Ancestrally Diverse Autologous Patient-Derived Organoid–Immune Cell Coculture Platform for Addressing Immunotherapeutic Outcome Disparities in High-Grade Endometrial Cancer

Abstract

High-grade endometrial cancers (HGEC) disproportionately affect women of African ancestry and often resist currently available immunotherapies. Defining the mechanisms driving this resistance is impeded by a lack of preclinical models that preserve ancestral diversity and patient-matched tumor–immune interactions without confounding alloreactivity. To address this gap, we established a biobank of 85 endometrial cancer patient-derived organoids (PDO) from a diverse cohort, enriched for HGEC PDOs from African American patients, and paired these with autologous immune cells to develop a patient-specific PDO–immune cell coculture platform with real-time live-imaging readouts. Using this system, we found that HGECs evade immune surveillance through pronounced suppression of major histocompatibility complex (MHC) class I and II antigen presentation pathways relative to their matched normal counterparts. Restoring antigen presentation, either by IFNγ stimulation or epigenetic reprogramming via enhancer of zeste homolog 2 inhibition, rescued MHC expression and sensitized HGEC PDOs to autologous T cell–mediated cytotoxicity. Extending the platform to NK cells revealed heightened killing of low–MHC-I PDOs. Consistent with clinical observations, mismatch repair (MMR)-deficient HGEC PDOs exhibited stronger immune engagement than their MMR-proficient counterparts. Finally, this platform enabled evaluation of the safety and efficacy of emerging immunotherapies, including protease-activatable bispecific T-cell engagers and EGFR-targeted chimeric antigen receptor T cells. Together, this sustainable, scalable, ancestrally diverse autologous PDO–immune cell coculture platform offers a robust resource for dissecting immune evasion mechanisms and accelerating the development of new immunotherapies to address disparities in endometrial cancer outcomes.

Significance:

The efficacy of immunotherapy for HGECs remains limited, partly because current preclinical models poorly capture tumor heterogeneity and patient-specific immune microenvironment. These cancers disproportionately affect women of African ancestry, yet most studies rely on European ancestry samples. We developed an autologous PDO and immune cell coculture platform from patients of diverse ancestries. This system enables patient-level analysis of tumor–immune interactions to support development and testing of novel immunotherapeutic strategies.

Introduction

Endometrial cancer is the most common gynecologic malignancy and one of the leading causes of cancer-related morbidity and mortality among women worldwide (1). Although advances in cancer research have led to decreased mortality rates in other cancers such as breast and ovarian, endometrial cancer has shown the highest increase in cancer death rates, underscoring an urgent need for improved treatment strategies. Endometrial cancer is broadly categorized into two types: type I, which is typically low-grade, estrogen-dependent, and linked to obesity and prolonged estrogen exposure and type II, or high-grade endometrial cancers (HGEC), which are more aggressive and include subtypes like serous carcinoma and carcinosarcoma (2–5). Notably, HGECs disproportionately affect women of African ancestry and are associated with poorer prognoses, highlighting significant racial disparities in endometrial cancer incidence and outcomes (6, 7).

Current treatment modalities for HGECs primarily involve surgical intervention followed by adjuvant chemotherapy and radiation, depending on the disease stage. However, the efficacy of these standard-of-care regimens remains suboptimal for HGECs, which are characterized by heterogeneity and high recurrence rates. The introduction of immune checkpoint inhibitors in combination with chemotherapy has provided new hope for patients with endometrial cancer, initially demonstrating efficacy in mismatch repair–deficient (MMRd) tumors (4, 5). Recently, the US Food and Drug Administration (FDA) expanded approval of this combination therapy to include advanced or recurrent endometrial cancers regardless of MMR status (8). Despite these advancements, significant challenges persist in effectively treating HGECs, necessitating the development of more effective therapeutic strategies.

Preclinical models are essential tools for understanding tumor biology and testing new therapies. Previous studies have delineated the molecular characteristics of endometrial cancer (9), and efforts have been made to establish cancer models, including patient-derived xenografts and organoid cultures (10–12). Patient-derived organoid (PDO) models, which are three-dimensional cultures derived from patient tumors, offer a promising avenue for recapitulating the histologic and genetic features of the original tumor. However, existing PDO models often lack the immune microenvironment, limiting their utility in studying tumor–immune interactions and immunotherapy responses. Moreover, there is a paucity of preclinical models that accurately represent HGEC subtypes from diverse patient populations, particularly among women of African ancestry who are disproportionately affected by HGECs.

Autologous models that integrate cancer cells and immune cells derived from the same patient are crucial for replicating the specific cellular dynamics and molecular features of individual tumors. These models facilitate in-depth investigation of tumor–immune interactions and the evaluation of novel immunotherapeutic approaches. Although autologous cancer–immune cell coculture systems have recently been developed for other cancer types (13, 14), such models have not been established for HGECs. To enhance the efficacy of immunotherapy in HGECs and address existing disparities, there is an urgent need to develop autologous models of HGECs from patients of diverse ancestries.

Several studies have described immunologic aspects of gynecologic cancers such as the alterations in antigen presentation pathways (15, 16) and the association between tumor-infiltrating lymphocytes and the tumor MMR status (17, 18). However, mechanistic studies are limited, in part due to the paucity of data generated from preclinical endometrial cancer models to study the cancer–immune cell interactions. A few long-established cell lines have been used for characterization and mechanistic studies (19) but, these cell lines do not fully recapitulate patient-specific clinical and molecular characteristics of the disease. In addition, cell lines for some types of HGEC are not available. Furthermore, the characterization of cancer–immune cell interactions using these cell lines is significantly limited by alloreactivity, particularly when assessing immunotherapeutics.

In this study, we address these gaps by establishing an autologous PDO–immune cell coculture platform for HGECs from patients of diverse ancestries. This platform enables the examination of cancer–immune cell interactions at the individual patient level and provides a personalized approach for developing and testing novel immunotherapeutic strategies. By incorporating the autologous immune microenvironment and capturing HGEC heterogeneity in a cohort that includes different ancestries, our models and resources hold significant potential for advancing our understanding of HGEC biology and improving therapeutic outcomes.

Materials and Methods

Human specimen collection

This study was approved by the Institutional Review Board at Northwell Health (IRB #18-0897). All participants provided written informed consent, and all procedures were conducted in accordance with recognized ethical guidelines. Tumor and matched nonmalignant endometrial tissues were collected from the same patients undergoing hysterectomy for benign or malignant indications at Long Island Jewish Medical Center. Matched peripheral blood from each patient was collected at the time of surgery. All surgical resections were performed by the Division of Gynecologic Oncology. Patients included in the study had been evaluated for endometrial cancer preoperatively by endometrial biopsy or curettage before definitive surgery. The age range for the patients is 30 to 85 years old (mean 63.8; SD 13.0). MMR status of primary tumors was determined by immunohistochemistry (IHC) for MLH1, MSH2, MSH6, and PMS2, interpreted by a board-certified pathologist at Long Island Jewish Medical Center. Sex as a biological variable was not considered because this study is focused on endometrial cancer in female patients. Because ancestry distribution in our cohort is confounded with tumor grade/subtype and sample sizes per ancestry are limited, we do not perform ancestry-stratified statistical comparisons.

Genetic ancestry estimation

Ancestry proportion is determined by the software ADMIXTURE v1.3.0 (RRID: SCR_001263; ref. 20), which uses a maximum likelihood–based method to estimate the proportion of reference population ancestries in a sample. Samples were genotyped by the reference markers generated from 1,964 unrelated 1000 Genomes project samples directly on the samples using GATK pileup (RRID: SCR_001876) with a cutoff value of 0.6. In the RNA sequencing (RNA-seq)-based genetic ancestry estimation, the reference individuals from populations Mexican ancestry from Los Angeles, United States, African Caribbean in Barbados, and African ancestry in Southwest United States) were excluded from the reference panel used for ADMIXTURE model fitting because these populations are putatively highly admixed. However, no patient samples were excluded from our cohort on the basis of self-reported ancestry. The reference was further filtered by using only SNP markers with a minimum minor allele frequency of 0.01 overall and 0.05 in at least one 1000 Genomes superpopulation. Variants are additionally linkage disequilibrium–pruned using PLINK v1.9 (RRID: SCR_001757; ref. 21) with a window size of 500 kb, a step size of 250 kb, and r2 threshold of 0.2. The analysis results in a proportional breakdown of each sample into five continental populations [African (AFR), admixed American (AMR), East Asian (EAS), European (EUR), and South Asian (SAS)] and 23 populations. Subcontinental classification of the AFR population was not performed due to high signal-to-noise ratio of RNA-based ancestry estimation (22).

Generation and culture of PDOs

PDO lines from endometrial tissue were derived and maintained as previously described (23–25). Briefly, endometrial tissues were minced until tissue was roughly 1 mm in diameter and washed twice with PBS. Tissue samples were digested in collagenase IV (Sigma-Aldrich, C9407) at 1 mg/mL concentration in the presence of 10 μmol/L Y-27632 dihydrochloride (Tocris, 1254) for 2 to 3 hours. Tissues were then incubated in TrypLE Express (Thermo Fisher Scientific, 12604) in the presence of Y-27632 dihydrochloride for 10 to 20 minutes and confirmed microscopically for isolated single-cells. Digestion was stopped with Advanced DMEM F-12 (Gibco, 12634028), and after centrifugation, the specimens were resuspended in 70% Matrigel (Corning, 356234) and 30% corresponding media (23) and plated on prewarmed plates. Media were refreshed every 3 days. PDOs were split every 10 to 20 days depending on the growth kinetics. For splitting, PDOs were retrieved from Matrigel by Cell Recovery Solution (Corning, 354253) treatment for 30 minutes and then digested with TrypLE supplemented with Y-Factor for 10 to 20 minutes. Digestion was stopped with Advanced DMEM F-12 (Gibco, 12634028), and after centrifugation, single cells were plated to a density of 5,000 cells/10 μL.

Drug treatment

Established PDO lines were split as described above when preparing for coculture and allowed 2 to 3 days of growth. Then PDOs were prestimulated with 2 ng/mL IFNγ (Peprotech, 300-02) for additional 2 days or with 0.5 nmol/L tazemetostat (TAZ; Selleckchem, S7128) for 10 days in respective experiments and harvested for either RT-qPCR or flow cytometry. For TAZ, culture media were refreshed every 2 days.

IHC and immunofluorescent staining

Tissues or PDOs examined as previously described (23). Briefly, tissues and PDOs were fixed in 10% formalin, embedded in paraffin, and sectioned. Antigen retrieval was performed with citrate buffer (Sigma-Aldrich, C9999) following the manufacturer’s protocol. Downstream IHC was performed using an IHC kit (Vector laboratory, 30130) and primary antibodies (Supplementary Table S2). Sections were then counterstained with hematoxylin. For immunofluorescence (IF) staining, the sections were incubated with secondary antibodies conjugated with fluorescent molecules (Supplementary Table S2) and washed with PBS three times. IF slides were then imaged using a confocal microscope.

Fluorescence in situ hybridization

Single-molecule in situ hybridization was performed to detect major histocompatibility complex (MHC)-II (HLA-DRA) with Advanced Cell Diagnostics RNAscope 2.5 HD Detection Kit (475891) by following the manufacturer’s protocol.

Peripheral blood mononuclear cell isolation

Peripheral blood mononuclear cells (PBMC) were isolated using SepMate-15 Kit (Stem Cell Technology, 85415) by following the manufacturer’s protocol from patients’ blood samples. Briefly, the whole blood samples were diluted at 1:1 ratio with PBS + 2% FBS and carefully poured into a 15-mL SepMate tube loaded with Lymphoprep solution (Stem Cell Technology, 07851). The tubes were then spun down at 1,200 × g for 15 minutes, and the top layers were transferred to new 15-mL tubes and spun down again at 300 × g for 3 minutes. The pellets were washed with PBS + 2% FBS and cryo-frozen with freezing medium (Gibco, 12648010) by following the manufacturer’s protocol.

Thawing cryo-frozen PBMCs

Cryo-frozen autologous PBMCs were thawed to 37°C, centrifuged, and resuspended in human T-cell media [ImmunoCult-XF T-cell expansion medium (Stem Cell Technology, 10981) supplemented with hIL2 (R&D Systems, 202-IL-010) and hIL7 (R&D Systems, 207-IL-010)] at 1 × 10⁶ cells/mL concentration. Culture medium was changed every 3 days until utilization.

T and NK cell isolation from PBMCs

For T-cell subtype assays, PBMCs were first expanded using CD3-CD28 antibody following the manufacturer’s protocol (Stem Cell Technology, 10971). Briefly, PBMCs were stimulated with CD3-CD28 antibodies at 25 μL/1 × 10⁶ PBMCs/mL concentration for 3 days with human T-cell media. The activators were then washed, and PBMCs were then cultured for additional 2 weeks. Upon entering lagging growth phase, CD8⁺ and CD4⁺ T cells were then isolated from expanded T cells using EasySep Human CD8⁺ T-cell (Stem Cell Technology, 17953) and CD4⁺ T-cell (Stem Cell Technology, 19852) isolation kits following the manufacturer’s protocols. NK cells were isolated from PBMCs from healthy donors using EasySep Human NK Isolation Kit (Stem Cell Technology, 17955). Isolated immune cell subtypes (CD8⁺, CD4⁺, and NK cells) were then used for coculture experiments immediately.

PDO–immune cell coculture system

PDO lines were split as described above when preparing for coculture and allowed 2 to 3 days of growth. Then PDOs were prestimulated with 2 ng/mL IFNγ (Peprotech, 300-02) for additional 2 days (day 4–5 PDOs) or with 0.5 nmol/L TAZ (Selleckchem, S7128) for 10 days (day 12–13 PDOs) in respective experiments. PDOs were then stained with 1 μmol/L Cell Tracker Red (Thermo Fisher Scientific, C34552), and immune cells were stained with 1 μmol/L Cell Tracker Green (Thermo Fisher Scientific, C7025) for 30 minutes. Cells were then washed with their corresponding media twice. Immune cells (200k cells) were directly mixed with the PDOs (50–100 PDOs; approximately 1:10 effector:target ratio) in a 45% organoid media/45% human T-cell media and 10% Matrigel with 500 nmol/L caspase-3/7 dye (Sigma-Aldrich, SCT104) and plated on individual wells of a 96-imaging well plate (Agilent, 204626-100). The plate was then briefly spun (100 g, 1 minutes) to bring the cells on the same plane. Plated coculture assays were imaged on the next day unless otherwise stated to assess cancer–immune cell interaction and organoid apoptosis. The same human T-cell media were used for all immune cell subtypes [PBMCs, CD8⁺, CD4⁺, NK, and untransduced/transduced EGFR chimeric antigen receptor (CAR)-T cells].

Bispecific T-cell engager coculture

Similar to the coculture setup above, both immune cells and PDOs were prepared before the last spin. Then, 100 nmol/L of each heavy and light chains of bispecific engagers [uncleavable, matriptase, urokinase plasminogen activator (uPA), and cathepsin B (CTSB)] were added to corresponding wells of 96-imaging wells and briefly spun down. Plated coculture assays were then imaged after a short recover time (3 hours).

EGFR–CAR-T coculture

Similar to the coculture setup above, PDOs, EGFR–CAR-T cells, and parental allogeneic untransduced CD8⁺ T cells were prepared. Due to the limiting number of immune cells, the number of both immune cells (100k) and PDOs (25–50) were adjusted. Plated coculture assays were then imaged on the next day.

Live-cell imaging and analysis

All imaging was performed using the Perkin-Elmer UltraVIEW VoX system, with the imaging time indicated in the figures and figure legends. The system is equipped with a high-speed spinning disk (Yokogawa CSU-X1) laser confocal microscope with optimal live-cell imaging capability, photokinesis unit (FRAP, photo activation), six laser lines (405, 440, 488, 514, 561, and 640 nm), high-end CCD camera, and image tiling with fully automated stage. Analysis was performed using Volocity version 6.3 (RRID: SCR_002668). First, Z-stack images were processed with maximum intensity projections for downstream analysis. Thresholds for each fluorophore were set to detect positive signals from corresponding targets (e.g., PDOs, immune cells, and caspase-3/7) by comparing images with paired brightfield images. Colocalization coefficient values (26)—representing the ratio of a colocalized fluorophore over total of the fluorophore— \were generated for individual time points. These values were then exported to MS Excel for further analysis. To quantify cancer–immune cell interactions, the immune cell colocalization coefficient (Immune_co-loc/Immune_total) at each time point was normalized to the coefficient value at initial time point (t = 0), in which both PDOs and immune cells were evenly distributed throughout in the wells. This normalization was done to compensate false positive values from noninteracting cells (i.e., when PDOs and immune cells were simply layered on top of each other). The normalized values were then plotted over the time course of coculture. For caspase-3/7 colocalization coefficient values, the ratio of PDOs colocalized by caspase-3/7 (Organoid_co-loc/Organoid_total) were first normalized to the coefficient values from the non-immune cell group at corresponding time point to account for non-immune cell–mediated caspase-3/7 activity, such as that is induced by starvation. Then the ratio between complete and incomplete apoptosis was multiplied by the coefficient values to quantify the magnitude of immune cell–mediated organoid apoptosis at the terminal time point. Data from multiple images from each of three wells for each condition were obtained for statistical analysis.

RNA-seq

PDOs were collected in Cell Recovery Solution (Corning), and their RNAs were harvested with TRIzol reagent (Thermo Fisher Scientific, 15596026). RNA was extracted from PDOs using the Zymogen Direct-zol RNA kits (Zymo Research, R2062) according to the manufacturer’s instructions. RNA depletion and library preparation was performed using NEBNext^Ⓡ Ultra II Directional RNA Library Prep Kit for Illumina (Illumina, E7765S/L) according to the manufacturer’s protocol. Prepared library was analyzed using the Agilent Bioanalyzer 2100 using high-sensitivity DNA chips (Agilent, 5067-4626). Sequencing was performed using NextSeq500. The index trimmed single-end 75 base pair reads were aligned to the human reference genome using Hisat2 (v2.1.0; RRID: SCR_015530) to generate BAM files. The gene-level count matrix was summarized from bam files using featureCounts. The count matrix was imported into the R Bioconductor (RRID: SCR_006442) DESeq2 package (RRID: SCR_000154) for differential gene expression analysis with a false discover rate <0.1. The differentially expressed genes obtained were then used to create heat maps by applying R package heat-map with scaling and clustering.

DQ-ovalbumin assay

Normal and HGEC PDOs were plated in Matrigel and corresponding media for 1 day. On the next day, the media were replaced with the media containing 20 μg DQ-ovalbumin (DQ-ova) or equivalent volume of PBS (Thermo Fisher Scientific, D12053), and wells were incubated at 37°C for 16 hours. Cells were harvested and analyzed by flow cytometry.

Flow cytometry

PDOs were collected in Cell Recovery Solution (Corning) for 30 minutes and then digested with TrypLE (Gibco) in the presence Y-27632 dihydrochloride as described above until they became single cells. Cells were then stained with the antibodies at 1:200 dilution (Supplementary Table S2) for 30 minutes and washed with PBS + 2% FBS. Prepared samples were then filtered through 40-μm mesh and analyzed by flow cytometry using BD LSRFortessa.

RT-qPCR

For qPCR analysis, PDOs were lysed with TRIzol, and RNA was isolated using the ZymoResearch RNA isolation kit according to the manufacturer’s protocol. RNA was then converted to cDNA using a cDNA synthesis kit (SuperScript IV VILO MasterMix, Invitrogen, 11756050). qRT-PCR was performed using TaqMan Gene Expression Assay (Thermo Fisher Scientific, Supplementary Table S2). PCR reactions were performed in duplicate twice. The expression levels of genes of interest were normalized to a HSP90, and relative expression was calculated. Expression fold expression of target genes was calculated by normalizing the relative expression of experimental groups with corresponding control groups.

Western blot and densitometry

For Western blot analysis, denatured proteins (20 μg) from treated or untreated HGEC PDOs were separated by NuPAGE gels (Thermo Fisher Scientific, NP0322BOX) and transferred to polyvinylidene difluoride membranes (Thermo Fisher Scientific, 88518). Membranes were blocked with 5% nonfat milk in PBST for 1 hour in room temperature and incubated with a rabbit monoclonal antibody (mAb) recognizing total histone H3 (Cell Signaling Technology, 4499, clone: D1H2, RRID: AB_10544537, Supplementary Table S2) or a rabbit mAb recognizing H3K27me3 (Cell Signaling Technology, 9733, clone: C36B11, RRID: AB_2616029 1:1,000 dilution, Supplementary Table S2) at 4°C overnight. After washing with PBST, the blots were incubated with a secondary antibody (donkey anti-rabbit, Invitrogen, A32790 RRID: AB_2762833, 1∶20,000 dilution Supplementary Table S2). The blots were washed again in PBST, and protein bonds were detected using Licor Odyssey Western blot imager. Total histone H3 were used as an equivalent loading control. The band densities were quantified using Image J software (RRID: SCR_003070). Band densities for protein of interest were normalized to that of the band for total histone H3 in the same sample.

Statistical analysis

All experiments were performed in triplicate unless otherwise stated. All samples represent biological replicates. All data were shown as the mean + SEM unless otherwise stated. Statistical significance was defined as P < 0.05. For the statistical difference between two groups, Student t test was used. Multiple comparisons were analyzed using one- or two-way ANOVA for unpaired samples. For matched samples, paired t test was used. All statistical analyses were performed using GraphPad Prism version 8.4.3 (RRID: SCR_002798).

Results

An ancestrally diverse PDO biobank reveals dampened antigen presentation in HGEC

To determine the ancestry composition of our endometrial cancer patient cohort, we first performed RNA-based ancestry prediction (20) on 85 PDOs generated from 44 individuals encompassing various endometrial cancer subtypes (Fig. 1A and B; Supplementary Fig. S1A). Consistent with prior studies reporting a higher prevalence of HGEC among African American patients (6), the majority of HGEC PDOs in our cohort originated from patients of African ancestry (16/27; Supplementary Table S1). Additionally, the distributions of body mass index and microsatellite stability status within our cohort (Fig. 1C) aligned with established findings (27, 28). We also observed that MMR mutational status across ethnic groups is not significantly different (Supplementary Fig. S1B), which is concordant with previous reports (29, 30).

Figure 1.

The antigen presentation and processing are downregulated in HGEC organoids. A, RNA-based ancestry estimation of endometrial cancer (EC) PDOs used in the experiments (n = 50). B, RNA-based ancestry estimation of EC PDOs used in the experiments classified based on the tumor grade. C, Bar plot showing endometrial cancer patients’ microsatellite stability status (n = 50) and associated BMI (cancer; n = 50, benign; n = 8). MSI, microsatellite instability; MSS, microsatellite stability. D, Heatmap generated from RNA-seq data using paired EC PDO lines (low-grade; n = 4, high-grade; n = 7, normal; n = 11). E, Violin plots highlighting MHC-II genes from (D) in HGEC (top, red) and LGEC (bottom, blue) PDOs and their matched normal pairs. F, Steady-state MHC-I, -II genes and PD-L1 mRNA expression levels among matched normal and HGEC PDOs analyzed by RT-PCR (n = 10 for each). Each dot represents a relative mRNA quantity of a target gene normalized to a reference gene, HSP90A from individual PDO line. G, Mean fluorescent intensity (MFI) of immunomodulatory protein expression levels of EC PDOs at steady-state measured by flow cytometry. (normal; n = 7, high-grade; n = 9). Ordinary one-way ANOVA (F and G) was used to calculate statistical significance (*, P < 0.05; **, P < 0.01; ***, P < 0.001).

Tumors reprogram their microenvironment to suppress immunosurveillance by preventing the activation of immune cells involved in tumor clearance such as T cells. A key component of this immune evasion involves the downregulation of MHC molecules, which are crucial for antigen presentation and the subsequent activation of T cells (31). We performed bulk RNA-seq on PDOs, including HGEC PDOs from AFR ancestry, to examine the expression of genes involved in antigen presentation and processing pathways (Supplementary Fig. S1C). These analyses revealed variable expression levels of MHC class I (HLA-A/B, B2M, NLRC5, TAPBP, and PDIA3), MHC class II (HLA-DQ, -DM, -DR, -DO, CD74, CIITA, and CTSB), and immunoproteasome (PSMB, PSME, and HHLA) components across normal endometrial tissues, low-grade endometrial cancer (LGEC), and HGEC PDOs. Given the observed heterogeneity in HGECs, we focused on matched normal–tumor PDOs for more direct comparisons. We noticed reduced expression of immunomodulatory genes (Fig. 1D), including downregulation of key MHC class II genes such as CIITA and HLA-DOB in HGEC PDOs, a pattern not observed in LGEC PDOs (Fig. 1E) along with other genes involved in MHC-II expression (Supplementary Fig. S1D). To validate these findings, we performed RT-qPCR on matched normal/benign and endometrial cancer PDO pairs from a larger cohort, measuring MHC class I and II gene expression at the individual patient level. In LGEC PDOs (4 EUR and 6 AFR), the expression levels of MHC class I and II genes were comparable with those of their matched normal PDOs (Supplementary Fig. S1E). In contrast, HGEC PDOs (1 EUR and 9 AFR) exhibited significantly reduced MHC class I and II gene expression relative to their matched normal counterparts, confirming the RNA-seq results (Fig. 1F).

Previous studies, including our own, have reported that MHC-II pathway genes are downregulated during tumor progression (32). To further assess whether MHC-II pathway is suppressed at protein levels in HGEC, we performed flow cytometry. Concordant with RNA levels, MHC class II protein levels were markedly lower in HGEC PDOs (9 AFR) when compared with matched normal controls (Fig. 1G), an effect not seen in LGEC PDOs (2 EUR and 1 admixed; Supplementary Fig. S1F). Using single-molecule in situ hybridization, we found that MHC-II expression was dampened in primary HGEC tissues and their corresponding PDOs compared with their matched normal counterparts (Supplementary Fig. S1G; AFR).

We next evaluated whether the observed downregulation of MHC class II pathway genes affected the functional capacity to process extracellular antigens in HGEC PDOs. We cultured HGEC and normal control PDOs with DQ-ova, an albumin conjugate that emits fluorescent light upon proteolytic degradation and used as a proxy for antigen processing for the MHC-II pathway (33, 34). HGEC PDOs (6 AFR) demonstrated significantly reduced capacity to process DQ-ova compared with normal controls (Supplementary Fig. S1H), including in one matched tumor–normal pair (Supplementary Fig. S1I). Collectively, these findings indicate that both gene expression levels and the functional capacity of the antigen presentation pathways are diminished in HGEC PDOs, potentially contributing to the immune evasion frequently observed in HGECs.

Establishing an autologous PDO–immune cell coculture platform to test immunomodulatory interventions

IFNγ signaling regulates the expression of both MHC-I and MHC-II antigen presentation pathways in epithelial and cancer cells (32, 35). Traditionally, IFNγ was considered to have antitumor roles due to MHC upregulation, but accumulating studies indicates that chronic exposure to IFNγ could have protumor roles by upregulating immunomodulatory molecules, such as PD-L1 in the cancer cells (36, 37). In addition, IFNγ-driven MHC-II upregulation can either enhance or hinder immune responses, depending on the specific tumor microenvironment (TME; refs. 32, 35), and its significance in HGEC remains unclear. To investigate IFNγ responsiveness in HGEC PDOs, we stimulated both HGEC and normal PDOs with IFNγ and observed a marked increase in MHC-I and MHC-II mRNA (Supplementary Fig. S2A; tumor: 9 AFR and normal: 2 EUR, 1 SAS, and 1 AMR) as well as protein (Supplementary Fig. S2B; tumor: 11 AFR and normal: 1 AFR, 3 EUR, 2 SAS, 1 AMR, and 1 N/A) levels. Next, to determine how IFNγ signaling and IFNγ-induced antigen presentation affect immune surveillance of HGEC, we developed a coculture platform with live-cell imaging to precisely track cancer–immune cell interactions and immune cell–mediated tumor cell death (Fig. 2A). In this system, cancer cells were labeled with red fluorescent dye and immune cells with green fluorescent dye; their colocalization was monitored by confocal microscopy over time. Additionally, we measured immune-mediated apoptosis in real-time using a live-cell imaging dye for caspase-3/7 enzyme activity. We first validated our approach using allogeneic PBMCs, CD8⁺ T cells, and CD4⁺ T cells. Although IFNγ pretreatment alone did not induce apoptosis in HGEC PDOs (Fig. 2B), it significantly promoted immune cell migration toward the cancer cells, resulting in enhanced allogeneic, immune-mediated cytotoxicity (Supplementary Fig. S3A–S3C). Building on these findings, we then evaluated how IFNγ stimulation influences immune surveillance of HGEC PDOs by autologous immune cells from the same patient. Whereas autologous PBMCs interacted with untreated HGEC PDOs, they did not elicit any significant cytotoxicity, consistent with the immune-evasive features of HGEC. Notably, IFNγ pretreatment of HGEC PDOs increased the migration and cytotoxicity of autologous PBMCs, CD8⁺ T cells, and CD4⁺ T cells. Immune cell colocalization analysis demonstrated that, even though immune cells interacted with the PDOs over time in untreated controls, this effect was substantially amplified in IFNγ-stimulated cultures for all three immune cell types (Fig. 2C–E). Similar results were obtained with additional autologous PDO–immune cell pairs (Supplementary Fig. S3D and S3E). Collectively, these proof-of-principle data show that augmenting antigen presentation in HGEC PDOs through IFNγ prestimulation can strengthen cancer–immune cell interactions and enhance immune-mediated cytotoxicity by autologous PBMCs, CD8⁺ T cells, and CD4⁺ T cells.

Figure 2.

IFNγ pretreatment enhances cancer–immune cell interactions. A, A schematic depicting the experimental flow of organoid–immune cell coculture. EC, endometrial cancer. B, Raw colocalization coefficient between PDOs and caspase-3/7 signal from vehicle- and IFNγ-treated group (no immune cells) at the end point (86 hours). C, Representative images of coculture using a HGEC PDO with or without IFNγ prestimulation and autologous PBMC, CD8, and CD4 T cells. Red and green fluorescence represent endometrial organoids and corresponding immune cells, respectively. Blue fluorescence represents active caspase-3 signals. Scale bar, 70 μm. D, Colocalization of PDOs and immune cells over the time of imaging. Colocalization coefficient (R1) was calculated as described previously (26) and normalized to the initial time point. The area under the curve was used for statistical significance (ordinary t test). E, The end point viability of PDOs in the coculture measured by colocalization coefficient between organoid and caspase-3 signals. Colocalization coefficient was calculated as described previously (26) and normalized to no immune cell control (vehicle). Cas3, caspase-3. Ordinary one-way ANOVA (E) was used to calculate statistical significance (**, P < 0.01; ***, P < 0.001; ****, P < 0.0001).

Autologous PDO–immune cell coculture system to assess the therapeutic significance of small-molecule drugs with immunomodulatory potential

Because IFNγ treatment restored MHC expression in HGEC PDOs, this rescue suggested that MHC suppression was not solely due to genetic alterations but could be driven by reversible epigenetic mechanisms. Indeed, tumor-suppressive pathways are frequently silenced through histone modifications in various cancers (38). Specifically, the histone methyltransferase EZH2—which catalyzes H3K27 trimethylation (H3K27me3)—is often upregulated and has been implicated in repressing genes involved in immune surveillance (39, 40). Previous reports indicate that EZH2 inhibition renders cancer cells immunogenic, thus enhancing antitumor immunity (41, 42), but its role in the context of modulating immune surveillance in endometrial cancer is not well established.

To evaluate whether inhibiting EZH2 could relieve a possible epigenetic blockade that restricts MHC expression in HGEC PDOs, we leveraged TAZ, an FDA-approved EZH2 inhibitor (43). We first confirmed that TAZ significantly reduced H3K27me3 levels in HGEC PDOs without affecting H3K27 acetylation (Supplementary Fig. S4A–S4D). Consistent with a role for EZH2 in silencing MHC pathways, TAZ treatment led to a pronounced upregulation of MHC-I and MHC-II—particularly in ARID1A-mutant PDOs (2 AFR and 1 EUR) without affecting the expression of PD-L1 (Supplementary Fig. S4E–S4H). Notably, TAZ pretreatment enhanced tumor–immune cell interactions in our autologous coculture assays, as indicated by greater immune cell colocalization with the PDOs and increased tumor cell apoptosis (Fig. 3A–C; Supplementary Fig. S4I). These findings suggest that epigenetic reprogramming via EZH2 inhibition can restore MHC expression and amplify T cell–mediated cytotoxicity in HGEC. By targeting an epigenetic enzyme already actionable with an FDA-approved inhibitor, this strategy opens promising avenues for combining EZH2 blockade with immunotherapeutic regimens to improve outcomes in endometrial cancer.

Figure 3.

Enhanced MHC expression mediated by EZH2 inhibition in HGEC organoids augments cancer–T cell interactions. A, Representative images of coculture using a HGEC PDO with or without TAZ prestimulation and autologous PBMC, CD8, and CD4 T cells. Scale bar, 70 μm. B, Colocalization of PDOs and immune cells over the time of imaging. Colocalization coefficient (R1) was calculated as described previously (26) and normalized to the initial time point. The area under the curve was used for statistical significance (ordinary t test). C, End point viability of PDOs in autologous coculture measured by colocalization coefficient between organoid and caspase-3 signals. Colocalization coefficient was calculated as described previously (26) and normalized to no immune cell control (vehicle). Cas3, caspase-3. Ordinary one-way ANOVA (C) was used to calculate statistical significance (****, P < 0.0001).

Establishing a PDO–NK cell coculture system

NK cells play an essential role in immunosurveillance and are attractive therapeutic targets due to their robust cytotoxicity against cancer (44, 45). Importantly, tumors that evade T cell–mediated immunity by downregulating MHC-I can still be recognized by NK cells through the “missing self” mechanism (46). To extend the capabilities of our coculture platform, we established a PDO–NK cell coculture system and assessed NK cell–mediated apoptosis in HGEC (Fig. 4A). Colocalization analyses revealed robust NK cell–cancer cell interactions (Fig. 4B) and cytotoxicity (Fig. 4C) in PDOs with low MHC-I expression. However, pretreating the PDOs with IFNγ to elevate MHC-I antigen presentation genes markedly reduced both NK cell colocalization and cytotoxicity. Collectively, these findings underscore the interplay between MHC-I modulation and NK cell cytotoxicity against HGEC PDOs. Moreover, this proof-of-concept PDO–NK cell coculture system provides a valuable framework for investigating and optimizing future NK cell–based immunotherapeutic strategies against HGEC.

Figure 4.

Organoid–NK cell coculture and autologous MMRp and MMRd organoid coculture validate our platform. A, Representative images of allogeneic NK cell coculture using a HGEC PDO line with or without IFNγ prestimulation. Red and green fluorescence represent HGEC PDOs and corresponding immune cells, respectively. Blue fluorescence represents active caspase 3 signals. Scale bar, 70 μm. B, Colocalization of PDOs and immune cells over the time of imaging. The area under the curve was used for statistical significance (ordinary t test). C, The end point viability of PDOs in the coculture measured by colocalization coefficient between organoid and caspase-3 signals. D, Representative images of coculture using MMRd and MMRp HGEC PDOs and autologous PBMCs. Scale bar, 70 μm. E, Quantification of MHC-I, -II, and PD-L1 protein expression levels of the MMRd and MMRp HGEC PDO lines measured by flow cytometry. MFI, mean fluorescent intensity. F, Colocalization of PDOs and immune cells at 14 hours of coculture period. G, Viability of the MMRd PDO in the coculture measured by colocalization coefficient between organoid and caspase-3 signals at 12 and 20 hours coculture period. H, Viability of the MMRp PDO in the coculture measured by colocalization coefficient between organoid and caspase-3 signals at 12 and 20 hours coculture period. Cas3, caspase-3. Ordinary one-way ANOVA (C, G, and H) and t test (F) were used to calculate statistical significance (*, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001).

Assessing the influence of MMR status on autologous PDO–immune cell interactions

Mutations in DNA MMR genes (MSH2, MSH3, MSH6, MLH1, PMS1, and PMS2) lead to the accumulation of DNA mutations, which may increase the neoantigen production. Consequently, MMRd tumors often exhibit higher levels of tumor-infiltrating lymphocytes and generally respond more favorably to anti–PD-1 therapy compared with MMR-proficient (MMRp) tumors (47). To interrogate the influence of MMR status on cancer–immune cell interactions in our autologous system, we utilized MMRp and MMRd HGEC PDOs and their respective PBMCs from our biobank (Fig. 4D). These PDOs had comparable expression levels of MHC-I, MHC-II, and PD-L1 (Fig. 4E). Consistent with previous reports showing that MMRd tumors correlate with enhanced immune infiltration in endometrial cancer (48, 49), MMRd PDOs demonstrated significantly greater immune cell interaction (Fig. 4F) and earlier onset of tumor cell apoptosis (Fig. 4G and H) during the coculture compared with MMRp PDOs. These findings indicate that our autologous system faithfully captures the enhanced immunogenicity observed in MMRd endometrial cancers, which often underlies their improved clinical responses to checkpoint blockade immunotherapy. By reproducing the MMR-dependent variations in tumor–immune dynamics, our platform provides a valuable resource for elucidating mechanisms of immune evasion and evaluating novel immunotherapeutic strategies tailored to the MMR status of HGEC.

Assessing therapeutic efficacy of next-generation bispecific T-cell engagers using autologous PDO–immune cell coculture system

Bispecific T-cell engagers (TCE) have emerged as a promising class of immunotherapies, particularly in hematologic malignancies, by simultaneously binding tumor antigens and the CD3 complex on T cells to induce targeted cytotoxicity. However, their application in solid tumors remains challenging due to the limited availability of truly tumor-specific antigens and the potential for on-target, off-tumor toxicities (50).

To test whether our autologous PDO–immune coculture platform can be used to evaluate the therapeutic efficacy of novel TCE modalities in HGEC, we utilized two component-guided antibody tumor engagers (TwoGATE). TwoGATE is a conditional TCE system based on a half-life extended VH- and VL-stabilized split-paratope design with protease-cleavable polypeptide linker (Supplementary Fig. S5). This design aims to confine T-cell activation to the TME, thereby mitigating systemic toxicity. We engineered constructs that target EpCAM on HGEC PDOs and CD3 on T cells, with linkers selectively cleaved by matriptase, uPA, or CTSB. These proteases are known to be significantly enriched in the TME (51). Using MMRd and MMRp HGEC PDO lines displaying varying Epcam levels (Fig. 5A), we cocultured each line with autologous PBMCs to assess immune-mediated tumor cell killing (Fig. 5B and C).

Figure 5.

Bispecific TCE constructs facilitate cancer–immune cell interactions. A, Quantification of EpCAM protein expression levels of the MMRd and MMRp PDO lines measured by flow cytometry. MFI, mean fluorescent intensity. B, Representative images of autologous coculture using the MMRd HGEC PDO line. Red and green fluorescence represent organoids and corresponding immune cells, respectively. Blue fluorescence represents active caspase 3 signals. Scale bar, 70 μm. C, Representative images of autologous cancer–immune cell coculture using the MMRp HGEC PDO line. Scale bar, 70 μm. D, Colocalization of the MMRd HGEC PDOs and immune cells at 7 hours of coculture period. E, Viability of the MMRd PDO in the coculture measured by colocalization coefficient between organoid and caspase-3 signals at 12 hours coculture period. F, Colocalization of the MMRp HGEC PDOs and immune cells at 42 hours of coculture period. G, Viability of the MMRd organoid in the coculture measured by colocalization coefficient between organoid and caspase-3 signals at 46 hours coculture period. Cas3, caspase-3. t test (A) and ordinary one-way ANOVA (D–G) were used to calculate statistical significance (*, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001).

All three protease-cleavable TwoGATE constructs significantly increased autologous tumor–immune cell colocalization in both MMRd (Fig. 5D) and MMRp (Fig. 5F) HGEC PDOs. This enhanced interaction was accompanied by robust cytotoxicity, evidenced by elevated caspase-3/7 activity in PDOs relative to untreated controls (Fig. 5E and G). In contrast, the noncleavable TwoGATE construct did not induce changes in colocalization or apoptosis, underscoring the necessity of protease-dependent activation of TwoGATE within the TME. Notably, the MMRd PDO exhibited an earlier onset of apoptosis (12 hours) than the MMRp PDO (46 hours), despite the latter’s higher Epcam expression. Taken together, these findings demonstrate the capacity of our autologous PDO–immune cell coculture system to assess the specificity and efficacy of next-generation TCEs in HGEC, thereby providing a valuable preclinical platform for optimizing immunotherapeutic strategies.

Assessing specificity and efficacy of CAR-T cells in HGEC PDOs

CAR-T cells are approved for treating certain types of hematologic malignancies and are now being actively investigated in solid tumors. Epidermal growth factor (EGFR) overexpression has been associated with poor prognosis in endometrial cancer (52, 53). Furthermore, The EGFR tyrosine kinase inhibitor erlotinib has shown therapeutic efficacy in HGEC models with high EGFR expression both in vitro and in vivo (54). Thus, to leverage our PDO–immune cell coculture platform as a preclinical testing tool for evaluating the efficacy of CAR-T cell therapies, we selected EGFR as our target.

We first screened 24 endometrial cancer PDO lines (11 AFR, 8 EUR, 4 SAS, and 1 admixed) from our biobank for their EGFR expressions using qPCR and flow cytometry. We found that EGFR expression levels were variable across the PDO lines (Supplementary Fig. S6A and S6B), with HGEC PDOs overall expressing higher EGFR compared with LGEC and normal PDOs (Supplementary Fig. S6C and S6D). For proof-of-concept experiments, we then selected one HGEC PDO line with high EGFR expression (AFR), one HGEC PDO line with low EGFR expression (admixed), and a normal endometrium PDO line with low EGFR expression (SAS; Supplementary Fig. S6E and S6F). Using these models, we evaluated the safety and efficacy of EGFR-targeting CAR-T cells (55) within our coculture system. Our results demonstrate that the cytotoxic efficacy of EGFR CAR-T cells is closely correlated with the level of EGFR expression in the target cells (Fig. 6A). Specifically, HGEC PDOs with high EGFR levels exhibited enhanced T-cell colocalization (Fig. 6B) and robust CAR-T cell–mediated tumor cell apoptosis (Fig. 6C), whereas HGEC PDOs with low EGFR expression showed markedly reduced responses. Importantly, normal endometrial PDOs cocultured with EGFR CAR-T cells displayed no significant cytotoxicity, confirming both the specificity and safety of the approach. These findings highlight the sensitivity and robustness of our autologous PDO–immune cell coculture platform, which effectively distinguishes therapeutic responses based on target antigen expression. Consequently, our system provides a powerful preclinical tool for optimizing personalized CAR-T cell therapies in advanced or recurrent endometrial cancers, in which effective targeted immunotherapies are urgently needed.

Figure 6.

CAR-T cells demonstrate target-dependent activity in endometrial organoid. A, Representative images of coculture using a normal and two HGEC PDOs (low and high EGFR) without immune cells, with untransduced CD8 T cells (mock), or EGFR-CAR-T cells (55). Red and green fluorescence represent endometrial organoids and corresponding immune cells, respectively. Blue fluorescence represents active caspase-3 signals. Scale bar, 70 μm. B, Colocalization of the PDOs and corresponding immune cells over the time of imaging. Colocalization coefficient (R1) was calculated as described previously (26) and normalized to the initial time point. The area under the curve was used for statistical significance (ordinary t test). C, The end point viability of the PDO in the coculture measured by colocalization coefficient between the PDOs and caspase-3 signals. Colocalization coefficient was calculated as described previously (26) and normalized to no immune cell control (vehicle). Cas3, caspase-3. D, A schematic depicting the utilization of autologous PDO–immune coculture system. MSI-H, microsatellite instability–high; MSS, microsatellite stability. Ordinary one-way ANOVA (C) was used to calculate statistical significance (*, P < 0.05; **, P < 0.01; ***, P < 0.001).

Discussion

Autologous PDO–immune coculture systems represent a significant advancement in personalized oncology, particularly in HGEC, in which treatment options remain limited and patient outcomes are poor (56). In the case of endometrial cancer, most previous PDO studies have focused on LGEC PDOs (10, 11). A few studies have established PDOs from HGCE and described their molecular characteristics, providing useful information for drug development (10, 12). However, these prior PDOs, derived primarily from patients of European ancestry, have limited representation of non-European populations (11, 12). In addition, the characterization of cancer–immune cells interactions (57, 58) is limited due to usage of nonautologous immune cells that elicit alloreactive immune responses (59).

Our platform incorporates HGEC PDOs from a racially diverse patient cohort, including a substantial representation of African ancestry, which is critically underrepresented in available models and datasets yet disproportionately affected by this aggressive disease (60, 61). Moreover, by integrating autologous immune cells into our cultures, we capture the patient-specific tumor–immune interactions and overcome the inherent limitations of allogeneic systems that are confounded by HLA mismatches. We believe that our autologous PDO–immune cell coculture platform for HGEC provides a clinically relevant scalable and sustainable model for studying cancer–immune cell interactions and assessing the efficacy of immunotherapeutic interventions.

First, endometrial cancer PDOs derived from diverse ancestries can be used to better represent steady-state levels of expression of immunologic markers, which may help predict immunotherapeutic responsiveness. Although our cohort includes patients of diverse ancestries, the current study is not powered for ancestry-stratified functional comparisons. We therefore present ancestry as cohort characterization and focus on providing proof-of-concept examples for molecular/functional analyses on HGECs and matched tumor–normal pairs. We identified reduced antigen presentation and processing, particularly MHC-II, among HGEC PDOs which may contribute to immune evasion within the host and poor response to immune checkpoint inhibitors. Next, we found that dampened MHC-I and -II expression can be rescued by stimulating the IFNγ signaling pathways or by inhibiting EZH2, which then promotes autologous immune cell–mediated killing of cancer cells. These results indicate that pharmaceutical agents that augment MHC expression may enhance immune responses in HGEC and highlight that our system can be used to screen novel immunotherapeutic agents that would not have been identified in conventional in vitro assays. Further studies are needed to decipher the causal epigenetic mechanisms that regulate the silencing and rescuing MHC expression in HGEC.

Cancer cell–intrinsic MHC-II plays a dual role in modulating antitumor immunity. On one hand, as we and others have recently shown, high levels of MHC-II in cancer cells facilitate effective antigen presentation to CD4⁺ T cells that promotes antitumor immune surveillance (32, 62–64). On the other hand, we recently demonstrated that, in the context of breast cancer lymph node metastasis, elevated MHC-II expression in the absence of adequate co-stimulatory signals can drive regulatory T-cell responses and inhibit antitumor immunity (35). Our current data using autologous HGEC PDO–immune coculture system reveal that elevating MHC-II expression enhances CD4⁺ T cell–mediated killing of tumor cells. Specifically, when HGEC PDOs with initially low MHC-II levels were treated with IFNγ to restore MHC expression, we observed significantly increased CD4⁺ T-cell engagement and subsequent tumor cell apoptosis. These findings suggest that, in HGEC, MHC-II elevation likely promotes immune surveillance rather than immune evasion. Further studies are needed to investigate whether elevating MHC-II expression in HGEC augments antitumor immunity by stimulating robust CD4⁺ T-cell responses.

Our platform effectively reveals the impact of tumor antigenicity on antitumor immunity and utility to assess the efficacy of a wide range of immunotherapeutic modalities in HGEC. Consistent with their clinical responsiveness to immunotherapy, MMRd HGEC PDOs exhibited an earlier and more pronounced immune cell–mediated apoptosis compared with their MMRp counterparts. We also validated our system using various immunotherapeutic modalities—including NK cell transfer, bispecific TCEs, and EGFR-targeted CAR-T cells—demonstrating that therapeutic efficacy correlates with key tumor characteristics such as target antigen expression levels. These comprehensive validations underscore the robustness and sensitivity of our autologous coculture platform, establishing it as a powerful preclinical tool for optimizing immunotherapy modalities in HGEC. Furthermore, our system can be used to generate tumor-specific autologous T cells from the coculture, which could be harnessed for identifying common HGEC neoantigens.

Although our system offers a robust platform to study cancer–immune cell interactions in endometrial cancer, it does not fully represent other key components of TME such as stromal cells (65) and other regulators of antitumor immunity such as microbiome (66, 67). Expanding our ex vivo coculture system into in vivo models using immunocompromised mice without MHC-I and -II (68), which exhibit minimal graft-versus-host disease, could provide a more comprehensive representation of HGEC’s complex TME. This in vivo autologous PDO–immune cell reconstitution model could be used to assess both safety and efficacy of immunotherapy modalities for HGEC. Equitable translation of PDO-enabled precision approaches will require more robust community engagement and communication. We propose routine monitoring of enrollment across diverse communities and the inclusion of community advisors to ensure these new resources can help close the existing gaps in cancer research and care.

In summary, our findings underscore the promise of autologous PDO–immune cell coculture systems (Fig. 6D) in advancing our understanding of tumor–immune dynamics and guiding the development of novel immunotherapeutic strategies against HGEC. By incorporating diverse patient samples and focusing on critical immunomodulatory pathways, our work lays a robust framework for optimizing therapies that harness the full potential of the immune system to combat this aggressive malignancy, while also addressing key gaps in current preclinical models.

Supplementary Material

Supplementary Figure 1

Endometrial tumor organoids display a heterogenic immunomodulatory gene expression.

Supplementary Figure 2

IFNγ upregulates MHC expression in HGEC PDOs.

Supplementary Figure 3

IFNγ pre-treatment in HGEC PDOs enhances cancer-immune cell interactions in both allogeneic and autologous setting.

Supplementary Figure 4

Sub-lethal Tazemetostat treatment induced histone modification associates to MHC expression in Arid1A mutant HGEC PDOs.

Supplementary Figure 5

Structure of TwoGATETM T cell engager.

Supplementary Figure 6

EC PDOs display heterogeneous EGFR expression.

Supplementary Table 1

Establishment of endometrial cancer patient derived organoids.

Supplementary Table 2

Reagents.

Data Availability

Bulk RNA-seq data can be accessed from Gene Expression Omnibus with the following accession number: GSE307568. All other data generated in this study are available upon request to the corresponding author.

Authors’ Disclosures

M. Frimer reports personal fees from AbbVie, Merck, AstraZeneca, and GSK during the conduct of the study. S. Beyaz reports grants from Revitope Oncology and Caper Labs outside the submitted work. W. Meier, K. Simon and M. Viola were employees of Revitope Oncology during the conduct of study. No disclosures were reported by the other authors.

Authors’ Contributions

C. Chung: Conceptualization, data curation, formal analysis, investigation, methodology, writing–original draft, writing–review and editing. B. Yueh: Data curation, methodology. S. Subhash: Formal analysis, visualization. O. Eskiocak: Data curation, investigation. A. Nizam: Data curation, investigation, methodology, writing–review and editing. M. Viola: Resources. P. Belleau: Formal analysis. A. Kredentser: Data curation, investigation. A. Kapedani: Resources. A. Krasnitz: Formal analysis. K. Simon: Resources. W. Meier: Resources, writing–review and editing. M. Frimer: Resources, writing–review and editing. G.L. Goldberg: Resources, writing–review and editing. S. Beyaz: Conceptualization, resources, supervision, funding acquisition, investigation, writing–review and editing.