MYC and p53 Alterations Cooperate through VEGF Signaling to Repress Cytotoxic T Cell and Immunotherapy Responses in Prostate Cancer

Abstract

Patients with castration-resistant prostate cancer (CRPC) are generally unresponsive to tumor targeted treatments and immunotherapies. Genetic alterations acquired during the evolution of CRPC may impact anti-tumor immunity and immunotherapy responses, which could inform personalized therapeutic strategies. Using our innovative electroporation-based mouse models, we generated distinct genetic subtypes of CRPC found in patients and uncovered unique immune microenvironments. Specifically, mouse and human prostate tumors with MYC amplification and p53 disruption had weak cytotoxic lymphocyte infiltration and an overall dismal prognosis. MYC and p53 cooperated to induce tumor intrinsic secretion of VEGF, which signaled through VEGFR2 expressed on CD8⁺ T cells to directly inhibit T cell migration and effector functions. Targeting VEGF-VEGFR2 signaling in vivo remodeled the immune suppressive prostate tumor microenvironment, leading to CD8⁺ T cell-mediated primary tumor and metastasis growth suppression and significantly increased overall survival in MYC and p53 altered CPRC. VEGFR2 blockade also led to induction of PD-L1 in tumors and produced anti-tumor efficacy in combination with PD-L1 immune checkpoint blockade in multiple preclinical CRPC mouse models. Thus, these results identify a genetic mechanism of immune suppression through VEGF signaling in prostate cancer that can be targeted to reactivate immune and immunotherapy responses in an aggressive subtype of CRPC.

INTRODUCTION

Prostate cancer is the leading cancer afflicting American men, with 1 in 8 males diagnosed with prostate cancer in their lifetime (1). The standard-of-care for advanced prostate cancer is a form of androgen-deprivation therapy (ADT), to which most patients initially respond well (2). However, up to 30% of prostate cancers will relapse with castration-resistant prostate cancer (CRPC), which is no longer responsive to hormone therapy and quickly becomes metastatic (3,4). While great strides have been made to develop next generation androgen receptor (AR) signaling inhibitors (ARSIs) that are now clinically approved (5), they generally offer temporary benefit, and metastatic CRPC (mCRPC) remains intractable. An alternative therapeutic avenue in prostate and other treatment-refractory solid tumor malignancies has been the use of immunotherapies to stimulate immune recognition and clearance of local and disseminated tumor cells. Indeed, Sipuleucel-T (Provenge) was the first cancer vaccine to be approved by the FDA and achieved designation in the setting of mCRPC (6). Still, its effects on overall survival remain marginal at-best for patients (7,8), and other immunotherapy modalities such as anti-CTLA-4 and PD-1/PD-L1 immune checkpoint blockade (ICB) that have been curative in other cancers are generally ineffective in prostate malignancies (9–11). This lack of durable immunotherapy responses is believed to be due to the inherently “cold” tumor microenvironment (TME) of prostate cancer that is devoid of the cytotoxic lymphocytes and enriched in suppressive myeloid cell populations (9,12,13). Thus, it will be critical to understand the mechanisms contributing to the immune suppressive prostate TME in order to design more effective immunotherapy strategies for CRPC.

Large-scale analyzes of patient samples have revealed genomic, molecular, and histological subtype classifications of CRPC (14–20). Though the majority of CRPCs remain AR-dependent through AR amplification or splice variants (21,22), a small subset become an AR-independent form of aggressive variant prostate cancer through acquisition of additional genetic or epigenetic alterations or lineage conversion into neuroendocrine prostate cancer (NEPC) (23,24). Such genetic alterations acquired in CRPC include amplification of oncogenic MYC and MYCN, mutations or deletions in tumor suppressor genes such as TP53, PTEN, RB1, and APC, and perturbations in DNA repair pathways (16,17). Interestingly, recent work has demonstrated that, albeit rare, CRPCs harboring alterations in DNA damage (CDK12) and mismatch repair (MSH2, MLH1) genes resulting in microsatellite instability present with an inflamed TME with increased antigen presentation and better response rates to anti-PD-1 ICB (15,25–27). In contrast, previous studies in prostate and other cancer types revealed a role for more prevalent genetic alterations such as MYC induction and TP53, PTEN, and APC inactivation in promoting the infiltration of myeloid-derived suppressor cells (MDSCs) and macrophages, reducing antigen presentation by tumor cells and dendritic cells (DCs), and suppressing interferon signaling necessary for both innate and adaptive immunity (28–35). As such, understanding how genetic alterations that frequently co-occur in prostate cancer impact the immune landscape could lead not only to better stratification of patients for precision medicine, but also to new therapeutic approaches to treat different subtypes of CRPC.

To model the complex and compound genetic alterations commonly associated with aggressive variant prostate cancer in a rapid and flexible manner, we previously developed electroporation-based non-germline genetically engineered mouse models (EPO-GEMMs), whereby oncogenes can be expressed by transposon-mediated transgenesis and tumor suppressor genes inactivated by CRISPR/Cas9-mediated genomic editing to generate prostate tumors de novo and in situ in adult animals (36,37). These models recapitulate the histological and molecular phenotypes associated with human aggressive variant prostate cancer, and display low AR expression and indifference to castration indicative of CRPC. Given this platform can be used to generate prostate tumors in their resident TME with an intact immune system, we generated a suite of genetically-defined EPO-GEMMs to identify the unique immune landscapes of different genetic subtypes and explore tumor intrinsic mechanisms of immune suppression. In doing so, we uncovered a novel mechanism of immune suppression in a lethal aggressive variant prostate cancer subtype driven by MYC and Tp53 (hereafter p53) co-alterations that presents an actionable target to remodel the “cold” prostate TME and potentiate ICB responses in CRPC.

MATERIALS AND METHODS

Sex as a biological variable

Our study exclusively examined male mice and patient samples because the disease modeled (prostate cancer) is only relevant in males.

Animal studies

All mouse experiments in this study were approved by the University of Massachusetts Chan Medical School Internal Animal Care and Use Committee (IACUC). Mice were maintained under specific pathogen-free conditions, and food and water were provided ad libitum. FVB and C57BL/6 male mice for transplantation models were purchased from Charles River Laboratories (RRID IMSR_CRL:207) and Jackson Laboratory (RRID: IMSR_JAX:005304), respectively.

Electroporation based non-germline genetically engineered mouse models (EPO-GEMMs)

The electroporation procedure was performed as previously described (36). Briefly, 8- to 12-week old WT C57BL/6 male mice were anesthetized with 2–3% isoflurane and a small incision made in the peritoneal cavity near the pelvic region. After locating one of the seminal vesicles and attached anterior lobe, 30 μL of plasmid mix (see specifications below) was injected into an anterior lobe of the prostate using a 27.5 gauge syringe. Tweezer electrodes were then placed around the injection bubble and two pulses of electrical current (60V) given for 35-millisecond lengths at 500-millisecond intervals were then applied using an in vivo electroporator (Nepa Gene NEPA21 Type II Electroporator). After electroporation, the peritoneal cavity was rinsed with 0.5 mL of prewarmed saline. The abdominal wall was then sutured with an absorbable Vicryl suture (Ethicon), and the skin was closed with wound clips (CellPoint Scientific Inc.) Mice were monitored for tumor development by palpation and ultrasound imaging. At study endpoint, prostate tumors were harvested and tissue divided for 10% formalin fixation for immunohistochemistry (IHC) or immunofluorescence (IF) analysis, or single cell suspensions for flow cytometry analysis.

To generate MYC; p53^−/− (MP) EPO-GEMM tumors, 5μg of a pT3-MYC transposon vector (RRID: Addgene_92046), 1μg of Sleeping Beauty transposase (SB13), and 20μg of a pX330 CRISPR/Cas9 vector with an sgRNA targeting the p53 locus (sequence: ACCCTGTCACCGAGACCCC) were injected into the anterior lobe of the prostate. To generate MYC; Pten^−/− (MPten) EPO-GEMM tumors, 5μg of a pT3-MYC transposon vector, 1μg of SB13, and 20μg of a pX330 CRISPR/Cas9 vector with an sgRNA targeting the Pten locus (sequence: GTTTGTGGTCTGCCAGCTAA) were injected into the anterior lobe of the prostate. To generate PtPRb (Pten^−/−;p53^−/−;Rb1^−/−) EPO-GEMM tumors, 20μg each of two pX330 CRISPR/Cas9 vectors, one harboring a sgRNA targeting the p53 locus and another harboring tandem sgRNA sequences targeting Pten and Rb1 (sequence: TGCGCGGGGTCGTCCTCCCG) were injected. The SB13 and pT3-EF1α transposon vector were a gift from Dr. Xin Chen at UCSF and pX330 vector a gift from Feng Zhang at the Broad Institute (Addgene #42230, RRID: Addgene_42230). Genome editing in resulting EPO-GEMM tumors was confirmed by Sanger sequencing.

Cell lines

MP, MPten, and PtPRb murine prostate cancer cell lines were previously generated from EPO-GEMM tumors with these genotypes (36). EPO-GEMM prostate tumors were minced, digested in DMEM containing 3 mg/mL Dispase II (Gibco) and 1 mg/mL Collagenase IV (C5138;Sigma) for 1 hour at 37°C, and then plated on 10-cm culture dishes coated with 100 μg/mL collagen (PureCol; 5005; Advanced Biomatrix). Cells that attached to the plate were passaged at least three times to remove non-tumor cell contaminants before bulk freezing at an early passage (P3–4). Sanger sequencing was performed to confirm that EPO-GEMM cell lines maintained the same genetic alterations as their respective EPO-GEMM tumors. Myc-CaP cells (RRID: CVCL_J703) were obtained from A.M. Mercurio at P4 and frozen down in bulk. All cell lines were cultured for experiments no further than P15 in a humidified incubator at 37°C with 5% CO₂ and grown in DMEM supplemented with 10% FBS and 100 IU/ml penicillin/streptomycin (P/S). For animal transplantation experiments, cells were passaged only once post-thaw before injection into mice. Cells tested negative for mycoplasma by PCR analysis prior to their use for in vitro and in vivo experiments.

Clonogenic assays

Bicalutamide was purchased from Selleck Chemicals (S1190), dissolved in dimethyl sulfoxide (DMSO) to yield 10 mM stock solutions, and stored at −80 °C. EPO-GEMM-derived cell lines were treated with varying concentrations of bicalutamide (or DMSO as a vehicle control) for 7 days, with growth media with or without drugs changed every 3 days. The remaining cells were fixed with methanol (1%) and formaldehyde (1%), stained with 0.5% Crystal Violet and photographed using a digital scanner.

CRISPR-mediated p53 KO in Myc-CaP cells

To knockout (KO) p53, Myc-CaP cells were transiently transfected using Lipofectamine^™ 3000 Transfection Reagent (Thermo Fisher; L3000008) according to manufacturer’s protocol with 20 μg of a pX330 CRISPR/Cas9 construct containing a sgRNA targeting p53 (sequence: ACCCTGTCACCGAGACCCC) or no sgRNA as a control. Cells with p53 deficiency were then selected by treatment with 10μM of the MDM2 inhibitor nutlin-3 (Selleck Chemicals; S1061) for 72 hours. Successful generation of Myc-p53KO cells was confirmed by loss of p53 expression by RT-qPCR analysis.

T cell co-culture assays

To isolate primary murine CD8⁺ T cells, spleens were harvested from male 8–10 week old C57BL/6 (for culturing with EPO-GEMM-derived cell lines) or FVB (for culturing with Myc-CaP-derived cell lines) naïve and tumor-bearing mice and passed through a 70μm cell strainer. Cells were centrifuged at 1500 rpm x 5 minutes before red blood cells were then lysed with ACK lysis buffer (Quality Biological) for 5 minutes. Samples were centrifuged and then resuspended in FACS buffer (PBS supplemented with 2% FBS) before CD8⁺ T cells were isolated by negative selection using a CD8 T cell Isolation Kit according to the manufacturer’s protocol (Miltenyi Biotec; 130–104-075).

For CD8⁺ T cells isolated from non-tumor-bearing mice, cells were incubated in RPMI media supplemented with 10% FBS and stimulated for 1 hour with PMA (20 ng/ml, Sigma-Aldrich), Ionomycin (1 μg/ml, STEMCELL technologies), and monensin (2 μM, Biolegend) in a humidified incubator at 37°C with 5% CO₂. CD8⁺ T cells were then added to a 96-well plate with 5×10³ prostate tumor cells of the same genetic background in triplicate at an effector to target ratio of 10:1 and incubated with or without a VEGFR2 (DC101; 1μg/mL, RRID: AB_1107766) blocking antibody. Some CD8⁺ T cells from C57BL/6 mice were directly exposed to 50ng/mL recombinant murine VEGF from R&D Systems (493-MV-005/CF) in the absence of tumor cell co-culture as a control condition.

For CD8⁺ T cells isolated from MP tumor-bearing C57BL/6 or Myc-p53KO tumor-bearing FVB mice, cells were stimulated for 1 hour with PMA (20 ng/ml, Sigma-Aldrich), Ionomycin (1 μg/ml, STEMCELL technologies), and monensin (2 μM, Biolegend), as well as incubated with a functional grade anti-CD3ε (145–2C11, 1μg/mL, RRID: AB_11150783) antibody to promote their activation and a LIVE/DEAD fixable dead cell stain in Aqua (ThermoFisher; L34957) to identify live cells in a humidified incubator at 37°C with 5% CO₂. In addition, some T cells were pre-treated with a VEGFR2 (DC101; 1μg/mL, RRID: AB_1107766) blocking antibody. CD8⁺ T cells were then spun down, washed, and resuspended in fresh RPMI media with 10% FBS before being added to a 96-well plate with 5×10³ prostate tumor cells of the same genetic background in triplicate at an effector to target ratio of 10:1.

Following 4 hour incubation of murine CD8⁺ T cell and prostate tumor cell co-cultures prepared as described above in a humidified incubator at 37°C with 5% CO₂, cells were trypsinized, resuspended in PBS supplemented with 2% FBS, and stained with cell surface antibodies against CD45 AF700 (30-F11; 1:320, RRID: AB_3172633), CD3 BV650 (17A2; 1:300, RRID: AB_11204249), CD8 FITC (53–6.7; 1:400, RRID: AB_312744), and VEGFR2 PE (AVAS12; 1:200, RRID: AB_1967093) for 30 minutes at 4°C. To assess Granzyme B (GZMB), IFNγ, and TNFα levels in CD8⁺ T cells, intracellular staining was performed using the Foxp3/transcription factor staining buffer set (eBioscience), where cells were fixed, permeabilized, and then stained with GZMB APC (GB11, Biolegend; 1:100, RRID: AB_2294995), IFNγ V450 (XMG1.2, TONBO Biosciences; 1:100, RRID: AB_2621970), and TNFα PE-Cy7 (MP6-XT22, eBioscience; 1:100, RRID: AB_11042728) antibodies. LIVE/DEAD Aqua (ThermoFisher; L34957) was used to distinguish live/dead cells. GZMB, IFNγ, and TNFα positivity was evaluated by gating on CD3⁺CD8⁺ T cells on a FACSymphony A5 flow cytometer and analyzed using FlowJo (TreeStar). Unstained and unstimulated CD8⁺ T cells were used for positive gating confirmation.

T cell migration assays

CD8⁺ T cells were isolated from the spleens of Myc-p53KO tumor-bearing FVB mice as described above for T cell co-culture assays. CD8⁺ T cells were then stimulated for 1 hour with PMA (20 ng/ml, Sigma-Aldrich), Ionomycin (1 μg/ml, STEMCELL technologies), and monensin (2 μM, Biolegend), as well as incubated with a functional grade anti-CD3ε (145–2C11, 1μg/mL, RRID: AB_11150783) antibody to promote their activation and a LIVE/DEAD fixable dead cell stain in Aqua (ThermoFisher; L34957) to identify live cells in a humidified incubator at 37°C with 5% CO₂. In addition, some T cells were pre-treated with a VEGFR2 (DC101; 1μg/mL, RRID: AB_1107766) blocking antibody. CD8⁺ T cells were then spun down and resuspended in serum-free RPMI without 10% FBS. Serum-free tumor cell conditioned media was collected from mouse prostate tumor cells after 72 hours of culturing, filtered through a 0.45μm syringe filter (VWR) to remove cellular debris, and diluted with serum-free media if necessary to normalize media based on prostate tumor cell counts.

A 24-well plate was prepared with 600μL of conditioned media from different cell lines, or normal RPMI with 10% FBS as a control, in triplicate wells before 8.0μm transwell plate inserts (Corning; 3422) were added to each well. CD8⁺ T cells at a concentration of 5×10⁴ per 100μL were then slowly added to the top chamber of the transwell inserts and incubated for 4 hours in a humidified incubator at 37°C with 5% CO₂. Following incubation, transwell inserts were removed and the plate spun down to ensure all cells were at the bottom before wells were imaged. Live cell numbers were evaluated by LIVE/DEAD Aqua staining (ThermoFisher; L34957) and quantified using a Celigo Imaging Cytometer (Nexcelom, RRID:SCR_018808).

Prostate orthotopic transplantation models

2.5×10⁵ MP, 5×10⁵ MPten, or 5×10⁵ PtPRb cells were resuspended in 15μl of Matrigel (Matrigel, BD) diluted 1:1 with cold DMEM/F12 media and transplanted into one anterior lobe of the prostate of 8-week-old C57BL/6 male mice. 1×10⁶ Myc-CaP or Myc-p53KO cells were resuspended in 15μl of Matrigel (Matrigel, BD) diluted 1:1 with cold DMEM/F12 media and transplanted into one anterior lobe of the prostate of 8-week-old FVB male mice. Following anesthetization using 2–3% isoflurane, an incision was made in the peritoneal cavity and the cell suspension was injected into an anterior lobe of the prostate using a Hamilton Syringe. The injection’s success was confirmed by the presence of a fluid bubble without any indications of leakage into the abdominal cavity. The abdominal wall was sutured with an absorbable Vicryl suture (Ethicon), and the skin was closed with wound clips (CellPoint Scientific Inc.). Mice were monitored for tumor development by ultrasound imaging and randomized into treatment groups upon tumor formation based on tumor volume. Following sacrifice, a portion of the prostate tumor tissue was preserved in 10% formalin for fixation, while another portion was used for flow cytometry analysis.

In vivo blocking antibody administration

To assess the impact of VEGFR2 and/or PD-1/PD-L1 antibody blockade on tumor and immune responses and overall animal survival, mice harboring genetically-defined EPO-GEMM or transplanted prostate tumors were randomized based on tumor size into different treatment cohorts and received vehicle (PBS), αVEGFR2 (DC101; 400μg, RRID: AB_1107766), αPD-L1 (10F.9G2; 200μg, RRID:AB_2934050), αPD-1 (RMP1–14; 200μg, RRID:AB_10949053), or combined VEGFR2 and PD-L1 blocking antibodies concurrently by intraperitoneal (i.p.) injection twice per week. To determine the impact of CD8⁺ T cell depletion on tumor progression and animal survival, mice were injected i.p. with an αCD8 (2.43: 200 μg, RRID:AB_1125541) depleting antibody twice per week. Antibodies were purchased from BioXcell and diluted in PBS.

Ultrasound imaging

High-contrast ultrasound imaging was performed on a Vevo 3100 System (RRID:SCR_022152) with a MS250 13- to 24-MHz scanhead (VisualSonics) to stage and quantify prostate tumor burden. Tumor volume was analyzed using Vevo 3100 software, version 5.50.

Flow cytometry

For analysis of MHC-I expression in prostate cancer cell lines cultured in vitro, cells were trypsinized, resuspended in PBS supplemented with 2% FBS, and stained with either an H-2Kb PE antibody (AF6–88.5.5.3, eBioscience; 1:200, RRID:AB_10598797) (for EPO-GEMM derived cell lines) or an H-2Kq Alexa Fluor 647 antibody (KH114, Biolegend; 1:200, RRID:AB_893562) (for Myc-CaP derived lines) for 30 minutes on ice. Flow cytometry was performed on a FACSymphony A5 cytometer (RRID:SCR_022538), and data were analyzed using FlowJo (TreeStar, RRID:SCR_008520).

To prepare single cell suspensions from in vivo tumor samples for flow cytometry analysis, tumors were minced with scissors into small pieces and placed in 5ml of collagenase buffer [1x HBSS w/ calcium and magnesium (GIBCO), 1 mg/ml Collagenase A (Roche) and 0.1 mg/ml DNaseI (DN25; Sigma)]. Samples were then transferred to C tubes and processed using program 37C_multi_A on a gentleMACS Octo dissociator with heaters (Miltenyi Biotec, RRID:SCR_020272). Dissociated tissue was passed through a 70μm cell strainer and centrifuged at 1500 rpm x 5 minutes. Red blood cells were then lysed with ACK lysis buffer (Quality Biological) for 5 minutes and samples were centrifuged and then resuspended in FACS buffer (PBS supplemented with 2% FBS). Samples were incubated with the following antibodies for 30 minutes at 4°C: CD45 AF700 (30-F11; 1:320, RRID: AB_493715), NK1.1 BV605 (PK136; 1:200, RRID: AB_2562273), CD3 BV650 (17A2; 1:300, RRID: AB_11204249), CD8 FITC (53–6.7; 1:400, RRID:AB_312744), CD4 PE-Cy5 (GK1.5; 1:200, RRID: AB_312695), VEGFR2 PE (AVAS12; 1:200, RRID: AB_1967093), CD44 BV785 (IM7; 1:100, RRID:AB_11218802), CD62L PE/Dazzle 594 (MEL-14; 1:200, RRID:AB_2566163), CD69 APC-Cy7 (H1.2F3; 1:200, RRID: AB_10679041), CTLA-4 PerCP/Cy5.5 (UC10–4B9; 1:100, RRID:AB_2564473), PD-1 PE-Cy7 (29F.1A12; 1:200, RRID:AB_10696422), LAG-3 BV711 (C9B7W; 1:100, RRID:AB_2876450), TIM-3 APC (RMT3–23; 1:200, RRID:AB_2561656), PD-L1 BV421 (10F.9G2; 1:200, RRID:AB_10897097), CD103 BV711 (2E7; 1:100, RRID:AB_2686970), CD86 APC-Cy7 (GL-1; 1:200, RRID:AB_2244452), CD206 BV650 (C068C2; 1:300, RRID:AB_2562445), F4/80 APC (BM8; 1:200, RRID:AB_893481), Gr-1 PE/Dazzle 594 (RB6–8C5; 1:200, RRID:AB_2564248), CD11c BV785 (N418; 1:100, RRID:AB_11219204), MHC-II PE (M5/114.15.2; 1:200, RRID:AB_313322) (Biolegend) and CD11b BUV395 (M1/70; 1:1,280) (BD Biosciences, RRID:AB_2721166). DAPI was used to distinguish live/dead cells. Flow cytometry was performed on a FACSymphony A5 cytometer. CD4⁺ and CD8⁺ CD3⁺ T cell, CD3⁻ NK1.1⁺ NK cell, CD11b⁺ F4/80⁺ macrophage, CD11c⁺CD11b⁻MHC-II⁺ and CD11c⁺CD11b⁺MHC-II⁺ dendritic cell, and CD11b⁺Gr-1⁺ MDSC numbers were analyzed using FlowJo (TreeStar).

To analyze FOXP3⁺ T_regs cells and Granzyme B (GZMB), IFNγ, and TNFα levels in NK and T cells, single cell suspensions from tumor tissue were resuspended in RPMI media supplemented with 10% FBS and 100 IU/ml P/S and incubated for 4 hours with PMA (20 ng/ml, Sigma-Aldrich), Ionomycin (1 μg/ml, STEMCELL technologies), and monensin (2 μM, Biolegend) in a humidified incubator at 37°C with 5% CO₂. Cell surface staining was first performed with CD45 AF700 (30-F11; 1:320, RRID: AB_493715), NK1.1 BV605 (PK136; 1:200, RRID: AB_2562273), CD3 BV650 (17A2; 1:300, RRID: AB_11204249), CD8 APC-Cy7 (53–6.7; 1:200, RRID:AB_312752), and CD4 PE-Cy5 (GK1.5; 1:200, RRID: AB_312695) (Biolegend) antibodies. Intracellular staining was then performed using the Foxp3/transcription factor staining buffer set (eBioscience), where cells were fixed, permeabilized, and then stained with GZMB APC (GB11, Biolegend; 1:100, RRID: AB_2294995), IFNγ V450 (XMG1.2, TONBO Biosciences; 1:100, RRID: AB_2621970), FOXP3 PE (FJK-16s, eBioscience; 1:100, RRID:AB_465935), and TNFα PE-Cy7 (MP6-XT22, eBioscience; 1:100, RRID: AB_11042728) antibodies. GZMB, IFNγ, and TNFα positivity was evaluated by gating on CD3⁻NK1.1⁺ NK cells and CD3⁺ CD4⁺ and CD8⁺ T cells and FOXP3 positivity was evaluated on CD4⁺ T cells on a FACSymphony A5 flow cytometer and analyzed using FlowJo (TreeStar) as described above.

Cytokine array

Murine prostate cancer cells were cultured in a humidified incubator at 37°C with 5% CO₂ and grown in fresh DMEM supplemented with 100 IU/ml penicillin/streptomycin (P/S). Conditioned media was then collected after 72 hours of culturing and cells trypsinized and counted using a Countess II cell counter (Invitrogen, RRID:SCR_025370). Media samples were normalized based on cell number by diluting with culture media. 60μl aliquots were analyzed using a multiplex immunoassay (Mouse Cytokine/Chemokine 44-Plex) from Eve Technologies.

RT-qPCR

Total RNA was isolated from mouse prostate cell lines or WT prostate tissue from 8–10 week old C57BL/6 mice using the RNeasy Mini Kit (Qiagen). Complementary DNA (cDNA) was synthesized using the TaqMan reverse transcription reagents (Applied Biosystems) according to the manufacturer’s instructions. Real-time qPCR was performed in triplicate using SYBR Green PCR Master Mix (Applied Biosystems) on the StepOnePlus Real-Time PCR system (Applied Biosystems, RRID:SCR_015805). Gene expression values were calculated using the ΔΔCT method and normalized to Gapdh levels as an endogenous reference gene. Primer sequences are listed in Supplementary Table S1.

Bulk RNA-seq analysis of EPO-GEMM prostate tumors

RNA-seq analysis was performed on bulk prostate tumors from MP, MPten, and PtPRb EPO-GEMM mice as previously described (36). Heatmaps were generated using pheatmap (RRID:SCR_016418). Gene set enrichment analysis (GSEA) was performed using the GSEAPreranked tool (RRID:SCR_003199) against Hallmark gene sets or previously published NE specific (24) or AR responsive (38) genesets.

Immunohistochemistry (IHC)

Murine prostate tissues were fixed overnight in 10% formalin and paraffin embedded. Formalin-fixed, paraffin-embedded (FFPE) blocks were then cut into 5μm sections. Hematoxylin and eosin (H&E) and IHC staining were performed using standard protocols. Sections were de-paraffinized, rehydrated, and boiled in a pressure cooker for 15 minutes in 10mM citrate buffer (pH 6.0) or 10 mM Tris base, 1 mM EDTA, 0.05% Tween 20 buffer (pH 9.0) for antigen retrieval. Endogenous peroxidases were quenched by incubating the slides in 3% hydrogen peroxide for 15 minutes. The sections were then washed 2x with PBS and blocked for 1 hour in 5% bovine serum albumin (BSA) in PBS solution at room temperature. Tissues were incubated overnight at 4°C in primary antibodies at respective dilutions (see Supplementary Table S2). HRP-conjugated secondary antibodies (Vectastain ImmPRESS^®: Rabbit, MP-7401–50; Mouse, MP-7402–15; Goat, MP-7405–15; Rat, MP-7444–15) were then applied for 30 minutes and visualized with DAB (Vector Laboratories; SK-4100). Images were obtained on an Aperio ScanScope (Leica Microsystems, RRID:SCR_018457). For immune cell and blood vessel quantifications, 10–20 high power 20x fields per section were counted and averaged using ImageScope v.12.3.2.8013 software from Leica Microsystems (RRID:SCR_014311).

Immunofluorescence (IF)

Prostate tissue sections were prepared for IF staining using standard protocols as described for IHC. Primary antibodies were incubated overnight at 4°C (see Supplementary Table S2). Secondary Alexa Fluor 488 or 647 dye-conjugated antibodies (Thermo Fisher; 1:150) were then applied for 1 hour at room temperature. Slides were mounted with Prolong Gold Antifade mountant (Prolong Molecular Probes; P36934) after counterstaining with DAPI. Fluorescent images were obtained on a Zeiss Axio Observer 7 microscope (RRID:SCR_023694) and quantified using Fiji (RRID:SCR_002285).

Prostate cancer patient samples

15 human prostate cancer specimens, including 5 of Gleason Score 6, 5 of Gleason Score 7, 2 of Gleason Score 8, and 3 of Gleason Score 9, were obtained from the UMass Center of Clinical and Translational Sciences Biorepository and derived retrospectively from patients undergoing surgery at UMass Memorial Hospital consented under the IRB approved protocol no. H-4721. De-identified FFPE tumor specimens were cut into 5μm sections and IHC staining performed as above (see Supplementary Table S2 for primary antibody information). IHC staining score grading was carried out by clinical pathologist S. Bai in a blinded manner. MYC staining was scored as low (<50% positive staining in tumor area) or high (>50% positive staining in tumor area) and VEGFR2 staining was scored on a 0–2 scale corresponding to low (little to no positive staining), intermediate (staining in some but not all tumor areas), and high (staining in majority of tumor areas). CD8⁺ T cell and NKp46⁺ NK cell numbers were counted by K.C. Murphy in a blinded manner and averaged from 20 high power 20x fields using Fiji (RRID:SCR_002285).

Immunofluorescence-based spatial proteomics analysis

Murine and resected human prostate tumor sections were prepared for IF staining using standard protocols as described for IHC. To perform multiplexed IF staining, primary antibodies (see Supplementary Table S2) were conjugated to fluorophores prior to incubation using the following kits according to the manufacturer’s protocol: VEGFR2 DyLight 488 (abcam; ab201799), CD8 DyLight 550 (abcam; ab201800), CD11c APC (abcam; ab201807), and PD-L1 iFluor Red 750 (abcam; ab312811). Conjugated primary antibodies were diluted 1:100 each in 5% donkey serum in PBS solution and slides incubated overnight at 4°C. Slides were then mounted with Prolong Gold Antifade mountant (Prolong Molecular Probes; P36934) after counterstaining with DAPI. Fluorescent whole tissue scans and tile-merged high power 20X images were obtained using a Leica THUNDER Imager Tissue Microscope (RRID:SCR_026034).

Quantification and cell centroid localization were performed using QuPath software (RRID:SCR_018257). First, images were annotated to exclude the edges of tissues and reduce background or auto-fluorescence. Then, automatic cell detection of DAPI⁺ nuclei was run on the selection, followed by single measurement thresholding and classifying of VEGFR2⁺, CD8⁺, CD11c⁺, and PD-L1⁺ staining. A composite classifier of all markers was generated and applied to all tissues. Finally, all detection measurements per sample were exported and used for subsequent spatial analysis.

Spatial proteomics analysis was performed in R studio using R packages FNN for nearest neighbor analysis and ggplot2 for plotting and visualization (RRID:SCR_014601, RRID:SCR_003005). For K-nearest neighbor analysis, the constant k was limited to 10 maximum, each dataset was proportionally sampled to 5000 cells if larger than, and the ‘kd_tree’ algorithm was used to avoid overload. Following analysis of the centroid X,Y positions and nearest neighbors, dot plots were generated to display the localization and connections of each cell Class using ggplot2. Interaction heatmaps were generated by analyzing the frequency of “Class1” Class to “Class2” Neighbor (and vice versa) and plotted with ggplot2 on a log scale to account for distance variability due to tissue size discrepancies. CD8⁺ T cell distances from the center were calculated by Euclidean distance after determining the center of each tissue from the mean of X and Y centroid values for every detection positive for markers and not. Distance analyses between Classes of cells were done by filtering for the desired Class interactions, normalizing by the number of interactions, and generating box and whisker plots of each interaction distance. Wilcoxon tests were performed to account for effects due to sample size.

The PD-L1 staining score for a given sample was calculated by averaging the intensities of PD-L1⁺ cells in a sample, then multiplying this value by the percent of PD-L1⁺ cells out of the total number of cells identified. The values for each sample were then ranked in a group-blind manner and assigned a value of 0, 1, or 2 for staining intensity based on distribution.

Analysis of liver metastasis and ascites burden

The incidence of metastasis to the liver was determined at study endpoint by analysis of H&E-stained liver sections at 10x magnification on an Aperio ScanScope (Leica Biosystems, RRID:SCR_018457) in a blinded manner by K.C. Murphy. Micrometastases were defined as <100 cells and macrometastases >100 cells. The presence of ascites in prostate tumor-bearing mice was assessed at study endpoint based on mild (<500μL) or full (>500μL) amounts of bloody liquid in the peritoneal cavity.

Human clinical data analysis

CBioPortal.org (RRID:SCR_014555) was used to construct an OncoPrint plotting the frequency of alterations in AR, MYC, P53, PTEN, and RB1 in primary prostate cancer patients from The Cancer Genome Atlas (TCGA) dataset (39) and mCRPC patients from a Stand Up to Cancer (SU2C) dataset (16), and generate a Kaplan-Meier survival curve of mCRPC patients harboring MYC, P53, and/or PTEN alterations from the SU2C dataset.

To analyze established immune signatures in prostate cancer patients with specific genetic alterations, we obtained gene expression dataset from the SU2C dataset (16) and corresponding genomic mutation datasets (40,41) from cBioPortal.org. Samples were categorized into groups based on the genomic status of MYC, TP53, and PTEN. Boxplots and Wilcoxon rank sum test were employed to compare the expression of inflamed/NK/T cell signatures (42–44) between groups using the R packages ggplot2 (RRID:SCR_014601) and ggpubr (RRID:SCR_021139).

For xCell analysis of immune-related transcripts in tumors stratified by high and low expression of KDR (gene encoding VEGFR2), we downloaded the prostate adenocarcinoma expression dataset from The Pan-Cancer Atlas at gdc.cancer.gov. Using the R package xCell (RRID:SCR_026446) (45), we conducted cell type enrichment analysis for 64 immune and stromal cell types. Based on these results, violin plots were generated and Wilcoxon rank sum tests were performed on two distinct groups of samples stratified by the mean expression of KDR, utilizing the R packages ggplot2 (RRID:SCR_014601) and ggpubr (RRID: SCR_021139) for graphic representation and statistical testing. Gene lists used to define total, naïve, memory, and effector memory CD8⁺ T cells are available here (45).

Statistical analysis

Statistical analyses were performed as described in the figure legend for each experiment. The indicated sample size (n) represents biological replicates and measurements were taken from distinct samples. Sample sizes were determined based on previous studies, and no statistical method was used to predetermine numbers. For in vivo studies, mice were randomized based on tumor burden as assessed by ultrasound imaging to achieve equal tumor volume between experimental groups. Sample allocation was performed randomly for in vitro experiments. Scoring of IHC/IF staining in mouse and human tumor samples was performed in a blinded manner. For other experiments, data collection and analysis were not performed blind to the conditions of the experiments. All representative experiments were repeated at least 2–3 times with similar results. All samples that met proper experimental conditions were included in the analysis. Normality was first assessed using the Shapiro-Wilk test or visualization by Q-Q plots. In the case of normal distributions, statistical significance was determined by two-sided unpaired t tests with Welch’s correction or Wilcoxon test when comparing two groups, or one- or two-way analysis of variance (ANOVA) with Tukey’s posttests or Šidák’s multiple comparisons, respectively, when comparing more than two groups. For nonnormal distributions, statistical significance was determined by Mann-Whitney tests when comparing two groups or the Kruskal-Wallis with Dunn’s multiple comparisons test when comparing more than two groups. For Kaplan-Meier survival plots, statistical significance was determined by the log-rank test or multiple pairwise comparisons with Bonferroni’s correction applied for multiple groups. Statistical significance was determined using Prism 10 Software (GraphPad Software, RRID:SCR_002798) or R as indicated. P values ≤ 0.05 were considered significant.*, p < 0.05, **, p < 0.01, ***, p < 0.001, ****, p < 0.0001.

Study Approval

All mouse experiments in this study were approved by the University of Massachusetts Chan Medical School Internal Animal Care and Use Committee (IACUC) under protocol nos. 202000106 and 202200034. Surgically resected prostate cancer patient samples were acquired from the UMass Center of Clinical and Translational Sciences Biorepository and consented under the IRB approved protocol no. H-4721. Written informed consent was received from participants prior to inclusion in the study. Studies were conducted in accordance with U.S. Common Rule (45 CFR 46).

Data Availability

Bulk RNA-seq data generated in this study are deposited in the Gene Expression Omnibus (GEO) (RRID:SCR_005012) database under accession number GSE271975. Previously published bulk RNA-seq data mined in this study were obtained from the GEO database under accession number GSE139340. Publicly available SU2C (16) and TCGA (39) RNA-seq and genomic datasets were obtained from cBioPortal.org. Markdown files of code used for spatial proteomics analysis can be found here: https://github.com/katcmurph/Imaging-Based-Spatial-Analysis. All other data, resources, and reagents not available within the article and its supplementary data files will be made available from the corresponding author upon request.

RESULTS

Prostate cancer EPO-GEMMs exhibit genotype-specific differences in immune landscape

We previously developed electroporation-based non-germline genetically engineered mouse models (EPO-GEMMs) harboring transposition-mediated human c-MYC (MYC) overexpression and CRISPR/Cas9-mediated Pten or p53 disruption that produce lethal and metastatic prostate cancer de novo and in situ in the anterior lobe of the prostate of adult mice with high penetrance (83% and 76%, respectively) (Fig. 1A and B) (36). Given the rapid nature and flexibility to engineer various compound oncogene and tumor suppressor gene alterations in these models, we used this platform to explore the impact of different genetic alterations commonly associated with human CRPC on the immune landscape (Fig. 1C). In addition to MYC-driven prostate cancer EPO-GEMMs harboring compound human MYC overexpression with Pten (hereafter MPten) or p53 (hereafter MP) inactivation we previously characterized (36), we also generated a new EPO-GEMM model defined by CRISPR-mediated disruption of three tumor suppressor genes, Pten, p53, and Rb1 (hereafter PtPRb), in the absence of MYC genetic alterations or upregulation (Fig. 1A; Supplementary Fig. S1A and B). PtPRb EPO-GEMMs developed lethal prostate cancer with a high penetrance (83%) but over a longer latency (median survival of 167 days) compared to MYC-driven MPten and MP genetic subtypes (median survival 74 and 90 days, respectively) (Fig. 1B). Like MP and MPten tumors, PtPRb tumors had low expression of the androgen receptor (AR), the luminal marker CK8, and a broad AR activity geneset as assessed by bulk RNA-seq analysis, as well as were unresponsive to ADT, indicative of a poorly differentiated and aggressive variant prostate cancer likely to be castration-resistant (20,24) (Supplementary Fig. S1C–E). Consistent with the association of these co-alterations with small cell or neuroendocrine (NE) differentiation in human CRPC (23,46), PtPRb tumors also expressed NE markers Synaptophysin (SYP), NEUROD1, and ASCL1, displayed small cell morphology, and were enriched for NE genesets compared to other EPO-GEMM prostate tumor genotypes or WT mouse prostates (Supplementary Fig. S1D and E).

Figure 1. EPO-GEMM prostate cancer models reveal genetically-defined changes in the immune TME.

(A) Schematic of electroporation-based non-germline genetically engineered mouse model (EPO-GEMM) generation and specific oncogene and tumor suppressor gene alterations engineered. Created in BioRender. Murphy, K. (2025) https://BioRender.com/az3f6vv. SB, sleeping beauty transposase. Sg, single guide RNA. (B) Kaplan-Meier survival curves of EPO-GEMMs produced in C57BL/6 mice harboring prostate tumors with indicated genotypes (n = 9–11 mice per group). MP, MYC;p53^−/−. MPten, MYC;Pten^−/−. PtPRb, Pten^−/−;p53^−/−;Rb1^−/−. (C) OncoPrint displaying types and frequencies of genomic alterations in AR, MYC, TP53, PTEN, and RB1 in primary prostate cancer (primary PCa) patients samples from the The Cancer Genome Atlas (TCGA) (39) dataset (n= 478 patient samples) or metastatic CRPC (mCRPC) patient samples from a Stand Up to Cancer (SU2C) dataset (16) (n= 444 patient samples). Generated on cBioPortal.org. (D and E) Representative immunohistochemical (IHC) (D) and immunofluorescence (IF) (E) staining of MP, MPten, and PtPRb EPO-GEMM prostate tumors harvested at endpoint. Arrowheads indicate positive staining for immune cells. Scale bars, 50μm. (F) Quantification of NKp46⁺ NK cells, CD8⁺ T cells, FOXP3⁺ regulatory T cells (Tregs), Arginase1⁺ (ARG1) suppressive myeloid cells, F4/80⁺ macrophages, CD11c⁺ dendritic cells (DCs), and Gr-1⁺ myeloid-derived suppressor cells (MDSCs) per field (n = 3–7 mice per group). (G) IF-based quantification of PD-L1 staining score in MP, MPten, and PtPRb EPO-GEMM prostate tumors (n = 3 mice per group). (H) Representative dot plots of K-nearest neighbor spatial analysis performed following multiplexed-IF staining for CD11c, CD8, and PD-L1 in MP, MPten, and PtPRb prostate tumors. Coordinates, μm. (I) Quantification of average distance of CD8⁺ T cells from the center of MP, MPten, and PtPRb prostate tumors (n = 3 mice per group). (J) Box and whisker plot displaying distances between CD8⁺ T cells and CD11c⁺ DCs in MP, MPten, and PtPRb prostate tumors normalized by number of interactions (n = 3 mice per group). The central line represents the median, the ends of the box the upper and lower quartiles, and whiskers represent the range of non-outlier values. Data represent mean ± SEM. P-values were calculated by one-way analysis of variance (ANOVA) with Tukey’s post-test (F, G and I) or Kruskal-Wallis test (J). NS, not significant, *, p < 0.05, **, p < 0.01, ***, p < 0.001, ****, p < 0.0001. IHC and IF experiments were repeated at least twice and representative images shown.

To determine the contribution of MYC and tumor suppressor gene alterations to the immune suppressive prostate cancer landscape, genetically-defined tumors of similar size harvested from these animals at survival endpoint were subjected to immunophenotyping by immunohistochemistry (IHC) or immunofluorescence (IF) analysis. Despite MPten and MP prostate tumors having the same driver oncogene, they had distinct immune infiltrates defined by their respective tumor suppressor gene alterations. Whereas both MPten and MP prostate tumors had accumulation of F4/80⁺ macrophages and Gr-1⁺ immature myeloid cells/MDSCs that express Arginase 1 (ARG1), MP tumors had reduced numbers of cytotoxic Natural Killer (NK) and CD8⁺ T cells and an increase in regulatory T cells (T_regs) compared to MPten tumors (Fig. 1D–F). Strikingly, PtPRb tumors formed in the absence of MYC induction displayed a significant increase in both cytotoxic NK and CD8⁺ T lymphocytes, as well as F4/80⁺ macrophages and Gr-1⁺ MDSCs expressing ARG1 compared to MP subtypes also harboring p53 alterations (Fig. 1D–F). In addition, PtPRb tumors had elevated numbers of CD11c⁺ dendritic cells (DCs) critical for antigen presentation to and priming of T cells, and expressed higher amounts of PD-L1 that are indicative of a more inflamed TME compared to MYC-driven subtypes (Fig. 1F and G). These genotype-specific immune phenotypes were further validated in prostate tumors from mice transplanted with MP, MPten, and PtPRb cell lines generated from EPO-GEMM tumors, with PtPRb tumors having the largest immune infiltrate and MP tumors displaying reduced numbers of lymphoid and myeloid cells in comparison as assessed by IHC analysis (Supplementary Fig. S1F and G).

To further determine the spatial localization of T cells and DCs within the prostate TME, we performed nearest neighbor analysis following multiplexed IF staining on genetically defined EPO-GEMM transplant tumors (Fig. 1H). CD8⁺ T cells in MYC-driven MP and MPten prostate tumors were not only predominantly confined in the periphery and excluded from the tumor center, but, in MP tumors specifically, were a greater distance from CD11c⁺ DCs as compared to CD8⁺ T cells in PtPRb tumors that robustly infiltrated the tumor and co-localized with DCs (Fig. 1I and J). This spatial analysis suggests that CD8⁺ T cells are likely not primed by DCs and as a result fail to infiltrate the tumor and mediate immune surveillance of MP prostate tumors. Together, these findings demonstrate that MYC, and to an even greater extent compound MYC activation and p53 loss, lead to cytotoxic lymphocyte suppression in prostate cancer.

Compound MYC and p53 alterations are associated with poor outcomes and immune suppression in human CRPC

To evaluate the clinical impact of MYC alterations on immune suppression in human prostate cancer, we first stained primary, surgically resected prostate cancer tumor samples of various Gleason Scores we obtained from the UMass Center for Clinical and Translational Science Biorepository for MYC and markers of NK and CD8⁺ T cells. Consistent with findings in our EPO-GEMM models, human tumors with high MYC expression had significantly fewer CD8⁺ T cell and NK cell infiltrates compared to tumors with low or absent MYC expression as scored by a clinical pathologist (Fig. 2A and B). Further multiplexed IF staining of human tumors showed that CD11c⁺ DC numbers and PD-L1 staining did not change significantly between low and high MYC-expressing patient tumors (Fig. 2C and D; Supplementary Fig. S2A). However, spatial analysis revealed significant exclusion of CD8⁺ T cells from both the tumor center as well as CD11c⁺ DCs in high as compared to low MYC-expressing prostate cancer patient samples (Fig. 2C, E and F).

Figure 2. MYC and p53 co-alterations result in immune suppression and more aggressive disease in human prostate cancer.

(A) Representative IHC staining in surgically resected primary prostate cancer patient samples stratified by MYC staining score into high and low groups. Arrowheads indicate positive staining for immune cells. Scale bars, 50μm. (B) Quantification of CD8⁺ T cells and NKp46⁺ NK cells in prostate cancer patient samples stratified by MYC staining score into high and low groups (n = 5–7 samples per group). (C) Representative dot plots of K-nearest neighbor spatial analysis performed following multiplexed-IF staining for CD11c, CD8, and PD-L1 in low and high MYC-expressing prostate tumors. Coordinates, μm. (D) Quantification of CD11c⁺ DCs in prostate cancer patient samples stratified by MYC staining score into high and low groups (n = 4–5 samples per group). (E) Quantification of average distance of CD8⁺ T cells from the center of low and high MYC prostate tumors (n = 4–5 samples per group). (F) Box and whisker plot displaying distances between CD8⁺ T cells and CD11c⁺ DCs in low and high MYC-expressing prostate tumors normalized by number of interactions (n = 4–5 samples per group). The central line represents the median, the ends of the box the upper and lower quartiles, and whiskers represent the range of non-outlier values. (G) Box and whisker plots showing gene expression analysis of magnitude of immune infiltration [MImmScore (42)], NK cell (43), and CD8⁺ T cell (44) signatures in metastatic CRPC patient samples with MYC or TP53 alterations alone or co-altered from SU2C datasets (16) (n = 10–35 samples per group). The central line represents the median, the ends of the box the upper and lower quartiles, and whiskers represent the range of non-outlier values. (H) Kaplan-Meier survival curves of metastatic CRPC patients harboring prostate tumors with MYC or TP53 alterations alone or co-altered from SU2C datasets (16) (n = 38–143 samples per group). Data represent mean ± SEM. P-values were calculated by two-tailed, unpaired t-test with Welch’s correction (B, D and E), Wilcoxon test (F and G), or multiple pairwise comparisons with Bonferroni’s correction (H). NS, not significant, *, p < 0.05, **, p < 0.01, ***, p < 0.001, ****, p < 0.0001. IHC and IF experiments were repeated at least twice and representative images shown.

To more comprehensively investigate the immune landscape of patients with MYC, P53, and/or PTEN alterations, we analyzed expression of immune-related gene signatures within sequenced metastatic CRPC (mCRPC) samples from the Stand Up to Cancer (SU2C) dataset (16). Whereas patients with MYC or TP53 alterations alone had no differences in the magnitude of immune infiltration [MImmScore (42)] as well as specific CD8⁺ T cell (44) and NK cell (43) gene signature expression, those with compound MYC and P53 alterations had significantly decreased expression of CD8⁺ and NK cell signatures (Fig. 2G). This decrease in immune-related transcripts was not observed in the context of PTEN alterations (Supplementary Fig. S2B). In addition, mCRPC patients harboring MYC and P53 co-alterations had significantly worse overall survival compared to those with MYC amplification or P53 mutation/deletion alone that was not observed with compound MYC and PTEN alterations (Fig. 2H; Supplementary Fig. S2C), substantiating that this genetic subtype marks an aggressive form of CRPC. These data further support our findings in animal models that MYC induction in combination specifically with p53 disruption, which is found in ~8–9% of patients (Fig. 1C), leads to an aggressive and immune suppressed subtype of CRPC.

MYC and p53 alterations combine to repress inflammatory signaling and induce VEGF secretion

We next wanted to determine the tumor cell intrinsic mechanisms by which MYC overexpression and p53 loss cooperate to mediate immune suppression. RNA sequencing (RNA-seq) analysis of bulk prostate tumor samples from genetically defined EPO-GEMM animals revealed that both MP and MPten tumors had reduced expression of antigen presentation and processing genes (B2m, H2-d1, H2-k1, Tap1/2, Erap1) necessary for effective antigen-dependent T cell responses, as well as stimulatory ligands necessary for NK cell engagement (Ulbp1, H60b/c, Raet1d/e) as compared to PtPRb tumors lacking MYC overexpression (Fig. 3A). MP and MPten cell lines propagated from EPO-GEMM tumors also had significantly reduced major histocompatibility complex (MHC) Class I (MHC-I) surface levels in comparison to PtPRb lines (Supplementary Fig. S3A and B). Gene Set Enrichment Analysis (GSEA) of RNA-seq data from primary tumors demonstrated significant enrichment of genes related to inflammatory, NF-κB, and Type I interferon signaling almost exclusively in PtPRb as compared MP and MPten tumors (Fig. 3B; Supplementary Fig. S3C). Interestingly, when comparing between MYC-driven subtypes, we observed reduced enrichment of these inflammatory response gene sets in MP compared to MPten tumors (Fig. 3B), indicating that p53 loss in the context of MYC induction may further suppress inflammatory pathways in prostate tumors that could contribute to an immune suppressed TME.

Figure 3. MYC induction and p53 disruption cooperate to repress inflammatory signaling and stimulate VEGF secretion from prostate tumor cells.

(A) Heatmap of major histocompatibility complex (MHC), antigen presentation, and NK cell ligand gene expression in MP, MPten, and PtPRb EPO-GEMM tumors from bulk RNA-seq analysis (n = 8–17 mice per group). (B) Gene Set Enrichment Analysis (GSEA) of inflammatory, NFκB, and interferon (IFN) signaling gene sets in indicated EPO-GEMM tumors (n = 8–17 mice per group). NES, normalized Enrichment Score. (C) Heatmap of cytokine array analysis results from Myc-CaP (Myc) cells and MPten, MP, and PtPRb EPO-GEMM-derived cell lines. Data shown as mean of 3 biological replicates. (D) Quantification of protein levels of factors differentially secreted in MP compared to other cell lines from cytokine array analysis in C (n = 3 biological replicates per cell line). (E) Cytokine array analysis of differentially secreted proteins in parental Myc-CaP (Myc) cells compared to those with CRISPR-mediated p53 knockout (Myc-p53KO) (n = 3 biological replicates per cell line). (F and G) Representative histograms (F) and quantification (G) of mean fluorescent intensity (MFI) of MHC-I (H-2Kq) expression on Myc-CaP (Myc) and Myc-p53KO tumor cells (n = 3 biological replicates per group). (H) RT-qPCR analysis of antigen presentation genes in Myc-CaP (Myc) and Myc-p53KO cells (n = 2 biological replicates associated with 3 technical replicates per group). A.U., arbitrary units. Data represent mean ± SEM. P-values were calculated by one-way analysis of variance (ANOVA) with Tukey’s post-test (D) or two-tailed, unpaired t-test with Welch’s correction (E and G). NS, not significant, *, p < 0.05, **, p < 0.01, ***, p < 0.001, ****, p < 0.0001. Flow cytometry and RT-qPCR experiments were repeated at least twice with similar results.

To assess differences in the secretory profile of prostate tumors across MYC-driven genotypes, we performed cytokine array analysis on EPO-GEMM-derived MP, MPten, and PtPRb tumor cell lines, as well as the previously characterized Myc-CaP cell line propagated from Hi-Myc mice harboring overexpression of a human MYC transgene downstream of androgen response elements (hereafter referred to as Myc for simplicity) (47). Consistent with our RNA-seq analysis, we observed genotype-specific differences in inflammatory chemokine and cytokine secretion, particularly in the MP subtype (Fig. 3C). Whereas quantities of some secreted factors, such as IFNβ, CX3CL1, and CCL17, were significantly decreased in MP tumor cells, a number of other proteins, including cytokines IL-6 and chemokines CXCL10 and CCL20, were preferentially induced in the MP setting as compared to either Myc, MPten, or even PtPRb cells (Fig. 3C and D). Interestingly, one of the most highly induced factors in MP as compared to other prostate tumor cell lines was VEGF-A (hereafter VEGF), which through binding to its canonical receptor VEGFR2 can have a pleiotropic effects on diverse immune and stromal cell types in the TME (48–50).

To further dissect the contribution of p53 loss specifically to changes in inflammatory signaling, we used CRISPR/Cas9 to knockout p53 in Myc-CaP prostate tumor cells (hereafter Myc-p53KO) (Supplementary Fig. S3D). When transplanted orthotopically into FVB mice, Myc-p53KO cell line-derived prostate tumors had decreased CD8⁺ T cell accumulation in the TME compared to parental Myc-CaP-derived tumors (Supplementary Fig. S3E), recapitulating the limited CD8⁺ T cell infiltration found in MP EPO-GEMM tumors with the same genotype (see Fig. 1, D, F, and I). In vitro, Myc-p53KO tumor cells had reduced secretion of IFNβ and chemokines such as CXCL1, CXCL2, CXCL5, and CX3CL1 important for both myeloid and lymphoid cell chemotaxis into the TME compared to parental Myc-CaP cells (Fig. 3E). P53 deletion in Myc-CaP cells also led to reduced expression MHC-I and antigen presentation/processing genes (Fig. 3F–H). Consistent with MP EPO-GEMM lines, the most differentially upregulated secreted factor in Myc-p53KO lines was VEGF (Fig. 3E). Overall, these data demonstrate that MYC overexpression and p53 loss of function cooperate in a tumor intrinsic manner to not only inhibit inflammatory signaling important for attracting both lymphocytes and myeloid cells and presenting antigen to T cells, but also produce VEGF that could have dynamic effects on the prostate TME.

VEGF-VEGFR2 signaling directly suppresses CD8⁺ T cell effector functions and migration in MP prostate tumors

Given our results showing increased VEGF expression in MP prostate cancer cell lines, we hypothesized that VEGF signaling may contribute to immune suppression in this aggressive subtype. Consistent with our in vitro findings, VEGF protein levels were significantly higher in MP EPO-GEMM primary tumors and Myc-p53KO cell line transplant tumors compared to tumors of other genetic subtypes as assessed by IHC analysis (Fig. 4A and B). Co-immunofluorescence (co-IF) analysis demonstrated that VEGF expression in these prostate cancer models was predominantly localized within MYC⁺ tumor cells as opposed to macrophages that are also known to secrete VEGF in the TME, confirming tumor intrinsic VEGF production (Supplementary Fig. S4A). Tumor-derived VEGF signaling was also associated with increased numbers of CD31⁺ blood vessels in Myc and p53 co-altered tumors (Fig. 4A and C), consistent with the canonical role of VEGF in angiogenesis. However, blood vessels in tumors of the MP genetic subtype were smaller and lacked visible open lumens, indicative of reduced vascular integrity that could contribute indirectly to poor extravasation of immune cells into the TME (51) (Fig. 4A; Supplementary Fig. S4B).

Figure 4. VEGF secretion from MP prostate tumors leads to suppression of the migration and effector functions of VEGFR2-expressing CD8⁺ T cells.

(A) Representative IHC staining of prostate tumors from FVB mice transplanted orthotopically with Myc-CaP (Myc) or Myc-p53KO cells or from C57BL/6 mice transplanted orthotopically with MP, MPten or PtPRb EPO-GEMM-derived cell lines. Arrowheads indicate VEGF positive cells in tumor areas. Scale bars, 50μm. (B and C) Quantification of VEGF⁺ cells (B) and CD31⁺ blood vessels (C) per field in A (n = 3–4 mice per group). (D) Representative co-IF staining for CD8 and VEGFR2 in indicated Myc-CaP or EPO-GEMM cell line-derived transplant prostate tumors. White arrowheads indicate VEGFR2⁺CD8⁺ double positive cells. Scale bars, 50μm. (E) Quantification of percentage of CD8⁺ T cells that are VEGFR2⁺ from co-IF analysis in D (n = 3 mice per group). (F) Schematic of ex vivo tumor-immune co-culture and migration assays using spleen-derived CD8⁺ T cells from prostate tumor-bearing mice and murine prostate cancer cell lines. Created in BioRender. Murphy, K. (2025) https://BioRender.com/uokmixl. (G) Flow cytometry analysis of VEGFR2 expression on spleen-derived CD8⁺ T cells from tumor-free (naïve) mice stimulated with PMA (20 ng/ml), Ionomycin (1 μg/ml), and monensin (2 μM) and cultured ex vivo with indicated prostate cancer cell lines (n = 3 biological replicates per group) or in the presence of recombinant VEGF (50ng/mL). (H and I) Flow cytometry analysis of IFNγ (H) and Granzyme B (GZMB) (I) expression in spleen-derived CD8⁺ T cells from Myc-p53KO or MP prostate tumor-bearing mice stimulated with anti-CD3 (17A2, 1μg/mL), PMA (20 ng/ml), Ionomycin (1 μg/ml), and monensin (2 μM) and pre-treated in the presence or absence of a VEGFR2 blocking antibody (DC101; 1μg/mL) prior to culturing with indicated prostate cancer cell lines (n = 3 biological replicates per group associated with 3 technical replicates). UNSTIM, unstimulated CD8⁺ T cells cultured alone; STIM, stimulated CD8⁺ T cells cultured alone. (J) Quantification of the migration of spleen-derived CD8⁺ T cells from Myc-p53KO prostate tumor-bearing mice stimulated with anti-CD3 (17A2, 1μg/mL), PMA (20 ng/ml), Ionomycin (1 μg/ml), and monensin (2 μM) and pre-treated in the presence or absence of a VEGFR2 blocking antibody (DC101; 1μg/mL) through transwell inserts toward conditioned media from Myc or Myc-p53KO cells or control FBS-containing media (n = 3 biological replicates per group associated with 3 technical replicates). Data represent mean ± SEM. P-values were calculated by one-way analysis of variance (ANOVA) with Tukey’s post-test (B, C, E, and G), or two-way ANOVA with Šidák’s post-test (H-J). NS, not significant, *, p < 0.05, **, p < 0.01, ***, p < 0.001, ****, p < 0.0001. IHC and IF experiments were repeated at least twice and representative images shown.

Recent evidence suggests that VEGF can also directly impact T cell phenotypes in cancer (52–54). Indeed, co-IF analysis revealed that the majority of CD8⁺ T cells in MP tumors, but not in other genetic tumor contexts, expressed the VEGF receptor VEGFR2 (Fig. 4D and E). Flow cytometry analysis further confirmed that a high percentage of CD8⁺ T cells expressed VEGFR2 in transplanted Myc-p53KO tumors, particularly within the naïve subset (Supplementary Fig. S4C and D). In contrast, only a small percentage of CD4⁺ T cells, NK cells, DCs, macrophages, and MDSCs were VEGFR2 positive (Supplementary Fig. S4C and D), suggesting that CD8⁺ T cells may be the predominant immune population responsive to VEGF signaling in this subtype. This prompted us to investigate whether VEGF produced by MP prostate tumor cells could have a direct effect on the functionality of CD8⁺ T cell expressing VEGFR2. To this end, we performed ex vivo tumor-immune co-culture and migration assays with prostate cancer cells and CD8⁺ T cells isolated from spleens of naïve or prostate tumor-bearing FVB and C57BL/6 male mice and stimulated in vitro (Fig. 4F). Similar to our in vivo findings, we observed increased expression of VEGFR2 on CD8⁺ T cells co-cultured with MP and Myc-p53KO tumor cells compared to other genetically-defined prostate cancer lines and even to a similar degree as T cells directly stimulated with recombinant VEGF, supporting a role for VEGF in directly upregulating VEGFR2 expression on CD8⁺ T cells (Fig. 4G; Supplementary Fig. S4E). Only a marginal percentage of prostate tumor cells, on the other hand, expressed VEGFR2, regardless of genotype (Supplementary Fig. S4F).

CD8⁺ T cells co-cultured with MYC and p53 co-altered tumor cells had reduced expression of effector cytokine IFNγ and cytotoxic protease GZMB than those cultured with Myc, MPten, and PtPRb cell lines (Fig. 4H and I; Supplementary Fig. S4G–I). Remarkably, while treatment with a VEGFR2 blocking antibody (α-VEGFR2; DC101) had no impact on T cell phenotypes following co-culture with Myc, MPten, and PtPRb cells, VEGFR2 blockade restored both IFNγ and GZMB expression in CD8⁺ T cells exposed to MP and Myc-p53KO tumor cells to similar levels as those cultured with other genetic prostate cancer subtypes (Fig. 4H and I). CD8⁺ T cells also exhibited significantly reduced migration through a transwell insert when exposed to conditioned media from Myc-p53KO tumor cells compared to control or condition media from Myc tumor cells, which was reversed by VEGFR2 blockade (Fig. 4J). These results demonstrate a direct functional role for VEGF-VEGFR2 signaling in suppression of cytotoxic T cell effector functions and migration specifically in MP altered prostate cancer.

VEGFR2 signaling suppresses CD8⁺ T cell surveillance in MYC-driven human prostate cancer

Analysis of patient samples confirmed a role for VEGFR2 signaling in CD8⁺ T cell suppression in MYC-driven human prostate cancer. IHC and co-IF analysis demonstrated a significantly higher VEGFR2 staining score, as well as percentage of CD8⁺ T cells that were VEGFR2⁺, in MYC^Hi as compared to MYC^Lo prostate cancer patient samples (Fig. 5A and B; Supplementary Fig. S5A and B). Patient tumors with high VEGFR2 staining had fewer CD8⁺ T cell and NK cell infiltrates (Fig. 5C and D). xCell analysis (45) of RNA-seq data from primary prostate tumor patient samples from the Pan-Cancer Atlas further confirmed these observations, where tumors with higher VEGFR2 (KDR) expression had fewer total as well as naïve and memory CD8⁺ T cell transcripts (Supplementary Fig. S5C).

Figure 5. VEGFR2 expression associated with cytotoxic T cell suppression in MYC-altered patient prostate tumors.

(A) Representative co-IF staining of CD8 and VEGFR2 expression in primary prostate cancer patient tumors stratified by MYC staining score into high and low groups. Scale bars, 50μm. (B) Quantification of percentage of CD8⁺ T cells that are VEGFR2⁺ in MYC^hi and MYC^lo patient prostate tumors in A (n = 5–7 samples per group). (C) Representative IHC staining in primary prostate cancer patient samples stratified by VEGFR2 staining score into low, intermediate, and high groups. Arrowheads indicate positive staining for immune cells. Scale bars, 50μm. Representative images of VEGFR2 staining in VEGFR2 low and high samples are the same displayed in Supplementary Fig. S5A for VEGFR2 staining in MYC^hi and MYC^lo samples. (D) Quantification of CD8⁺ T cell and NKp46⁺ NK cell numbers in primary prostate cancer patient samples stratified by VEGFR2 staining score in C (n = 4–6 samples per group). (E) Representative dot plots of K-nearest neighbor spatial analysis performed following multiplexed-IF staining for CD11c, CD8, VEGFR2, and PD-L1 in low and high MYC-expressing prostate tumors. Coordinates, μm. (F) Representative heatmaps of indicated cell-cell interaction frequency from K-nearest neighbor analysis in low and high MYC-expressing prostate tumors. Boxes with an X indicate no interactions between indicated cell Class and Neighbor cell Class. (G) Box and whisker plots displaying distances between CD8⁺ T cells and DCs based on their VEGFR2 and PD-L1 expression in low and high MYC-expressing prostate tumors normalized by number of interactions (n = 4–5 samples per group). The central line represents the median, the ends of the box the upper and lower quartiles, and whiskers represent the range of non-outlier values. Data represent mean ± SEM. P-values were calculated by two-tailed, unpaired t-test with Welch’s correction (B), one-way analysis of variance (ANOVA) with Tukey’s post-test (D) and Kruskal-Wallis test with Dunn’s correction (G). NS, not significant, *, p < 0.05, **, p < 0.01, ***, p < 0.001, ****, p < 0.0001. IHC and IF experiments were repeated at least twice and representative images shown.

VEGFR2 can also be expressed on DCs (see Supplementary Fig. S4C), where it has been shown to limit DC migration, maturation, and function (55–57). Spatial analysis of multiplexed IF stained images revealed not only that a higher percentage of DCs expressed VEGFR2 in MYC^Hi human tumors, but that those VEGFR2⁺ DCs also had elevated positivity for the immune checkpoint PD-L1 (Supplementary Fig. S5D–F). Though VEGFR2⁻ CD8⁺ T cells and DCs were physically more separated in high MYC tumors, VEGFR2⁺ CD8⁺ T cells and DCs co-expressing VEGFR2 and PD-L1 were in significantly closer proximity (Fig. 5E–G), suggestive of a VEGF-VEGFR2 signaling hub suppressive to CD8⁺ T cell priming. Collectively, these data demonstrate that VEGF-VEGFR2 signaling orchestrated by MYC induction and p53 inactivation in prostate cancers of mice and humans can suppress CD8⁺ T cell priming and effector functions.

VEGF signaling blockade reactivates anti-tumor CD8⁺ T cell immunity in MP-driven prostate cancer

As VEGFR2 blockade could enhance CD8⁺ T cell effector functions and migration in vitro, we next asked whether VEGF signaling inhibition could restore anti-tumor T cell immunity in vivo. FVB mice were transplanted orthotopically with Myc-p53KO tumor cells, and upon tumor development, randomized into treatment groups where they received a VEGFR2 blocking antibody (DC101) or vehicle control. Following two-week treatment, prostate tumors were harvested, dissociated into single cell suspensions, and immunophenotyped by flow cytometry (Supplementary Fig. S6). α-VEGFR2 treatment led to a significant increase in both total CD3⁺ T cells, as well as cytotoxic CD8⁺ T cells (Fig. 6A). CD8⁺ T cells in the TME also had significantly higher levels of activation marker CD69, effector cytokines IFNγ and TNF⍺, and cytotoxic GZMB expression following α-VEGFR2 treatment (Fig. 6B). Though CD4⁺ T cell and NK cell numbers did not change upon VEGFR2 blockade, they displayed elevated expression of CD69 and a trend toward increased effector differentiation and cytokine production, respectively (Fig. 6A; Supplementary Fig. S7A–C). Deeper phenotypic characterization of CD8⁺ T cell subsets revealed a shift from naïve to effector CD8⁺ T cells that had reduced expression of exhaustion markers PD-1, CTLA-4, and TIM3 following VEGFR2 blockade (Supplementary Fig. S7D and E).

Figure 6. VEGFR2 neutralization can restore anti-tumor CD8⁺ T cell immunity in MYC and p53 altered prostate cancer.

(A) Flow cytometry analysis of CD3⁺, CD8⁺, and CD4⁺ T cell numbers in Myc-p53KO transplant prostate tumors from mice treated with vehicle or a VEGFR2 blocking antibody (DC101; 400μg) for 2 weeks (n = 8–9 mice per group). (B) Flow cytometry analysis of expression of IFNγ, GZMB, TNFα, and CD69 in CD8⁺ T cells in Myc-p53KO transplant prostate tumors from mice treated as in A (n = 8–9 mice per group). (C and D) Flow cytometry analysis of CD11c⁺MHC-II⁺CD11b⁺ cDC2 (C) and CD11c⁺MHC-II⁺CD11b⁻ cDC1 (D) numbers in Myc-p53KO transplant prostate tumors from mice treated as in A (n = 8–9 mice per group). (E and F) Quantification (E) of mean fluorescent intensity (MFI) of CD86, CD8α, and CD103 and representative histogram (F) of CD103 expression on cDC1s in Myc-p53KO transplant prostate tumors from mice treated as in A (n = 8–9 mice per group). (G) Representative dot plots of K-nearest neighbor spatial analysis performed following multiplexed-IF staining for CD11c, CD8, VEGFR2, and PD-L1 in Myc-p53KO transplant prostate tumors from mice treated as in A. Coordinates, μm. (H) Representative heatmaps of indicated cell-cell interaction frequency from K-nearest neighbor analysis in Myc-p53KO transplant prostate tumors from mice treated as in A. Boxes with an X indicate no interactions between indicated cell Class and Neighbor cell Class. (I and J) Box and whisker plots displaying distances between VEGFR2⁻ CD8⁺ T cells and CD8⍺⁺CD11c⁺ activated DCs (I) and VEGFR2⁺ CD8⁺ T cells and CD11c⁺ DCs (J) in Myc-p53KO transplant prostate tumors from mice treated as in A normalized by number of interactions (n = 3 mice per group). The central line represents the median, the ends of the box the upper and lower quartiles, and whiskers represent the range of non-outlier values. (K) Waterfall plot of response of Myc-p53KO transplant tumors to 2-week treatment with vehicle, α-VEGFR2 (DC101; 400μg), and/or a CD8 neutralizing antibody (2.43; 200μg) (n = 8–14 mice per group). Vehicle and α-VEGFR2 tumor volumes are pooled from two independent replicate experiments. (L) Line plot of Myc-p53KO transplant tumor response to treatment as in K plotted over time as change in rate of tumor growth between treatment groups (n = 8–14 mice per group). Vehicle and α-VEGFR2 tumor volumes are pooled from two independent replicate experiments. (M) Quantification of the percentage of Myc-p53KO prostate tumor-bearing mice with micro- or macrometastases in the liver at endpoint following treatment as in K (n = 8–9 mice per group). (N) Quantification of percentage Myc-p53KO prostate tumor-bearing mice with mild or full ascites at endpoint following treatment as in K (n = 8–9 mice per group). (O) Kaplan-Meier survival curve of Myc-p53KO prostate tumor-bearing mice treated as in K (n = 8–9 mice per group). Data represent mean ± SEM. P-values were calculated by two-tailed, unpaired t-test with Welch’s correction (A-E), Kruskal-Wallis test with Dunn’s correction (K), two-way analysis of variance (ANOVA) with Šidák’s post-test (L-N), Wilcoxon test (I and J), and multiple pairwise comparisons with Bonferroni’s correction (O). NS, not significant, *, p < 0.05, **, p < 0.01, ***, p < 0.001, ****, p < 0.0001. IHC and IF experiments were repeated at least twice and representative images shown. Flow cytometry and tumor and metastasis analysis experiments were repeated at least twice with similar results.

There were significant changes in other immune populations after treatment that likely contributed to the enhanced CD8⁺ T cell activation observed. Suppressive Gr-1⁺ immature myeloid cells/MDSCs and FOXP3⁺ T_regs that inhibit T and NK cell activity were also reduced by VEGFR2 blockade (Supplementary Fig. S7F–H). Moreover, treatment shifted macrophage populations from an immune suppressive CD206⁺ M2 phenotype toward an M1 phenotype characterized by upregulation of MHC-II and co-stimulatory molecule CD86 that contribute to antigen presentation to T cells (Supplementary Fig. S7I). Similarly, VEGFR2 blockade led to increased expression of CD103, CD86, and CD8⍺ on conventional cDC1s indicative of enhanced cross-presentation to CD8⁺ T cells (Fig. 6C–F). Indeed, spatial imaging analysis confirmed that VEGFR2 blockade brought CD8⁺ T cells back into closer proximity with total and activated (CD8⍺⁺) DCs, where antigen-presentation and T cell priming occur, regardless of VEGFR2 or PD-L1 expression (Fig. 6G–J; Supplementary Fig. S7J and K). To investigate whether this remodeling of the tumor-immune landscape following VEGFR2 blockade was specific to MP tumors, we also performed the same experiment using transplanted Myc-CaP parental cells that have the p53 locus intact. We did not observe the same increase in T cell numbers and activation of CD8⁺ T cell responses in Myc-CaP-derived tumors that occurred in Myc-p53KO tumors following VEGFR2 antibody administration (Supplementary Fig. S7L and M). Moreover, there was no change in blood vessel density, lumen structure, or expression of endothelial activation markers such as ICAM-1 and VCAM-1 that are important for T cell extravasation after VEGFR2 antibody treatment of Myc-p53KO tumor-bearing mice, suggesting that the effects of VEGFR2 inhibition on immunity were likely independent of vascular remodeling (Supplementary Fig. S7N–Q). Together these findings suggest that VEGFR2 blockade can prime and activate CD8⁺ T cell effector responses either directly or by enhancing antigen presentation and overcoming immune suppressive barriers in prostate tumors with compound MYC and p53 alterations in vivo.

We further assessed the short- and long-term impact of VEGFR2 blockade on tumor growth, metastasis, and animal survival. Myc-p53KO prostate tumors from α-VEGFR2-treated mice had significantly reduced growth after two-week treatment compared to those from control vehicle-treated mice (Fig. 6K and L). Moreover, α-VEGFR2 treatment resulted in a reduction in visceral metastases to liver, as well as the incidence of ascites that is a common result of metastatic seeding of tumor cells to peritoneum and other organs at endpoint (Fig. 6M and N; Supplementary Fig. S7R). This reduction in primary tumor as well as metastatic burden resulted in significantly enhanced overall survival of prostate tumor-bearing mice treated with VEGFR2 blocking antibodies (Fig. 6O). Finally, to determine if CD8⁺ T cells activated following treatment functionally contributed to the anti-tumor effects of VEGFR2 blockade, we also administered a CD8 depleting antibody (2.43) to some animals. CD8⁺ T cell ablation significantly reduced the survival benefits of α-VEGFR2 blockade and led to an increased tumor and metastatic burden following treatment (Fig. 6K–O; Supplementary Fig. S7R). Thus, VEGF signaling blockade can potentiate CD8⁺ T cell-mediated tumor control and increased survival outcomes in MP-driven CRPC.

VEGFR2 blockade improves anti-PDL-1 ICB efficacy in preclinical CRPC models

The immune suppressive prostate TME devoid of cytotoxic lymphocytes is thought to contribute to de novo resistance to anti-PD1/PD-L1 and anti-CTLA-4 ICB that has been curative in other malignancies (9–11,58,59). Indeed, we found that anti-PD-1 (RMP1–14) regimens had no impact on tumor growth or survival in preclinical MP-driven prostate cancer models (Supplementary Fig. S8A–D). Interestingly, we observed strong induction of PD-L1 expression in tumor areas of Myc-p53KO prostate lesions following VEGFR2 blockade (Fig. 7A). Though a small proportion of myeloid cells such as macrophages, DC, and MDSCs expressed PD-L1 basally in Myc-p53KO tumors, their PD-L1 positivity did not increase after treatment (Supplementary Fig. S8E and F). This increase in activated T cells coupled with induction of PD-L1 on tumor cells following α-VEGFR2 administration provided strong rationale for combining VEGFR2 blocking antibodies with anti-PD-L1 ICB.

Figure 7. Combined VEGFR2 and anti-PD-L1 immune checkpoint blockade produces anti-tumor efficacy in preclinical prostate cancer models.

(A) Representative IHC staining for PD-L1 in Myc-p53KO transplant prostate tumors from FVB mice treated with vehicle or a VEGFR2 blocking antibody (DC101; 400μg) for 2 weeks. Scale bars, 50μm. (B) Representative IHC staining for immune markers in Myc-p53KO transplant prostate tumors harvested at endpoint from mice treated with vehicle, VEGFR2 (V) (DC101; 400μg), and/or PD-L1 (P) (10F.9G2; 200μg) blocking antibodies. Arrowheads indicate positive staining for immune cells. Quantification of NKp46⁺ NK cells, CD8⁺ T cells, and GZMB⁺ cytotoxic lymphocytes per field is shown inset (n = 5 mice per group). Scale bars, 50μm. (C) Kaplan-Meier survival curve of Myc-p53KO prostate tumor-bearing mice treated in B (n = 5–8 mice per group). (D) Quantification of percentage of Myc-p53KO prostate tumor-bearing mice with mild or full ascites at endpoint following treatment as in B (n = 5–8 mice per group). (E) Representative H&E staining of liver metastases (outlined in black) in Myc-p53KO prostate tumor-bearing mice treated as in B. Scale bars, 100μm. (F) Quantification of percentage of Myc-p53KO prostate tumor-bearing mice with micro- or macrometastases in the liver at endpoint following treatment as in B (n = 5–8 mice per group). (G) Waterfall plot of response of autochthonous MP EPO-GEMM prostate tumors to 2-week treatment as in B (n = 5–6 mice per group). (H) Line plot of response of autochthonous MP EPO-GEMM prostate tumors to treatment as in G plotted over time as change in rate of tumor growth between treatment groups (n = 5–6 mice per group). † indicates tumor-bearing mice that reached humane endpoint prior to the 2-week timepoint. (I) Representative H&E staining of liver metastases (outlined in black) in autochthonous prostate tumor-bearing MP EPO-GEMMs treated as in B. Scale bars, 100μm. (J) Quantification of percentage of autochthonous prostate tumor-bearing MP EPO-GEMM animals with micro- or macrometastases in the liver at endpoint following treatment as in B (n = 5–6 mice per group). (K) Kaplan-Meier survival curve of autochthonous prostate tumor-bearing MP EPO-GEMM animals treated as in B (n = 5–6 mice per group). (L) Representative IHC staining for immune markers in autochthonous prostate tumors harvested at endpoint from MP EPO-GEMM mice treated as in B. Arrowheads indicate positive staining for immune cells. Scale bars, 50μm. Data represent mean ± SEM. P-values were calculated by two-way analysis of variance (ANOVA) with Šidák’s post-test (D, F, H, J), one-way ANOVA with Tukey’s post-test (G), and multiple pairwise comparisons with Bonferroni’s correction (C and K). NS, not significant, *, p < 0.05, **, p < 0.01, ***, p < 0.001, ****, p < 0.0001. IHC experiments were repeated at least twice and representative images shown.

We first treated FVB mice harboring transplanted Myc-p53KO prostate tumors with vehicle or antibodies targeting VEGFR2 or PD-L1 (10F.9G2) alone or in combination to assess the impact on tumor and immune responses by IHC analysis. Consistent with its lack of efficacy as a monotherapy in patients, PD-L1 ICB had no impact on NK and CD8⁺ T cell frequencies and cytotoxicity, as well as on numbers of suppressive FOXP3⁺ T_regs in the prostate TME (Fig. 7B; Supplementary Fig. S8G). In contrast, combining PD-L1 with VEGFR2 blockade led to an increase in NK and CD8⁺ T cell accumulation and GZMB expression and reduction in T_regs in the prostate TME compared to not only PD-L1 ICB monotherapy but even to anti-VEGFR2 single agent treatment (Fig. 7B; Supplementary Fig. S8G). Animals were subsequently treated continuously with single or dual antibody treatment to assess the long-term effects on metastatic progression and survival. Whereas single arm α-PD-L1 dosing resulted in comparable animal survival to control vehicle treatment, combined VEGFR2 and PD-L1 blockade achieved robust and significant increases in overall survival compared to either antibody regimen alone (Fig. 7C). This survival advantage following dual VEGFR2/PD-L1 blockade also corresponded with a reduction in the presentation of ascites and metastases to the liver (Fig. 7D–F).

To further validate the preclinical efficacy and immune remodeling capacity of VEGFR2 and PD-L1 antibody treatment, we evaluated these regimens in autochthonous MP EPO-GEMM models. Treatment of prostate tumor-bearing MP EPO-GEMMs with the combination of VEGFR2 and PD-L1 blockade resulted in reduced primary tumor growth and increased areas of tumor necrosis after two weeks of treatment, as well as diminished metastatic spread to the liver, compared to either single treatment alone (Fig. 7G–J; Supplementary Fig. S8H). These effects on primary and metastatic tumors culminated in significantly greater overall survival for MP EPO-GEMMs treated with combined VEGFR2 and PD-L1 blockade compared to those receiving single agent treatment in these aggressive models (Fig. 7K). Importantly, we found that VEGFR2 blockade alone or in combination with PD-L1 ICB had similar effects on remodeling the immune suppressive prostate TME in autochthonous MP EPO-GEMM models as found in transplanted Myc-p53KO models. We observed an increased accumulation of NK cells and CD8⁺ T cells, induction of cytotoxic GZMB expression, and reduction in suppressive T_reg populations within the TME of MP EPO-GEMM tumors treated with α-VEGFR2 alone that was further enhanced by dual VEGFR2 and PD-L1 blockade (Fig. 7L; Supplementary Fig. S8I and J). Collectively, our results demonstrate that VEGFR2 blockade can potentiate the effects of PD-L1 blockade to enhance anti-tumor immunity, extend overall survival, and even block metastatic progression in multiple aggressive and late-stage preclinical models of CRPC.

DISCUSSION

For prostate cancer patients that relapse on hormone therapy and develop CRPC, there are still no durable treatment options. Immune checkpoint blockade (ICB) regimens that can produce curative responses in treatment-refractory melanoma and lung cancer have been generally ineffective in prostate malignancies, owing to their immune suppressed or “cold” tumor microenvironment (TME) that is devoid of cytotoxic T cells (9–11,58,59). Here we used an innovative in vivo electroporation approach to engineer genetic alterations that commonly arise in human CRPC, including amplification of MYC and deletion or mutation of tumor suppressor genes P53, PTEN, and RB1, in mouse models of prostate cancer. While MYC overexpression alone was associated with some reduction in CD8⁺ T cell and NK cell numbers, the most potent immune suppression was observed in human prostate cancers harboring MYC and p53 (MP) co-alterations, which portended significantly worse survival outcomes for CRPC patients. Mechanistically, MYC induction and p53 deficiency cooperated to promote tumor intrinsic secretion of VEGF, which can bind to its cognate receptor VEGFR2 that is expressed substantially on CD8⁺ T cells in MP prostate tumors to inhibit their infiltration, priming, and effector functions. Treatment of MP prostate tumor-bearing mice with VEGFR2 blocking antibodies resulted in CD8⁺ T cell-mediated tumor and metastasis control. VEGF-VEGFR2 signaling inhibition also led to robust PD-L1 upregulation in prostate tumors, and combined VEGFR2 and PD-L1 antibody treatment produced significant anti-tumor T cell responses and survival outcomes in multiple aggressive and late-stage preclinical prostate cancer models resistant to ICB alone. As such, our results unveil VEGF-VEGFR2 signaling as a novel tumor intrinsic mechanism and biomarker of immune suppression in prostate cancer and promising target to potentiate immunotherapy in an aggressive subtype of CRPC lacking in effective treatment options.

MYC has long been described as a driver of oncogenesis, but more recently its role in generating an immunosuppressive TME has begun to surface (60), including in primary prostate cancer (61). Our findings using an innovative in vivo genetic engineering approach further expand on this concept by demonstrating that while MYC induction can indeed suppress inflammatory signaling networks and NK and T cell responses in prostate cancer, this immune suppression is further enhanced in combination with p53 disruption in the CRPC setting. Alterations in MYC and p53 converge to induce expression of VEGF in cancer cells as well as VEGFR2 on T cells that directly inhibits CD8⁺ T cell migration and activation. In contrast to a lung adenocarcinoma study demonstrating that MYC, through induction of CCL9 expression, can recruit macrophages that secrete VEGF (62), we find that VEGF is directly produced by tumor cells rather than macrophages. Still, the molecular mechanisms by which MYC induction and p53 loss cooperate to drive VEGF expression warrant further investigation. Given that interferon signaling important for both MHC-I and PD-L1 expression is suppressed in MYC and p53 co-altered tumors, along with previous studies demonstrating that VEGF can repress the Type I interferon receptor IFNAR1 and its downstream signaling (63,64), it will be of interest to explore the role of IFN signaling modulation in prostate cancer immune responses in future studies. Importantly, the EPO-GEMM platform can be leveraged to generate genetic alterations in prostate tumors in mice with different host immune backgrounds (e.g. NU/NU, Ifnar1⁻/⁻) in order to further dissect the inflammatory signaling mechanisms and tumor-immune interactions contributing to tumor progression and therapy responses.

VEGF has a well-established role in regulating angiogenesis through activating VEGFR2 expressed on endothelial cells that fuels tumor invasion and metastasis (65). In addition, the leaky and poor vascular integrity mediated by chronic VEGF signaling can inhibit effective extravasation of T lymphocytes into tissues (66). Indeed, recent pan-cancer meta-analysis of angiogenic and immune signatures demonstrated that greater than 80% of prostate cancers are associated with an inversely related high angiogenic and low T cell activity score predictive of poor responses to ICB therapy (67). Vascular normalization through administration of low doses of antibodies targeting VEGF or VEGFR2 has been pursued as a strategy to increase immune cell infiltration and function, as well as potentiate immunotherapy responses in various cancer settings (68,69). Here, we find that though VEGF secretion is associated with an increased number of poorly formed blood vessels in tumors with MYC and p53 co-alterations, the vasculature per se does not seem to directly contribute to immune suppression in our model as VEGFR2 antibodies at the administered doses have no effect on blood vessel numbers or vascular phenotypes that could impact immune functions. In contrast, VEGF had a direct effect on the functions of CD8⁺ T cells expressing VEGFR2, leading to repression of their effector functions and migratory capacity that could be reversed by VEGFR2 inhibition. In addition, VEGFR2 blockade significantly diminished the frequencies of immune suppressive regulatory T cells (T_regs) and MDSCs, as well as increased the antigen presentation capacity of macrophages and DCs that can also express VEGFR2. More specifically, VEGFR2 blockade brought CD8⁺ T cells back into closer proximity with DCs where effector T cell priming can occur. This suggests VEGFR2 blockade may target multiple immune suppressive mechanisms to inflame the prostate TME and restore anti-tumor T cell immunity. Still, CD8 depletion did not fully rescue the anti-tumor effects of VEGFR2 blockade, suggesting that other non-immune mechanisms could also contribute to treatment efficacy. In the future, utilization of single cell sequencing approaches, coupled with mouse models where VEGF or VEGFR2 could be conditionally deleted in different immune, tumor, and stromal populations, can be employed to determine the contribution of VEGF/VEGFR2 signaling in different cellular compartments to prostate cancer progression, immune suppression, and immunotherapy outcomes.

VEGF signaling was first identified as a potential therapeutic target for solid cancer types over fifty years ago, and many attempts have been made since to block this pro-tumorigenic axis (70). In prostate cancer, despite VEGF expression being associated with disease progression and stage, neither VEGF neutralization with antibodies such as bevacizumab nor VEGFR2 inhibition through use of multi-receptor tyrosine kinase (RTK) inhibitors has led to a significant benefit in overall survival in combination with standard chemotherapy in phase III clinical trials in mCRPC patients (71,72). As VEGF expression has been shown to be androgen regulated (73,74), one consideration could be to combine VEGFR2 blockade with AR signaling inhibitors as a means to synergistically block VEGF/VEGFR2 signaling. Our results indicate that the ~8–9% of CRPC patients harboring co-alterations in MYC and p53 may particularly benefit from clinically approved VEGF (e.g. bevacizumab) and VEGFR2 (e.g. Sunitinib) targeting therapies despite overall treatment failure in the broader population. Moreover, though this patient population presents with a severely immune suppressed TME and de novo resistance to PD-1/PD-L1 ICB, we find that VEGFR2 blockade induces robust PD-L1 expression and sensitivity to anti-PD-L1 ICB regimens in combination. Indeed, atezolizumab (anti-PD-L1) and bevacizumab (anti-VEGF) combinatorial therapy was recently FDA-approved for hepatocellular carcinoma (HCC), and preclinical studies demonstrate its effectiveness in a MYC-driven mouse model of HCC (75). Collectively, our results pave a clear translational path for the implementation of VEGFR2 and PD-L1 blocking antibodies clinically approved in other malignancies for the treatment of an aggressive CRPC subtype driven by MYC and p53 alterations. More broadly, similar approaches could be taken to implement targeted immunotherapy regimens based on the genetics of a tumor in prostate and other cancer types for “precision immunotherapy”.

Statement of Significance.

VEGFR2 blockade inhibits VEGF-mediated T cell suppression and potentiates the effects of PD-L1 immune checkpoint blockade to treat castration-resistant prostate cancer driven by MYC and p53 alterations.