Single Domain Antibody-Mediated Blockade of Programmed Death-Ligand 1 on Dendritic Cells Enhances CD8 T-cell Activation and Cytokine Production

Dendritic cell [DC] vaccines can induce durable clinical responses, at least in a fraction of previously treated, late stage cancer patients. Several preclinical studies suggest that shielding programmed death-ligand 1 [PD-L1] on the DC surface may be an attractive strategy to extend such clinical benefits to a larger patient population. In this study, we evaluated the use of single domain antibody [sdAb] K2, a high affinity, antagonistic, PD-L1 specific sdAb, for its ability to enhance DC mediated T-cell activation and benchmarked it against the use of the monoclonal antibodies [mAbs], MIH1, 29E.2A3 and avelumab. Similar to mAbs, sdAb K2 enhanced antigen-specific T-cell receptor signaling in PD-1 positive (PD-1pos) reporter cells activated by DCs. We further showed that the activation and function of antigen-specific CD8 positive (CD8pos) T cells, activated by DCs, was enhanced by inclusion of sdAb K2, but not mAbs. The failure of mAbs to enhance T-cell activation might be explained by their low efficacy to bind PD-L1 on DCs when compared to binding of PD-L1 on non-immune cells, whereas sdAb K2 shows high binding to PD-L1 on immune as well as non-immune cells. These data provide a rationale for the inclusion of sdAb K2 in DC-based immunotherapy strategies.


Introduction
Dendritic cell [DC] vaccination is extensively studied as a strategy to activate cancer-specific cytotoxic T lymphocytes [CTLs]. To induce potent anti-tumor CTLs three requirements need to be fulfilled. First, the peptide/MHC-I complex on the surface of DCs must be correctly recognized by the T-cell receptor [TCR] expressed on CD8 pos T cells. Second, co-stimulatory molecules, like CD80 and CD86, expressed on DCs, need to bind with their receptors, like CD28, expressed on CD8 pos

Stimulation of CD8 pos Melan-A-Specific T Cells by DCs
CD8 pos T cells were plated in triplicate in a 96 well round bottom plate at 10 5 cells per well in 100 µL IMDM containing 1% heat-inactivated human AB serum [Innovative Research, Novi, MI, USA], 100 U/mL penicillin, 100 µg/mL streptomycin, 2 mM L-Glutamine and non-essential amino acids [Sigma Aldrich]. moDCs were electroporated with Melan-A mRNA to generate so called DC-MEL [19]. DC-MEL were added to the T cells at an effector:stimulator ratio of 10:1 in 100 µL culture medium. Co-cultures with DC-MEL were performed for 7 days at 37 • C, 5% CO 2 in the presence of 10 µg/200 µL blocking anti-PD-L1 mAbs or sdAb K2, or isotype matched control mAbs or sdAb R3B23. T cells were restimulated on day 7. Analysis of the activation of Melan-A-specific T cells was performed 7 days after the last stimulation. The number of viable, Melan-A/HLA-A2 dextramer pos CD8 pos T cells was determined by flow cytometry. The production of interferon γ [IFN-γ] was evaluated with ELISA [Thermo Scientific, Brussels, Belgium] according to the manufacturer's instructions. The production of interleukin [IL]-2, IL-4 and tumor necrosis factor-α [TNF-α] was determined using BioPlex Pro Human Cytokines [Biorad, Temse, Belgium] according to the manufacturer's instructions.

Proliferation Assay
PBMCs depleted from CD14 pos cells from healthy donor were labeled with 0.5 µM Cell Trace Violet (Invitrogen). These cells (10 5 ) were co-cultured for 6 days with or without DC-MEL (4.8 µg Melan-A mRNA) at a effector:stimulator ratio of 10:1 in 200 µL IMDM containing 1% heat-inactivated human AB serum, PS, L-Glu and non-essential amino acids. T-cell proliferation was measured in flow cytometry as the dilution of the CellTrace Violet dye in the CD8 pos T-cell population. Proliferation observed in cultures without DC-MEL was considered as background.

Evaluation of DC Maturation in Response to Endotoxins Present in sdAb Preparations
To evaluate the effect of any endotoxins in the sdAb solutions, we incubated moDCs for 24 h with 10 µg sdAb K2 or sdAb R3B23 at 37 • C and 5% CO 2 . Untreated moDCs and moDCs treated with 1 ng/mL lipopolysaccharide [LPS] served as negative and positive controls, respectively. Up-regulation of maturation markers was evaluated in flow cytometry.

Flow Cytometry
The procedure for staining of surface markers was previously described [20]. All cells were acquired on the LSRFortessa flow cytometer [BD Biosciences, Erembodegem, Belgium] and data were analyzed with FACSDiva [BD Biosciences] or FlowJo [Tristar Inc.] software.

Statistical Analysis
Results are expressed as mean ± standard error of the mean [SEM]. A paired student t-test was carried out to compare data sets. Sample sizes and number of independent experiments are indicated in the figure legends. The number of asterisks in the figures indicates the statistical significance as follows: * p < 0.05; ** p < 0.01 and *** p < 0.001.

Inhibition of TCR Signaling in PD-1 pos 2D3 Cells Activated with PD-L1 pos DCs is Alleviated by sdAb K2
To evaluate whether sdAb K2 can be used as a therapeutic agent in combination with DC-vaccination, we first performed a functional assay using PD-1 pos and PD-1 neg 2D3 cells. The latter are derived from the established Jurkat T-cell line and are characterized by CD8 expression, lack of endogenous TCR expression and expression of eGFP under the control of the nuclear factor of activated T cells [NFAT] promoter. We previously showed that TCR modified PD-1 pos and PD-1 neg 2D3 cells can be used to validate the blocking capacity of PD-1 and PD-L1 blocking mAbs [16].
PD-1 pos and PD-1 neg 2D3 cells, electroporated with mRNA encoding the TCR recognizing gp100 280-288 in the context of HLA-A2 (Figure 1a,b), were co-cultured with HLA-A2 pos PD-L1 pos moDCs pulsed with gp100 280-288 peptide ( Figure 1c). The expression of eGFP was determined 24 h later as a measure of TCR signaling. We observed that the expression of eGFP by 2D3 cells was inhibited upon PD-1/PD-L1 interaction (Figure 1d). This inhibition could be alleviated through addition of the anti-PD-L1 mAb MIH1 [IgG1] or sdAb K2, but not through addition of an isotype matched control mAb or the control sdAb R3B23 (Figure 1d). Comparable results as with mAb MIH1 and sdAb K2 were obtained with the IgG1 mAb avelumab. To ensure that the activation of 2D3 cells in the context of moDC stimulation and sdAb K2 mediated PD-1/PD-L1 blockade was not due to maturation of the moDCs as a result of endotoxins present in the sdAb preparations, we compared the phenotype of moDCs that were untreated or matured with LPS to the phenotype of moDCs treated with sdAb K2 or sdAb R3B23. Up-regulation of maturation associated phenotypic markers like CD40, CD80 and the antigen presenting molecule HLA-I were only observed when moDCs were treated with LPS ( Figure 1e). These results indicate that the increase in TCR signaling in PD-1 pos 2D3 cells during antigen presentation by PD-L1 pos moDCs in the presence of sdAb K2 is most likely due to inhibition of the PD-1/PD-L1 interaction and not due to an increase in HLA-I expression, therefore antigen presentation.
In conclusion, sdAb K2 potently enhances TCR-signaling during antigen presentation by moDCs, as shown by the NFAT-mediated up-regulation of eGFP in PD-1 pos 2D3 cells. In conclusion, sdAb K2 potently enhances TCR-signaling during antigen presentation by moDCs, as shown by the NFAT-mediated up-regulation of eGFP in PD-1 pos 2D3 cells.

Activation of Melan-A-Specific CD8 pos T Cells by DC-MEL Is Significantly Enhanced by sdAb K2-Mediated PD-1/PD-L1 Blockade
To address whether PD-L1 blockade enhances the activation of antigen-specific CD8 pos T cells by DC-MEL, i.e., moDCs electroporated with mRNA encoding Melan-A, we performed T-cell stimulation experiments. Herein DC-MEL were co-cultured with CD8 pos T cells at a 1 to 10 ratio. Two rounds of stimulation were performed in the presence of the blocking anti-PD-L1 mAbs 29E.2A3 or avelumab, isotype control mAbs, sdAb K2 or sdAb R3B23 (control) to obtain sufficient Melan-Aspecific T cells for analysis. Of note, avelumab was chosen for comparison to sdAb K2 as we previously showed that avelumab and sdAb K2 compete for binding to PD-L1 [13].

Activation of Melan-A-Specific CD8 pos T Cells by DC-MEL Is Significantly Enhanced by sdAb K2-Mediated PD-1/PD-L1 Blockade
To address whether PD-L1 blockade enhances the activation of antigen-specific CD8 pos T cells by DC-MEL, i.e., moDCs electroporated with mRNA encoding Melan-A, we performed T-cell stimulation experiments. Herein DC-MEL were co-cultured with CD8 pos T cells at a 1 to 10 ratio. Two rounds of stimulation were performed in the presence of the blocking anti-PD-L1 mAbs 29E.2A3 or avelumab, isotype control mAbs, sdAb K2 or sdAb R3B23 (control) to obtain sufficient Melan-A-specific T cells for analysis. Of note, avelumab was chosen for comparison to sdAb K2 as we previously showed that avelumab and sdAb K2 compete for binding to PD-L1 [13].
First, we phenotyped the DC-MEL using flow cytometry, analyzing the expression of the co-inhibitory molecule PD-L1 as well as the co-stimulatory molecules CD70, CD80 and CD86, and the antigen presenting molecules HLA-A2 and HLA-DR. We observed expression of all evaluated surface markers on DC-MEL with the exception of the co-stimulatory molecule CD70 (Figures 2a  and 3a). Furthermore, we used flow cytometry to analyze the expression of PD-1 on the CD8 pos T cells before they were co-cultured with DC-MEL, showing low expression of PD-1 (Figures 2b and 3b). Detection of Melan-A-specific CD8 pos T-cell activation upon co-culture with DC-MEL in the presence of blocking anti-PD-L1 mAbs 29E.2A3 or avelumab, isotype control mAbs, sdAb K2 or sdAb R3B23 (control) was performed using flow cytometry. Together with the total number of cells this allowed us to determine the number of Melan-A-specific T cells that were expanded in these co-cultures. We showed that compared to respective control compounds, the presence of sdAb K2 but not the anti-PD-L1 mAb 29E.2A3 or avelumab resulted in a significantly higher number of Melan-A-specific T cells (Figure 3c). We further analyzed the capacity of the activated Melan-A-specific CD8 pos T cells to produce multiple cytokines and to proliferate using multiplex cytokine analysis and evaluation of Cell Trace Violet dilution in flow cytometry, respectively. We observed that Melan-A-specific CD8 pos T cells stimulated in the presence of sdAb K2 secreted more IL-2, IL-10, IFN-γ and TNF-α when compared to CD8 pos T cells stimulated in the presence of sdAb R3B23. Of these cytokines IL-2 and IL-10 were produced at pg-levels, while IFN-γ and TNF-α were produced at ng-levels ( Figure 3d). Moreover, these CD8 pos T cells showed a higher proliferative capacity as well ( Figure 3e). These data suggest that sdAb K2, however not mAb 29E.2A3 or avelumab, can block the interaction between PD-L1 and PD-1 during antigen presentation by DC-MEL to CD8 pos T cells, as such resulting in enhanced activation of functional antigen-specific CD8 pos T cells.
It was unexpected that the mAbs 29E.2A3 and avelumab were unable to enhance the activation of Melan-A-specific CD8 pos T cells by DC-MEL, as both have been described as blocking mAbs to enhance activation of antigen-specific CD8 pos T cells by PBMCs [21,22]. Moreover, avelumab has proven its efficiency in several clinical trials [23]. In search for an explanation for this lack of effect, we evaluated mAbs 29E.2A3 and avelumab as reagents to detect and bind human PD-L1 on moDCs. Staining of moDCs with mAb MIH1 and sdAb K2 were performed for comparison. We observed that mAb MIH1 was most efficient as a reagent to detect PD-L1 on moDCs followed by sdAb K2, avelumab and mAb 29E.2A3 (Figure 4a). In fact, in flow cytometry mAb 29E.2A3 was not proficient as a reagent to stain PD-L1. In contrast, efficient staining of PD-L1 on PD-L1 pos 293T cells was observed (Figure 4b), suggesting a different efficiency of mAb 29E.2A3 and avelumab to bind to PD-L1 on immune cells versus non-immune cells. Such differences in sensitivity of reagents to discriminate PD-L1 on immune cells versus tumor cells in immunohistochemistry was previously described [24].
In conclusion, these results suggest that the low efficacy of 29E.2A3 and avelumab to bind PD-L1 on moDCs when compared to sdAb K2 may impact on the efficacy of mAb 29E.2A3 and avelumab to enhance DC-mediated T-cell activation. Furthermore, the results generated with sdAb K2 in the context of TriMixDC-MEL and DC-MEL mediated CD8 pos T-cell activation suggest that in the absence of strong co-stimulatory signals, PD-L1 is a major determinant of T-cell activation. Finally, the data generated with DC-MEL and sdAb K2 provides a rationale for the inclusion of sdAb K2 in DC-based immunotherapy strategies.
Vaccines 2019, 7, 85 7 of 13 oreover, these CD8 pos T cells showed a higher proliferative capacity as well (Figure 3e). These da ggest that sdAb K2, however not mAb 29E.2A3 or avelumab, can block the interaction betwee -L1 and PD-1 during antigen presentation by DC-MEL to CD8 pos T cells, as such resulting i hanced activation of functional antigen-specific CD8 pos T cells.  It was unexpected that the mAbs 29E.2A3 and avelumab were unable to enhance the activation of Melan-A-specific CD8 pos T cells by DC-MEL, as both have been described as blocking mAbs to enhance activation of antigen-specific CD8 pos T cells by PBMCs [21,22]. Moreover, avelumab has proven its efficiency in several clinical trials [23]. In search for an explanation for this lack of effect, we evaluated mAbs 29E.2A3 and avelumab as reagents to detect and bind human PD-L1 on moDCs. Staining of moDCs with mAb MIH1 and sdAb K2 were performed for comparison. We observed that mAb MIH1 was most efficient as a reagent to detect PD-L1 on moDCs followed by sdAb K2, avelumab and mAb 29E.2A3 (Figure 4a). In fact, in flow cytometry mAb 29E.2A3 was not proficient as a reagent to stain PD-L1. In contrast, efficient staining of PD-L1 on PD-L1 pos 293T cells was observed (Figure 4b), suggesting a different efficiency of mAb 29E.2A3 and avelumab to bind to PD-L1 on immune cells versus non-immune cells. Such differences in sensitivity of reagents to discriminate PD-L1 on immune cells versus tumor cells in immunohistochemistry was previously described [24]. In conclusion, these results suggest that the low efficacy of 29E.2A3 and avelumab to bind PD-L1 on moDCs when compared to sdAb K2 may impact on the efficacy of mAb 29E.2A3 and avelumab to enhance DC-mediated T-cell activation. Furthermore, the results generated with sdAb K2 in the context of TriMixDC-MEL and DC-MEL mediated CD8 pos T-cell activation suggest that in the absence

Discussion
The inhibitory function of PD-1/PD-L1 interaction during antigen presentation by DCs to T cells is generally recognized, pinpointing this inhibitory pathway as an attractive therapeutic target to enhance the potency of DC-vaccines. Several strategies have been successfully employed in preclinical studies to interfere with PD-1/PD-L1 interactions during antigen presentation by DCs to CD8 pos T cells.
In particular the use of mAbs in combination with DC-vaccination has found its way to the clinic, as evidenced by a number of clinical trials in a range of malignancies [25]. However, the immune synapse clears and even excludes molecules above a certain size, including mAbs [26]. Therefore, the use of small sized, blocking PD-1/PD-L1 agents might be more advantageous. Recently, we described a human PD-L1 specific sdAb, sdAb K2, obtained after alpaca immunizations and biopannings, as a PD-1/PD-L1 neutralizing moiety with high target specificity and affinity [13]. Its small size [≈15 kDa, 10 times smaller than a mAb] was shown to penetrate much more efficiently within cell-cell interfaces like immune synapses [26], making it an interesting candidate for implementation in combination therapy with DCs.
We showed that sdAb K2 could shift the balance between stimulatory and inhibitory signals during the early stage of T-cell activation when using DCs with a low stimulatory profile. Although several mAbs targeting PD-L1 have shown clinical efficacy, we were unable to show any benefit of adding mAbs to the in vitro DC-MEL CD8 pos T-cell co-cultures [23]. This is in contrast to other studies reporting on the use of avelumab and/or mAb 29E.2A3 to enhance the activation of human T-cell populations of healthy donors [21,22]. Several reasons can explain this discrepancy. Grenga et al. [21] studied interaction between PBMCs and CD8 pos T cells, rather than moDCs and CD8 pos T cells, showing that in this setting, activation of virus-specific CD8 pos T cells was most pronounced in the presence of avelumab when compared to mAb 29E.2A3. The use of viral peptides as antigens is a major difference, as in this case most likely memory CD8 pos T cells are activated rather than naïve T cells. They claim that avelumab is able to shift Th2 cells to Th1 cells, demonstrated by an increase in IFN-γ concentration. The maximum amount of IFN-γ and the hereby increase in IFN-γ (±2 ng/mL) was comparable with the increases in IFN-γ after adding sdAb K2 to the DC-MEL co-culture (±3 ng/mL). On the other hand, Brown et al. [22] used moDCs as stimulator cells, however, performed allogeneic mixed lymphocyte reactions, evaluating CD4 pos T cell activation in the presence of the mAb 29E.2A3. In this setting the presence of allogeneic HLA-antigens may serve as a danger signal, also inducing overall T-cell activation, including memory T-cell activation [27]. Re-activation of antigen experienced effector memory T cells was suggested to be the driver of the efficacy of PD-L1/PD1 blockade in human cancer therapy, while activation of CD8 pos effector T cells was not reported [28]. Nonetheless, here as well, the level of IFN-γ concentration observed after treatment with mAb 29E.2A3 (5 ng/mL; 40 µg/mL mAb 29E.2A3) was at the same level as observed in the DC-MEL co-cultures after treatment with 50 µg/mL sdAb K2 (± 3 ng/mL). Moreover, the concentration of blocking mAbs used, differs from the above described studies. Both Grenga et al. and Brown et al. observed an effect when using 20 µg/mL of anti-PD-L1 mAbs, whereas we could not even observe an effect when using 50 µg/mL. Moreover, in the in vitro cell-based assay, DC-MEL were exclusively co-cultured with CD8 pos T cells, therefore not taking into account the role of PD-L1 blockade on other immune cells such as myeloid-derived suppressor cells, macrophages, etc. It was previously shown that avelumab exerts its effect via antibody-dependent cellular cytotoxicity [ is pinpointed as the main player in ADCC as it is expressed predominantly by natural killer cells, which were not represented in the in vitro cell culture assays [29,30]. The source of stimulator cells might also contribute to the difference in experimental outcome. We observed that mAb 29E.2A3 was unable to detect PD-L1 in flow cytometry on the moDCs we generated, and that detection of PD-L1 with avelumab was less evident than with mAb MIH1 and sdAb K2. Staining of PD-L1 expressed on 293T cells precludes that this observation is a technical artefact, as avelumab, mAbs 29E.2A3 and MIH1 as well as sdAb K2, were able to stain PD-L1 on these cells. The reason for this different sensitivity to PD-L1 expressed on moDCs versus 293T cells is at present unclear, however, might explain as to why de novo activation of antigen-specific CD8 pos T cells was not observed in the presence of these mAbs in our study. It is conceivable that our moDCs differ from the moDCs used by Brown et al. [22], as different culture conditions were used [e.g., culture medium]. For sure, the moDCs used in this study are different from the PBMCs used by Grenga et al. [21]. Further studies are required to assess binding of different mAbs to different PD-L1 pos immune and non-immune cell populations, including DCs. Such studies have already been performed with other antibodies in the context of immunohistochemical detection of PD-L1 in tumor tissue and lymph nodes, showing that different antibodies indeed have different propensities to bind PD-L1 on tumor cells versus immune cells, and sometimes even discriminate between lymphocyte-like cells versus DCs [24]. In our study, sdAb K2, similar to mAb MIH1, did not make the distinction between PD-L1 expressed on moDCs versus 293T cells.
We previously showed that both sdAb K2 and avelumab bind with high affinity to PD-L1 and that binding of K2 hampers subsequent binding of avelumab and vice versa [13]. Studies showed that avelumab binds mainly with its V H domain on the strands of the front β sheet face of the IgV domain of PD-L1, which is different from the epitopes bound by other PD-L1 targeting mAbs, such as durvalumab, atezolizumab and BMS-936559 [31,32]. As such sdAb K2 is a unique small-sized, biological inhibitor when compared to other small molecule inhibitors such as the anti-PD-L1 sdAb KN035 and the non-peptide anti-PD-L1 inhibitors BMS-202 and BMS-8, which show similar binding to PD-L1 as durvalumab [32,33]. We previously showed that sdAb K2 blocks PD-1/PD-L1 interactions at the protein level and tumor cell T-cell level [13]. We now show that sdAb K2 can also block PD-1/PD-L1 interactions at the immunological synapse created when DCs interact with CD8 pos T cells. The single domain nature of sdAbs offers interesting perspectives in view of DC-vaccine development. Many protocols are available to deliver tumor antigens and activation stimuli to DCs. Many of these are based on genetic engineering using viral and non-viral vectors [34,35]. While cloning of classical mAbs or mAb fragments offers serious challenges, cloning of sdAbs [i.e., coded by a single 360 bp DNA fragment] is straightforward, therefore can be easily incorporated into existing DC-engineering protocols. Several of these ex vivo DC-engineering strategies have also been used to specifically engineer DCs in situ, even in the tumor environment [36][37][38][39][40]. The targeted delivery of sdAb K2, and its release in the immunological synapse offers attractive safety considerations compared to systemic mAb or sdAb-administration. It will tip the balance from immune inhibitory to stimulatory signals only between antigen and sdAb K2-engineered DCs and cognate T cells, thereby ensuring increased on-target T-cell responses with little to no off-target T-cell activation.
In conclusion, we report on the use of sdAb K2, a versatile PD-L1/PD-1 blocking moiety, to enhance the capacity of DCs to stimulate T-cell activation and cytokine production. Inclusion of sdAb K2 in DC-vaccination protocols may have therapeutic potential in the clinical setting where several technologies to modify DCs for T-cell activation are investigated in the setting of cancer as well as infectious disease.

Patents
The use of moDCs electroporated with tumor antigen and TriMix mRNA is the topic of a patent application [WO2009/034172] on which K.T. is filed as an inventor. This patent is licensed to eTheRNA Immunotherapies NV. The use of sdAb K2 in onco-immunology is the topic of a patent application