Direct and indirect engagement of dendritic cell function by antibodies developed for cancer therapy

Abstract

Dendritic cells (DC) are crucial for the priming of T cells and thereby influence adaptive immune responses. Hence, they also represent important players in shaping anti-tumour immune responses. Cancer immunotherapy has been driven over many years by the aim to harness the T-cell stimulatory activity of these crucial antigen-presenting cells (APC). Efficient antigen delivery alone is not sufficient for full engagement of the T-cell stimulatory activity of DC and the inclusion of adjuvants triggering appropriate DC activation is essential to ensure effective anti-tumour immunity induction. While the direct engagement of DC function is a powerful tool for tumour immunotherapy, many therapeutic antibodies, such as antibodies directed against tumour-associated antigens (TAA) and immune checkpoint inhibitors (ICI) have been shown to engage DC function indirectly. The induction of anti-tumour immune responses by TAA-targeting and immune checkpoint inhibitory antibodies is thought to be integral to their therapeutic efficacy. Here, we provide an overview of the immunotherapeutic antibodies in the context of cancer immunotherapy, that has been demonstrated to directly or indirectly engage DC and discuss the current understanding of the functional mechanisms underlying anti-tumour immunity induction by these antibody therapies. In the future, the combination of therapeutic strategies that engage DC function directly and/or indirectly with strategies that allow tumour infiltrating immune effector cells to exert their anti-tumour activity in the tumour microenvironment (TME) may be key for the successful treatment of cancer patients currently not responding to immunotherapeutic antibody treatment.

The article is a minireview on direct and indirect engagement of DC function by immunotherapeutic antibodies for treatment of cancer.

Graphical Abstract

Graphical Abstract.

Introduction

Dendritic cells (DC) are sentinels of the immune system that detect signs of infection in the periphery, sample antigens associated with these infectious signals, and initiate the induction of an adaptive immune response. Innate immune activation of DC in response to infection mediated via conserved pattern recognition receptors (PRR) is crucial for the initiation of adaptive immune responses [1]. These PRR sense a variety of pathogen-associated molecular patterns (PAMP) which are characteristic of whole classes of pathogens. Upon PRR-mediated activation, DC migrate to the draining lymph node where they interact with T cells and induce antigen-specific T-cell activation by supplying costimulatory signals and cytokines in a cognate manner. In addition to PAMP, PRR can also sense endogenous damage-associated molecular patterns (DAMP) as signs of cellular stress or cell death to be investigated by innate immune cells including DC [2]. The stimulatory properties of DC are further influenced by tissue-derived factors which in combination with PAMP and DAMP together impact the phenotype of the T cell response and the associated immune defence mechanisms that are mounted [3, 4]. In the absence of appropriate innate immune activation signals, antigen-presenting DC fails to provide T-cell stimulatory signals required for the initiation of antigen-specific immunity, which can lead to tolerance induction by T-cell depletion, anergy induction, or induction of regulatory T-cells [5]. There is an ongoing debate about the tolerogenic nature of immature DC versus DC that were alternatively activated in the absence of PAMP or DC that have encountered an immunosuppressive environment [6–8].

Due to their crucial role in T-cell priming and in modulating cellular and humoral immune responses, DC are important players not only in naturally unfolding immune responses but also in immunotherapy [3]. Strategies to harness the T-cell stimulatory function of DC for cancer immunotherapy aim at inducing antigen-specific T-cell responses of a T-helper type 1 (Th1) phenotype accompanied by priming of cytotoxic T lymphocytes (CTL) with the ability to eradicate tumour cells. Skewing of T-cell priming towards a Th1 phenotype requires appropriate innate activation signals [4]. As professional antigen-presenting cells (APC), DC express a variety of cell surface, endosomal and cytoplasmic PRR which enable detection of innate stimuli in the environment, associated with infected cells or upon infection of the DC themselves. PRR includes C-type lectin receptors (CLR), Toll-like receptors (TLR) and cytoplasmic NOD-like receptors, retinoic acid-inducible gene 1-like receptors, and cytoplasmic DNA sensors. CLR are a class of surface receptors involved in the recognition of pathogen-associated and endogenous carbohydrate structures which have been widely explored for DC targeting [9]. TLR are located either at the cell surface or in a specialized endosomal compartment. Since TLR agonists triggering TLR3-, TLR4-, TLR7/8-, and TLR9-mediated activation promote Th1 responses and CTL priming, they have attracted great interest as adjuvants in the context of cancer immunotherapy [10, 11].

Generally, DC can be divided into conventional DC subsets 1 and 2 (cDC1 & cDC2) and plasmacytoid DC (pDC). More in-depth classification of DC is complex with different subsets having specialized yet at the same time overlapping roles in immunity induction [12, 13]. Functionally, the cDC1 subset is specialized in cross-presentation of exogenous antigen on major histocompatibility complex I (MHCI) and was shown to be crucial for priming of anti-viral and anti-tumoural CTL [14]. In contrast, cDC2 are potent inducers of T helper responses. pDC are highly specialized in producing high levels of type I interferon (IFN-I) in response to viral PAMP and thereby promote the induction of anti-viral immunity [15]. However, their role as antigen-presenting cells in tolerance induction, T-cell priming, and backup of ongoing protective responses needs further clarification [15, 16]. Importantly, the functional specialization of individual DC subsets is partially a consequence of their PRR expression pattern and has crucial implications for immunotherapeutic strategies [17, 18]. Unfortunately, phenotypic and functional differences between human and mouse DC subsets complicate the investigation of strategies targeting different DC subsets, and insight from mouse models cannot be directly translated into the clinical setting [1]. In addition to the species-specific differences, the expression of phenotypic markers expressed by cDC1, cDC2, and pDC differs between DC located in various peripheral tissues and resident in lymphoid tissues. Furthermore, there are species-specific differences in TLR and CLR expression by DC [8]. For example, TLR9 is more widely expressed on mouse DC subsets whereas human TLR9 is expressed more exclusively by pDC and human TLR8 is expressed by cDC1 recognizes viral RNA whereas the mouse TLR8 homologue is functionally different [19, 20]. Similarly, there are species-specific differences in CLR expression between humans and mice. For example, the mouse homologue of the human pDC-specific marker BDCA-2 and the mouse orthologue of DC-SIGN have not been identified [21, 22]. Hence, the existing experimental mouse model systems for investigating the activation and antibody-mediated engagement of DC are often suboptimal predictors for clinical efficacy.

Early strategies for cancer immunotherapy focused on DC vaccines. These consisted of ex vivo generated monocyte-derived DC pulsed with tumour lysate or tumour antigen-specific peptides or proteins in the presence of adjuvants [23, 24]. While numerous clinical trials demonstrated that DC vaccines are safe, it also became evident that they are labour- and cost-intensive and show low efficacy as stand-alone treatment [25, 26]. Hence, strategies for antigen-loading of DC in vivo were explored as an alternative approach for the successful treatment of a wider number of cancer patients with off-the-shelf vaccines [27]. Antibodies targeting DC receptors involved in antigen uptake seemed a natural choice for efficient antigen-loading of DC in vivo.

Direct antibody-mediated targeting of dendritic cells for tumour immunotherapy

One of the first and most widely explored receptors targeted for antigen delivery to DC is the CLR DEC-205 (CD205) which plays a role in antigen presentation [28]. Antigen conjugated to the DEC-205 specific antibody NLDC-145 was found to be presented 400 times more efficiently than free antigen and to reduce tumour growth in a mouse melanoma model [29, 30]. Subsequently, other CLR have been explored for antibody-mediated DC targeting including Clec4A, Clec7A (Dectin-1), Clec8A (LOX1), Clec9A (DNGR-1, CD370), Clec12a, DC-SIGN (CD209), DCIR2, BDCA-2, langerin, and the mannose receptor (MR). Similar to DEC-205, targeted antibody-mediated delivery of antigen to these CLR was shown to promote T cell activation and anti-tumour immunity induction [31, 32].

CLR are a diverse class of PRR that bind and capture foreign pathogen-associated or endogenous carbohydrate structures and are involved in innate immune activation, antigen uptake, and processing but also mediate cellular interactions [9]. CLR vary in their expression patterns with some CLR such as DEC-205 being widely expressed on different cell types and DC subsets whereas others such as Clec9a are expressed more exclusively on specific DC subsets [9]. Consequently, the choice of CLR used for antibody-targeting can have functional implications. Targeting of cDC1 is thought to be preferable over the targeting of CLR exclusively expressed on the cDC2 subset. Antibody-mediated targeting of cDC1-expressed CLR was shown to promote priming of CD4 and CD8 T cells whereas targeting of cDC2-specific CLR was shown to predominantly lead to induction of CD4 T cell responses [33–35]. However, while cDC1 are crucial for cross-priming of CTL responses, a clear advantage of exclusively targeting this DC subset could not be established experimentally. Similarly, antibody-mediated targeting of pDC has been shown to induce CD4 and CD8 T cells as well as humoral responses [36]. However, the role of pDC in T cell priming is not well understood and may be restricted under physiological conditions [15, 16]. It is currently unclear whether specific targeting of cDC1 versus pDC offers a specific advantage for tumour immunotherapy. Overall, strategies that allow for efficient targeting of cross-priming cDC1 and additional DC subsets may be preferential for optimal CD4 and CD8 T cell priming. Importantly, no detrimental effects of targeting other cell types in addition to DC have been observed, further supporting the notion that the requirement to limit targeting to a specific DC subset is not essential as long as DC is efficiently loaded with antigen and the cross-priming activity of DC is sufficiently engaged [31, 37].

What all antibody-mediated DC targeting strategies seem to have in common is the need to ensure appropriate innate immune activation to fully engage the T-cell stimulatory activity of DC and, thereby, enable optimal cellular immunity induction. In the absence of adjuvants, DC-targeted antibody-mediated delivery of antigen fails to induce effective cellular immunity but allows limited induction of humoral responses in some cases [38]. For clinical application, the addition of adjuvants has also been identified to be crucial for the efficacy of antibody-mediated DC targeted therapy [39]. Targeting of CLR such as Clec9A, DEC-205, DC-SIGN, and mannose receptor in the absence of adjuvants can be employed therapeutically for tolerance induction showing the crucial role of appropriate innate immune activation for therapeutic outcome [40].

The adjuvants of choice for activating DC capable of inducing Th1 and CTL priming are TLR agonists [41]. TLR agonists trigger NFκB activation which in turn leads to the expression of DC-derived cytokines including IL-6, TNFα, IL-12, and IFN-I. The exact cytokine profile that is induced in response to specific TLR agonists depends on the expression profile of the TLR on different DC subsets and on cell-intrinsic differences in the downstream signalling pathway with pDC for example being specialized in producing high levels of IFN-I in response to several nucleic acid TLR agonists [12]. DC-produced IFN-I and IL-12 are instrumental in inducing strong Th1 responses and enabling CTL priming [42, 43]. Several TLR agonists explored for antibody-mediated DC targeting have been demonstrated to be suitable adjuvants for anti-tumour immunity induction in pre-clinical and clinical studies [10].

CLR are not the only receptors that have been explored for DC targeting. Among the list of non-CLR that have been targeted with antibodies are the integrin CD11c which is widely expressed on DC and activated T cells, the scavenger receptor CD36, the lectin Siglec-H, the tumour necrosis superfamily member CD40 which is up-regulated on DC upon activation as well as BST-2 which is exclusively expressed by pDC [27, 31, 36, 37]. Interestingly, in a comparative study exploring targeted nanoparticles, targeting of CD11c, CD40, and DEC-205 on DC was equally efficient in inducing activation of CD8 T cell responses [44]. As for targeting CLR, the therapeutic outcome of targeting other receptors for direct DC engagement is influenced by the expression pattern of the targeted receptor on different DC subsets, by the intrinsic properties of the endosomal compartment into which the therapeutic reagent is internalized and by the adjuvants which are used to promote the induction of cellular immunity. Interestingly, different antibody isotypes as well as Fab fragments and single-chain variable fragments have been used successfully for CLR-directed antibody-mediated direct DC engagement in mouse models and there is currently no indication that Fc-receptor mediated mechanisms influence antibody-mediated antigen delivery [36, 38].

Another form of antibody therapy targeting DC directly is immunostimulatory monoclonal antibody therapy. The anti-tumour activity of agonistic antibodies against CD40 is the result of various effector mechanisms including but not limited to the induction of antibody-dependent cellular cytotoxicity (ADCC) and CTL-mediated cytotoxicity [45]. Upon direct engagement of DC, anti-CD40 antibodies enable the activated tumour-associated antigen (TAA)-presenting cells to prime anti-tumour CTL responses in the absence of T cell help [46–48]. The isotype of anti-CD40 antibodies was shown to play a crucial role in their agonistic potency in pre-clinical models [49, 50].

While antibodies are established tools for in vivo targeting of cell surface receptors, there are other means for DC targeting. Natural ligands such as chemokines or natural and engineered glycan molecules have been used for shuttling antigen to DC [51–54]. In terms of efficacy, it is impossible to generalize whether antibodies or natural ligands are the better strategies for cancer immunotherapeutic engagement of DC function.

Indirect engagement of dendritic cells with therapeutic antibodies targeting tumour-associated antigens

Monoclonal antibodies targeting membrane-bound TAA have been developed for tumour therapy, with several agents approved for clinical use including Trastuzumab (anti-HER2), Cetuximab (anti-EGFR), Rituximab (anti-CD19), and Alemtuzumab (anti-CD52) [55, 56]. Anti-tumour monoclonal antibodies typically interact with tumour cells via the antigen-binding site and with Fc receptors on immune cells via the antibody constant region (Fc), thereby mediating a variety of anti-tumour immune effects. Among antibody types and subtypes, IgG1 antibodies distinguish themselves with superior capacity for Fc-domain mediated immune effects, which is why most targeted anti-cancer antibodies are of the IgG1 isotype [57, 58].

Classical Fc-mediated immune effects of such antibodies, including ADCC largely accomplished by natural killer (NK) cells, antibody-dependent cellular phagocytosis (ADCP) mostly mediated by monocytes and macrophages, and complement-dependent cytotoxicity (CDC) have been extensively described elsewhere [59].

Interestingly, recent data showed that in addition to classically described Fc-mediated anti-tumour immune effects, antigen presentation by DC resulting in the generation of anti-tumour T cell responses can be enhanced by treatment with clinically used anti-tumour antigen antibodies such as Cetuximab [60, 61] and Trastuzumab [62, 63]. Here, we elaborate on these observations and their underlying mechanisms.

The capacity of tumour-targeted antibodies to promote DC activity resulting in enhanced T cell responses was extensively investigated for Cetuximab, an anti-EGFR antibody of the IgG1 isotype. Treatment with Cetuximab in colon cancer patients increased tumour infiltration with T cells and NK cells, including at metastatic sites, as demonstrated by immunohistochemistry staining for T cell markers CD3, CD8, as well as NK cell marker CD56 [61]. Higher numbers of activated T cells and NK cells were also detected in the blood of colon cancer patients receiving Cetuximab treatment [64]. Similarly, higher levels of CTL specific for EGFR and the TAA MAGE-3 were observed in patients treated with Cetuximab for head and neck malignancies [60]. In vitro experiments have shown that NK cells activated by tumour cells opsonized with Cetuximab are key to promoting DC maturation, Th1 cytokine production [65] and CTL priming [60, 66]. DC activation was dependent on IFN-γ produced by activated NK cells, and was not observed with IgG2 isotype anti-EGFR antibody [60]. In contrast, NK-cell mediated DC priming was observed using Nimotuzumab, an alternative IgG1 isotype anti-EGFR antibody. Treatment with Nimotuzumab also increased the number of EGFR-specific CTL in patients with head and neck cancers [67].

Increased T cell responses after anti-cancer TAA-targeted antibody therapies have been observed in breast cancer patients treated with Trastuzumab [62, 63] or trastuzumab antibody-cytotoxic drug conjugate [68]. A separate group observed increased levels of MHC-II positive population in the breast cancer biopsies from patients with a partial response to treatment with Trastuzumab [69]. Literature investigating the mechanisms behind these observations is, however, sparse. A few mouse models showed that anti-HER2 antibodies clear tumours in immunocompetent mice, but not in mice in which CTL were depleted [70, 71]. Similar results were obtained with other TAA-targeted antibodies and different tumour models [72, 73].

Taken together, TAA-targeted antibody therapies have been shown to promote DC maturation and subsequent T cell responses. There are a number of mechanisms (Fig. 1) with the potential to drive this indirect engagement of DC function alone or in combination: (1) Tumour antigens may become more readily available for DC sampling through increased ADCC-induced cell death, (2) expression or release of DAMP by dying tumour cells may aid DC activation and (3) opsonization of tumour cells with therapeutic antibodies may promote FcγR-mediated uptake of tumour cell material by DC [74]. However, specific Fc receptor subtypes and environmental conditions leading to antigen presentation and T cell responses are complex, and currently incompletely defined [75]. In vitro studies showed that Trastuzumab enhances HER2 uptake by via Fc-FcγR interactions, leading to increased cross-presentation and priming of HER2-specific T cells [76]. Currently, it is unclear to which extent and under what conditions these mechanisms contribute to expansion of anti-tumour CTL in patients.

Figure 1:

Comparison of direct versus indirect antibody-mediated DC engagement targeting receptors on DC, surface TAA, or ICI, respectively. Please note, antibody-TAA conjugates are used as a representative form of antibody-mediated DC-targeted TAA delivery. Also, ICI therapy can target immune checkpoint molecules on T cells, tumour cells, and/or DC.

In addition, multiple studies reported that inflammatory factors such as IFN-I, Flt3L, XCL1, CCL4, and CCL5 support the antigen-presenting and T-cell stimulatory function of DC [77–80]. Many of these inflammatory factors are produced by NK cells, however, their role in promoting anti-tumour immunity in response to TAA-targeted antibody therapy remains unclear. Similarly, it is still unclear which immunological parameters distinguish patients that mount effective anti-tumour T cell responses in response to TAA-targeted immunotherapy from non-responders, nor which therapeutic conditions may promote such responses.

Dendritic cells and immune checkpoint inhibitors

In recent years, the clinical success of immune checkpoint inhibitors (ICI) has revolutionized cancer immunotherapy. ICI are monoclonal antibodies targeting molecules that represent immune checkpoints for effector cells infiltrating the tumour microenvironment (TME) such as CTL-associated protein 4 (CTLA-4), programmed cell death protein 1 (PD-1), or programmed death-ligand 1 (PD-L1) molecules. ICI antibodies block inhibitory signals regulating T cell function, thereby unleashing T cell anti-tumour activity in the TME. The impact of immune checkpoint inhibition on T cell function has been extensively described elsewhere [81, 82].

ICI therapy leads to tumour regression in a percentage of patients with solid or haematological malignancies. Unfortunately, it is still impossible to predict or understand why some patients respond to treatment while others do not [83, 84]. This prompted an ongoing wider investigation into the mechanisms mediating ICI-induced anti-tumour immune responses, as well as their effects on multiple cell types in the TME including DC.

An emerging body of evidence from mouse [78, 85–87] and human studies [88] shows that DC, and especially the cDC1 subset, are crucial for achieving a therapeutic response to ICI. Recent insight from experimental mouse studies highlights cDC1 as essential for effector T-cell priming [85] and also emphasizes the crucial role of cDC1 residing in the TME in recruiting effector T cells to the TME [78]. In vivo expansion of cDC1 in response to Flt3L administration and TLR-mediated cDC1 activation enhances ICI-induced anti-tumour immune responses in several tumour mouse models [86, 87]. Importantly, recent studies showed that mouse and human cDC1 are similar, but not identical at the genetic and at functional level. This implies that mechanistic insights from experimental studies are likely to apply to clinical settings [89].

Evidence supporting the role of DC in response to ICI therapy stems from a transcriptomic analysis of patient tissue samples. High cDC1 infiltration into lung adenocarcinoma patients correlated with increased survival, as well as response to anti-PD-1 treatment [88]. Similarly, DC intratumoral signature was associated with better outcomes for PD-L1 therapy in patients with renal cell carcinoma or non-small lung cell cancer [90]. Furthermore, the ratio of cDC1 to other DC subsets was correlated with a better prognosis in a number of tumours [91].

While emerging evidence suggests that the presence of functional cDC1 in the tumour microenvironment is essential for ICI efficiency, ICI therapy in turn influences DC activity through direct and indirect mechanisms [92]. Tumour-infiltrating DC expresses PD-L1 in mice and cancer patients and contribute to PD-1 mediated regulation of tumour-infiltrating lymphocyte (TIL) function [90, 93, 94]. Direct ICI-mediated blockade or deletion of PD-L1 on DC promotes CTL-mediated control of tumour growth in the mouse model [93, 94]. In mice, ICI therapy initiates T-cell-derived IFN-γ production and thereby supports IL-12 induction by DC, which in turn supports CTL function [95]. By supporting cross-talk between DC and T-cells ICI indirectly promote CTL induction [95].

Together, these findings support the crucial role of DC and particularly cDC1 in determining the response to ICI. Whether this effect is tumour-specific remains to be investigated. Similarly, it is currently unclear whether cDC1 recruitment and activation would improve response to ICI in non-responsive tumours. An important limitation is that most of the human data is limited to transcriptomic analysis of patient tumours in observational studies. With the emergence of sophisticated humanized mouse models, enabling the investigation of human DC subsets in mice, it will be interesting to see if such in vivo experiments with human cDC1 validate the findings described here.

Conclusions

Overall, there is compelling evidence from pre-clinical and clinical studies highlighting the crucial role of DC in general and cDC1 in particular in effective anti-tumour immunity induction in response to immunotherapeutic intervention strategies including direct antibody-targeting of DC, antibody-targeting of surface TAA and ICI (Fig. 1). It is important to recognize the multiple possibilities by which combination therapy can promote therapeutic efficacy and may convert patients not responding to single treatment modalities into patients responding successfully to cancer immunotherapy. Optimal direct engagement of the T-cell stimulatory activity of cross-priming DC paired with therapeutic intervention strategies aimed at altering the TME may hold the key to widening the successful clinical application of tumour immunotherapy. Alteration of the TME for promotion of anti-tumour immunity can be achieved by ICI, by targeting surface TAA with therapeutic antibodies, by local administration of adjuvants, and by combinations of these strategies. Future clinical trials should focus on identifying combinations with the highest improved efficacy in the therapeutic outcome, which may differ for different cancers.

Glossary

Abbreviations

ADCC

antibody-dependent cellular cytotoxicity

ADCP

antibody-dependent cellular phagocytosis

APC

antigen-presenting cell

CDC

complement-dependent cytotoxicity

cDC1 & 2

conventional dendritic cell subsets 1 & 2

CLR

C-type lectin receptor

CTL

cytotoxic T lymphocyte

CTLA-4

CTL-associated protein 4

DAMP

danger-associated molecular pattern

DC

dendritic cell

ICI

immune checkpoint inhibitor

IFN-I

type I interferon

MHCI

major histocompatibility complex I

PAMP

pathogen-associated molecular pattern

pDC

plasmacytoid dendritic cell

PD-1

programmed cell death protein 1

PD-L1

programmed death-ligand 1

PRR

pattern recognition receptor

TAA

tumour-associated antigen

TLR

Toll-like receptor

TME

tumour microenvironment

Funding

D.C. was funded by a NIBSC studentship. S.S.D, and D.C. did not receive extramural funding.

Conflicts of interest

S.S.D. and D.C. declare no conflict of interest.

Author contributions

S.S.D and D.C. contributed equally to the outline and the writing of the manuscript.

Data availability

This review article does not contain any experimental data generated by the authors.