Monocytes, pan T cells, naïve CD4 T cells, and NK cells were obtained from peripheral blood mononuclear cells (PBMCs), isolated by Ficoll density gradient centrifugation (Lymphoprep, Serumwerk Bernburg) from buffy coats derived from healthy blood donors (Sanquin, the Netherlands). Monocytes were isolated with MACS CD14+ isolation kit (130-050-201, Miltenyi Biotec). After the positive CD14 isolation, the negative fraction containing the peripheral blood lymphocytes (PBLs) was used to isolate autologous pan T cells and NK cells. Pan T cells and NK cells were isolated using the MACS Pan T cell isolation kit (130-096-535, Miltenyi Biotec) and the NK cell isolation kit (130-092-657, Miltenyi Biotec) following manufacturer’s instructions. Naïve CD4 T cells were isolated from PBLs using the MACS Naïve CD4 T cell isolation kit (130-094-131, Miltenyi Biotec) following the manufacturer’s instructions. Pan T cells, NK cells, and naïve CD4 T cells were cryopreserved until use in freezing media, composed of 50% X-VIVO medium (Lonza), 40% FBS (HyClone) and 10% DMSO (WAK-Chemie).

moMDSCs and moDCs were differentiated in vitro by culturing monocytes at a concentration of 0.5 million cells per ml in T25 flasks containing a final volume of 4 ml X-VIVO medium (Lonza) supplemented with 2% human serum (HS) (Sanquin, the Netherlands) for 7 days. For the generation of moMDSCs, monocyte cultures were supplemented with 60 U/ml of GM-CSF (130-093-867, Miltenyi Biotec) and 10 ng/ml of IL-6 (130-093-933, Miltenyi Biotec). For moDC generation, monocyte cultures were supplemented with 450 U/ml of GM-CSF (130-093-867, Miltenyi Biotec) and 300 U/ml of IL-4 (130-093-924, Miltenyi Biotec). On day 3 of differentiation, moMDSCs and moDC cultures were supplemented with 1 ml of X-VIVO supplemented with 2% HS and 5x concentrated IL-6 and GM-CSF in the case of moMDSCs, and 5x concentrated IL-4 and GM-CSF in the case of moDCs. To generate mature moDCs, on day 6 of differentiation, moDC cultures were further stimulated with 2 µg/ml of poly (I:C) (Invivogen) for 24 hours prior to further use in functional experiments.

Human melanoma cells (A375) were tested to be mycoplasma-free and cultured in Dulbecco’s modified Eagle’s medium (DMEM, Gibco) supplemented with 10% fetal bovine serum (FBS) 1% Antibiotic-Antimycin (AA, Gibco). To generate the melanoma cell line conditioned media (CM), A375 cells were cultured in T75 flasks at a concentration of 0.3 million cells per ml in T75 flasks with a final volume of 14 ml. After three days of culture, the supernatant was harvested and centrifuged for 5 minutes at 1500 rpms to get rid of cellular debris present in the media. Resulting CM was aliquoted and stored at -20 C until use.

To determine the individual role of EP2 and EP4 on moMDSCs, the specific EP2 antagonist AH6809 (Cayman chemicals) and specific EP4 antagonist L161-982 (Cayman chemicals) were used. For EP2 blockade, 40 µM of EP2 antagonist was used, whereas 3 µM was used for the blocking of EP4 receptor. The same concentrations were used when performing the combined EP2/4 blockade. On day 5 of moMDSC differentiation, cultures were treated with the indicated antagonist two hours prior to exposure to the melanoma-derived CM. After 3 hours, 2 ml of CM were added to each T25 flask.

To establish the role of CM addition, moMDSCs were instead provided with 2 ml of fresh media, depicted as untreated moMDSCs. On day 7, after 48 hours of CM treatment, moMDSCs were harvested. For harvesting, moMDSCs culture media was removed and replaced by cold PBS supplemented with 0.1% BSA and 0.2 mM of EDTA followed by incubation for 1 hour at 4°C. Thereafter, flasks were tapped and moMDSCs cell suspension was harvested and centrifuged prior to further analysis.

The impact of CM on monocyte development was tested by culturing monocytes at a concentration of 0.5 million cells per ml in 6 well plates with 2 ml of X-VIVO media supplemented with 2% HS. To determine the role of EP2/4 during differentiation, monocyte cultures were supplemented with 40 µM and 3 µM of EP2 antagonist and EP4 antagonist respectively. After 2 hours, monocyte cultures were treated with 0.5 ml of CM. On day 3, the differentiated monocytes were harvested for flow cytometric analysis. For day 6 of differentiation, day 3 monocytes were re-stimulated with the indicated concentrations of aEP2/4 and exposed to a second dose of 0.5 ml CM. On day 6, the resulting differentiated monocytes were harvested and either analyzed by flow cytometry or replated for the performance of functional experiments.

Flow cytometry was used to determine the phenotype of moMDSCs and the differentiated monocytes. Cells were harvested from culture flask or plates, washed with cold PBS, and incubated with blocking buffer (PBS + 0.1% BSA + 0.01 NaH3 + 5% HS). After 15 minutes, cells were stained with the specific antibodies for CD14 APC-H7 (1:100, 560180, BD Biosciences), CCR5 BV421 (1:50, 359118, Biolegend), CD86 APC (1:100, 555660, BD Biosciences), CD11b Pe (1:100, 301306, Biolegend), HLA-DR BV510 (1:100, 307646), and MerTK Pe-Cy7 (1:100, 367610, Biolegend). After staining for 20 minutes, cells were washed twice using blocking buffer prior to flow cytometric analysis (see Supplementary Figures 1B and 2 for the gating strategy). Flow cytometric analysis was performed on the BD FACS Verse, and data were analyzed in FlowJo software version (BD, Franklin Lakes, NJ, USA).

For the detection of IL-6, IL-10, and PGE2 production, harvested moMDSCs were replated in 96 U bottom plates at a concentration of 0.5 million cells per ml at a final volume of 200 µL of X-VIVO medium supplemented with 2% HS and 100 ng/ml LPS. After 24 hours of culture, the moMDSCs supernatants were harvested and analyzed by ELISA for the presence of IL-6 (88-7106-77, Invitrogen), IL-10 (88-7066-77 Invitrogen), and PGE2 (EHPGE2 ThermoFisher) according to manufacturer’s instructions.

The suppression of autologous T cell responses by moMDSCs was assessed by coculturing moMDSCs in 96 U bottom plates with autologous T cells. Prior to coculture, T cells were cultured for 30 minutes in X-VIVO supplemented with 2% HS at a concentration of 1 million cells per ml with human activator anti-CD3/CD28 dynabeads (1:100, 11161D, ThermoFisher). For T cell proliferation assays, autologous pan T cells were stained with 5 µM carboxyfluorescein diacetate succinimidyl ester (CFSE, C34554, ThermoFisher). moMDSCs were cultured with CFSE-labelled T- cells in a ratio 1:1 for three days at 37°C, 5% CO2. After three days, T cells were harvested and stained in cold PBS with the viability dye eFluor 780 (1:2000, ThermoFisher) for 15 minutes followed by CD4 PE staining (1:100, 555347 BD Biosciences) and CD8 APC (1:100, 555369, BD Biosciences) for 20 minutes. T cell proliferation was determined as loss of CFSE dye in T cells (see Supplementary Figure 3A)

To determine cytokine production by T cells, moMDSCs were cocultured with autologous pan T cells non-CFSE labelled at a ratio 1:1. After three days, T cells were stimulated with 25 ng/ml of PMA (Calbiochem), 0.5 mg/ml of Ionomycin (I0634, Sigma-Aldrich) and 10 ng/ml of brefeldin A (Cayman chemicals). After 3 hours of stimulation, T cells were washed and stained with the viability dye eFluor 780 (1:2000, ThermoFisher) for 15 minutes in cold PBS. Next, cells were fixated and permeabilized by using the FOXP3 kit (00-5523-00, ThermoFisher) according to manufacturer’s instructions. Cells were then blocked in wash and permeabilization buffer (provided by the kit) supplemented with 5% HS. After 15 minutes, T cells were washed and stained for 20 minutes with antibodies for IFNγ BV421 (1:100, 562988, BD Biosciences), IL-2 PerCP-Cy5.5 (1:100, 500322, Biolegend), TNFα APC (1:50, 130-117-382, Miltenyi Biotec), and CD8 BV510 (1:100, 344732, Biolegend). Next, cells were washed twice using blocking buffer prior to flow cytometric analysis (see Supplementary Figure 3B for gating strategy). Flow cytometric analysis was performed on the BD FACS Verse, and data were analyzed in FlowJo software version (BD, Franklin Lakes, NJ, USA).

The suppression of autologous NK cell responses by moMDSCs was assessed by coculturing moMDSCs in 96 U bottom culture plates with autologous NK cells. Prior to coculture, NK cells were cultured for 1 hour in X-VIVO supplemented with 2% HS and 100 U/ml of IL-2 (130-097-773, Miltenyi Biotec) at a concentration of 1 million cells per ml. Then moMDSCs were cocultured with NK cells at a ratio 1:2 (M-MDSC: NK cell ratio) for three days at 37°C, 5% CO2. For the characterization of activation markers, NK cells were harvested and stained in cold PBS with the viability dye eFluor 780 (1:2000, ThermoFisher) for 15 minutes. Next, NK cells were incubated with blocking buffer (PBS + 0.1% BSA + 0.01 NaH3 + 5% HS). After 15 minutes, cells were stained with the specific antibodies for NKG2D BV510 (1:50, 320816, Biolegend), NKp46 Pe (1:50, IM3711, Beckman), CD25 APC (1:100, 555434, BD Biosciences), CD69 PerCP (1:50, 340548, Biolegend), and CD56 BV421 (1:100, 562751, BD Biosciences). After staining for 20 minutes, cells were washed twice using blocking buffer prior to flow cytometric analysis (see Supplementary Figure 4 for gating strategy). Flow cytometric analysis was performed on the BD FACS Lyric, and data were analyzed in FlowJo software version (BD, Franklin Lakes, NJ, USA).

For the characterization of intracellular cytokine levels, NK cells were stimulated with 25 ng/ml of PMA (Calbiochem), 0.5 mg/ml of Ionomycin (Sigma-Aldrich) and 10 ng/ml of brefeldin A (Cayman chemicals) for 3 hours. Next, NK cells were washed and stained with the viability dye eFluor 780 (1:2000, ThermoFisher) for 15 minutes in cold PBS. Then, cells were fixated and permeabilized by using the FOXP3 kit (00-5523-00, ThermoFisher) according to manufacturer’s instructions. Prior to staining, cells were incubated in wash and permeabilization buffer (provided by the kit) supplemented with 5% HS. After 15 minutes, T cells were washed and stained for 20 minutes with antibodies specific for IFNγ BV421 (1:100, 562988, BD Biosciences), granzyme b Pe (1:100, 372208, Biolegend), TNFα PerCP-Cy5.5 (1:100, 45-7349-42, eBioscience), and CD56 APC (1:50, 341027, BD Biosciences). Next, cells were washed twice using blocking buffer prior to flow cytometric analysis (see Supplementary Figure 4 for the gating strategy). Flow cytometric analysis was performed on the BD FACS Verse, and data were analyzed in FlowJo software version (BD, Franklin Lakes, NJ, USA).

The ability of moMDSCs to modulate the expansion of pro-inflammatory T cell populations by activated DCs was determined by coculturing moDCs, allogeneic naïve CD4 T cells, and moMDSCs at a ratio 1:10:5 in 96 U bottom plates. On day 6 of coculture, half of the culture medium containing the expanding T cells was removed. Equal volume of fresh X-VIVO medium supplemented with 2% HS was added and T cell cultures were further supplemented with 20 U/ml of IL-2 (130-097-773, Miltenyi Biotec). From this day onwards, cell culture media was refreshed as described above with new media containing 20 U/ml of IL-2 every 2 days until day 12 of coculture. Resulting T cell populations were harvested and characterized either for the presence of Tregs or the intracellular cytokine expression. For determining the presence of Tregs, T cells were stained with the viability dye eFluor 780 (1:2000, ThermoFisher) in cold PBS for 15 minutes. T cells were then fixed using the FOXP3 staining kit (00-5523-00, ThermoFisher) according to manufacturer’s instructions. After fixation, cells were incubated in wash/permeabilization buffer supplemented with 5% HS for 15 minutes. Then, cells were washed and stained for CD127 PE (1:100 12-12780-42, eBioScience), CD25 APC (1:100 555434, BD Biosciences), and FOXP3 A488 (1:50 53-4776-42, eBioscience) for 20 minutes. Stained cells were washed twice prior to flow cytometric analysis (see Supplementary Figure 5B for gating strategy).

For the characterization of intracellular cytokines, T cells were stimulated with 25 ng/ml of PMA (Calbiochem), 0.5 mg/ml of Ionomycin (Sigma-Aldrich) and 10 ng/ml of brefeldin A (Cayman chemicals) for 3 hours. Next, T cells were washed and stained with the viability dye eFluor 780 (1:2000, ThermoFisher) for 15 minutes in cold PBS. Then, cells were fixated and permeabilized by using the FOXP3 kit (00-5523-00, ThermoFisher) according to manufacturer’s instructions. T cells were blocked in wash and permeabilization buffer (provided by the kit) supplemented with 5% HS. After 15 minutes, T cells were washed and stained for 20 minutes with antibodies specific for IFNγ BV421 (1:100, 562988, BD Biosciences), TNFα Pe (1:100, 502909, Biolegend), or IL-2 PerCP-Cy5.5 (1:100, 500322, eBioscience). Next, cells were washed twice and analyzed by flow cytometry (see Supplementary Figure 5A for gating strategy). Flow cytometry was performed on the BD FACS Verse, and data were analyzed in FlowJo software version (BD, Franklin Lakes, NJ, USA).

CRC-PDTOs (originally named PDO013 in the biobank (56) were established and cultured as described before (57). The generation of cocultures between PDTOs and moMDSCs in a 3D collagen gel was performed as previously described for DC studies (57). In short, PDTOs and moMDSCs on day 5 of differentiation were embedded in a collagen matrix at a cell ratio of 1:1. Collagen drops (of 25 µL containing 50 000 PDTOs and moMDSCs) were loaded into 24 well plates and incubated at 37°C to enable collagen polymerization and solidification for 45 minutes. Collagen drops were cultured with 300 µL X-VIVO media supplemented with 2% HS. Next day, cultures were either treated with or without the EP2/4 antagonists. For this experiment, 120 µM and 9 µM of EP2 and EP4 antagonists respectively were used. After 24 hours, collagen was degraded using collagenase I (Sigma-Aldrich, C0130). Dissociated cell suspension was then filtered, and the resulting cell suspension was washed and stained with antibodies specific for CD11b Pe (1:100, 301306, Biolegend), HLA-DR BV510 (1:100, 307646), MerTK Pe-Cy7 (1:100, 367610, Biolegend), CD14 APC-H7 (1:100, 560180, BD Biosciences), CCR5 BV421 (1:50, 359118, Biolegend), and CD86 APC (1:100, 555660, BD Biosciences). Next, cells were washed and analyzed by flow cytometry using the BD FACS Verse (see Supplementary Figure 6 for gating strategy), and data were analyzed in FlowJo software version (BD, Franklin Lakes, NJ, USA).

Statistical analysis was performed with Graphpad Prism 8 (Version 8.0.2. GraphPad Software San Diego, CA, USA). Unless indicated otherwise, data are depicted as mean ± SEM in bar scattered dot plots. Significance was determined as indicated in the legend of each figure. MoMDSCs phenotype data were analyzed using either a one-way analysis of variance (ANOVA) followed by multiple comparison Tukey correction, or a Friedman test followed by Dunn’s multiple comparison test. Significance across the functional experiments was determined using either a one-way ANOVA followed by a Dunnett’s test, or a Friedman test followed by a Dunn’s multiple comparison test. Differences among the CM differentiated monocytes or moMDSCs cocultured with CRC-PDTOs were performed using either a paired t-test or a Wilcoxon test. The statistical significance is depicted as follows: *p < 0.05, **p < 0.01, ***p < 0.001.

Monocytes exposed to tumor signals acquire a suppressive phenotype and function that recapitulate that of MDSCs (24, 25, 55). As model to determine the role of tumor-derived PGE2 signaling during monocyte development into MDSC-like cells, monocytes were cultured for 6 days with conditioned media (CM) derived from the melanoma cell line A375 (Figure 1A), detected to produce PGE2 (see Supplementary Figure 1A). To establish the effect of PGE2-EP2/4 signaling during MDSC-like differentiation, monocytes were incubated with specific antagonists for EP2 and EP4 (aEP2/4). Phenotype analysis after 3 and 6 days showed that EP2/4 blockade limited the expression of CD14, MerTK, CCR5, and CD11b (Figures 1B, C). No significant changes were detected for the markers HLA-DR and CD86. To assess the functional relevance of targeting PGE2-EP2/4 axis, monocytes cultured for 6 days in the presence of CM were harvested and cocultured with autologous NK and T cells. aEP2/4 attenuated the suppressive function of CM-educated monocytes as seen by the increased T cell proliferation and increased expression of the pro-inflammatory cytokines TNFa and IFNγ in NK cells (Figure 1D). Overall, these data illustrate the role PGE2-EP2/4 axis in the priming of a suppressive MDSC-like state of monocytes.

PGE2 is well known as a modulator of the phenotype of myeloid cells, including different cell models of mouse and human MDSCs (23, 24, 53, 54). As a human M-MDSC model, we used moMDSCs, generated from monocytes by culturing them for 5 days in the presence of IL-6 and GM-CSF, previously reported to express typical MDSC markers such as CD11b or CD33 (21, 31). To distinguish the effects of tumor-derived PGE2 signaling trough the different PGE2 receptors, A375-derived CM was added to moMDSCs that were preincubated for 2 hours with aEP2, aEP4, or the combination designated as aEP2/4 (Figure 2A). After 48 hours, CM treatment led to the upregulation of CD14, MerTK, and CCR5, and the downregulation of HLA-DR and CD86 (Figures 2B, C). Although not significant, we noted a tendency towards the upregulation of CD11b. EP blockade demonstrated both EP2 and EP4 to contribute to the induction of this phenotype. Of note, aEP4 more robustly than aEP2 prevented the acquisition of this phenotype. Combined aEP2/4 blockade prevented the acquisition of the phenotype yielded by the tumor-derived CM on moMDSCs.

Suppressive M-MDSCs exploit a variety of mechanisms to establish a suppressive environment, such as the secretion of suppressive mediators (22). To determine the effect of PGE2-EP2/4 axis on the secretome of moMDSCs, we measured the levels of IL-6, IL-10, and PGE2. MoMDSCs exposed to CM displayed a robust upregulation of these three soluble mediators. aEP2 modestly inhibited the production of these mediators, whereas aEP4 significantly impaired their production (Figure 3). Similar to aEP4 treatment, the combined aEP2/4 inhibited the secretion of these three mediators, recapitulating the production levels of moMDSCs unexposed to CM. Overall, these results identify the PGE2-EP2/4 axis as a major modulator of IL-10, IL-6, and PGE2 production by moMDSCs, with EP4 more robustly than EP2 mediating this feature.

M-MDSCs are potent suppressors of T cell immune responses as they inhibit T cell proliferation and the production of pro-inflammatory cytokines such as IFNγ (20, 21, 24–26, 28). To assess the ability of moMDSCs to modulate T cell proliferation, moMDSCs were cocultured with autologous T cells labelled with CFSE dye for three days (Figure 4A). Additionally, moMDSCs were cocultured with T cells not labelled with CFSE for the intracellular detection of cytokines. Prior to coculture, T cells were primed to proliferate by using activating anti-CD3/CD28 dynabeads. Exposure to CM led to moMDSCs with a superior ability to suppress the proliferation of autologous T cells (Figure 4B). Cytokine analysis demonstrated that CM treated moMDSCs yielded T cells with a reduced production of IFNγ, TNFα, and IL-2, cytokines associated with active pro-inflammatory T cell responses (Figure 4C).

Given the strong phenotypical modulation of moMDSCs achieved by blocking EP2 and EP4, we further assessed whether this inhibition could recover T cell function upon moMDSC coculture. aEP2 tended to partly recover T cell proliferation, whereas aEP4 significantly restored T cell proliferation to the levels of moMDSCs unexposed to the CM. As expected, combined aEP2/4 restored T cell proliferation to a level similar to that induced by aEP4 (Figure 4B). Although not significantly, cytokine analysis showed both EP receptors similarly contribute to the downregulation of IFNγ, IL-2, and TNFα production by T cells after moMDSC coculture. The combined aEP2/4 further upregulated the expression of these cytokines compared to the individual blockade of EP2 and EP4 (Figure 4C). Further characterization of CD4 and CD8 T cells within the cocultures demonstrated these suppressive features to be similarly present across CD4 and CD8 T cells (see supplementary Figures 3C, D).

Tumor cell clearance requires active immune responses by cytotoxic lymphocytes, including NK cells. PGE2 signaling on M-MDSCs has demonstrated to mediate NK cell dysfunction in the tumor tissue, enabling tumor progression (28, 29). To investigate the relevance of M-MDSCs on NK cells, moMDSCs were cocultured with autologous NK cells pre-stimulated with IL-2 (Figure 5A). After three days, NK cells were characterized for the expression of different activation surface markers and the production of pro-inflammatory cytokines. CM treated moMDSCs suppressed the production of IFNγ and TNFα, whereas no changes were detected for granzyme B (Figure 5B). NK activation marker analysis showed CM-treated moMDSCs to suppress the expression of CD25, CD69, and NKG2D, whereas NKp46 levels remained unchanged (Figure 5C). Of note, a significant downregulation of the lineage marker CD56 on NK cells was observed. EP2 and EP4 modulation demonstrated aEP4 rather than aEP2 to primarily recover the expression of CD25, CD69, NKG2D, IFNγ, and TNFα. Combined aEP2/4 further recovered the expression of these markers compared to aEP4 only, resembling the NK suppression level induced by untreated moMDSCs. Overall, these results indicate the synergistic contribution of EP2 and EP4 on moMDSCs during NK cell suppression, with EP4 primarily mediating such feature.

DCs are key initiators of anti-tumor immunity via priming tumor antigen specific T cell responses (58). However, this anti-tumoricidal DC feature is often impaired in cancer patients due to the suppressive TME. Although this outcome likely arises from tumor-mediated DC dysfunction, we hypothesize that environmental M-MDSCs can further steer or compromise the development of pro-inflammatory T cell responses regardless of the activation status of DCs. To address this question, moMDSCs were cocultured with moDCs previously treated with poly I:C and allogeneic naïve CD4 T cells (Figure 6A). After coculture, the resulting T cell populations were analyzed for the expression of the pro-inflammatory cytokines IFNγ, TNFα, and IL-2 (Figures 6B, 6C). Additionally, Treg occurrence was assessed by determining the expression of CD127, CD25, and FOXP3 (Figure 6C). T cell analysis showed CM-treated MDSCs in coculture with moDCs to robustly suppress IFNγ responses. Although not significant we observed a tendency towards the downregulation of TNFα and IL-2, and a modest increase in Tregs. EP2 and EP4 blockade demonstrated both receptors contribute to this phenotype, although we noted a higher EP4 contribution compared to that of EP2. Combined aEP2/4 on moMDSCs recovered IFNγ and TNFα production to the level of untreated moMDSCs. Additionally, aEP2/4 significantly diminished the enrichment of Tregs compared to CM treated moMDSCs. Collectively, these results show that PGE2-EP2/4 signaling in moMDSCs skew moDC-mediated T cell responses towards the expansion of T cells with a suppressive rather than a pro-inflammatory profile.

To assess the impact of targeting EP2/4 signaling in a model that better recapitulates the TME, we performed a 3D coculture of moMDSCs with a CRC-PDTO derived from liver metastasis. MoMDSCs were co-cultured in a collagen matrix together with the CRC-PDTOs to allow PDTOs and moMDSCs interaction (Figure 7A). After 24 hours, cocultures were treated with aEP2/4. Next day, collagen cultures were dissociated, and the resulting cell suspension was stained and analyzed for flow cytometry. MoMDSCs were distinguished based on positivity for CD11b (Figure 7B). aEP2/4 treatment of the cocultured cells led to M-MDSCs with lower expression levels of CD11b, MerTK, and CCR5, and higher levels of CD86 (Figure 7C). Although not significant, we noted aEP2/4 tended to downregulate the expression of CD14. Overall, aEP2/4 can significantly attenuate the suppressive phenotype of moMDSCs cultured in a 3D TME model.