs' function is unclear. Some studies clearly showed that, during treatment, PD-L1 blockade promotes cDC1-mediated reactivation of tumor-infiltrating T cells to control tumors (47, 48). PD-L1 on cDC1s might be essential for limiting T-cell reactivation. Another study found that anti-PD-L1–mediated tumor control depends on the cGAS-STING pathway, which suggests that cGAS may trigger type I IFNs to regulate DC cross-priming during ICI (49). Overall, these studies showed that CD8+ T-cell reactivation induced by ICI treatment requires cDC1 cross-priming and occurs within tumors in addition to the lymph nodes. However, how type I IFN is induced and involved inside the TME remains to be determined.
Conventional Therapies Trigger Innate Sensing and DC Activation to Enhance Cancer Immunotherapy
Conventional anticancer therapies, such as radiotherapy and chemotherapy, have been developed to interfere with the tumor cell cycle, such as the synthesis of DNA or RNA, mitotic spindle formation, and specific oncogenic signaling pathways for cancer survival (50–52). These therapies have been thought to clear tumors through direct killing. However, accumulating studies show that these conventional therapies depend on both innate and adaptive immunity. This treatment-induced tumor cell damage triggers innate immune sensing, including type I IFN production, which activates DCs and further CD8+ T-cell responses, which contribute to complete tumor control (Fig. 1).
Tumor cell DNA damage sensed by the cGAS-STING pathway is an important signal that triggers innate sensing during both tumor establishment and conventional therapies (53–56). Radiotherapy is a traditional tumor treatment modality that is widely used in the clinic because of its direct tumoricidal effect. Recent studies found that radiotherapy induces the leakage of DNA from the nucleus or mitochondria into the cytosol; the cGAS-STING pathway recognition of this DNA in the cytosol is one of the important innate sensing mechanisms. Upon encountering double-stranded DNA, cGAS catalyzes the synthesis of cyclic-di-GMP-AMP, which in turn binds the adapter protein STING on the endoplasmic reticulum and promotes TBK1-dependent IRF3 and NF-κB activation to further produce type I IFNs, proinflammatory cytokines, and chemokines to initiate adaptive antitumor responses (57, 58). Radiotherapy-induced type I IFNs further enhance DC cross-priming and CD8+ T-cell response, which are required for tumor clearance (59–61). Moreover, radiotherapy and anti-PD-L1 treatment synergistically control tumor growth in a CD8+ T-cell–dependent manner (62).
Some drugs might preferentially induce DNA fragments in a unique way. For example, 6-thio-2′-deoxyguanosine (6-thio-dG), a nucleoside analogue, can be incorporated into de novo–synthesized telomeres by telomerase to induce damage on telomeric DNA (63). This damage can activate cGAS-STING and type I IFN production and thus, synergize with PD-L1 blockade therapy (64). However, there are negative mechanisms within apoptotic tumor cells that allow their nucleic acids to escape innate immune sensing. The injured DNA could induce DNA damage response (DDR) factors to initiate DNA repair. Inhibiting the DDR pathway by DDR inhibitors, like drugs targeting PARP and checkpoint kinase I, impairs DNA repair, and inducing more damage can trigger the cGAS-STING pathway and type I IFN production within tumor cells. DDR inhibitors can synergize with PD-L1 blockade to enhance antitumor CD8+ T-cell responses (65). Another study showed that activated caspases, especially CASP9, can promote degradation of genomic DNA without triggering PRR recognition and type I IFNs, and can lead to nonimmunogenic tumor cell death (66). Irradiation can induce tumors to release mtDNA fragments, and pan-caspase inhibitor emricasan or knocking out Casp9 significantly increases tumor cell–derived type I IFNs via cGAS-STING pathway activation (66).
The involvement of RNA damage in triggering innate sensing and type I IFN production during chemotherapy is under debate. There are reports that treatment with anthracycline drugs, such as doxorubicin and mitoxantrone, can induce type I IFNs in tumors through the TLR3 pathway (24). However, the ligand here remains unknown; further research is needed to clarify this.
HMGB1 is another danger signal that has often been reported to be released after conventional therapies. Certain chemotherapeutics, including anthracycline and oxaliplatin, induce immunogenic cell death with the release of HMGB1. HMGB1 is recognized by TLR4 in DCs and activates downstream MyD88 signaling and type I IFN production (67).
Type I IFN–Based Tumor Therapy and Its Limitations
IFNα has been used to treat tumors like melanoma and lymphoma. However, the clinical utilization of type I IFNs has been limited because of severe peripheral toxicity when it is delivered systemically (68). To specifically deliver type I IFNs into tumor tissues, some studies fused type I IFNs with tumor-targeting antibodies like anti-EGFR or anti-PD-L1 (69, 70). These fusion proteins increase DC cross-priming and tumor-specific CD8+ T-cell responses and synergize with anti-PD-L1 to control tumors. Various pattern recognition receptor agonists, which can induce type I IFNs, have also been put forward, including TLR agonists, STING agonists, and RIG-I agonists. Unfortunately, the therapeutic effect of many of these IFN inducers has not met expectations. Here, we summarize the progress of several type I IFN inducers in clinical trials (Table 1).
Target PRRs . | IFN-1 inducer . | Relevant cancer type . | Current phase . | Remarks . | References . |
---|---|---|---|---|---|
TLR7 | Lipo-MERIT | Melanoma | Phase I | It is RNA-lipoplexes and used as vaccine. Systemic IFNα and IP-10 were induced. | (80, 81) |
RIG-1 | Synthetic RNA oligonucleotide (MK-4621) | Advanced solid tumors, leukemia | Phase I | Combined with pembrolizumab, modest antitumor activity. | (82, 83) |
STING | ADU-S100 | Advanced solid tumors; lymphoma | Phase I/II | Combined with anti-CTLA4 or pembrolizumab. | (84, 85) |
MK-1454 | HNSCC | Phase I/II | Single agent or combined with pembrolizumab. | (86) | |
BMS-986301 | Advanced solid tumors | Phase I | Single or combined with nivolumab and ipilimumab. | ||
TLR2/4 | Bacillus Calmette–Guérin | Superficial bladder carcinoma | FDA approved (1997) | Have clinical trials for other cancers. | (87) |
TLR3/MDA5 | Poly IC:LC | Gliomas; breast, pancreatic, and ovarian cancer | Phase I/II | (88, 89) | |
TLR4 | Monophosphoryl lipid A | HPV-associated cervical cancer | FDA approved (2009) | Used as adjuvant to Cervarix. | (87) |
TLR7 | Imiquimod | Basal cell carcinoma | FDA approved (2004) | (87) | |
TLR9 | CpG | Low-grade B-cell lymphoma, SCLC, NSCLC, CRC | Phase I/II | Limited antitumor effect. | (90–92) |
Target PRRs . | IFN-1 inducer . | Relevant cancer type . | Current phase . | Remarks . | References . |
---|---|---|---|---|---|
TLR7 | Lipo-MERIT | Melanoma | Phase I | It is RNA-lipoplexes and used as vaccine. Systemic IFNα and IP-10 were induced. | (80, 81) |
RIG-1 | Synthetic RNA oligonucleotide (MK-4621) | Advanced solid tumors, leukemia | Phase I | Combined with pembrolizumab, modest antitumor activity. | (82, 83) |
STING | ADU-S100 | Advanced solid tumors; lymphoma | Phase I/II | Combined with anti-CTLA4 or pembrolizumab. | (84, 85) |
MK-1454 | HNSCC | Phase I/II | Single agent or combined with pembrolizumab. | (86) | |
BMS-986301 | Advanced solid tumors | Phase I | Single or combined with nivolumab and ipilimumab. | ||
TLR2/4 | Bacillus Calmette–Guérin | Superficial bladder carcinoma | FDA approved (1997) | Have clinical trials for other cancers. | (87) |
TLR3/MDA5 | Poly IC:LC | Gliomas; breast, pancreatic, and ovarian cancer | Phase I/II | (88, 89) | |
TLR4 | Monophosphoryl lipid A | HPV-associated cervical cancer | FDA approved (2009) | Used as adjuvant to Cervarix. | (87) |
TLR7 | Imiquimod | Basal cell carcinoma | FDA approved (2004) | (87) | |
TLR9 | CpG | Low-grade B-cell lymphoma, SCLC, NSCLC, CRC | Phase I/II | Limited antitumor effect. | (90–92) |
Abbreviations: CRC, colorectal cancer; HNSCC, head and neck squamous cell carcinoma; HPV, human papillomavirus; SCLC, small cell lung cancer.
Better understanding the dual effects of type I IFNs in antitumor immunity may help in developing more effective therapies. Even though it is clear that type I IFNs can activate DCs and promote antitumor CD8+ T-cell responses, sustained exposure to type I IFNs can sometimes impair the host's immunity. First, in both ICI therapy and radiotherapy, the extended exposure of tumor cells to type I IFNs can induce a PD-L1–independent multigenic resistance program (71–73). Second, type I IFNs are strong agonists of inhibitory PD-L1 on DCs and other immune cells, which could induce T-cell exhaustion and even apoptosis (74–76). TMEs with type I IFN activation in plasmacytoid DCs exhibit an immunosuppressive signature, marked by PD-L1, IDO expression, and Treg recruitment, which also supports findings on the negative effects of type I IFNs on antitumor immunity (77). In the future, a potential strategy for exploiting type I IFNs may be to target or induce type I IFNs to specific DC subsets (cDC1s), but not cancer cells within the TME. The construction of an attenuated IFNα and antibody fusion protein is a smart design to fulfill this purpose (78, 79). In the meantime, it is better to combine type I IFNs and inducers with ICIs to overcome the resistance caused by PD-L1 induction.
Conclusions
cDC1 cross-priming is involved not only in initiating a CD8+ T-cell response in the host against tumor establishment, but also in reactivating CD8+ T cells inside the TME. To overcome ICI resistance, one needs to explore treatment modalities that can not only kill tumors, but also trigger innate sensing for more type I IFNs inside the TME. The next generation of IFN-based therapy must be designed to target tumor tissues or even specific cells (cDC1s) efficiently while avoiding a peripheral sink effect and side effects.
Authors' Disclosures
No disclosures were reported.
Acknowledgments
This work was supported, in part, by Cancer Prevention & Research Institute of Texas grant RP180725 to Y.-X. Fu and R. Hannan. We thank Drs. Damiana Chiavolini and Jonathan Feinberg for scientific editing.