Perspective Chapter: Dendritic Cells in the Tumor Microenvironment

Abstract

Tumor infiltrating dendritic cells (DCs) play a critical role in initiating the process of anti-tumor immune responses. They can uptake tumor antigens either directly at the tumor site or from circulating antigens, and elicit T cell activation and adaptive immunity in secondary lymphoid organs. Subtypes of dendritic cells have various roles in immunity and tumor rejection. In this chapter, we will summarize the role of dendritic cell populations on mounting anti-tumor immunity. Conversely, we will discuss tumor-mediated dysfunction of dendritic cells that aid immune evasion including prevention of recruitment, impairment in antigen presenting and mediation of tolerance. At last, we briefly introduced the progress in DC vaccine applications in clinic.

1. Introduction

Dendritic cells (DC) are responsible for activating effector responses and mediating adaptive immunity. Immune responses are dependent on multiple factors including the DC type, maturation status, and immunogenicity of antigens. DCs have the capacity of inducing protective immunity as well as generating a tolerogenic immune environment. The complexity of how cancer impacts the spectrum of response varies depending on the cancer type, largely on the cancer immunophenotype. Here we discuss how different DC subtypes interact between cancer and adaptive immunity. We also touch on various cancer-mediated immune evasion strategies that alter DC function. Lastly we evaluate immunotherapeutic strategies that employ DCs to elicit anti-tumor T cell responses.

2. Anti-tumor roles of different dendritic cell sub-populations

DCs are composed of heterogenous sub-populations with each subtype possessing unique functions to compensate for each other. They cooperate to elicit both innate and adaptive immunity. In this section, we will generally review the roles of different DC sub-populations in anti-tumor immunity (Figure 1).

Figure 1.

Roles of DC subpopulations in anti-tumor response. Immature DCs in peripheral tissue can be recruited into tumor site through CCR5/CCL3/4 chemoaxis. Both cDC1 and cDC2 uptake tumor antigens in situ and migrate to tdLN in a CCR7-dependent way, trafficking tumor antigens into tdLN. In tdLN, antigen-loaded cDC1 primes both naïve CD8 T cells and CD4 T cells. Primed CD4 T cells further boost CD8 T cell activation through licensing cDC1 in a CD40/CD40L dependent way. Antigen-loaded cDC2 predominantly primes naïve CD4 T cells. It can alternatively activate CD8 T cells through transferring antigens to LN resident cDC1(rcDC1), which primes naïve T cells in LN. Activated T cells migrate to tumor site in a CXCR3 dependent chemokine recruitment way. Effector T cells can be further stimulated by antigen activated cDCs and moDCs in tumor and exert cytotoxic tumor killing. moDCs exert CD4 and CD8 T priming function to compensate for cDCs when they are dysfunctional or depleted. Activated pDCs secrete CCL chemokines to recruit and activate NK cells through type I IFN and OX40-OX40L interaction. Activated NK cells promote cytotoxic CD8 T priming through activation of cDC in an IFNγ-dependent way, meanwhile, activated cDCs secrete cytokines like IL12, IL18 that enhance NK cell activation.

2.1 Conventional DCs

Conventional DCs (cDCs) are derived from common DC precursors (CDP) in the bone marrow, which are comprised of two main subsets: cDC1 and cDC2. The infiltration of cDCs in the tumor has a positive correlation with patient survival in some solid cancers [1]. cDC1 development is specified by transcriptional factors BATF3, IRF8, and ID2, while cDC2 development depends on transcriptional factors RELB, IRF4, ZEB2 and KLF4 [2, 3]. BATF3 is required for maintaining IRF8 expression during cDC1 commitment in specified cDC1 progenitor [4]. BATF3 is also required for cDC1 cross-presentation function and cross-presentation independent anti-tumor immunity functions [5, 6]. BATF-dependent cDC1 is specified by its unique role to initiate naïve CD8⁺ T cell activation in tumor-draining lymph nodes(tdLN), as well as enhance both T cell accumulation and local CD8 T cell cytotoxicity. The abundance of cDC1 in the tumor microenvironment positively correlates with cancer patient survival and response to immunotherapy across different cancer types [7, 8]. CD103⁺ cDC1s sample tumor antigen in tumor mass and migrate to tumor-draining lymph nodes via CCR7, where they prime T cell responses [9]. Tumor-resident BATF3⁺ cDC1s secret CXCL9 and CXCL10 to recruit CXCR3 expressing effector T cells and NK cells [10, 11]. In turn, IFNγ produced by tumor effector T cells and NK cells induce CXCL9/10/11 production by myeloid cells, creating a feedback loop in this response [12]. cDCs also secrete IL-12, IL-18 and IL-2 provoking NK cells to produce IFNγ, TNFα, or GM-CSF, which further promotes DC activation [13].

Unlike cDC1s, cDC2s have limited capacity to cross-present tumor antigens to CD8 T cells. The function of cDC2s are largely restricted to priming of CD4 T cells in tdLN or in tumor [14, 15, 16]. cDC2s mediate cross-presentation of soluble antigens and is enhanced by TLR7 agonist [17, 18]. cDC2s complement the function of cDC1s by also activating CD8 T cells. Migratory cDC2 capture antigens in tumor and transfer antigens to LN resident cDC1s through antigen vesicles and synaptic transfer, which is capable of activating CD8 T cells [19]. In the absence of cDC1s, activating cDC2s by type I IFN can stimulate CD8 T activation in tumor [20]. In preclinical models where cDC1 function is impaired, deletion of cDC1 population improves cDC2 migration into tdLN and CD4 T activation [1].

2.2 Plasmacytoid dendritic cells

Plasmacytoid dendritic cells(pDC) are largely regarded as immunomodulating cells through secretion of massive amounts of type-I interferon during anti-virus immune responses. The role of pDC in anti-tumor immunity is controversial. pDC in tumors have been found to have impaired response to Toll-like receptor activation and decreased type-I IFN production. They recruit and expand immune regulatory T cells in the tumor microenvironment (TME) and are associated with poor prognosis [15, 21]. As an escape mechanism, tumor cells attract pDCs to induce an immunosuppressive environment through secreting chemokine CXCL12 [22].

On the contrary, in some solid tumors, pharmacological agents can be used to overcome immunosuppression. For example, imiquimod stimulation can induce pDC mediated tumor killing via secretion of TRAIL and granzyme B independent of adaptive immunity [23]. pDCs can also drive anti-tumor response by activating adaptive T cell immunity mediated by cDC activation dependent on type-I IFN [24]. Direct injection of TLR9 activated pDC into B16 melanoma tumor bearing mice induces robust cytotoxic T lymphocytes (CTL) cross-priming against tumor, leading to tumor regression. TLR9 activated pDCs produce large amounts of chemokines CCL3, CCL4, and CCL5 within the tumor, which recruits CCR5⁺ NK cells. Recruited NK cells are activated by pDC through cell-to-cell interaction via OX40/OX40L and type I IFN secreted by pDC. Tumor cells lysated by NK cells cause tumor antigen release into cDCs and IFNg secreted by activated NK cells also help activate CTL in dLN [25]. Such an activated subset of pDC with higher levels of OX40 is also found in head and neck squamous cell carcinoma (HNSCC) tumor with distinct immunostimulatory and cytolytic function and can synergize with cDCs in generating tumor antigen-specific CD8⁺ T cell responses [26].

2.3 Monocyte-derived dendritic cells

Monocyte-derived dendritic cells (moDCs) are differentiated from monocytes under inflammatory conditions. Activation of p53 in MDSCs and monocytic progenitors can induce moDC-like population differentiation in tumor, which potentiates the anti-tumor response [27]. Increase of moDC in tdLN can be a measurable indication of immune activation, particularly after treatment with pharmacological agents such as TLR agonists [28]. moDCs in tumor are essential for CD8 T activation and antitumor response after local immunostimulatory agent treatment [29]. In mice with Zbtb46-DTR bone marrow chimeras, which are deficient in cDC production after diphtheria toxin (DT) treatment, moDC compensate for the loss of cDCs and account for intratumoral CTL expansion and function [30]. Compared with cDC, moDCs are less efficient at inducing CD4 T cell proliferation but more efficient at inducing Th1 and Th17 differentiation [31]. However, moDCs are also able to cross-present antigens through the vacuolar pathway and activate naïve CD4 T and CD8 T cells [32]. In vitro differentiated moDCs have been used in clinical trials as vaccines for cancer patients and encouraging responses have been shown when combined with other cancer therapies [33, 34, 35, 36, 37], which will be further discussed in the last section of this chapter.

3. Tumor-mediated immune evasion: impact on dendritic cells

3.1 DC tumor infiltration and migration to LN

cDCs in tumor are found to be sparse among tumor infiltrated immune populations [8, 38]. Increased cDC amount within tumor is associated with improved prognosis and response to check-point inhibitor immunotherapy [8, 39]. Tumor cells secret soluble factors that suppress DCs infiltrating to tumor site and migrating to LN (Figure 2).

Figure 2.

Impact of DC recruitment and migration by tumor. Tumor cells suppress CCL4/5 expression by inducing β-catenin signaling pathway, which inhibits cDC recruitment via CCL4/5/CCR5 chemoaxis. PGE2 produced from tumor directly acts on intratumoral NK cells through EP2/4 receptors. PGE2 inhibits CCL5 and XCL1 secretion by NK cells, which results in decreased cDC recruitment. PGE2 can also inhibit CCR5 and CXR1 expression on cDC to impair cDC recruitment. High concentration of PGE2 inhibits moDC migration into dLN. TGFβ from tumor inhibits LC migration into LN.

Tumor cells suppress chemokine CCL4 production through activating beta-catenin signaling, and beta-catenin activation induces ATF3 expression. ATF3 binds the promoter of CCL4 gene and suppresses CCL4 expression. Decreased CCL4 leads to decreased intratumoral cDC recruitment by CCL4/CCR5 axis [40]. beta-catenin signaling suppresses CCL5 level in tumor loci. CCL5/CCR5 chemoaxis recruits cDC1 into tumor. Increased CCL5 expression in tumor recruits cDC tumoral infiltration and promotes anti-tumor immune response and promotes efficacy when combined with anti-PD1 [41]. Prostaglandin E2 (PGE2), a prostanoid lipid catalyzed by enzyme cyclooxygenase (COX), is highly produced in tumor [42, 43, 44]. cDC1s are absent from PGE2 producing tumor. PGE2 suppresses NK cells mediated cDC1 recruitment in tumor. Intratumoral NK cells secret cDC1 chemoattractant CCL5, XCL1, which are inhibited by tumor derived PGE2 through PGE2 receptor EP2, EP4 on NK cells. However, expressing CCL5, or XCL1 in tumor is insufficient to reverse intratumoral DC exclusion in PGE2 producing tumor. Further study shows that PGE2 can also downregulate CXR1 and CCR5 expression in DC, which leads to impairment of response to chemokine even in the existence of chemoattractant [21, 44, 45].

DC cells uptake antigens in tumor site and then migrate to dLN for priming T cells. TGFβ secreted from tumor inhibits DC migration to dLN in both autocrine and paracrine way. Langerhans cells (LCs), skin-resident DCs, play critical role in eliciting immune response in skin disease. Besides of tumor cells, LCs are also active TGFβ producer. Knock-out of TGFβ or its receptor in LCs induces mass migration of LCs to regional LN in both steady and inflammation states [46]. In skin tumor model, TGFβ inhibits tumor infiltration and migration to skin-dLN by LCs [47]. Role of PGE2 in regulating DC migration relies on its concentration. High concentration of PGE2 suppresses DC migration while it has also been shown as a positive regulator of CCR7 expression and migration of moDCs [48, 49].

3.2 Antigen presentation

Tumor cells develop mechanisms to impair antigen capture and presenting by DCs (Figure 3). Molecules released or exposed from dying cancer cells can act as danger signals to circulating DCs. DCs recognize dying/dead tumor cells or tumor derived debris through danger-associated molecular patterns (DAMPs) mediated by pattern recognition receptors (PRRs) like TLRs, and phagocyze dying-tumor cells or tumor derived debris [50]. DAMPs include ATP, heat shock proteins (HSPs), HMGB1, calreticulin, annexin A1, dsDNA, but are not limited to these [50]. Antigen uptake will stimulate DC maturation and migration to dLN. Internalized antigens will be processed and presented on the DC surface by MHC-I and MHC-II molecules. MHC-I molecules used to be thought for intracellular peptide presenting, while MHC-II is for exogenous peptide. However, this is not always the case. Cross-presenting is termed for presenting exogenous antigens by MHC-I, which plays a critical role in eliciting anti-tumor immune response by DCs [51]. Proteins internalized by DCs are degraded in phagosomes into peptides [52]. Peptides are then translocated into endoplasmic reticulum (ER) by transporter associated with antigen presentation (TAP). The MHC-I heterodimer is assembled in ER from a polymorphic heavy chain and a light chain β2-microglobulin (β2m) and stabilized by chaperone proteins like calreticulin and tapasin when peptide is not loaded. Chaperones will be exchanged when peptide is loaded. Peptides fit into the MHC-I peptide binding groove which stabilizes the peptide-MHC-I complex. MHC-II comprises transmembrane α- and β-chains and an invariant chain. MHC-II will be transported to a MHC-II compartment, an endosomal compartment, for invariant chain digestion, resulting a class II-associated lipid peptide (CLIP). With the help of H2-DM/HLA-DM, CLIP is exchanged with antigen peptide [53]. In LN, DCs present tumor antigens to CD8 T cells and CD4 T cells dependent on MHC-I and MHC-II respectively. Tumor proteins will be processed into immunogenic peptides and loaded on MHC-I or MHC-II molecules on cell surface.

Figure 3.

Tumor induced impairment of DC on tumor antigen capture and presentation. Tumor derived factors mediate lipid body accumulation in DC by inducing MSR1 expression. ROS, induced by tumor microenvironment, leads to increase lipid peroxidation in DC. Byproduct of lipid peroxidation, 4-HNE, induces ER stress, which activates XBP1 transcription factor. XBP1 then increases genes synthesizing triglyceride. Increased triglyceride leads to increased oxidized lipid bodies. The lipid bodies competitively bind HSP70 with pMHC, preventing pMHC been translocated onto cell surface. TIM3 competitively binds HMGB1 with dying tumor derive DNA, inhibiting DNA stimulated immune response via preventing DNA been internalized into DC. Tumor inhibits TIM4 expression on DC, thus inhibiting TIM4 mediated tumor associated antigen capture.

Tumor cells evolved multiple immune escape strategies to prevent recognition by DCs. For example, tumor-derived stanniocalcin 1(STC1) interacts with DAMP signal, calreticulin (CRT), to prevent CRT membrane from exposing to APCs, thereby abrogating membrane CRT-directed phagocytosis by DCs. High expression of STC1 in tumor is significantly correlated with poor responses to immunotherapy in patients [54]. In another immune escape mechanism, the mevalonate (MVA) pathway, which is highly activated in tumor cells, reduces F-actin exposure, a DAMP signal, on tumor cells, evading recognition mediated by Clec9A on cDC1. The MVA increases protein geranylgeranylation on Rac1, a small GTPase controlling actin cytoskeleton, resulting reduced F-actin in tumor [55, 56, 57]. The immune modulator TIM-3 is also highly expressed by DCs and has been shown to play an inhibitory role on DC activation. TIM-3 inhibits tumor derived DNA uptake by cDC1 through inhibiting endocytosis. HMGB1, a ligand of TIM-3, also acts as a DAMPS signal and binds tumor-derived DNA and is taken up by DCs. TIM-3 inhibits this process through sequestering HMGB1 bound DNA on cell surface [58]. TIM4, another T- cell immunoglobulin and mucin domain gene as TIM3, is also expressed on APCs like macrophages and dendritic cells. DAMP signal induces TIM4 expression on intratumoral macrophages and DCs [59]. Though TIM4 on tumor associated macrophage (TAM) has been shown impedes tumor antigen presentation through activating autophagy in TAM upon tumor antigen uptake in mouse melanoma model [59], TIM4 on lung resident cDCs in lung adenocarcinoma model shows a positive role in promoting anti-tumor immune activation [60]. TIM4 expression is downregulated in cDC1 from advanced lung tumor. Blocking TIM4 or knocking out of TIM4 abolishes tumor antigen uptake by lung resident cDC1 and impairs antigen presenting to CD8 T cells in vitro and in vivo.

DCs from tumor-bearing host have been found with accumulated lipids. Tumor derived factors induce oxidized lipid accumulation in cDCs from tumor bearing host. DCs have high oxidized lipids shows impaired cross-presentation while not affecting presenting endogenous antigens, and nor affecting the level of MHC-I. Scavenger receptor, MSR1, induced by tumor derived factors, accounts for the lipid accumulation in DC [61, 62]. ER stress signaling also involved in oxidized lipid accumulation in DC from tumor bearing host. 4-HNE is a byproduct from lipid peroxidation mediated by ROS and triggers ER stress and XBP1 activation. XBP1, a multitask transcription factor in response to ER stress, induces triglyceride biosynthesis. Elevated triglyceride biosynthesis leads to accumulated abnormal lipids and suppresses DC function [63]. HSP70 is a chaperon protein that binds with pMHC, and facilitates pMHC trafficking onto cell surface. Oxidative lipid bodies, not non-oxidized lipid bodies, competitively bind HSP70 covalently, preclude HSP70 interaction with pMHC, thus affect pMHC trafficking to cell surface [64].

3.3 Tolerance

Tumor cells have evolved different mechanisms to promote DC tolerance to facilitate immune escape (Figure 4). Tolerized DCs experience higher co-inhibitory markers, including PD-L1, PD-L2 and higher arginase activity, and lower MHC-II and co-stimulation markers, including CD80, CD86, CD40 [39, 65]. Tolerized DCs are not capable of activating T cells, while promote immune suppression through mechanisms like, for example, Treg upregulation.

Figure 4.

DC tolerance mechanism. Secreted factors or metabolites derived from tumor environment induce dendritic cell tolerance through activating or inhibiting cell-intrinsic signaling pathways in DC. DC tolerance leads to induced inhibitory molecular expression, Arg and IDO1 upregulation, anti-inflammation cytokine production and suppress co-stimulation and pro-inflammation cytokine production. And also result into dysregulated chemokine secretion. DC tolerance abolishes anti-tumor immune response through inhibiting T cell and NK recruitment and mediated tumor killing, and promoting immunosuppressive Treg differentiation and recruitment.

3.3.1 Secreted factors from tumor environment

Secreted tumor-derived factors is one of the major ways of driving DC tolerance, these include PGE2 and TGFβ which lead to subsequent induction of other immune modifiers. Tumor derived PGE2 suppresses cDCs activation by suppressing co-stimulation, IL-12 production, and increasing PD-L1 and Arg1 [44, 66]. PGE2 is the main inducer of arginase-1 during tumor induced DC tolerization [67]. TGFβ from tumor also induces DC tolerization [67, 68]. TGFβ induces IDO expression in pDC and enhances expression of CCL22 by myeloid DCs in tumor. IDO suppresses effector T cell activity and promotes Treg differentiation and activation. DC-derived CCL22 chemokine promotes CCR4-dependent recruitment of Tregs to the tumor microenvironment [69]. pDCs in tumor environments are associated with poor survival. Co-culture with TGFβ containing medium inhibits pDC activation and type I IFN secretion. Tolerized pDCs promote tumor growth through inhibiting NK cell infiltration and recruitment of Treg cells [70]. DCs with high TGFβ expression are poor at eliciting the activation of naive CD4 T cells and sustaining their proliferation and differentiation into Th1 effectors. Vascular endothelial growth factor (VEGF) inhibits LPS induced DC maturation via Nrp-1 receptor on DC. NRP1 interacted with LPS receptor TLR4 and suppressed downstream ERK and NF-κβ signaling, resulting in increased expression of MHC-II and costimulatory molecules (CD40, CD86) as well as proinflammatory cytokine production inhibition [71]. Infiltrating macrophages were the primary source of IL-10 within tumors, blocking IL-10 signaling increases intratumoral dendritic cell expression of IL-12 during chemotherapy in breast cancer [65].

LPA is a bioactive lipid produced by tumor cells. Blocking LPA-generating enzyme autotaxin in ovarian cancer cells elicits anti-tumor immune response driven by type-I IFN. LPA induces PGE2 synthesis by DCs, which suppressed type-I IFN production and response in DC via autocrine EP4 engagement [72]. Lactate is an oncometabolite resulted from metabolic adaption in cancer cells via Warburg effect. Lactate in tumor attenuates pDC activation in response to TLR9 ligand and consequent type I IFN induction. pDC tolerization by lactate is partially through activating GPR81, a cell surface G-protein coupled receptor of lactate. GPR81 activation induces intracellular Ca2⁺ mobilization and activates calcineurin phosphatase (CALN) expression. Inhibition of CALN reverses the inhibitory effect by lactate. Extracellular lactate can also influence pDCs through intracellular import via the monocarboxylate transporters (MCT). Inhibition of MCT genes resulte in significant reversal of the lactate-mediated inhibition of IFNα. Thus, both GPR81 triggering and cytosolic import via the MCT transporters mechanism are involved in lactate induced pDC tolerization. Lactate treated pDCs have enhanced tryptonphan metabolism, leading to excessive production of kynurenines which in turn induces Treg cell differentiation [73].

3.3.2 Cell-intrinsic mechanism

Tumor-derived Wnt5a induces β-catenin signaling activation-dependent IDO expression in DCs. DCs conditioned by wnt5a promote Treg development and suppresses effector T cell activation [74, 75]. β-Catenin complexed with PPAR-γ upon wnt5a stimulation and transcriptionally activates fatty acid oxidation (FAO) synthesis gene, CPT1a, inducing the synthesis of heme prosthetic group, protoporphyrin IX, which is required for IDO enzymatic activity [75]. Wnt1/β-catenin signaling in DC suppresses chemokine production, leading to T cell exclusion in tumor and decreased T cell activation [76]. β-catenin signaling in DC also impairs CD8 T priming through inducing IL-10 secretion via mTOR activation. Even though the negative regulation of initial CD8 T priming by β-catenin/mTOR/IL10 in DC, β-catenin–regulated IL-10 also shown has an opposite anti-tumor immunity role through maintenance of primed CD8 T cells after clonal expansion [77, 78]. β-catenin can also interact with TCF4 and activates gene expression of Aldh1, an enzyme to produce retinoic acid (RA) from vitamin A, resulting increased RA in DC [79]. Aldh1 expression in mature DC significantly correlated with immunoregulatory module including genes like PD-L1, PD-L2, CD83, and CCL22 [80]. RA induces Treg generation in vitro and in vivo [81, 82, 83]. DC maturation suppression could be mediated by E-cadherin based DC-DC adhesion. Disrupting this contact activates DC maturation through activating β-catenin/TCF, leading to increase of co-stimulatory molecules, MHC-II and chemokine receptors. However, such DC maturation is not coupled with proinflammation cytokine secretion and failed to prime CD4 T cells, coupled with a distinct transcriptional profile from those induced by TLR activation. DC matured by E-cadherin disruption also leads to Treg production. The data suggests a DC function regulatory role of E-cadherin/β-catenin/TCF axis [84].

Nuclear factor-κB (NF-κB) is an important transcription factor that participating in cancer inflammation. There are two general types of NF-κB signaling pathways: canonical and non-canonical pathways [85]. Canonical and non-canonical NF-κB pathways play different roles in DC functional regulation. Lung cancer patient derived tumor sera induce canonical NF-κB pathway inhibition, while activates non-canonical NF-kB pathway in human mo-DC [86]. IFNγ has been shown important for myeloid activation [87]. Canonical NF-κb/IRF1 mediated IFNγ response pathway is required for intra-tumoral cDC1 activation. IFNγ knock out or IFNGR1 knock out in cDC1 abolished IL12 production [88]. Impaired NF-κB or IRF1 loses control of tumor growth and expression of maturation markers and chemokines (CXCL9/10) for recruiting T cells [89]. Inhibiting NF-κB in BMDC has no effect on MHC-II or co-stimulation molecules, while promotes Treg differentiation in vitro [80]. VEGF mediated inhibition on LPS stimulated BMDC activation is dependent on the inhibition of canonical NF-κB signaling pathway [66]. Noncanonical NF-kB signaling in dendritic cells is required for IDO induction in the late stage of DC activation by CD40 ligation [90].

Inhibitory molecular expression on DC suppresses T cell activation and induces Treg differentiation. PD-L1 upregulation in tolerized DC is not dependent on the presence of type I and type II IFN signaling, nor is dependent on inflammasome or TRIF/MyD88 signaling. Instead, PD-L1 upregulation is dependent on phagocytic cell-surface receptor AXL activation upon antigen uptake. IL-4 signaling negatively regulates IL-12 production on DC. Blocking IL-4 signaling can increase IL-12 production without upregulating PD-L1 [88]. IRF4 plays a dural role of upregulating antigen presenting capability and tolerization of BMDC. Depletion of IRF4 reduces Aldh1 and PD-L2 expression, coupled with elevated cytokine IL-12 an TNF expression. IRF4-deficient DC is impaired for Treg generation in vivo. TIM-3 is predominantly found expressed in cDC cells in tumor. TIM-3 expression on DC can be induced by IL-10 or VEGF [91]. Blocking TIM-3 improve survival when combined with chemotherapy. The regulatory effect by TIM-3 blocking is neither through affecting cDC infiltration nor through regulating cDC activation. However, TIM-3 blocking increases CXCL9 secretion by cDC1, which is a ligand for CXCR3. CXCL9/CXCR3 chemoaxis attracts T cells into tumor [92]. TIM-3 on DC impaires DC recognition and response to tumor derived nucleic acids. TIM-3 serves as a receptor for DNA sensor, HMGB1, completing with nucleic acids for binding to th A-box domain of HMGB1. The binding of TIM-3 on HMGB1 inhibits nucleic acids to be internalized into endosomes [87].

DC activation is accompanied by an increased glycolysis metabolic process, which is required by both survival and effector function of activated DC. Bioactive gas nitric oxide (NO) is synthesized and secreted by activated DC, playing an immunomodulating role of DC. Cellular production of NO is catalyzed by NOS enzymes, which converts substrates L-arginine, NADPH and O2 to L-citrulline, NADP+, and NO [93]. Inducible NOS (iNOS) is the primary NO-synthesizing enzyme expressed by DC. iNOS expression in CD103⁻CD11b⁺ intratumoral DC is required for tumor suppressive Th17 T cell differentiation in PDA model [94]. Glucose could inhibit DC function through mTOR/HIF1a/iNOS signaling axis, inhibiting co-stimulation molecular expression and IL12 secretion and restricting T cell activation. When T cells encounter DCs, they compete for glucose availability, which suppress the glucose sensitive pathway resulting T cell activation [95]. Monocyte-derived tumor associated DCs are prominent in tumor antigen uptake, but lack of strong T-cell stimulatory capacity due to NO-mediated immunosuppression [96].

4. Application of DC vaccine in tumor immunotherapy

4.1 DC vaccination

DCs are the most efficient professional antigen-presenting cells that can initiate an adaptive immune response by presenting antigens to T cells [97, 98]. In the past 25 years, many groups have exploited this characteristic to create dendritic cell vaccines to direct the immune system to fight cancer. DC cell-based vaccine approaches have been proved safe for their minimal toxicity, and their low association rates with autoimmunity [99, 100]. The general process of DC vaccine preparation including DC generation, antigen loading and DC maturation. To date, different strategies have been developed to generate DC vaccine for clinical applications.

The most commonly used approach to generate DCs is through ex-vivo differentiation from peripheral blood. The advantage of this method is the easy generation of sufficient autologous DCs for vaccination. However, therapeutic outcomes still have a lot of room for improvement, with less than 15% the patients showing objective response [101]. Due to the artificial in vitro differentiation process, moDCs have compromised functionality compared with naturally-occurring DCs with different transcriptional profiles. The limitation of using naturally-occurring DCs is the low frequency of DCs in peripheral blood, resulting in a highly labor-intensive process in DC isolation for clinical use. To overcome this, a growing effort in the field has been exerted to facilitate the developing a feasible protocol, for example, an automatic system that can prepare DCs [102]. For DC vaccine production, DCs are then be loaded with total tumor lysate or RNAs and tumor associated antigens. The loading methods include pulsing by co-culturing, electroporation, viral transduction or DC-tumor fusion [103]. Maturation cocktails used in the clinic consist of TLR agonists and cytokines, often in combination with co-stimulatory proteins like CD40L. Introducing mRNAs coding constitutively-active TLR4, CD40L and CD70 via electroporation has shown clinical success [33, 104].

4.2 DC vaccination clinical trial in glioblastoma

DC vaccination in the context of glioblastoma has shown both positive and negative results in clinical trials. Even though a phase III clinical trial aiming to assess DC vaccine targeting the EGFR deletion mutation EGFRvIII in newly diagnosed EGFRvIII-expressing GBM patients failed [105], some other clinical trials have shown promising results. Another phase III clinical trial utilizing an autologous tumor lysate-pulsed dendritic cell vaccine combined with standard therapy showed significant overall survival benefit from 15 to 17 months to 23.1 months [36]. In another phase II clinical trial, ICT-107 (autologous dendritic cells (DC) pulsed with six synthetic peptide epitopes targeting GBM tumor/stem cell-associated antigens MAGE-1, HER-2, AIM-2, TRP-2, gp100, and IL13Rα2 was given to newly diagnosed glioblastoma patients in addition to standard therapy. Results showed progression free survival (PFS) increased 2.2 months in ICT-107 cohort compared with matched DC control cohort. HLA-A2 subgroup patients achieved a meaningful therapeutic benefit with ICT-107, in both the MGMT methylated and unmethylated prespecified subgroups, whereas only HLA-A1 methylated patients had an OS benefit [106, 107]. Combination with other intervention methods could help increase DC vaccine efficacy. Pre-conditioning at vaccinated site can improve DC vaccination efficacy. Mitchell et al. (2015) showed that glioblastoma patients pre-exposed to tetanus/diphtheria (Td) toxoid in the vaccine site before vaccination with pp65 RNA-pulsed DCs had improved tumor-antigen-specific DC migration and improved survival compared to the ones that were not pre-exposed to the toxoid through increasing DC migration to dLN [108]. Three phase II clinical trials (ATTAC; ELEVATE; NCT00639639, NCT00639639, NCT02366728) aim to test pp65 DC with Td vaccine in newly diagnosed GBM patients. Results to date have shown that despite a small cohort, three successive trials demonstrate consistent survival outcomes, supporting the efficacy of cytomegalovirus DC vaccine therapy in GBM [109].

5. Conclusions

Dendritic cells, as the most professional APCs, play key roles in mediating the bridge between innate and adaptive immunity in anti-tumor immunity. DC subpopulations, through use of different action mechanisms in activating adaptive immunity, collaborate with each other to elicit anti-tumor immunity. In the battle with tumor, DC functions become regulated by tumor cells or other components in the tumor microenvironment, leading to DC dysfunction. These include impairments on antigen uptake, antigen presentation, migration to LN, and DC tolerance. Secreted factors from the tumor environment play a key role in mediating DC regulation. These suppressive signals act on DCs inducing DC dysfunction through different cellular intrinsic pathways. DC vaccine development for tumor treatment has made significant progress in the last decades, but still faces challenges in achieving a wide and significant therapeutic success. Deepening our understanding on DC function and regulation in the tumor environment will help the field in developing new and more powerful therapeutic intervention approaches.

Acknowledgments

This work was funded by NIH NINDS 5R01NS111033-03.

Conflict of interest

CF is a founder of iOncologi.