Activated T Cells Break Tumor Immunosuppression by Macrophage Reeducation

Abstract

In this study, we observe that in human and murine melanomas, T-cell activation abates hematopoietic prostaglandin-D2 synthase (HPGDS) transcription in tumor-associated macrophages (TAM) through TNFα signaling. Mechanistically, HPGDS installs a prostaglandin D2 (PGD2) autocrine loop in TAMs via DP1 and DP2 activation that sustains their protumoral phenotype and promotes paracrine inhibition of CD8+ T cells via a PGD2-DP1 axis. Genetic or pharmacologic HPGDS targeting induces antitumoral features in TAMs and favors CD8+ T-cell recruitment, activation, and cytotoxicity, altogether sensitizing tumors to αPD1. Conversely, HPGDS overexpression in TAMs or systemic TNFα blockade sustains a protumoral environment and αPD1 resistance, preventing the downregulation of HPGDS by T cells. Congruently, patients and mice resistant to αPD1 fail to suppress HPGDS in TAMs, reinforcing the evidence that circumventing HPGDS is necessary for efficient αPD1 treatment. Overall, we disclose a mechanism whereby T-cell activation controls the innate immune system, and we suggest HPGDS/PGD2 targeting to overcome immunotherapy resistance.

Significance: In this study, we show a mechanism whereby T-cell activation controls the innate immune system and shapes the tumor microenvironment by reducing PGD2 production in TAMs. We suggest HPGDS inhibition as a promising strategy to treat refractory tumors to current immunotherapies or to overcome acquired resistance to immune checkpoint blockade.

Figure 1.

Identification of HPGDS in a subset of TAMs. A, Volcano plot extracted from an in-house scRNA-seq dataset, showing DEGs (out of 45 genes in the arachidonic acid metabolism) in total macrophages of NR vs. R patients with melanoma early OT with ICBs. B, Uniform Manifold Approximation and Projection (UMAP) representing the expression of HPGDS in different immune cell populations of patients with melanoma, extracted from an in-house scRNA-seq dataset. C and D, Violin plot from two publicly available scRNA-seq datasets of patients with melanoma, showing the expression of HPGDS in different tumor cell populations. E,HPGDS expression pattern in TAMs (CXCL9, CCL3, LYVE1, PLA2G2D, and SPP1) and monocytes from patients with melanoma extracted from an in-house scRNA-seq dataset. F, Volcano plot depicting DEGs between HPGDS^high and HPGDS^low macrophages in the TME of patients with melanoma. G, Colocalization of the macrophage-specific CD163 marker and HPGDS in primary melanomas, spreading melanomas, and melanoma metastases (i.e., in-transit metastasis, cutaneous metastasis, and lymph node or liver metastasis) of 14 patients with melanoma. H, Quantification of HPGDS colocalization with CD163⁺ TAMs. I, Dot plot of normalized expression of HPGDS and other protumoral genes in Rs, NRs with melanoma before treatment (BT) and early OT with ICB. J, Representative images of HPGDS, CD163, and CD8⁺ T-cell abundance in primary melanomas in Rs and NRs early OT with ICB. The colocalization of HPGDS with CD163 is indicated by the yellow arrowheads. P values were assessed by F, Wilcoxon test using the Seurat R package. G and J, Scale bar: 20 μm. FC, fold change.

Figure 2.

HPGDS expression in macrophages is sustained byprotumoral cytokines and is downmodulated by activated T cells. A, Violin plot comparing Hpgds expression between Ifn⁺ and Mrc1⁺ macrophages, extracted from a publicly available scRNA-seq dataset. B,Hpgds expression assessed by qRT-PCR in monocytes (CD45⁺, CD11b⁺, and CD115⁺) and TAMs (CD45⁺, CD11b⁺, and F4/80⁺) from melanoma-bearing mice (n = 3). C, qRT-PCR analysis of the expression of Hpgds in BMDMs Ctrl or stimulated with antitumoral stimuli (i.e., IFNγ with LPS), protumoral cytokines (i.e., IL-4, IL-6, and IL-10) or with IL-4 and after IL-4 washout, with antitumoral stimuli (e.g., TNFα, IFNγ, and LPS, alone or in combination; n = 3). D, Extracellular levels of PGD₂ measured by ELISA in BMDMs Ctrl or stimulated with IL-4 or first with IL-4 and then with TNFα (n = 3). E, qRT-PCR analysis showing the regulation of Hpgds expression in IL-4–polarized BMDMs alone (Ctrl) or cocultured with preactivated CD8⁺ T cells or CD8⁺ T-cell CM with an isotype or an αTNFα blocking antibody (n = 3–4). F and G, Tumor growth of resistant (YUMM 1.7; F) vs. responsive (YUMMER 1.7; G) melanoma tumors treated with IgG control, αCD8 antibody, αPD1, or the combination of αCD8 and αPD1 (n = 3–4). H and I, Quantification (H) and representative images (I) of HPGDS⁺ macrophages in resistant (YUMM 1.7) vs. responsive (YUMMER 1.7) melanoma tumors treated with IgG control, αCD8 antibody, αPD1, or the combination of αCD8 and αPD1 (n = 3–4). J, Quantification of CD206⁺ immunosuppressive macrophages in resistant (YUMM 1.7) vs. responsive (YUMMER 1.7) melanoma tumors treated with IgG control, αCD8 antibody, αPD1 or the combination of αCD8 and αPD1 (n = 3–4). K, Quantification of vessel density in resistant (YUMM 1.7) vs. responsive (YUMMER 1.7) melanoma tumors treated with IgG control, αCD8 antibody, αPD1, or the combination of αCD8 and αPD1 (n = 3–4). L and M, Quantification (L) and representative images (M) of αSMA⁺ pericyte blood vessel coverage of YUMM 1.7 or YUMMER 1.7 melanoma-bearing mice treated with IgG, αCD8, αPD1, or the combination of αCD8 and αPD1 (n = 3–4). N, Tumor growth of YUMMER 1.7 melanoma-bearing mice treated with IgG, αPD1, or αPD1 combined with Enbrel (n = 8–9). O, Quantification of F4/80⁺ macrophages in YUMMER 1.7 melanoma tumors treated with IgG, αPD1, or αPD1 in combination with Enbrel. In the group treated with αPD1, three out of nine tumors regressed completely; thus, the staining was not performed (n = 6–9). P and Q, Quantification (P) and representative images (Q) of HPGDS⁺ macrophages in YUMMER 1.7 melanoma tumors treated with IgG, αPD1, or αPD1 in combination with Enbrel. In the group treated with αPD1, three out of nine tumors regressed completely; thus, further analyses were not performed (n = 6–9). P values are calculated by (A) two-sided Wilcoxon test, (B) unpaired, two-tailed Student t test, (C) one-way ANOVA with Dunnett correction, and (D, E, H, J, K, –L, O, and P) one-way ANOVA with the Tukey multiple comparison test. F, G and N, two-way repeated measures ANOVA. Graphs show the mean ± SEM. I, L and Q, Scale bar: 20 μm.

Figure 3.

Hpgds overexpression in macrophages promotes tumor progression and confers ICB resistance. A and B, Tumor growth and tumor weight of YUMMER 1.7 melanoma-bearing Ctrl or Hpgds KI mice overexpressing Hpgds selectively in TAMs, treated with IgG or with αPD1 (n = 5–6). C, Quantification of CD8⁺ T cells in Ctrl and Hpgds KI mice treated with IgG or with αPD1 (n = 5–6). D–F, Quantification by qRT-PCR (D) of Hpgds in macrophages or histology of HPGDS⁺ macrophages (E) and representative tumor section images (F) from YUMMER 1.7 melanoma-bearing Ctrl and Hpgds KI mice treated with IgG or with αPD1 (n = 5–6). G and H, Quantification of CD105⁺ vessel (G), density (G), and size (H) in Ctrl and Hpgds KI mice treated with IgG or with αPD1 (n = 5–6). I, Quantification of αSMA⁺ pericyte blood vessel coverage from YUMMER 1.7 melanoma-bearing Ctrl and Hpgds KI mice treated with IgG or with αPD1 (n = 5–6). J and K, Quantification (J) and representative images (K) of Lectin-FITC⁺ perfused vessels in Ctrl and Hpgds KI mice treated with IgG or with αPD1 (n = 5–6). L and M, Quantification (L) and representative images (M) of the hypoxic area (PIMO⁺) in Ctrl and Hpgds KI mice treated with IgG or with αPD1 (n = 5–6). N, Number of lung metastasis in Hpgds KI mice treated with IgG or with αPD1 (n = 5–6). O, Representative scheme showing the interaction between TNFα secreted by cytotoxic CD8⁺ T cells upon activation and HPGDS expression in macrophages in responsive and resistant patients. P values are calculated by (A) two-way repeated-measure ANOVA, (B, C, E, G–J, L and N) one-way ANOVA with the Tukey multiple comparison test, and (D) multiple unpaired Student t test. Graphs show the mean ± SEM. F, K, and M, Scale bar: 20 μm. O, Created with BioRender.com. Trotta, R. (2025) (https://BioRender.com/y35l375).

Figure 4.

Hpgds deletion in TAMs reshapes the TME and inhibits tumor growth. A, Extracellular levels of PGD₂ measured by LC/MS in the interstitial fluid of YUMM 1.7 melanoma tumors from Ctrl and Hpgds^ΔMø mice (n = 4–5). B, Intratumoral concentration of PGD₂ measured by LC/MS in Ctrl and Hpgds^ΔMø mice (n = 3). C and D, Tumor growth (C) and tumor weight (D) of YUMM 1.7 CD90.1⁺ melanoma-bearing mice implanted in Ctrl or Hpgds^ΔMø mice (n = 9–11). For the analyses from H–U, 4 of 10 tumors in Hpgds^ΔMø mice were not included due to their total regression. FACS and immunofluorescence (IF) were performed by processing the same number of tumors in both groups. E, FACS analysis of the percentage of CD90.1⁺ cells (out of viable) in the blood collected from Ctrl and Hpgds^ΔMø YUMM 1.7 melanoma-bearing mice (n = 3). F, Quantification of CD90.1⁺ area out of Hoechst⁺ area per lung cross-section from Ctrl and Hpgds^ΔMø YUMM 1.7 melanoma-bearing mice (n = 4). G, qRT-PCR analysis of Cd90.1 expression in lungs from Ctrl or Hpgds^ΔMø YUMM 1.7 melanoma-bearing mice (n = 5). H and I, FACS analysis for MHC-II^high, CD11c⁺, and CD86⁺ (H) or MHC-II^low, CD206⁺, and CD204⁺ (I) TAMs from YUMM 1.7 melanoma-bearing Ctrl or Hpgds^ΔMø mice (n = 4). J, Quantification from Ctrl and Hpgds^ΔMø mice of melanoma sections costained for F4/80 and CD80 or CD206 (n = 5). K, FACS analysis for CD8⁺ T-cell percentages in melanoma tumors from Ctrl and Hpgds^ΔMø mice (n = 4). L, Flow cytometric quantification of activated CD8⁺ T cells (% of CD69⁺, IFNγ⁺, and GZMB⁺ out of CD8⁺ T cells) in melanoma tumors from Ctrl and Hpgds^ΔMø mice (n = 5). M–O, Quantification of CD105⁺ vessel density (M), perimeter (N), and size (O) in melanoma tumors from Ctrl and Hpgds^ΔMø mice (n = 5). P, Representative images of CD105⁺ vessels in melanoma tumors from Ctrl and Hpgds^ΔMø mice (n = 5). Q, Quantification of αSMA⁺ pericyte blood vessel coverage in melanoma tumors from Ctrl and Hpgds^ΔMø mice (n = 5). R and S, Quantification (R) and representative images (S) of Lectin-FITC⁺ perfused vessels in melanoma tumors from Ctrl and Hpgds^ΔMø mice (n = 5). T and U, Quantification (T) and representative images (U) of the hypoxic area (PIMO⁺) in YUMM 1.7 melanoma of Ctrl and Hpgds^ΔMø mice (n = 4). V and W, Tumor growth (V) and tumor weight (W) of YUMM 1.7 CD90.1⁺ melanoma tumors injected in Ctrl or Hpgds^ΔMø mice treated with IgG control or a CD8-depleting antibody (n = 5–6). X, Quantification of YUMM 1.7 melanoma sections stained for F4/80 and CD80 from Ctrl and Hpgds^ΔMø tumor-bearing mice treated with IgG or αCD8 (n = 5). Y, qRT-PCR analysis for Cd90.1 in lungs collected from Ctrl and Hpgds^ΔMø melanoma-bearing mice treated with IgG or αCD8 antibody (n = 3–4). P values were assessed by (A, B, D–G, K, M–O, Q, R, and T) unpaired, two-tailed Student t test; (C and V) two-way repeated-measure ANOVA; (H–J and L) multiple unpaired Student t test; and (W–Y) two-way ANOVA with the Tukey multiple comparison test. Graphs show the mean ± SEM. Scale bar: P and S, 20 μm; U, 50 μm.

Figure 5.

Synergism of Hpgds deletion with ICB and contribution of PGD₂ signaling to TAM reeducation and CD8⁺ T-cell biology. A and B, Tumor growth (A) and tumor weight (B) from Ctrl and Hpgds^ΔMø tumor-bearing mice treated with IgG or αPD1 when the average tumor of the group was 150 mm³. Tumors in 2 of 13 Hpgds^ΔMø mice that did not reach 150 mm³ were not included in the experiment. C, qRT-PCR analysis for Cxcl10 in IL-4–polarized Ctrl or Hpgds^ΔMø BMDMs treated or not with 1 μmol/L PGD₂ for 24 hours (n = 4). D, qRT-PCR analysis for Cd206 in IL-4–polarized Ctrl or Hpgds^ΔMø BMDMs treated or not with 1 μmol/L PGD₂ for 24 hours (n = 4). E, Migration through 5-μm-pore polycarbonate membranes of CD8⁺ T cells in the presence of IL-4–polarized Ctrl or Hpgds^ΔMø BMDMs pretreated or not with 1 μmol/L PGD₂ for 24 hours (n = 3). F, Quantification of the total sprout length of HUVEC spheroids embedded in collagen I in coculture with Ctrl or Hpgds^ΔMø BMDMs and supplemented with 1 μmol/L PGD₂, 50 ng/mL VEGF, or the combination of both (n = 4). G, HUVEC migration assay (through 8-μm-pore polycarbonate membranes) in which the top chamber was seeded with HUVECs pretreated for 2 hours with Ctrl or Hpgds^ΔMø macrophage CM with or without the CXCR3 inhibitor AMG 487 (500 nmol/L); 10 ng/mL VEGF were then placed in the bottom chamber. HUVEC medium without or with VEGF were used as negative and positive controls, respectively. H, YUMM 1.7 melanoma cancer cell migration (through 8-μm-pore polycarbonate membranes) in the presence of Ctrl or Hpgds^ΔMø BMDMs in the lower chamber (n = 3). I–K, FACS analysis of TNFα (I), IFNy (J), and GZMB (K) production in activated CD8⁺ T cells pretreated for 4 hours with 1 μmol/L PGD₂ (n = 3). L and M, FACS analysis of CD69⁺ (L) and PD1⁺ (M) activated CD8⁺ T cells pretreated for 4 hours with 1 μmol/L PGD₂ (n = 3). N, Flow cytometric quantification of dividing (CellTrace VioletCTV) OT-I CD8⁺ T cells prestimulated for 4 hours with 1 μmol/L PGD₂ and cocultured for 24 hours with OVA-expressing YUMM 1.7 in a ratio 2:1 (n = 6). O, Migration through 5-μm-pore polycarbonate membranes of activated CD8⁺ T cells (prestimulated for 4 hours with 1 μmol/L PGD₂) toward medium, 50 ng/mL CXCL10, 1 μmol/L PGD_2, or the combination of both (n = 5). P and Q, Tumor growth (P) and weight (Q) of YUMM 1.7 melanoma tumors in nontargeting control (NT), Ptgdr1, Ptgdr2, and Ptgdr1 Ptgdr2 KO macrophage chimeras (n = 8–14). R, Histologic quantification of CD80 or CD206 macrophages (F4/80⁺) in melanoma from NT and Ptgdr1 Ptgdr2 (dualKO) macrophage chimeras (n = 8–14). S and T, Histologic quantification of CD105⁺ vessel density (S) and αSMA pericyte coverage (T) in melanoma tumors from NT and dualKO macrophage chimeras (n = 8–14). U and V, YUMM 1.7 melanoma tumor growth (U) and tumor weight (V) in chimeric mice NT, Ptgdr1, and Ptgdr2 KO in CD8⁺ T cells (n = 7–9). W, Histologic quantification of CD80 or CD206 macrophages (F4/80⁺) in YUMM 1.7 melanoma-bearing mice from NT and Ptgdr1 KO CD8⁺ T-cell chimeras (n = 7–9). X, Quantification of αSMA pericyte coverage in YUMM 1.7 melanoma-bearing mice from NT and Ptgdr1 KO CD8⁺ T-cell chimeras (n = 7–9). Y, Histologic quantification of CD8⁺ T cells in YUMM 1.7 melanoma-bearing mice from NT and Ptgdr1 KO CD8⁺ T-cell chimeras (n = 7–9). P values are assessed by (A, P, and U) two-way repeated measures ANOVA, (B–E) two-way ANOVA with Tukey multiple comparison test, (F and G) two-way ANOVA with Tukey multiple comparison test, (H–N, S, T, and X) unpaired, two-tailed Student t test, (O, Q, and V) one-way ANOVA with Tukey multiple comparison test, (R and W) multiple unpaired Student t test, and (Y), multiple paired Student t test. Graphs show the mean ± SEM. P–T, Data show a pool of two independent experiments. Nr, number.

Figure 6.

Pharmacologic HPGDS inhibition phenocopies the effect of its genetic deletion in TAMs. A–C, Tumor growth (A), tumor weight (B), and tumor representative pictures (C) of YUMM 1.7 CD90.1⁺ melanoma-bearing mice treated by oral gavage with HQL-79, 15 mg/kg BID or with vehicle control (n = 6). The treatment window is highlighted in pink (A). D, Fold change of Cd80, Cxcl10, Arg1, and Cd206 expression in TAMs sorted from mice treated with HQL-79 or vehicle (n = 4). E and F, Quantification (E) and representative images (F) of F4/80 and CD80 or CD206 staining in melanoma tumors from vehicle or HQL-79–treated mice (n = 5). G, FACS analysis of YUMM 1.7 CD90.1⁺ cancer cell intravasation into the bloodstream in melanoma-bearing mice treated by oral gavage with HQL-79 or with vehicle control (n = 6). H, qRT-PCR analysis of Cd90.1 expression in lungs from YUMM 1.7 melanoma-bearing mice treated with vehicle or HQL-79 (n = 5–6). I and J, Tumor growth (I) and tumor weight (J) of YUMM 1.7 CD90.1⁺ melanoma-bearing Ctrl or Hpgds^ΔMø-ERT2 mice treated with vehicle or HQL-79 (n = 4). The treatment window is highlighted in pink (I). K–M, Tumor growth (K), tumor weight (L), and tumor representative pictures (M) of NRAS^{Q61 K}; Ink4a^−/− melanoma tumor-bearing mice treated by oral gavage with vehicle or HQL-79 15 mg/kg BID (n = 6). The treatment window is highlighted in pink (K). N and O, Tumor growth (N) and tumor weight (O) of YUMM 1.7 CD90.1⁺ melanoma-bearing mice treated by oral gavage with Cmpd1y 0.1 – 0.3 – 1 – 3 mg/kg BID or with vehicle (n = 7–9). Data show a pool of two independent experiments. The treatment window is highlighted in pink (N). P and Q, Tumor growth (P) and weight (Q) of YUMM 1.7 CD90.1⁺ melanoma-bearing mice treated by oral gavage with TAS-205 30 mg/kg BID or vehicle (n = 9). The treatment window is highlighted in pink (P). P values are assessed by (A, I, K, N, and P) two-way repeated measures ANOVA, (B, G, H, L, and Q), unpaired, two-tailed Student t test, (E) multiple unpaired Student t test, (J) two-way ANOVA with Tukey multiple comparison test, and (O) one-way ANOVA with Tukey multiple comparison test. Graphs show the mean ± SEM. Scale bar: F, 20 μm. BID, twice daily.

Figure 7.

HPGDS inhibition overcomes immunotherapy resistance and is effective in human settings. A–C, Tumor growth (A), tumor weight (B), and representative images (C) of Braf^V600EPten^−/− genetically engineered mice (GEM) treated with vehicle, HQL-79 15 mg/kg twice daily, αPD1, or the combination of HQL-79 and αPD1 (n = 6). The treatment window is highlighted in pink (A). D, Percentage of CD11c⁺ macrophages in Braf^V600EPten^−/− GEM treated with vehicle, HQL-79, αPD1, or the combination of HQL-79 and αPD1 (n = 6). E, Flow cytometric quantification of PD1 expressing CD8⁺ T cells in Braf^V600EPten^−/− GEM treated with vehicle, HQL-79, αPD1, or HQL-79 in combination and αPD1 (n = 6). F, Histologic quantification of CD8⁺ T cells at the tumor margin vs. core in Braf^V600EPten^−/− GEM treated with vehicle/IgG, HQL-79, αPD1, or HQL-79 in combination with αPD1 (n = 6). G and H, Tumor weight (G) and representative pictures (H) of KPC FC1245 PDAC-bearing mice treated with vehicle/IgG, HQL-79, αPD1, or HQL-79 in combination with αPD1 (n = 11–16). I, qRT-PCR analysis of the expression of HPGDS in hMDMs Ctrl or stimulated with protumoral cytokines (i.e., IL-4 and IL-10) or first with IL-4 and then with TNFα (n = 3). J, Two hours migration of CD8⁺ T cells through 5-μm-pore polycarbonate membranes of CD8⁺ T cells in the presence of IL-4–polarized or Ctrl hMDMs treated with 1 μmol/L HQL-79 or with vehicle control (negative control: bottom chamber medium without macrophages; n = 3–6). K and L, Calcein area (K) and CD8⁺ T-cell increase (L) of PDOTS treated with 1 μmol/L HQL-79 or with vehicle control for 72 hours. M–O, Flow cytometric quantification of total CD8⁺ T cells (CD8⁺ T cells out of CD45⁺ cells; M), TNFα produced by CD8⁺ T cells (MFI TNFα on CD8⁺ T cells; N), and HLA-DR (HLA-DR^high out of CD68⁺ cells; O) in PDOTS treated with 1 μmol/L HQL-79 or with vehicle control for 96 hours. P, Schematic representation of HPGDS contribution to the TME and immunotherapy response in melanoma. HPGDS-mediated PGD₂ production by TAMs results in autocrine activation of DP1 and DP2, which sustains their protumoral features, whereas PGD₂ binding to CD8⁺ T cells blocks their recruitment and activation (1). TNFα released by activated T cells suppresses HPDGS transcription, as observed in ICB Rs. In turn, HPGDS downregulation and a consequent reduction in PGD₂ production favor an immune-permissive and antitumoral environment (2). In NRs, HPGDS-expressing TAMs inhibit T-cell activation but also promote vessel abnormalization and metastasis (3). Blocking HPGDS induces vascular normalization and immune surveillance, overcoming ICB resistance (4). P values are assessed by A, two-way repeated measures ANOVA, (B, D, E, and G) one-way ANOVA with Tukey multiple comparison test, (F) multiple paired Student t test, (I) one-way ANOVA with Tukey multiple comparison test, and (J) two-way ANOVA with Tukey multiple comparison test. Graphs show the mean ± SEM. P, Created with BioRender.com. Trotta, R. (2025) (https://BioRender.com/y35l375.)