EZH2 inhibition remodels the inflammatory senescence-associated secretory phenotype to potentiate pancreatic cancer immune surveillance

Abstract

Immunotherapies that produce durable responses in some malignancies have failed in pancreatic ductal adenocarcinoma (PDAC) due to rampant immune suppression and poor tumor immunogenicity. We and others have demonstrated that induction of the senescence-associated secretory phenotype (SASP) can be an effective approach to activate anti-tumor natural killer (NK) cell and T cell immunity. In the present study, we found that the pancreas tumor microenvironment suppresses NK cell and T cell surveillance after therapy-induced senescence through enhancer of zeste homolog 2 (EZH2)-mediated epigenetic repression of proinflammatory SASP genes. EZH2 blockade stimulated production of SASP chemokines CCL2 and CXCL9/10, leading to enhanced NK cell and T cell infiltration and PDAC eradication in mouse models. EZH2 activity was also associated with suppression of chemokine signaling and cytotoxic lymphocytes and reduced survival in patients with PDAC. These results demonstrate that EZH2 represses the proinflammatory SASP and that EZH2 inhibition combined with senescence-inducing therapy could be a powerful means to achieve immune-mediated tumor control in PDAC.

Extended Data Fig. 1.. T/P treatment induces cellular senescence across tumor conditions in vivo.

a, Representative Haematoxylin and eosin (H&E) (top) and Masson’s trichrome (bottom) staining of indicated KPC1 PDAC (PIP, PIL, PILiver) and KP1 LUAD (LIL, LIP, LILiver) derived-tumors from 2-3 independent experiments. Scale bars, 100μm. b, Immunohistochemical (IHC) staining of indicated KPC1 PDAC (PIP, PIL, PILiver) and KP1 LUAD (LIL, LIP, LILiver) derived-tumors treated with vehicle (V) or combined trametinib (1mg/kg) and palbociclib (100 mg/kg) (T/P) for 2 weeks. Quantification of the percentage of SA-β-gal⁺ area and the number of Ki67⁺ and pRb⁺ cells per field are shown inset (n=2-4 per group). Scale bar, 50μm. c, Immunofluorescence staining of indicated KPC1 PDAC (PIP, PIL, PILiver) and KP1 LUAD (LIL, LIP, LILiver) derived-tumors grown in different organs and treated as in (b). Quantification of the percentage of GFP⁺ (green) tumor cells expressing p21 (cyan) is shown inset (n=2-4 per group). Scale bar, 50μm. d, GFP⁺ tumor cells were FACS sorted from indicated tumors and extracted RNA subjected to RNA-seq analysis (n=2-4 per group). Gene Set Enrichment Analysis (GSEA) of RNA-seq data using an established senescence gene set is shown. NES, normalized enrichment score. P values in d were calculated using two-sided, Kolmogorov-Smirnov test. Error bars, mean ± SEM.

Extended Data Fig. 2.. Suppression of NK immunity specific to pancreas TME following T/P-induced senescence.

a-b, KPC2 PDAC or KP2 LUAD tumor cells expressing GFP were injected i.v. or orthotopically into the pancreas of 8-12 week old C57BL/6 female mice. Following tumor formation, mice were treated with vehicle (V) or combined trametinib (1mg/kg) and palbociclib (100 mg/kg) (T/P) for 2 weeks. Flow cytometry analysis of NK cell numbers and degranulation in PDAC (PIP, PIL) (a) and LUAD-derived tumors (LIL, LIP) (b) grown in different organs are shown (a-b, PIP V, n=3; PIP TP, n=4; PIL V and PIL TP, n=8 independent mice). c, Flow cytometry analysis of NK cell numbers and degranulation in spleens of mice with KPC1-derived PIP tumors treated as in (a) (n=5 independent mice per group). d, Kaplan-Meier survival curve of mice with KPC2-derived PIP tumors treated with vehicle, combined trametinib (1mg/kg) and palbociclib (100mg/kg), and/or depleting antibodies against NK1.1 (PK136; 250 μg) or CD8 (2.43; 200 μg) (d, V, n=5; TP; TP+αNK1.1 and TP+αCD8, n=8). e, IVIS images showing luciferase signaling in KPC1-derived PIL tumors following treatment as in (a). Right, quantification of total luminescence in the thoracic region (e, V, n=5; TP and TP+αNK1.1, n=8 independent mice). f, Waterfall plot of the response of KPC1-derived PIP tumors following 2 week treatment with vehicle, combined trametinib (1mg/kg) and palbociclib (100mg/kg), and/or an NK1.1 depleting antibody (PK136; 250 μg) (f, V and TP, n=5, TP+αNK1.1, n=6 independent mice). g, Waterfall plot of the response of KPC2-derived PIP tumors following 2 week treatment with vehicle, combined trametinib (1mg/kg) and palbociclib (100mg/kg), and/or an NK1.1 (PK136; 250 μg) or CD8 (2.43; 200 μg) depleting antibody (g, V, n=5; TP, n=7; TP+αNK1.1 and TP+αCD8 ,n=8 independent mice). h, Kaplan-Meier survival curve of mice with KPC1-derived PIL tumors treated with vehicle, combined trametinib (1mg/kg) and palbociclib (100mg/kg), and/or depleting antibodies against CD8 (2.43; 200 μg) or CD4 (GK1.5; 200 μg) (h, V, n=5; TP, TP+αNK1.1 and TP+αCD8, n=7 independent mice). i, Flow cytometry analysis of CD4⁺ and CD8⁺ T cell numbers and degranulation in KPC1 PDAC (PIP, PIL, PILiver) and KP1 LUAD-derived tumors (LIL, LIP, LILiver) grown in different organs and treated as in (a) (i, PIP V, PIP TP, PIL TP, n=10; PIL V, n=9; LIL V, n=3; LIL TP, n=5; LIP V, n=13; LIP TP, n=15; PILiver V, n=9; PILiver TP, n=10; LILiver V, n=8; LILiver TP, n=10 independent mice). Data represents pool of 3 independent experiments. P values in a-c, e-g, and i were calculated using two-tailed, unpaired Student’s t-test, and those in d and h were calculated using log-rank test. Error bars, mean ± SEM.

Extended Data Fig. 3.. Repression of pro-inflammatory SASP gene expression specific to the pancreas TME following T/P treatment.

a, Heatmaps showing fold change in IFNα (left), IL-12 (middle), and TNFα pathway genes (right) following T/P treatment in indicated tumor settings from RNA-seq data in Figure 2a (n=2-4 per group). b, IHC staining of indicated KPC1 PDAC (PIP, PIL) and KP1 LUAD (LIL, LIP) derived-tumors grown in different organs and treated with vehicle (V) or combined trametinib (1mg/kg) and palbociclib (100 mg/kg) (T/P) for 2 weeks. H-score quantification of CCL2 and CXCL10 staining intensity is shown inset (n=2-3 per group). Scale bars, 50μm. Error bars, mean + SEM. c, Transcription factor enrichment analysis showing transcriptional regulators whose targets are differentially expressed in tumors in the lungs (LIL, PIL) following T/P treatment.

Extended Data Fig. 4.. Tumors in the pancreas TME are enriched for H3K27me3 repressive chromatin marks at SASP gene loci.

a, Heatmaps of normalized genome-wide H3K27me3 signaling intensities of consensus peaks from CUT&Tag analysis of KPC1 cells treated with vehicle or trametinib (25 nM) and palbociclib (500 nM) in vitro for 8 days, or KPC1 cells FACS sorted from transplanted PDAC tumors in C57BL/6 mice treated with vehicle or trametinib (1 mg/kg) and palbociclib (100 mg/kg) for 2 weeks (n=2-4 per group). b, Normalized H3K27me3 peak intensities of 87 SASP genes (see Table 47) from CUT&Tag analysis samples in (a) (n=2-4 per group). c, Genome browser tracks showing H3K27me3 occupancy at pro-inflammatory SASP gene loci from CUT&Tag analysis samples in (a) (n=2-4 per group).

Extended Data Fig. 5.. PDAC and LUAD tumor cells have a similar pro-inflammatory SASP response to T/P-induced senescence in vitro.

a, Cytokine array analysis of pro-inflammatory SASP genes in murine PDAC and LUAD cell lines treated with vehicle or combined trametinib (25 nM) and palbociclib (500nM) for 8 days (n=2 per group). #, outside the detectable limit. b, Normalized expression levels of pro-inflammatory SASP genes in human PDAC and LUAD cell lines following treatment as in (a) from analysis of RNA-seq data generated in Ruscetti et al. (2018) (n=2 per group).

Extended Data Fig. 6.. Suppression of EZH2-mediated H3K27me3 induces pro-inflammatory SASP and immunomodulatory cell surface molecules following T/P treatment in PDAC cells.

a, Representative clonogenic assay images (from 3 biological replicates) of KPC1 PDAC cells harboring shRen or shEzh2 shRNAs replated in the absence of drugs after an 8-day pre-treatment with vehicle or combined trametinib (25 nM) and palbociclib (500 nM) (top). Bottom, representative SA-β-gal staining (from 3 biological replicates) of KPC1 PDAC cells harboring shRen or shEzh2 shRNAs and treated with vehicle or combined trametinib (25 nM) and palbociclib (500 nM) for 8 days. b, qRT-PCR analysis of senescence and SASP gene expression in KPC1 PDAC cells harboring shRen or shEzh2 shRNAs treated with vehicle or combined trametinib (25 nM) and palbociclib (500 nM) for 8 days (n=3 biological replicates per group). A.U., arbitrary units. c, Representative histograms (top) and quantification of mean fluorescent intensity (MFI) of MHC-I (H-2k^b) expression (bottom) on KPC1 PDAC cells harboring shRen or shEzh2 shRNAs (left) or parental KPC1 PDAC cells (right) treated with vehicle, combined trametinib (25 nM) and palbociclib (500nM), and/or tazemetostat (5 μM) for 8 days (shRen V, shRen TP, shEzh2 V, shEzh2 TP, n=3; V, TP, TAZ, TP/TAZ, n=6 independent mice). d, Representative clonogenic assay images (from 3 biological replicates) of KPC1 PDAC cells replated in the absence of drugs after an 8-day pre-treatment with vehicle, combined trametinib (25 nM) and palbociclib (500 nM), and/or tazemetostat (5 μM) (top). Bottom, representative SA-β-gal staining (from 3 biological replicates) of KPC1 PDAC cells treated with vehicle, combined trametinib (25 nM) and palbociclib (500 nM), and/or tazemetostat (5 μM) for 8 days. e, qRT-PCR analysis of SASP gene expression in human PANC-1 PDAC cells treated with vehicle, trametinib (25 nM), palbociclib (500 nM), and/or GSK126 (1 μM) for 8 days (n=3 per group). A.U., arbitrary units. f, Heatmaps of normalized genome-wide H3K27me3 signaling intensities from CUT&Tag analysis of KPC1 PDAC cells harboring Ren or Ezh2 shRNAs treated with vehicle or trametinib (25 nM) and palbociclib (500 nM) for 8 days (n=2-4 per group). g, Genome browser tracks showing H3K27me3 occupancy at pro-angiogenic SASP gene loci (n=2-4 per group). P values in b, c, and e were calculated using two-tailed, unpaired Student’s t-test. Error bars, mean ± SEM.

Extended Data Fig. 7.. EZH2 knockdown in the KPC2 PDAC orthotopic transplant model potentiates anti-tumor NK and CD8⁺ T cell immunity and long-term tumor regressions following T/P treatment.

a, Ultrasound quantification of initial PDAC tumor volume 1-week post-transplantation of KPC1 or KPC2 cells harboring shRen or shEzh2 shRNAs into 8-12 week old C57BL/6 female mice prior to enrollment in treatment cohorts (n=18-75 per group). Data represents pool of 6 independent experiments. b, Immunoblots of shRen or shEzh2 KPC1 orthotopic PDAC tumors treated with vehicle or trametinib (1 mg/kg) and palbociclib (100 mg/kg) for 2 weeks. c, IHC staining of KPC1 and KPC2 orthotopic PDAC tumors harboring shRen or shEzh2 shRNAs treated as in (b). H-score quantification of EZH2 expression is shown inset (n=2-3 per group). Scale bars, 50μm. d-e, Flow cytometry analysis of NK cell (d) and T cell (e) numbers and activation markers in KPC2 orthotopic PDAC tumors harboring indicated shRNAs treated as in (b) (d,e shRen V, n=6; shRen TP, n=7; shEzh2 V, n=5; shEzh2 TP, n=7 independent mice). f, Flow cytometry analysis of F4/80⁺ macrophages in KPC1 orthotopic PDAC tumors harboring indicated shRNAs treated as in (b) (f, shRen V, n=9; shRen TP, n=8; shEzh2 V, n=6; shEzh2 TP, n=9 independent mice). Data represents pool of 2 independent experiments. g, Waterfall plot of the response of KPC2 orthotopic PDAC tumors harboring indicated shRNAs to treatment as in (b) (g, shRen V, n=9; shRen TP, n=12; shEzh2 V, n=7; shEzh2 TP, n=13 independent mice). Data represents pool of 2 independent experiments. h, Kaplan-Meier survival curve of mice with shEzh2 KPC2 orthotopic PDAC tumors treated with vehicle, combined trametinib (1mg/kg) and palbociclib (100mg/kg), and/or depleting antibodies against NK1.1 (PK136; 250 μg) or CD8 (2.43; 200 μg) (h, shEzh2 V, n=5; shEzh2 TP, n=8; shEzh2 TP+αNK1.1, n=8; shEzh2 TP+αCD8, n=8 independent mice). Dotted line indicates timepoint when mice were taken off of treatment. P values in a were calculated using One-way ANOVA followed by Tukey’s multiple comparison test, d-g using two-tailed, unpaired Student’s t-test, and h using log-rank test. Error bars, mean ± SEM.

Extended Data Fig. 8.. Combined EZH2 knockdown and T/P treatment reduces pancreatic metastasis growth and enhances NK and T cell immune surveillance in the lung.

a, KPC1 PDAC cells harboring shRen or shEzh2 shRNAs were injected i.v. into 8-12 week old C57BL/6 female mice. Following tumor formation in the lungs, mice were treated with vehicle (V) or combined trametinib (1mg/kg body weight) and palbociclib (100 mg/kg body weight) (T/P) for 2 weeks. Quantification of lung tumor burden after 2 weeks of treatment is shown (a, shRen V, n=4; shRen TP, n=4; shEzh2 V, n=3; shEzh2 TP, n=5 independent mice). b-h, Flow cytometry analysis of total CD45⁺ immune cells, F4/80⁺ macrophages, NK cells, T cells, and their expression of GZMB in shRen or shEzh2 KPC1 PDAC tumors in the lung following treatment as in (a) (b-h, shRen V, n=6; shRen TP, n=7; shEzh2 V, n=8; shEzh2 TP, n=8 independent mice). P values in a-g were calculated using two-tailed, unpaired Student’s t-test. Error bars, mean ± SEM.

Extended Data Fig. 9.. EZH2 blockade reduces T/P-induced blood vessel formation and promotes CCL2 and CXCL9/10 secretion that increases NK and CD8⁺ T cell infiltration into PDAC.

a, IHC staining of KPC1 orthotopic PDAC tumors harboring shRen or shEzh2 shRNAs treated with vehicle or combined trametinib (1mg/kg) and palbociclib (100 mg/kg) (T/P) for 2 weeks. Quantification of blood vessels per field are shown on inset (a, shRen V, n=4; shRen TP, n=4; shEzh2 V, n=2; shEzh2 TP, n=4 independent tumors). Scale bar, 50μm. b-c, Flow cytometry analysis of NK cell activation markers (b) and CD4⁺ and CD8⁺ T cell numbers (c) in KPC1 orthotopic PDAC tumors expressing control Empty or Ccl2 vectors and treated as in (a) (b-c, Empty V, n=4; Empty TP, n=9; Empty CCL2O/E V, n=11; CCL2 O/E TP, n=12 independent mice). Data represents pool of 3 independent experiments d, Flow cytometry analysis of CD4⁺ and CD8⁺ T cell numbers in shEzh2 KPC1 orthotopic PDAC tumors following treatment with vehicle, combined trametinib (1mg/kg) and palbociclib (100mg/kg), and/or a CCL2 depleting antibody (2H5; 200 μg) for 2 weeks (d, shEzh2 V, n=3; shEzh2 αCCL2, n=3; shEzh2 TP, n=7; shEzh2 TP+αCCL2, n=9 independent mice). Data represents pool of 2 independent experiments. e, Flow cytometry analysis of NK cell numbers in shEzh2 KPC1 orthotopic PDAC tumors following treatment with vehicle, combined trametinib (1mg/kg) and palbociclib (100mg/kg), and/or a CXCR3 depleting antibody (CXCR3-173; 200 μg) for 2 weeks (d, V, n=8; TP, n=8; αCCXR3, n=5; TP+αCCXR3, n=12 independent mice). Data represents pool of 2 independent experiments. P values in a-e were calculated using two-tailed, unpaired Student’s t-test. Error bars, mean ± SEM.

Fig. 1.. NK cell immunity is induced in the lung but not pancreas TME following therapy-induced senescence.

a-b, KPC1 PDAC (a) or KP1 LUAD (b) tumor cells expressing luciferase-GFP were injected i.v. or orthotopically into the pancreas of 8-12 week old C57BL/6 female mice. Following tumor formation in the lungs or pancreas, mice were treated with vehicle (V) or combined trametinib (1mg/kg body weight) and palbociclib (100 mg/kg body weight) (T/P) for 2 weeks (left). Right, flow cytometry analysis of NK cell numbers and degranulation in each condition (a) NK cell numbers PIP V, PIP TP and PIL TP, n=10; PIL V, n=9) and degranulation (n=5 independent mice per group). (b) NK cell numbers LIL V, n=3: LIL TP, n=5; LIP V, n=13; LIP TP, n=15 and degranulation LIP V, n=5; LIP TP, n=7 independent mice). Data represents pool of 3 independent experiments. c, KPC1 PDAC or KP1 LUAD cells expressing luciferase-GFP were injected orthotopically into the livers of 8-12 week old C57BL/6 female mice and treated as in (a) following tumor formation (left). Right, flow cytometry analysis of NK cell numbers Data represents pool of 2 independent experiments. (c) NK cell numbers and NK degranulation PILiver V, n= 9; PILiver TP, n=10 and LILiver V, n=8; LILiver TP, n=10 independent mice). d, Kaplan-Meier survival curve of C57BL/6 mice harboring KPC1 PDAC tumors in pancreas (PIP) treated with vehicle or trametinib (1 mg/kg) and palbociclib (100 mg/kg) in the presence or absence of a NK1.1 depleting antibody (PK136; 250 ug) (V, n=5; TP and TP+αNK1.1, n=7 independent mice). e, Kaplan-Meier survival curve of C57BL/6 mice harboring KPC1 PDAC tumors in lungs (PIL) and treated as in (d) (V, n=10; TP, n=9 and TP+αNK1.1, n=8 independent mice). f, Kaplan-Meier survival curve of C57BL/6 mice harboring KP2 LUAD tumors in the lungs (LIL) and treated as in (d) (V, n=6; TP, n=7 and TP+αNK1.1, n=8 independent mice). g, Kaplan-Meier survival curve of C57BL/6 mice harboring KP1 LUAD tumors in pancreas (LIP) and treated as in (d) (V, n=7; TP and TP+αNK1.1, n=8 independent mice). P values in a-c were calculated using two-tailed, unpaired Student’s t-test, and those in d-g calculated using log-rank test. Error bars, mean ± SEM.

Fig. 2.. The pro-inflammatory SASP is transcriptionally and epigenetically repressed in the pancreas TME.

a, KPC1 PDAC or KP1 LUAD tumor cells expressing luciferase-GFP were injected i.v. or orthotopically into the pancreas of 8-12 week old C57BL/6 female mice. Following tumor formation in the lungs or pancreas, mice were treated with vehicle (V) or combined trametinib (1mg/kg) and palbociclib (100 mg/kg) (T/P) for 2 weeks. GFP⁺ tumor cells were FACS sorted and extracted RNA subjected to RNA-seq analysis (n=2-4 per group). b, KEGG pathway analysis of pathways enriched in tumors in the lungs (LIL, PIL) compared to tumors in the pancreas (PIP, LIP) following T/P treatment. c, Heatmap showing fold change in SASP gene expression following T/P treatment in indicated tumor settings. d, Fold change in expression of select SASP chemokines following T/P treatment in indicated tumor settings (n=2-4 per group). Error bars, mean ± SEM. e, Transcription factor enrichment analysis showing transcriptional regulators whose targets are differentially expressed in tumors in the pancreas (PIP, LIP) following T/P treatment. f, Gene Set Enrichment Analysis (GSEA) of EZH2 transcriptional targets. NES, normalized enrichment score. g, Immunofluorescence staining of indicated KPC1 PDAC (PIP, PIL) and KP1 LUAD (LIL, LIP) derived-tumors grown in different organs and treated with vehicle (V) or combined trametinib (1mg/kg) and palbociclib (100 mg/kg) (T/P) for 2 weeks. Quantification of the percentage GFP⁺ (green) tumor cells expressing H3K27me3 (cyan) is shown inset (PIP V; PIP TP; PIL TP; LIL TP and LIP TP, n=3, PIL V; LIL V and LIP V, n=2 independent tumors). Scale bar, 50μm. P values in b were calculated using two-sided, hypergeometric test and those in f using two-sided, Kolmogorov-Smirnov test.

Fig. 3.. SMA⁺ fibroblasts in the pancreas TME constrain SASP-mediated NK and T cell immunity and promote EZH2 activation in PDAC.

a, Schematic of KPC PDAC syngeneic orthotopic transplantation into 8-16 week old male and female SMA-TK mice and treatment regimens. b, Immunohistochemical (IHC) staining of KPC1 orthotopic PDAC tumors propagated in SMA-TK mice treated with vehicle, trametinib (1 mg/kg) and palbociclib (100 mg/kg), and/or ganciclovir (GCV) (50 mg/kg) for 2 weeks. Quantification of number of SMA⁺ cells per field is shown inset (b, V, n=5; TP, n=4; GCV, n=4; and TP/GCV, n=5 independent tumors). Scale bar, 50μm. c, Waterfall plot of the response of KPC1 orthotopic PDAC tumors propagated in SMA-TK mice to treatment as in (b) (c, V, n=6; TP, n=10; GCV, n=8, and TP/GCV, n=15 independent mice). Data represents pool of 2 independent experiments. d-e, Flow cytometry analysis of NK (d) and T cell (e) numbers and activation markers in KPC1 orthotopic PDAC tumors propagated in SMA-TK mice following treatment as in (b) (n=8-16 per group). (d-e, V, n=8; TP, n=9; GCV, n=8, and TP/GCV, n=16 independent mice). Data represents pool of 2 independent experiments. Error bars, mean ± SEM. f, KEGG (left) and REACTOME (right) pathway analysis of RNA-seq data generated from FACS sorted GFP⁺ tumor cells from SMA-TK mice harboring KPC1 orthotopic PDAC tumors and treated as in (b) (n=4-5 per group). g, Heatmap of RNA-seq analysis of SASP gene expression in PDAC cells from KPC1 orthotopic PDAC tumors propagated in SMA-TK mice and treated as in (b) (n=4-5 per group). h, Transcription factor enrichment analysis showing transcriptional regulators whose targets are differentially expressed in tumor cells from KPC1 orthotopic PDAC propagated in SMA-TK mice treated with T/P alone compared with combined T/P/GCV treatment. i, Gene Set Enrichment Analysis (GSEA) of EZH2 transcriptional targets. NES, normalized enrichment score. j, Immunofluorescence staining of KPC1 orthotopic PDAC tumors propagated in SMA-TK mice treated as in (b). Quantification of the percentage GFP⁺ (green) tumor cells expressing H3K27me3 (cyan) is shown inset (n=2 per group). Scale bar, 50μm. P values in c-e were calculated using two-tailed, unpaired Student’s t-test, f using two-sided, hypergeometric test, and i using two-sided, Kolmogorov-Smirnov test.

Fig. 4.. Targeting EZH2 expression or its methyltransferase activity reactivates the pro-inflammatory SASP in PDAC.

a, Immunoblots of KPC1 and KPC2 PDAC cells harboring Renilla (Ren), Ezh2, or Suz12 shRNAs. b, Cytokine array results from KPC1 and KPC2 PDAC cells with indicated shRNAs treated for 8 days with vehicle or trametinib (25 nM) and palbociclib (500 nM) (n=3 per group). c, Immunoblots of KPC1 PDAC cells treated with vehicle, trametinib (25 nM), palbociclib (500 nM), GSK126 (1 μM), and/or tazemetostat (5 μM) for 8 days. d, qRT-PCR analysis of senescence and SASP gene expression in KPC1 PDAC cells treated as in c for 8 days (d, V, n=6; TP, n=6 for all genes except Cxcl10 where n=3, GSK126; TP/GSK126; TAZ and TP/TAZ, n=3 biological replicates). A.U., arbitrary units. Data represents pool of 2 independent experiments. e, Genome browser tracks from CUT&Tag analysis showing H3K27me3 occupancy at pro-inflammatory SASP gene loci in KPC1 PDAC cells harboring Ren or Ezh2 shRNAs treated with vehicle or trametinib (25 nM) and palbociclib (500 nM) for 8 days (n=2-4 per group). P values in b and d were calculated using two-tailed, unpaired Student’s t-test. Error bars, mean ± SEM.

Fig. 5.. EZH2 blockade activates NK and T cell-mediated long-term tumor control following therapy-induced senescence in PDAC models.

a, Schematic of KPC PDAC syngeneic orthotopic transplant model in 8-12 week old C57BL/6 female mice and treatment regimens. b, Representative flow cytometry plots of CD45⁺CD3⁻NK1.1⁺ NK cells in KPC1 orthotopic PDAC tumors harboring indicated shRNAs from mice treated with vehicle (V) or combined trametinib (1mg/kg) and palbociclib (100mg/kg) (T/P) for 2 weeks. SSC, side scatter. c-e, Flow cytometry analysis of total CD45⁺ immune cells (c), NK cell numbers and activation markers (d), and T cell numbers and activation markers (e) in KPC1 orthotopic PDAC tumors harboring indicated shRNAs following treatment as in (b) (c-e, shRen V, n=9; shRen TP, n=8; shEzh2 V, n =6; shEzh2 TP, n=9 independent mice). Data represents pool of 2 independent experiments. f, Waterfall plot of the response of KPC1 orthotopic PDAC tumors with indicated shRNAs to treatment as in (b) (f, shRen V n=10, shRen TP, n=14; shEzh2 V, n =17; shEzh2 TP, n=21 independent mice) . Data represents pool of 3 independent experiments. g, Representative ultrasound images of a shEzh2 KPC1 orthotopic PDAC tumor prior to treatment and after 2 or 15 weeks of treatment with combined trametinib (1 mg/kg) and palbociclib (100 mg/kg). PDAC tumors are outlined in white. h, Kaplan-Meier survival curve of mice with KPC1 orthotopic PDAC tumors harboring indicated shRNAs treated with vehicle, combined trametinib (1mg/kg) and palbociclib (100mg/kg), and/or an NK1.1 depleting antibody (PK136; 250 μg) (h, shRen V, n=5; shRen TP; shRen TP+αNK1.1 and shEzh2 TP+αNK1.1, n=8; shEzh2 V and shEzh2 TP, n=7 independent mice) . i, Kaplan-Meier survival curve of mice with shEzh2 KPC1 orthotopic PDAC tumors treated with vehicle, combined trametinib (1mg/kg) and palbociclib (100mg/kg), and/or a CD8 depleting antibody (2.43; 200 μg) (i, shEzh2 V, n=6; shEzh2 TP, n=8 and shEzh2 TP+αCD8, n=10 independent mice). Dotted line indicates timepoint when mice were taken off of treatment. P values in c-f were calculated using two-tailed, unpaired Student’s t-test, and those in h and i calculated using log-rank test. Error bars, mean ± SEM.

Fig. 6.. EZH2 suppression reinstates SASP-associated chemokines to drive NK and T cell accumulation in PDAC.

a, KEGG pathway analysis of RNA-seq data showing enriched pathways in shEzh2 compared to shRen KPC1 orthotopic PDAC tumor cells FACS sorted from C57BL/6 female mice treated with trametinib (1mg/kg) and palbociclib (100mg/kg) for 2 weeks (n=5-6 per group). b, Heatmap of RNA-seq analysis of SASP gene expression in tumor cells FACS sorted from KPC1 orthotopic PDAC tumors harboring indicated shRNAs and treated as in (a) (n=5-6 per group). c, qRT-PCR analysis of Ccl2 expression in KPC1 PDAC cells engineered to overexpress (O/E) a Ccl2 cDNA or Empty control vector (n=3 per group). A.U., arbitrary units. d, NK cell migration assay in the presence of conditioned media from KPC1 PDAC cells engineered to overexpress Ccl2 or Empty vector and treated with vehicle or trametinib (25nM) and palbociclib (500nM) for 8 days (n=3 per group). e, Flow cytometry analysis of NK cell numbers in KPC1 orthotopic PDAC tumors expressing control Empty or Ccl2 vectors following treatment as in (b) (e, Empty V, n=4; Empty TP, n=9; CCL2 O/E, n=11 and CCL2 O/E/TP, n=12 independent mice). Data represents pool of 3 independent experiments. f, Kaplan-Meier survival curve of mice with KPC1 orthotopic PDAC tumors expressing control Empty (left) or Ccl2 (right) vectors treated with vehicle, combined trametinib (1mg/kg) and palbociclib (100mg/kg), and/or an NK1.1 depleting antibody (PK136; 250 μg) (f, Empty V, n=12; Empty TP, n=14; Empty TP+αNK1.1, n=13; CCL2O/E V, n=10; CCL2 O/E TP, n=11 and CCL2 O/E TP+αNK1.1, n=12 independent mice). Data represents pool of 2 independent experiments. g, Flow cytometry analysis of NK cell numbers in shEzh2 KPC1 orthotopic PDAC tumors following treatment with vehicle, combined trametinib (1mg/kg) and palbociclib (100mg/kg), and/or a CCL2 depleting antibody (2H5; 200 μg) for 2 weeks (g, V, n=3; αCCL2, n=3; TP, n=7 and TP αCCL2, n=9 independent mice). Data represents pool of 2 independent experiments. h, Waterfall plot of the response of shEzh2 KPC1 orthotopic PDAC tumors to treatment as in (g) (h, shEzh2 TP, n=22 and shEzh2 TP αCCL2, n=23 independent mice). Data represents pool of 3 independent experiments. i, Kaplan-Meier survival curve of mice with shEzh2 KPC1 orthotopic PDAC tumors treated as in (g) (i, shEzh2 V, n=13; shEzh2 TP, n=15 and shEzh2 TP αCCL2, n=10 independent mice) Values for shEzh2 V and T/P treated cohorts are the same displayed in Figs. 5h and 5i. Dotted line indicates when mice were taken off of treatment. Data represents pool of 2 independent experiments. j, Flow cytometry analysis of CD4⁺ and CD8⁺ T cell numbers in shEzh2 KPC1 orthotopic PDAC tumors following treatment with vehicle, combined trametinib (1mg/kg) and palbociclib (100mg/kg), and/or a CXCR3 depleting antibody (CXCR3-173; 200 μg) for 2 weeks (j, V, n=8; αCXCR3, n=5; TP, n=8; and TP+αCXCR3, n=12 independent mice). Data represents pool of 2 independent experiments. k, Waterfall plot of the response of shEzh2 KPC1 orthotopic PDAC tumors to treatment as in (j) (k, V, n=11; αCXCR3, n=7; TP, n=13; and TP+αCXCR3, n=15 independent mice). Data represents pool of 2 independent experiments. P values in a were calculated using two-sided, hypergeometric test, c,d,e,g,h,j,k using two-tailed, unpaired Student’s t-test, and f and i using log-rank test. Error bars, mean ± SEM.

Fig. 7.. Pharmacological EZH2 methyltransferase inhibition in combination with T/P reactivates cytotoxic NK and T cell immunity and enhances tumor control in preclinical PDAC models.

a, Immunoblots of KPC1 orthotopic PDAC tumors from C57BL/6 female mice treated with vehicle, trametinib (1 mg/kg) and palbociclib (100 mg/kg), and/or tazemetostat (Taz) (125 mg/kg) for 2 weeks. b, Immunohistochemical (IHC) staining of KPC1 orthotopic PDAC tumors treated as in (a). H-score quantification of H3K27me3 expression is shown inset (n=3 per group). Scale bar, 50μm. c, Waterfall plot of the response of KPC1 orthotopic PDAC tumors following 2 week-treatment with vehicle, trametinib (1 mg/kg) and palbociclib (100 mg/kg), and/or low (125 mg/kg) or high (400 mg/kg) doses of tazemetostat (c, V, n=15; Taz low, n=16; TP, n=9; TP/TAZ low, n=10 and TP/TAZ high, n=7 independent mice). Data represents pool of 3 independent experiments. d-e, Flow cytometry analysis of NK (d) and T cell (e) numbers and activation markers in KPC1 orthotopic PDAC tumors following treatment as in (c) (d-e, V, n=15; Taz low, n=15; TP, n=9; TP/TAZ low, n=12 and TP/TAZ high, n=7 independent mice). Data represents pool of 3 independent experiments. f, IHC staining of KPC GEMM tumors treated as in (a). Quantification of the number of NKp46⁺ NK cells, CD3⁺ T cells, and GZMB⁺ and Cleaved Caspase-3 (CC3)⁺ cells per field, and H-scores for H3K27me3 expression, are shown inset (f, V; TP; TAZ, n=3 and TP/TAZ, n=3 independent tumors). Scale bars, 50μm. g, Waterfall plot of the response of KPC GEMM tumors to treatment as in (a) (g, V, n=10; TP, n=7; TAZ, n=10 and TP/TAZ, n=17 independent mice). h, Kaplan-Meier survival curve of KPC GEMM male and female mice treated as in (a) (h, V, n=8; TP, n=7; TAZ, n=9 and TP/TAZ, n=10 independent mice). P values in c-e and g were calculated using two-tailed, unpaired Student’s t-test, and those in h calculated using log-rank test. Error bars, mean ± SEM.

Fig. 8.. EZH2 is associated with suppression of inflammatory chemokine signaling, reduced NK and T cell immune surveillance, and poor survival in PDAC patients.

a, Pearson’s correlation analysis plots comparing signatures of EZH2 and PRC2 repressed genes with inflammatory response gene sets, CCL2, CXCL9, and CXCL10 expression, and NK and CD8⁺ T cell signatures in human primary PDAC transcriptomic data (n=145 samples). Line represents line of best fit. b, Representative IHC staining of surgically resected human PDAC tumors (n=34). Scale bars, 50μm. c, Scoring of EZH2, NKp46, and CD8 expression from IHC staining in (b) (n=34). Percentage of samples with indicated scores are shown, with the total number of samples in parentheses. d, Kaplan-Meier survival curve of human PDAC patients stratified based on EZH2 expression levels in (b) (n=13, 15, and 6 for EZH2 Lo, Int, and Hi, respectively). e, Kaplan-Meier survival curve of human PDAC patients stratified based on NKp46 expression levels in (b) (n=15 and 19 for NKp46 Lo and Hi/Int, respectively). f, Kaplan-Meier survival curve of human PDAC patients stratified based on CD8 expression levels in (b) (n=10 and 24 for CD8 Lo and Hi/Int, respectively). P values in a were calculated using two-tailed, unpaired Student’s t-test, and those in d-f were calculated using log-rank test. Error bars, mean ± SEM.