Blood DCs activated with R848 and poly(I:C) induce antigen-specific immune responses against viral and tumor-associated antigens

Monocyte-derived Dendritic cells (DCs) have successfully been employed to induce immune responses against tumor-associated antigens in patients with various cancer entities. However, objective clinical responses have only been achieved in a minority of patients. Additionally, generation of GMP-compliant DCs requires time- and labor-intensive cell differentiation. In contrast, Blood DCs (BDCs) require only minimal ex vivo handling, as differentiation occurs in vivo resulting in potentially better functional capacities and survival. We aimed to identify a protocol for optimal in vitro activation of BDCs including the three subsets pDCs, cDC1s, and cDC2s. We evaluated several TLR ligand combinations and demonstrated that polyinosinic:polycytidylic acid [poly(I:C)] and R848, ligands for TLR3 and TLR7/8, respectively, constituted the optimal combination for inducing a positive co-stimulatory profile in all BDC subsets. In addition, TLR3 and TLR7/8 activation led to high secretion of IFN-α and IL-12p70. Simultaneous as opposed to separate tailored activation of pDCs and cDCs increased immunostimulatory capacities, suggesting that BDC subsets engage in synergistic cross-talk during activation. Stimulation of BDCs with this protocol resulted in enhanced migration, high NK-cell activation, and potent antigen-specific T-cell induction. We conclude that simultaneous activation of all BDC subsets with a combination of R848 + poly(I:C) generates highly immunostimulatory DCs. These results support further investigation and clinical testing, as standalone or in conjunction with other immunotherapeutic strategies including adoptive T-cell transfer and checkpoint inhibition. Supplementary Information The online version contains supplementary material available at 10.1007/s00262-021-03109-w.


Introduction
Generating, enhancing, and maintaining tumor-specific immune responses is a major challenge in the search for cures for cancer. Dendritic cells (DCs) orchestrate innate and adaptive immunity, thereby inducing tailored, strong, and durable immune responses. They have therefore been extensively evaluated as an immunotherapeutic tool, including various in vivo targeting approaches [1,2]. However, most pre-clinical and clinical studies have focused on ex vivo DC activation and antigen-loading approaches [3,4].
Numerous clinical trials have evaluated monocytederived DCs (moDCs) against different cancer entities, including multiple myeloma and Acute myeloid leukemia (AML). Their overall safety and induction of tumor-specific immune responses have been demonstrated [4][5][6][7][8]. However, objective clinical responses have only been obtained for a minority of patients [9,10]. In addition, the generation of moDCs is labor-and time-consuming as the differentiation of monocytes into DCs takes several days. Furthermore, transcriptional comparisons of moDCs and naturally occurring DC subsets indicate great functional differences [11].
Therefore, other sources of DCs have recently been evaluated. Primary blood DCs (BDCs) are hypothesized to be ideal candidates for inducing anticancer immune responses since they differentiate in vivo and require only brief ex vivo handling, resulting potentially in better preservation of functional capacities and longer in vivo survival [12]. At least three BDC populations can be distinguished: CD11c + CD141 + CLEC9A + conventional DCs (cDC1s), CD11c + CD1c + cDC2s, and CD11c − CD303 + CD123 + plasmacytoid DCs (pDCs) [13].
pDCs are specialized to recognize viral infections via tolllike receptor (TLR) 7 or TLR9, which results in strong production of type I interferons (IFNs). These cytokines modulate adaptive and innate immunity by inducing, for example, NK-cell activation, B-cell differentiation, and Th1 polarization [14]. In contrast, cDC2s express TLRs that recognize lipopolysaccharides (TLR4), flagellin (TLR5), and the lipoproteins of bacteria and fungi (TLR2 and TLR6). They also express TLR8 and TLR9 that are activated by singlestranded RNA and DNA, respectively. Thus, different CD4 + T helper subsets and CD8 + T cells are induced depending on the circumstances [15]. cDC1s represent a rare BDC subset that expresses, among others, TLR3 that is activated by viral double-stranded RNAs leading to secretion of interleukin (IL) 12 [16]. They also express TLR8, which is closely related to TLR7, and the endocytic receptor CLEC9A that promotes antigen cross-presentation of intracellular pathogens or necrotic cells on MHC class I, thereby activating cytotoxic CD8 + T cells [17]. cDC1s have therefore attracted much interest in the context of cancer immunotherapy [18].
Owing to their multifaceted properties, BDCs are considered to induce a greater diversity of immune responses compared to moDCs when used in vaccination approaches. The first clinical trials evaluating pDCs and cDCs in solid tumors have already demonstrated safety and, to some extent, induction of tumor-specific immune responses. Tickborne encephalitis vaccine-activated pDCs demonstrated a mature phenotype and produced large amounts of type I IFNs in melanoma patients, inducing tumor-specific immune responses [19]; this is in contrast to antigen-loaded, but nonactivated cDC2s administered to prostate cancer patients [20]. In a clinical trial of GM-CSF-activated cDC2s, three of 14 metastatic melanoma patients exhibited functional tumor-specific T cells. However, the authors postulated that immune responses might be optimized by a better-tailored activation stimulus [21]. In line with this hypothesis, we evaluated different protocols for in vitro activation of BDCs based on TLR ligands.
Here, we report the first protocol for achieving simultaneous in vitro activation of all BDC subsets based on activation of TLR3 and TLR7/8 with polyinosinic:polycytidylic acid [poly(I:C)] and R848 (Resiquimod), respectively. Our data suggest this to be the optimal combination for inducing a positive co-stimulatory profile in all BDC subsets, high secretion of IFN-α, and maximal secretion of IL-12p70. Activating all BDC subpopulations together increased immunostimulatory capacities compared to separately activating pDCs and cDCs with tailored protocols, indicating a synergistic cross-talk between BDC subsets during activation. Moreover, we demonstrate that activation of BDCs with this protocol results in enhanced migration, high NK-cell activation, and potent antigen-specific T-cell induction.

Healthy donors
Heparinized peripheral blood was collected from healthy donors after informed consent in accordance with the Declaration of Helsinki and approval by the Institutional Review Board of the Ludwig-Maximilians-Universität (Munich, Germany).

Isolation and culturing of BDCs
BDCs were isolated from peripheral blood mononuclear cells (PBMCs) using the human Blood Dendritic Cell Isolation Kit II (Miltenyi Biotec, Bergisch Gladbach, Germany) and cultured in DC medium [X-Vivo 15 medium (Lonza, Verviers, Belgium), 2% human serum (HS;
Protamine-RNA complexes were formed as previously described [22] using mMESSAGE mMACHINE T7 Transcription Kit (Thermo Fisher Scientific, Vilnius, Lithuania) for mRNA generation from the supplied control template.

Flow cytometry and cytokine quantification
Expression of immune checkpoint molecules on BDCs was assessed using Aqua-LIVE/DEAD (Life Technologies, Eugene, OR, USA) and the antibodies and isotype controls listed in Table 1. Median fluorescence intensity (MFI) ratios were calculated based on respective isotype controls.

Migration assay
The lower part of a 96-transwell plate (5 µm pore size, Corning, Kennebunk, ME, USA) was filled with 200 µL X-Vivo15 medium supplemented with 2% HS and either 200 ng/mL CCL19 (R&D Systems), 200 ng/mL CCL21 (Biolegend) or no chemokine. BDCs were seeded in the upper chamber in technical duplicates. Cells were harvested after 3 h incubation at 37 °C from both chambers, stained for CD11c and CD123, and analyzed by flow cytometry together with Precision Count Beads (Biolegend).

A combination of TLR ligands is required to induce a positive co-stimulatory profile on all BDCs
To identify a protocol for activating all BDC subsets simultaneously, we analyzed the expression of immune checkpoint molecules by BDCs in response to different TLR ligands and combinations thereof (Fig. 2a-d). Supplementary Table 1 reports expression of the immune checkpoint molecules CD80, CD40, PD-L1, and the migratory receptor CCR7 for all BDC subsets in response to different combinations of TLR ligands. Activation with TLR3 ligand poly(I:C) caused high expression of CD80 and CD40 and the highest expression of CCR7 on cDC1s and cDC2s. However, expression of CD80 and CD40 was further increased on cDC subsets if poly(I:C) was combined with other TLR ligands. The combination of poly(I:C) with the TLR9 ligand CpG consistently caused the highest expression of these markers on pDCs (CD80: 191.1 ± 34.8; CD40: 129.2 ± 44.7; CCR7: 70.9 ± 13.2). The combination of poly(I:C) with the TLR7/8 ligand R848 resulted in consistently high expression of CD80, CD40, and CCR7 on all BDC subsets (Fig. 2a-c).
Since R848 + poly(I:C) caused an overall positive costimulatory expression on BDCs, high IL-12 secretion, and intermediate levels of IFN-α secretion, we focused on this combination for subsequent experiments.

IL-12 and IFN-α are secreted at different time points
To determine when cytokines are secreted by BDCs, we analyzed cytokine secretion 1, 3, 6, and 20 h after TLR activation. Owing to time constraints, we introduced an additional resting step of 16 h between BDC isolation and activation. As a control, we also evaluated a 20-h activation without the resting step.

A shorter BDC activation time does not increase T-cell responses
To determine if a shorter duration of BDC activation improves T-cell responses, we prepared co-cultures of T cells and autologous BDCs activated with R848 + poly(I:C) for 1, 3, 6, and 20 h after 16 h "resting" or for 20 h directly after cell isolation. We analyzed T-cell proliferation and IFN-γ secretion after five and four days, respectively. BDCs activated for 20 h induced a T-cell proliferation of 59.0 ± 6.0%. This was significantly higher compared to shorter activation periods of between 1 and 6 h (range: 32.3-36.9%; p < 0.001). "Resting" versus "no resting" prior to 20-h activation did not result in a significant difference (p = 0.781) (Fig. 3c). In contrast, maximum IFN-γ secretion by T cells was detected with BDCs activated for 20 h directly after cell isolation (1.6 × 10 3 ± 4.3 × 10 2 pg/mL), compared to BDCs rested for 16 h (for 20 h: 5.2 × 10 2 ± 5.6 × 10 1 pg/ mL; p = 0.007). No significant differences in IFN-γ secretion between BDCs activated for 1, 3, 6, or 20 h after the resting step was observed (Fig. 3d).

Individual activation of cDCs and pDCs with a tailored protocol is not superior to a combined activation of all BDCs
An important translational question is whether separate activation of cDCs and pDCs with a tailored activation protocol is superior to the combined activation of all BDC subsets. To address this question, we isolated pDCs, cDCs, and entire BDCs from the same donors and activated cDCs and BDCs with R848 + poly(I:C) for 20 h immediately after isolation. By contrast, pDCs were rested for 16 h before exposing them to CpG + poly(I:C) for 3 h, as described for optimal IFN-α secretion above. Secretion of IL-12 was higher by BDCs than by cDCs (3.5 × 10 3 ± 1.5 × 10 3 and 1.9 × 10 3 ± 7.4 × 10 2 pg/mL, respectively). Activation with CpG + poly(I:C) resulted in strong secretion of IFN-α by pDCs (4.9 × 10 3 ± 9.2 × 10 2 pg/mL), while observed cytokine concentrations for R848 + poly(I:C)-activated BDCs were lower (2.9 × 10 3 ± 8.1 × 10 2 pg/mL). We did not observe any cytokine secretion in controls without TLR ligands (Fig. 4a,  b).

TLR-activated BDCs expand antigen-specific CD8 + T cells
In order to test the capacity of BDCs to induce antigenspecific T-cell responses, autologous CD8 + T cells were co-cultured with BDCs pulsed with viral peptides derived from CMV, EBV, and influenza A (CEF peptides) or SARS-CoV-2, or peptides derived from the tumor oncogene WT1 (Fig. 5a, b).
R848 + poly(I:C)-activated BDCs demonstrated enhanced expansion of CEF-specific CD8 + T cells, compared to nonactivated BDCs (mean fold change of Cumulative secretion of IL-12p70 (a) and IFN-α (b) by BDCs after 1, 3, 6, and 20 h of activation with R848 and poly(I:C). BDCs were "rested" for 16 h before the TLR ligand was added. In addition, BDCs were activated directly after isolation for 20 h (n = 3-6). c Induction of T-cell proliferation by BDCs activated for various time peri-ods (n = 9). d Secretion of IFN-γ in co-cultures of T cells and BDCs that were activated for various periods of time (n = 6). Bars represent mean ± SEM. Statistical differences compared to 20-h activation with TLR ligands after resting for 16 h were analyzed by paired one-way ANOVA with Bonferroni's multiple comparison test. Only significant differences are shown: **p < 0.01, ***p < 0.001 1 3 antigen-specific T cells compared to respective controls with unpulsed BDCs: 21.3 ± 6.5 vs. 13.3 ± 1.9; ± SEM). Similarly, expansion of T cells specific for SARS-CoV-2 from donors that had been previously infected by the virus was higher by R848 + poly(I:C)-activated BDCs compared to nonactivated BDCs (9.0 ± 2.1 vs. 8.2 ± 1.2). For WT1specific T cells, R848 + poly(I:C)-activated BDCs similarly caused an improved expansion of antigen-specific T cells compared to non-activated BDCs (2.9 ± 1.1 vs. 1.6 ± 0.5).

Discussion
BDCs have become a promising alternative to moDCs in cancer immunotherapies owing to their multifaceted properties. However, results from first clinical trials indicate the need for improved activation protocols. Here, we systematically established a protocol for the simultaneous activation of all BDC subsets in vitro using a combination of two TLR ligands. Although BDCs represent only about 1% of Fig. 4 Comparison between tailored activation protocols for cDCs and pDCs and simultaneous activation of BDCs. a, b Secretion of IL-12p70 and IFN-α by pDCs, cDCs, and BDCs upon activation with tailored protocols for TLR stimulation (n = 6). c, d T-cell proliferation and IFN-γ secretion induced in co-culture experiments with autologous pDCs, cDCs, BDCs, or a 1:1 mixture of pDCs and cDCs (n = 6). Bars represent mean ± SEM. Statistical test: paired one-way ANOVA with Bonferroni's multiple comparison test; *p < 0.05, **p < 0.01. e Effect of IFN-α addition on IL-12p70 secretion by cDCs activated with R848 + poly(I:C) (n = 10). Bars represent mean ± SEM. Statistical test: Wilcoxon matched-pairs signedrank test, *p < 0.05 all PBMCs, automated systems allow their rapid isolation from buffy coats or leukapheresis products to high purity in a closed system [23].
We observed low to intermediate expression of positive co-stimulatory immune checkpoint molecules on BDCs after isolation from peripheral blood. Stimulation with TLR ligands induced high expression of the co-stimulatory molecules CD80 and CD40 and the migratory receptor CCR7, whereby a combination of a TLR3 with a TLR7/8 ligand proofed to be optimal for all BDC subsets. We also observed upregulation of PD-L1 in all BDC subsets, providing the rationale to combine BDC vaccination with PD-1/ PD-L1 blockade. This has already been successfully evaluated in pre-clinical studies, however, results from clinical trials are scarce [24][25][26]. Nevertheless, the feasibility of a similar approach has been demonstrated in metastatic melanoma patients by blockade of CTLA-4 in combination with moDCs [27].
TLR ligands also caused strong secretion of IL-12 and IFN-α by BDCs. Whereas IFN-α secretion was at its highest level upon activation with CpG + poly(I:C), IL-12 secretion was maximal with R848 + poly(I:C). In contrast, single TLR stimulation-induced generally lower BDC responses. This is in line with previous results, describing TLR synergy between TLR3 or TLR4 and TLR8 in human moDCs and cDC2s resulting in potent IL-12p70 induction [28]. Addition of IFN-γ to the combination of R848 and poly(I:C) led to further enhanced secretion of IL-12p70, similar to previous results in moDCs [28].
We decided to focus on R848 + poly(I:C) to optimize Th1 differentiation, which is induced by IL-12. Similar to our results, R848 + poly(I:C) has been reported to upregulate co-stimulatory immune checkpoints in BDC subsets and to induce cytokine and chemokine secretion (i.e., type I IFNs, IL-12) in a humanized mouse model [29]. Addition of CpG to R848 + poly(I:C) provided no benefit with respect to immune checkpoint expression and cytokine secretion by BDCs. This reflects the previous finding that single-stranded DNA oligonucleotides such as CpG inhibit the activation of moDCs and nonhematopoietic cells by poly(I:C) [30].
By analyzing the kinetics of cytokine secretion, we found that IFN-α production by BDCs starts only a few hours after stimulation with R848 + poly(I:C), whereas the onset of IL-12 secretion appears to be later. This is consistent with results from a humanized mouse model [29]. Unexpectedly, the introduction of a resting step before TLR activation of BDCs strongly increased IFN-α secretion, whereas IL-12 was reduced. Spontaneous apoptosis of cDCs during ex vivo handling might explain this observation, whereas cultivation with IL-3 during the resting step might have promoted survival of pDCs. Conversely, preincubation with IL-3 for longer periods was reported to decrease the IFN-α secretion capacity of pDCs [31], however, as appropriate controls were missing in that study, spontaneous apoptosis might have been at play. Furthermore, we observed that reducing the time span of BDC activation and introducing the resting step resulted in reduced T-cell responses compared to BDC activation for 20 h immediately after their isolation.
Evaluation of tailored activation protocols for BDC subsets allowed us to elucidate the individual contributions of pDCs and cDCs to induce immune responses. cDCs induced stronger T-cell responses than pDCs, correlating with IL-12 secretion. The role assigned to pDCs in shaping the immune response in oncological malignancies remains controversial, with contradicting correlations between pDC infiltration and disease prognosis [32,33]. Recently, the results of a clinical trial of protamine-RNA-activated pDCs, cDCs, or a combination of pDCs and cDCs in prostate cancer were reported. In all three treatment arms, antigen-specific T-cell responses were induced and correlated with radiographic progressionfree survival, but no significant differences were observed [34]. Interestingly, when the authors used protamine-RNA for DC activation, cDCs did not secrete IL-12, but pDCs did, albeit at a low level. This might have been caused by a rare contaminating cDC population sharing common pDC markers [35], as the purity of the pDC vaccine was only modest. However, in our study, BDCs elicited higher immunostimulatory capacities than a combination of pDCs and cDCs activated with tailored protocols, underlining the rationale for activating all BDC subsets simultaneously.
DC vaccines aim to elicit and maintain T-cell responses against tumor-specific antigens. WT1 is a universal tumor target antigen due to its expression pattern in multiple different cancer entities, thus having great potential for developing immunotherapies [36]. However, inducing sufficient immune responses is a major challenge owing to the low immunogenicity of nonmutated tumor antigens and clonal deletion of self-reactive T cells [37]. R848 + poly(I:C)activated BDCs were able to expand WT1-specific T cells, providing a rationale for their further development as an immunotherapeutic tool. In addition to their Fig. 5 Migration, NK-cell activation, and T-cell expansion by BDCs. a Expansion of T cells specific for CEF, WT1, and SARS-CoV-2 by BDCs activated with TLR ligands (n = 3-4). b Representative flow cytometry analysis of antigen-specific T-cell expansion by BDCs activated with R848 + poly(I:C). BDCs were either not pulsed (left) or pulsed with peptides (right) prior to co-culturing with autologous T cells. c Activation of NK cells in co-cultures with autologous BDCs activated with TLR ligands (n = 9). d Representative example for NK-cell activation by BDCs. Color coding: NK cells only (light green), BDCs activated without TLR ligands (green) and with R848 + poly(I:C) (dark green). e Secretion of IFN-γ in co-cultures of NK cells and autologous BDCs (n = 7). Migration of BDC subsets towards CCL-19 (f) and CCL-21 (g) upon activation with TLR ligands in a transwell assay (n = 6). Bars represent mean ± SEM. Statistical tests: paired one-way ANOVA with Bonferroni's multiple comparison test (c, e), Wilcoxon matched-pairs signed-rank test (f, g), *p < 0.05, ***p < 0.001 ◂ 1 3 use as monotherapy, DC vaccines might enhance existing immune responses in the context of Chimeric antigen receptor (CAR) T-cell therapies. Accordingly, moDCs have been shown to boost WT1-specific CAR T cells in a humanized mouse model, resulting in enhanced inhibition of tumor growth [38].
DC vaccination trials to date have mainly focused on induction of T-cell responses, however, NK-cell activation by DCs constitutes an interesting upside for immunotherapies. Not only do NK cells recruit and activate further DCs, promoting T-cell responses, they can also eliminate tumor cells directly. Thus, tumor-cell material is released that can be processed by further DCs and be presented to T cells [39]. As activation of NK cells is mediated by IL-12 and type I interferons [40], BDCs activated with R848 + poly(I:C) demonstrated potent NK-cell activation with respect to CD69 expression and IFN-γ secretion. Importantly, NK-cell activation has been shown to correlate with clinical outcomes in several clinical trials, including those of a DC vaccine targeting WT1 in AML [39,41,42].
Migration of DCs to draining lymph nodes relies on the migratory receptor CCR7 and is central to induce adaptive immune responses. CCL-19 and CCL-21 promote the migration of DCs to the lymph nodes themselves and to the T-cell zone within the lymph nodes, enabling T-cell activation by DCs [43]. Only a few clinical trials have monitored the migration of their DC vaccine, revealing that only a small percentage of the injected DCs successfully migrated to the lymph nodes, underlining the need for enhanced DC migration capacities [44][45][46]. Activation with R848 + poly(I:C) upregulated expression of CCR7 on all BDCs. As a consequence, we observed improved migration of BDCs toward CCL-19 and CCL-21. Importantly, pDCs migrated only upon TLR activation.
Our investigations show R848 + poly(I:C) to be an optimized cocktail for ex vivo activation of all BDC subsets. Our findings support the further investigation and usage in early clinical trials -as standalone or in conjunction with other immunotherapeutic strategies including adoptive T-cell transfer and checkpoint inhibition.