Very low endotoxin RPMI 1640 medium (FG 1415; Biochrom) supplemented with 1.5% human serum (serum pool of AB positive adult males; Institute for Transfusion Medicine)—hereafter named DC medium—was used for the generation of DCs and all coculture experiments. The following reagents were used to generate DCs: GM-CSF (300-03), rhIL-4 (200-04), IFN-γ (300-02; all PeproTech), rhIL-1β (201-LB), TNF-α (210-TA/CF; both R&D Systems), PGE2 (P5640; Sigma-Aldrich), and R848 (tlrl-r848; InvivoGen).

After written informed consent, peripheral blood (PB) samples were collected from healthy donors (HDs) under a clinical protocol entitled “in vitro studies to establish new immunotherapies for AML and other hematological neoplasias.” Both the consent form and the protocol were approved by the institutional review board (Ethikkommission bei der LMU München). Both cell isolation and generation of DCs were performed as described previously for TLR-3-DCs (9) with the exception of polyI:C, which was not included in the maturation cocktail.

Dendritic cells were pulsed with a mixed CMV, EBV, influenza, and tetanus (CEFT) peptide pool (2 µg/ml; PM-CEFT; JPT) for 2 h at 37°C, 5% CO₂, incubated for 10 min on ice and subsequently washed. CD3⁺ T cells were isolated from autologous non-adherent cells (NACs) by magnetic activated cell sorting (MACS, 130-050-101; Miltenyi Biotec) according to the manufacturer’s protocol. CEFT-pulsed DCs and CD3⁺ T cells were cocultured at a ratio of 1:10 in 96-well round bottom plates for 4 days at 37°C, 5% CO₂. For blocking experiments, the following monoclonal blocking antibodies were added at 10 µg/ml: α-CD244 (PP35; 16-2449-81; eBioscience), α-HVEM (122; 318802), α-TIM-3 (F38-2E2; 345003), α-PD-1 (EH12.27H7; 329911; all BioLegend), α-LAG-3 (17B4; AG-20B-0012PF; AdipoGen or ab40466; Abcam). The blocking antibody concentration of 10 µg/ml that we used was based on prior experiments demonstrating antibody blockade of immune checkpoints (21). Reducing the antibody concentration to 5 µg/ml did not alter our results (data not shown).

Dendritic cells were pulsed with the Epstein–Barr nuclear Ag 3 A peptide FLRGRAYGL (FLR) (2 µg/ml; JPT) for 2 h at 37°C, 5% CO₂ and subsequently washed. FLR-pulsed DCs and autologous NACs were cocultured at a ratio of 1:80 in 96-well round bottom plates for 6 days at 37°C, 5% CO₂. For blocking experiments, α-PD-1 and α-LAG-3 were added as above.

Peripheral blood mononuclear cells (PBMCs) were loaded with FLR and cultured in 96-well round bottom plates (5 × 10⁵/well) in the presence or absence of α-PD-1 and α-LAG-3 for 6–8 days at 37°C, 5% CO₂.

Immunofluorescent staining of DC surface antigens was performed using a panel of fluorescence-conjugated monoclonal antibodies: CD80 (PE, L307.4; 560925), CD83 (APC, HB15e; 551073) CD86 (FITC, 2331 (FUN-1); 557343), CD273 (APC, MIH18; 557926), CD274 (FITC, MIH1; 558065; all BD Biosciences), Galectin-9 (PE, 9M1-3; 348906), CD48 (FITC, BJ40; 336706), HLA-DR (Pacific Blue, LN3; 327016; all BioLegend), HVEM (APC, 94801; FAB356A; R&D Systems). Corresponding isotype controls were used.

Immunofluorescent staining of T-cell surface antigens was performed using the following fluorescence-conjugated monoclonal antibodies: CD244 (PE, C1.7; 329507 or APC, C1.7; 329511), PD-1 (Brilliant Violet 421, EH12.7H7; 329919), CD3 (FITC, UCHT1; 300406), CD45RA (Brilliant Violet 421, HI100; 304129; all BioLegend), CD160 (APC, 688327; FAB6700A), TIM-3 (PE, 344823; FAB2365P; both R&D Systems), CD8 (PerCP-eFluor 710, SK1; 8046-0087; eBioscience), CD4 (APC-H7, RPA-T4; 560158; BD Biosciences), LAG-3 (ATTO 647N, 17B4; AG-20B-0012TS AdipoGen), CCR7 (CD197, APC, FR 11-11E8; 130-098-125; Miltenyi Biotec). Corresponding isotype controls were used. Intracellular FoxP3 staining was performed according to the manufacturer’s instructions (APC, 3G3; Miltenyi Biotec).

Cells were analyzed using a FACS LSR II (BD Biosciences). Post-acquisition analysis was performed using FlowJo software (version 9.7.6; Tree Star). The median fluorescence intensity (MFI) ratio was calculated by dividing the MFI of the measured population by the MFI of cells stained with the isotype-matched antibody. For the upregulation of checkpoint molecules, the percentage of positive cells (% positive) was obtained by setting the gate at or below 1% in the respective isotype control.

Secretion of IFN-γ and TNF-α was quantified by cytometric bead array (CBA) Flex Set (560111; BD Biosciences) according to the manufacturer’s instructions.

Isolated CD3⁺ T cells were labeled with carboxyfluorescein N-succinimidyl ester (CFSE, C34554; Life Technologies) and cultured in the presence of autologous DCs. Unstimulated T cells served as negative control. Harvested cells were then stained with antibodies for CD3 (APC, UCHT1; 300412; BioLegend), CD4 (APC-H7), and CD8 (PerCP-eFluor 710). The percentage of divided cells (% divided) was analyzed using FlowJo software.

Magnetic activated cell sorting-enriched CD3⁺ T cells were sorted according to CCR7 and CD45RA expression levels into naive T cells (T_naive), central memory T cells (T_CM), effector memory T cells (T_EM), and effector memory RA T cells (T_EMRA) using an Aria III (BD Biosciences).

Wilms Tumor 1 antigen VLD (VLD = WT1 peptide VLDFAPPGA)-specific T cells were generated as previously described (33). Briefly, DCs were matured as described above. Autologous CD8⁺ T cells were isolated from NACs using the CD8⁺ T Cell Isolation Kit (130-096-495; Miltenyi Biotec) and incubated overnight in VLE-RPMI medium supplemented with 5% human serum and 5 ng/ml of IL-7 (200-07; Peprotech). DCs were pulsed with 2.5 µL/ml of the HLA-A*02:01-restricted VLD peptide (VLDFAPPGA; JPT) for 90 min and irradiated with 30 Gy. CD8⁺ T cells and DCs were cocultivated in a 4:1 T cell:DC ratio and incubated with 30 ng/ml of IL-21 (200-21; Peprotech) in the presence or absence of 10 µg/ml LAG-3 or PD-1 blocking antibodies for 72 h. On day 3, cocultures were expanded 1:1 by adding medium supplemented with 10 ng/ml IL-15 and IL-7 (200-07, 200-15; both Peprotech) and 10 µg/ml blocking antibodies. On days 6–7, cells were analyzed by flow cytometry using VLD multimer (WB3469; Immudex) and fluorescence-conjugated monoclonal antibodies (see above).

Data were analyzed using Prism 6 (GraphPad Software). All results are presented in box-and-whisker plots, with boxes representing the lower quartile, the median and the upper quartile, while the whiskers show the minimal and the maximal value. The significance of differences for pairwise comparison was determined using the two-tailed Wilcoxon signed rank test. p < 0.05 was considered statistically significant (* in all figures), while p < 0.01 is termed highly significant (** in all figures).

TLR-3-DCs were generated from PB of HDs. The characteristic phenotype of these DCs, with high expression of CD83, CD86, and CD80 and downregulation of CD14 is shown in Figure 1A. Expression of various inhibitory checkpoint molecules on DCs was analyzed by flow cytometry on 3–10 of these samples. HLA-DR was added to the panel as ligand for lymphocyte activation gene 3 (LAG-3) on T cells. MFI ratio of the expression data is presented in Figure 1B, statistical significance was tested against a theoretical median of 1.5. The expression of PD-L1 (median 6.2; n = 7; p = 0.004) and HLA-DR (median 184.5; n = 7; p = 0.016) on TLR-3-DCs was found to be (highly) significant. By contrast, HVEM (median 2.0; n = 10), CD48 (median 2.5; n = 7), Gal-9 (median 0.8; n = 7), and PD-L2 (median 0.9; n = 3) were not significantly expressed (Figure 1B).

Expression of the respective co-inhibitory ligands was determined on T cells with and without stimulation by DCs. TLR-3-DCs were generated from PB of HDs and pulsed with CEFT peptide pool. CD3⁺ T cells were isolated from PB of the same HDs and cocultured with autologous DCs or with CEFT peptide pool alone for 96 h. Expression of various inhibitory checkpoint molecules was analyzed on T cells by flow cytometry for 7–14 HDs. The percentage of positive cells is presented for CD4⁺ (Figure 2A; Figure S1 in Supplementary Material) and CD8⁺ (Figure 2B; Figure S1 in Supplementary Material) T cells. Statistical significance was tested between stimulation with pulsed DCs and CEFT stimulation alone as a control. CD4⁺ T cells showed a (highly) significant upregulation of CD244 (median of 2.3 vs. 1.5%; n = 7; p = 0.047), TIM-3 (median of 24.3 vs. 4.2%; n = 7; p = 0.016) and PD-1 (median of 16.4 vs. 5.9%; n = 13; p = 0.003) after stimulation with TLR-DCs, while expression of CD160 (median of 3.3 vs. 5.9%; n = 7) and LAG-3 (median of 1.8 vs. 0.7%; n = 9) were not changed (Figure 2A). On CD8⁺ T cells, we found (highly) significant upregulation of CD244 (median of 30.2 vs. 13.9%; n = 8; p = 0.008), TIM-3 (median of 30.8 vs. 3.9%; n = 8; p = 0.008), PD-1 (median of 21.5 vs. 13.4%; n = 14; p < 0.001) and LAG-3 (median of 5.4 vs. 0.4%; n = 9; p = 0.027), but not of CD160 (median of 4.5% vs. 5.0%; n = 8) (Figure 2B).

In order to determine the functional relevance of co-inhibitory molecule interaction between TLR-3-DCs and T cells, we first tested the influence of checkpoint blockade on proliferation of T cells after DC stimulation. CD3⁺ T cells isolated from PB of HDs were labeled with CFSE and cocultured with autologous CEFT-pulsed TLR-3-DCs for 5 days in the presence or absence of respective blocking antibodies. The percentage of divided cells was determined by flow cytometry. The ratio between the percentages of divided cells with and without blocking antibody was calculated. Data for 4–13 samples is presented in Figure 3A for CD4⁺ T cells and in Figure 3B for CD8⁺ T cells, original data is shown in Table S1 in Supplementary Material. Statistical significance was calculated against a fold change of 1.0, equal to no effect of the blocking antibody on proliferation. For CD4⁺ T cells, no effect of checkpoint blockade on proliferation was found for HVEM (fold change 0.91; n = 6), CD244 (fold change 1.05; n = 4) and TIM-3 (fold change 1.02; n = 4). Blockade of PD-1 resulted in slightly enhanced proliferation (fold change 1.15; n = 13; p = 0.002), and blockade of LAG-3 lead to markedly enhanced proliferation (fold change 1.44; n = 9; p = 0.002), both statistically highly significant (Figure 3A). Similarly, for CD8⁺ T cells, blockade of PD-1 resulted in slightly enhanced proliferation (fold change 1.08; n = 13; p = 0.003), and blockade of LAG-3 lead to markedly enhanced proliferation (fold change 1.24; n = 9; p = 0.002), both statistically highly significant, while no effect of checkpoint blockade on proliferation was found for HVEM (fold change 0.88; n = 6), CD244 (fold change 0.96; n = 4) and TIM-3 (fold change 0.91; n = 4) (Figure 3B).

Next, we determined whether checkpoint blockade also influenced IFN-γ and TNF-α secretion by T cells after DC stimulation. CD3⁺ T cells isolated from PB of HDs were cocultured with autologous CEFT-pulsed TLR-3-DCs for 96 h in the presence or absence of respective blocking antibodies. The concentration of IFN-γ and TNF-α in the culture supernatant was determined by CBA. IFN-γ and TNF-α fold change was calculated by dividing the concentration of the coculture with blocking antibody by the concentration of the control coculture without antibody. Data for 5–14 samples is presented for IFN-γ in Figure 3C, statistical significance was calculated against a fold change of 1.0, original data are shown in Table S1 in Supplementary Material. No effect of checkpoint blockade on IFN-γ secretion was found for HVEM (fold change 0.63; n = 7), CD244 (fold change 0.86; n = 5), and TIM-3 (fold change 1.01; n = 5). Blockade of PD-1 resulted in enhanced IFN-γ secretion (fold change 1.50; n = 14; p = 0.002) and blockade of LAG-3 lead to markedly enhanced IFN-γ secretion (fold change 5.00; n = 9; p = 0.004), both statistically highly significant (Figure 3C). Similarly, no effect of checkpoint blockade on TNF-α secretion was found for HVEM (fold change 0.89; n = 7), CD244 (fold change 1.01; n = 5) and TIM-3 (fold change 0.92; n = 5), while blockade of PD-1 (fold change 1.69; n = 14; p = 0.002), and blockade of LAG-3 (fold change 5.29; n = 9; p = 0.008) resulted in enhanced TNF-α secretion, both statistically highly significant (Figure S2 in Supplementary Material).

We tested the hypothesis that blockade of PD-1 and LAG-3 has additive or synergistic effects on proliferation or IFN-γ secretion by T cells after stimulation with TLR-3-DCs. For proliferation assays, CD3⁺ T cells isolated from PB of 7 HDs were labeled with CFSE and cocultured with autologous TLR-3-DCs for 5 days in the presence or absence of blocking antibodies for PD-1 and LAG-3, both alone and in combination. As above, the percentage of divided cells was determined by flow cytometry for the different conditions, and the ratio between the percentages of divided cells with and without blocking antibody was calculated. Data are presented in Figure 3D for CD4⁺ T cells and in Figure 3E for CD8⁺ T cells, statistical significance was calculated for the combination of blocking antibodies vs. single antibody blockade, original data are shown in Table S1 in Supplementary Material. For the combination of PD-1 and LAG-3 blockade (median fold change of 1.37 for CD4⁺ and 1.26 for CD8⁺), we found significantly higher T cell proliferation compared to PD-1 blockade alone (median fold change of 1.02 for CD4⁺; p = 0.016; 1.02 for CD8⁺; p = 0.016), but no difference to LAG-3 blockade alone (median fold change of 1.31 for CD4⁺; p = 0.094; 1.20 for CD8⁺; p = 0.250).

Similarly, for IFN-γ secretion assays, CD3⁺ T cells isolated from PB of 8 HDs were cocultured with autologous CEFT-pulsed TLR-3-DCs for 96 h in the presence or absence of blocking antibodies for PD-1 and LAG-3, both alone and in combination. The concentration of IFN-γ in the culture supernatant was determined by CBA. IFN-γ fold change was calculated as a ratio between the IFN-γ concentration of the coculture with and without blocking antibody. Statistical significance was calculated for the combination of blocking antibodies vs. single antibody blockade. For the combination of PD-1 and LAG-3 blockade (median fold change of 2.80), the increase in IFN-γ secretion compared to PD-1 blockade alone was statistically highly significant (median fold change of 1.41; p = 0.008). In comparison to LAG-3 blockade alone, we found a slight, but statistically significant enhancement (median fold change of 2.70; p = 0.016) (Figure 3F). Taken together, LAG-3 blockade alone resulted in strong enhancement of T cell proliferation and IFN-γ secretion. The effect on IFN-γ secretion was slightly increased by the combination with PD-1 blockade, while no additional effect was seen for T cell proliferation.

Next, we analyzed the differential effect of PD-1 and LAG-3 blockade on T cell subpopulations. MACS-enriched CD3⁺ T cells were sorted according to CCR7 and CD45RA expression levels into T_naive, T_CM, T_EM, and T_EMRA T cells (Figure 4A). The various T cell populations were cocultured with autologous CEFT-pulsed TLR-3-DCs for 96 h in the absence of presence of blocking antibodies for PD-1 and LAG-3. Again, the concentration of IFN-γ in the culture supernatant was determined by CBA. IFN-γ fold change was calculated by dividing the concentration of the coculture with blocking antibody by the concentration of the control coculture without antibody. Data for six samples is presented for PD-1 (Figure 4B) and for LAG-3 (Figure 4C) blockade. Statistical significance was calculated against a fold change of 1.0. We found that PD-1 blockade lead to significantly increased IFN-γ secretion of T_naive (median fold change of 1.41; p = 0.031), T_CM (median fold change of 1.43; p = 0.031), and T_EM (median fold change of 1.47; p = 0.031), while the increased secretion of T_EMRA was not statistically significant (median fold change of 1.96; p = 0.156) (Figure 4B). By contrast, LAG-3 blockade had significant effects on IFN-γ secretion of T_naive (median fold change of 2.04; p = 0.031) and T_CM (median fold change of 1.71; p = 0.031), but not on T_EM (median fold change of 1.34; p = 0.094) and T_EMRA (median fold change of 1.33; p = 0.094) (Figure 4C). With respect to the CD25⁺/FoxP3⁺ regulatory T cell subpopulation of CD4⁺ T cells, we saw a tendency toward a higher percentage after LAG-3 blockade (Figure S3 in Supplementary Material).

Next, we tested whether blockade of PD-1 and LAG-3 also enhances the proliferation of antigen-specific T cells after stimulation with TLR-3-DCs. NACs (mainly consisting of T cells) of 9 HDs were cocultured with autologous FLR-pulsed TLR-3-DCs for 144 h in the presence or absence of blocking antibodies for PD-1 and LAG-3, both alone and in combination. The percentage of FLR tetramer positive (Tet⁺) cells within the CD8⁺ T cell population was determined by flow cytometry. Tet⁺ fold change was calculated by dividing the percentage in the condition with blocking antibody by the percentage in the condition without any antibody. Statistical significance was calculated against a fold change of 1.0. Blockade of LAG-3 resulted in a significantly increased percentage of Tet⁺ CD8⁺ T cells (median fold change 1.69; p = 0.039), while blockade of PD-1 (median fold change 0.79) and the combination of LAG-3 and PD-1 blockade (median fold change 0.61) did not enhance the percentage of antigen-specific T cells (Figure 5A). This was not due to a lack of PD-1 expression on T cells, as further analysis of the Tet⁺ CD8⁺ T cells after stimulation with FLR-pulsed DCs revealed that PD-1 was expressed on 92.6% of the T cells, while LAG-3 was found on only 49.1% of T cells (Figure 5B).

We then asked if the effect of LAG-3 blockade on proliferation and IFN-γ secretion also holds true, if T cells are not stimulated by TLR-3-DCs, but by the various APCs naturally occurring within PBMCs. PBMCs of 8 HDs were pulsed with FLR peptide and cultured for 6 days in the presence or absence of blocking antibodies for PD-1 and LAG-3, both alone and in combination. Thereafter, the percentage of FLR tetramer positive cells (Tet⁺) within the CD8⁺ T cell population was determined by flow cytometry. Tet⁺ fold change was calculated by dividing the percentage in the condition with blocking antibody by the percentage in the condition without any antibody. Statistical significance was calculated against a fold change of 1.0. Blockade of LAG-3 resulted in a significantly increased percentage of Tet⁺ CD8⁺ T cells (median fold change 1.80; p = 0.023), while blockade of PD-1 (median fold change 1.05) did not enhance the percentage of antigen-specific T cells. The combination of LAG-3 and PD-1 blockade also significantly enhanced the percentage of Tet⁺ CD8⁺ T cells (median fold change 1.74; p = 0.016), but this was not different from LAG-3 alone (p = 0.461) (Figure 6A).

The concentration of IFN-γ was determined in the culture supernatant after 6–8 days of coculture by CBA. IFN-γ fold change was calculated by dividing the concentration of the coculture with blocking antibody by the concentration of the control coculture without antibody. Data for the same eight samples is presented in Figure 6B, statistical significance was calculated against a fold change of 1.0. Blockade of PD-1 (median fold change 0.96) did not enhance IFN-γ secretion, while increase of IFN-γ secretion after blockade of LAG-3 was highly significant (median fold change 4.07; p = 0.008). The combination of LAG-3 and PD-1 blockade also enhanced IFN-γ secretion highly significantly (median fold change 6.88; p = 0.008), but the difference to LAG-3 blockade alone was not significant (p = 0.188) (Figure 6B).

Further analysis of the Tet⁺ CD8⁺ T cells after stimulation with FLR-pulsed PBMCs revealed that PD-1 was expressed on almost all of the T cells (93.3%), while LAG-3 was found on only 10.5% of T cells (Figure 6C). Therefore, the non-existent effect of PD-1 blockade in this setting was not due to an absence of PD-1 on the T cell surface.

Finally, we tested the hypothesis that the effect of LAG-3 blockade on proliferation of antigen-specific T cells can also be transferred to tumor antigen specificity. CD8⁺ T cells of 3 HDs were cocultured with autologous TLR-3-DCs pulsed with a WT1 antigen (VLD peptide) for 6–7 days in the presence or absence of blocking antibodies for PD-1 and LAG-3. The percentage of VLD tetramer positive (Tet⁺) cells within the CD8⁺ T cell population was determined by flow cytometry. Results for all three donors are presented in Figure 7. Blockade of LAG-3 resulted in an increased percentage of Tet⁺ CD8⁺ T cells in two of three cases, while blockade of PD-1 resulted in an increase in Tet⁺ CD8⁺ T cells in only one case, and to a lesser extent.