2.2.1. Type 1 Classical/Conventional Dendritic Cells

cDC1s are specialized in cross-presentation and require several key transcription factors. Hacker et al. used inhibitor DNA binding 2 (Id2) KO mice to show that ID2 affects the development of CD8α⁺ DCs and Langerhans cells and that transforming growth factor beta (TGF-β) contributes to ID2 activation [26]. Hildner et al. showed reduction of CD8α⁺ DCs in basic leucine zipper ATF-like 3 transcription factor (Batf3) KO mice, which resulted in impaired immunity against viral infections and cancer [27]. However, cDC1s can develop in Batf3 KO mice mediated by interleukin (IL)-12 via a BATF- and BATF2-related compensatory mechanism [28]. In addition, interferon-regulatory factor (IRF) 8 (IRF8) and nuclear factor interleukin 3 regulated (NFIL3) are required for cDC1 development [29,30]. Recently, using a CRISPR-Cas9 genome editing system, cDC1 development was shown to be dependent on site-specificity of IRF8 [31]. Thus, cDC1 development involves many complicated transcription factor interactions, and more of these interactions remain to be discovered. The surface markers for mouse cDC1s are CD8α, CD103, XC chemokine receptor 1 (XCR1), C-type-lectin (CLEC) domain containing 9A (CLEC9A), and DEC205 [32,33,34]. The markers for human cDC1s are CD141, XCR1, CLEC9A, and DEC205 [35]. Both mouse and human cDC1s express XCR1, and XCR1 expression contributes to cDC1 chemotaxis via the XCR1/XC chemokine ligand 1 (XCL1) axis [3,36]. cDC1s in lymphoid tissue express CD8α, whereas cDC1s in non-lymphoid tissue and migratory cDC1s express CD103 [3]. cDC1s directly activate CD8⁺ T cells via cross-presentation of exogenous Ags using MHC-I [3,16]. Joke et al. injected irradiated β2m^−/− splenocytes loaded with ovalbumin. Although cell-associated ovalbumin was injected, the cytotoxic T lymphocyte (CTL) response occurred via the MHC-I receptors of CD8⁺ DCs [37], suggesting that exogenous Ags can be presented by MHC-I molecules through an unconventional pathway. cDC1s also activate CD8⁺ T cells indirectly via a type 1 T helper (Th1) response. Intracellular parasite infection induces Th1 immunity, and the main IL-12 producers during infection are BATF3-dependent CD103⁺ cDC1s or CD8⁺ cDC1s [38,39]. In addition to the contribution of cDC1s to adaptive immunity by T cell activation through Ag presentation, cDC1s express PRRs, including Toll-like receptor (TLR) 3 (TLR3) and TLR11 [39,40,41]. In cDC1s, TLR3 detects dsRNA of pathogens; then, TLR3 signaling upregulates the expression of cytokines and co-stimulatory molecules [40]. TLR11 in cDC1s detects intracellular parasites, and TLR11 signaling activates a Th1 response by stimulating secretion of IL-12 [39]. cDC1 cross-presentation and Th1 activation have been exploited in cancer immunotherapy by loading tumor-associated Ags (TAAs) into cDC1s [42]. In summary, cDC1s specialize in the activation of CTLs and a Th1 response.

2.2.2. Type 2 Classical/Conventional Dendritic Cells

cDC2s are differentiated from pre-DCs through continuous stimulation of FLT3L [3,43], and this process requires several key transcription factors. Using genetically deficient mouse models, RelB [44], IRF2 [45], IRF4 [46,47], and recombination signal binding protein for immunoglobulin kappa J region (RBPJ) [48] have been identified as important factors for cDC2 development. RBPJ is a transcription factor involved in Notch signaling, so Notch signaling is important for cDC2 development. Notch2 and Krüppel-like factor 4 (KLF4) are also crucial for cDC2 development [49]. Although cDC1s express different markers depending on their location in lymphoid (CD8α⁺) or non-lymphoid (CD103⁺) tissue, cDC2s dominantly express CD11b regardless of location and also express CD4 [3,43,50]. Signal regulatory protein-α (SIRPα) and CD301b are also markers of cDC2s [49]. Endothelial cell-selective adhesion molecule (ESAM) is highly expressed in cDC2s in lymphoid tissues, and its expression is dependent on Notch2 signaling [51]. In humans, cDC2s express CD1a, CD1c, CD172a, and CLEC4C. Compared with human cDC1s, human cDC2s are more numerous and are therefore yet to be fully characterized [49]. cDC2s are involved in CD4⁺ T cell activation [3,43,50]. During a type 2 T helper (Th2) response, Tussiwand et al. showed that KLF4 is essential for cDC2 development, which is different from the essential Notch2 signaling pathway in cDC2s. The authors infected mice with pathogens to stimulate Th1, Th2, and Th17 responses and only the Th2 response was impaired. Thus, a house dust mite-induced asthma model was prevented in cDC2-specific KLF4-deleted mice [52]. In addition to a type II response, cDC2s also contribute to a type III response (Th17). Satpathy et al. used mice expressing zinc finger and BTB domain-containing 46 (ZBTB46) and green fluorescent protein (GFP) to show that intestinal cDC2s express CD103. They identified that CD103⁺CD11b⁺ cDC2s are dependent on Notch2 signaling, and DC-specific deletion of Notch2 decreases CD103⁺CD11b⁺ cDC2s. They demonstrated that CD103⁺CD11b⁺ cDC2s are essential for Citrobacter rodentium infection, which induces a Th17 response through IL-23 production [53]. cDC2s mainly activate CD4⁺ T cells, but some studies reported that cDC2s can also directly activate CD8⁺ T cells. Bosteels et al. reported that inflammatory cDC2s are induced under inflammatory conditions, and they share characteristics of both cDC1s and monocytes. Induction of inflammatory cDC2s activates CD8⁺ T cells more than cDC2s do [54]. Duong et al. also reported that cDC2s activate CD8⁺ T cells. They showed that type I interferon (IFN)-responsive cDC2s (IFN-stimulated genes [ISG⁺] DCs) externally express tumor-derived MHC class I molecules (“MHC-I dressing”), and MHC-I-dressed ISG⁺ DCs can activate CD8⁺ T cells [55].

2.2.3. Plasmacytoid Dendritic Cells

pDCs were identified in 1999 by two research groups and were named for their plasmacytoid shape [56,57]. pDCs develop from pDC progenitors under continuous Flt3L stimulation, and transcription factor 4 (TCF4 [E2-2]) is the most important factor driving pDC differentiation. TCF4 also inhibits differentiation into the cDC lineage [58,59]. Then, transcription factor Spi-B restricts egress of non-dividing pDCs from the bone marrow until the pDCs are fully matured [60], when Runt-related transcription factor 2 (RUNX2) is activated. RUNX2 represses C-X-C chemokine receptor type 4 (CXCR4), which also inhibits egress from the bone marrow, and RUNX2 induces IRF7, which participates in terminal differentiation of pDCs [61]. pDCs express B220, sialic acid-binding Ig-like lectin H (SIGLEC-H), and CD317. In addition, CCR9 and lymphocyte antigen 6C (Ly6C) are also expressed in pDCs. Human pDCs also express CD45RA, CD123, and CD2 [50,62]. Although pDCs only make up 0.1–0.5% of the blood cells, pDCs are the main producer of type I IFN [62]. pDCs produce type I IFN by recognizing the viral genome during virus infection through endosomal TLRs. Compared with other immune cells, only pDCs have a high sensitivity for production of type I IFN, which is dependent on IRF7. When IFN-β is produced for the first time, IRF7 is induced by interferon alpha/beta receptor (IFNAR) signaling, and the IRF7/IFNAR axis forms a positive feedback loop. IFNAR signaling is not essential for pDC activity but is necessary to maintain their full functionality [62,63,64,65]. In relation to type I IFN production, TLRs expressed by pDCs primarily recognize pathogens, but pDCs also recognize infected cells. pDCs dock with infected cells using lymphocyte function-associated antigen-1 (LFA-1), and the TLR ligand is transferred to the pDCs from infected cells as an exosome or viral particle [62,66,67]. pDCs can also activate an adaptive immune response. Transcriptomics analysis of human peripheral blood mononuclear cells (PBMCs) enabled identification of Axl⁺ pDCs. Although Axl⁺ pDCs have a similar expression signature to pDCs, they produce less IFN-α in response to cytidine-phospho-guanosine and lipopolysaccharide stimulation than normal pDCs and activate CD4⁺ and CD8⁺ T cells more efficiently than normal pDCs [68,69,70]. Therefore, pDCs mainly protect the host by secreting type I IFN during viral infection; they also affect autoimmunity by acting on T cell and B cell responses. However, in a tumor environment, almost all pDC activity is suppressed by tumor-derived TGF-β [62].

4.2.2. Stimulation and Inhibition

Some patients with cold tumors and who have limited TILs or have TILs that localize at the edge of tumors fail to respond to ICI therapy. Salmon et al. analyzed Ag transport and investigated CD103⁺ DCs because they are uniquely able to transport Ags to tumor-associated lymph nodes. The authors injected FLT3L and identified accumulation and expansion of CD103⁺ DCs in the tumor mass. FLT3L injection followed by poly I:C promoted tumor regression in an ICI-resistant mouse tumor model and had a synergistic effect with anti-PD-L1 therapy [157]. Their study also showed that an immune response in an ICI-resistant tumor model was dependent on type I IFN [157]. The intrinsic tumor pathway also contributes to antibody resistance. However, specific delivery of IFN-β to the tumor site enhances antibody therapy, and linkage of antibody and IFN-β restores the function of DCs and causes dramatic regression of tumor growth [158]. Aside from cytokines, chemokine treatment also directly enhances the migration of DCs. Fusion of an Ag and XCL1 to create an XCL1-based vaccibody turned a poorly immunogenic tumor microenvironment into a highly immunogenic one. The XCL1-based vaccibody increased CD103⁺ DCs and the CD8⁺ T cell response [159]. In a clinical trial, DCs transduced with CCL21 using an adenoviral vector (Ad-CCL21-DC) were administrated to patients with lung cancer who had low infiltration of CD8⁺ T cells and were resistant to PD-1 therapy to attract T cells and DCs. Vaccination with Ad-CCL21-DC induced systemic Ag-specific immune responses and enhanced infiltration of CD8⁺ T cells into the tumor site [160]. As tumor PD-L1 expression was also upregulated, it is conceivable that Ad-CCL21-DC reverses ICI resistance and will be synergistic with ICI therapy.

DCs express other co-stimulatory molecules besides cytokines and chemokines, and stimulation of co-stimulatory molecules can be also an efficient method to reverse an immunosuppressive tumor microenvironment. CoAT (which includes an agonistic CD40 antibody, soluble Ag, and a TLR3 agonist) increased efficacy of DC immunization [161]. The CD40/CD40L axis is also important in an immunosuppressive tumor microenvironment. To overcome a cold tumor microenvironment, nitric oxide production by infiltrating myeloid cells (especially DCs that release TNF and nitric oxide) is essential. CD40/CD40L signaling strengthens Nitric oxide synthase 2 (NOS2) expression, and production of nitric oxide and TNF-α through CD40 contributes to tumor elimination by adoptive cell transfer [162]. Furthermore, NOS2, CD40, and TNF expression have a strong correlation with survival of patients with colorectal cancer [162]. In patients with RCC, treatment with MoDCs co-electroporated with tumor RNA and synthetic CD40L was well tolerated and resulted in immunologic responses coupled with extension of patient survival [163].

Inhibition of immunosuppressive factors also enhances the efficacy of DC-based therapy. In ovarian cancer, some patients were resistant to anti-PD-1 therapy because of tumor-soluble factors that block the migration of TILs to the tumor site. VEGF-A establishes a vascular endothelial cell barrier to prevent T cells homing to tumors [164]. Thus, bevacizumab, a VEGF-A-blocking antibody, had a synergistic effect with a DC-based vaccine in ovarian cancer patients [165]. Tumor-derived cytokines also directly inhibit the functionality of DCs. Haiyan et al. identified that tumor-associated cytokines induced signal transducer and activator of transcription 3 (STAT3) activity in DCs, and expression of STAT3 repressed Id2 transcription required for DC-based anti-tumor immunity [166]. Using a conditional KO mouse model with ID2 overexpression in DCs, the authors corroborated that STAT3 deficiency and expression of ID2 in DCs enhanced the anti-tumor response. In particular, ID2 expression downregulated TNF-α, which induced proliferation of Tregs at the tumor site. As a result, there was a shift in the proportion of effector T cells to Tregs through ID2 expression in DCs, which reversed the poorly immunogenic tumor microenvironment [166]. Similarly, Yong et al. converted the proportion of effector T cells to Tregs by inhibition of p38 mitogen-activated protein kinase (MAPK) during DC differentiation [167]. The authors showed that inhibition of p38 in DCs upregulated OX40L and induced effector T cell proliferation in the presence of Tregs. They further showed that a p38 inhibitor repressed peroxisome proliferator-activated receptor gamma (PPARγ) and this downregulation induced expression of p50, the transcriptional activator of OX40L [167].

4.2.3. Combination Therapy

Chemotherapy induces immunogenic cell death in tumor cells [6], and combination therapy together with immunogenic cell death increases DC functionality. Anthracycline-induced phosphorylation of eukaryotic translation initiation factor 2A (eIF2α) causes exposure of calreticulin in tumor cells, which acts as an “eat me” signal; thus, DCs phagocytose calreticulin-exposed tumor cells. Injection of recombinant calreticulin or blockade of eIF2α dephosphorylation increased immunogenicity of tumor cells and had a therapeutic benefit in tumor-bearing mice [168]. Vacchelli et al. identified single nucleotide polymorphisms that affected overall survival of chemotherapy-treated patients. They demonstrated that formyl peptide receptor 1, which recognizes annexin-1 in dying tumor cells, was affected only by single nucleotide polymorphisms in the formyl peptide receptor 1 (Fpr1) gene. Mutation of formyl peptide receptor 1 correlated with reduced survival, although the mutation of TLR3 and TLR4 were unaffected. During chemotherapy, activation of the FPR1/annexin-1 axis induced interactions with dying tumor cells and resulted in T cell activation, but migration of DCs to the tumor site was not affected by the activation of formyl peptide receptor 1 [169].

Chemotherapy also affects migration of DCs. Dying tumor cells treated with chemotherapy release ATP, which induces migration of CD11b⁺Ly6C^hi DCs to the tumor site [72]. Recruited CD11b⁺ DCs present TAAs to activate T cells. ATP also activates DCs via purinergic receptors. Using various KO mouse models, Ghiringhelli et al. demonstrated that the purinergic receptors in DCs activate the inflammasome of DCs and induce release of IL-1β, which further activates CD8⁺ T cells [170]. Chemotherapy also directly affects DCs. Platinum-based chemotherapy enhanced the T cell-stimulatory potential of DCs, but co-stimulatory molecules and cytokine production from DCs was not altered. Instead, co-inhibitory molecules, such as PD-L2, were significantly downregulated through STAT6 dephosphorylation [171].

Radiation therapy also induces immunogenic cell death in tumor cells. In particular, irradiated tumor cells are recognized by the cGAS-STING pathway in DCs. Activation of DCs via the STING pathway leads to secretion of IFN-β and activation of CD8⁺ T cells. Thus, treatment with exogenous IFN-β or a STING agonist promotes effective adaptive immune responses [172]. Photoradiation therapy-based DC vaccines induced tumor regression in mice with high-grade gliomas. Radiated high-grade gliomas released DAMP molecules and increased DC vaccine efficacy. Activation of DCs modified the ratio of T cells by increasing Th1 cells and decreasing Th17 cells, which resulted in an anti-tumor response [173].

Ineffective T cell responses can be reversed using monoclonal antibodies targeting PD-1 (anti-PD-1 antibody) and CD137 (anti-4-1BB antibody). However, Alfonso et al. showed that this therapy had a poor response in Batf3^−/− mice [174]. The authors suggested that the efficacy of ICI therapy is dependent on DCs, and ICI therapy in combination with a DC vaccine can induce a synergistic effect. In poorly immunogenic tumor models, the use of CTLA-4-blocking antibodies alone was not sufficient to improve patient survival/shrink tumor mass, but in combination with a DC vaccine, patients with melanoma had increased TILs, specific T cell responses, and clinical benefit in a pre-clinical trial [175]. An advanced form of a DC vaccine, GVAX (GM-CSF-transduced prostate cancer vaccine), gave clinical benefit and had a good safety profile when used in combination with ipilimumab (anti-CTLA-4 antibody) in a phase I dose-escalation trial [176]. Another form of DC vaccine, TriMix-DC (MoDCs electroporated with mRNA encoding CD70, CD40L, and TLR4), had a synergistic effect with ipilimumab in patients with advanced melanoma [177]. SD-101, a synthetic oligonucleotide with cytidine-phospho-guanosine motifs, stimulates pDCs through TLR9. Some patients with advanced melanoma who did not have pre-existing immune responses did not respond to anti-PD-1 monotherapy, but injection of SD-101 changed the tumor microenvironment to a hot tumor, which resulted in a CTL response [178]. In a mouse tumor model, CXCL9/10 recruited and activated CD8⁺ T cells. Anti-PD-1 therapy did not enhance the anti-tumor response in Cxcr3^−/− mice that were non-responsive to CXCL9/10. Epigenetic modulation to reverse the repression of CXCL9/10 in intratumoral CD103⁺ DCs reversed the efficacy of anti-PD-1 therapy [179]. The authors suggested that their pre-clinical data support activation of the CXCR3 system as a viable approach in anti-PD-1 therapy.

DC vaccines combined with adoptive cell transfer show clinical benefit. Patients with stage IV melanoma were treated with a DC vaccine followed by adoptive cell transfer, and this combination therapy was shown to be feasible with some patients experiencing tumor regression [180]. CCL22 is expressed in pancreatic cancer cells. Rapp et al. transduced T cells to overexpress CCR4 (the CCL22 receptor). Activation of the CCR4/CCL22 axis increased T cell recruitment and mediated interactions with intercellular adhesion molecule 1 in DCs. As a result, the CTL response was upregulated and mouse survival was enhanced [181]. Adoptive cell transfer increases the efficacy of DC-based immunotherapy. However, DCs also have an important role in adoptive cell transfer efficacy. Herranz et al. identified that vancomycin treatment with adoptive cell transfer inhibited mouse tumor growth. They revealed that the gut microbiota modulated migration of CD8⁺ DCs and affected IL-12 production in CD8⁺ DCs [182], which increased the efficacy of adoptive cell transfer.

The efficacy of DC vaccines can be enhanced when combined with modified bacteria or viruses. Killed but metabolically active Listeria monocytogenes were engineered to express tumor Ags. The modified L. monocytogenes activated MoDCs, and the Ags were successfully presented by MHC-I molecules on DCs. Targeting of in vivo DCs by modified L. monocytogenes induced the activation of CD4⁺ and CD8⁺ T cells and controlled mouse tumor growth [183]. Recently, Chen et al. incorporated cell membranes from Escherichia. coli and tumor-cell membranes in nanoparticles. They vaccinated tumor-bearing mice with the nanoparticles and showed improved activation of DCs and Ag-specific T cells. The nanoparticle vaccination enhanced mouse survival and provided a long-term anti-tumor effect [184]. Similar to stimulation by bacteria, tumor inflammation stimulated by a virus enhances the efficacy of DC vaccines. A hTERT-Ad was injected followed by a DC vaccine, and tumor-specific replicating virus infection induced oncolysis and tumor inflammation. Tumor inflammation induced by a virus provides all necessary signals to activate DCs and can induce greater anti-tumor responses compared with TLR activators [185]. It is suggested that combination therapy of a DC vaccine with an oncolytic virus is a good candidate for treatment of cold tumors.