Article Text

Original research
Cancer cell genotype associated tumor immune microenvironment exhibits differential response to therapeutic STING pathway activation in high-grade serous ovarian cancer
  1. Noor Shakfa1,2,
  2. Deyang Li1,2,
  3. Gwenaelle Conseil1,2,
  4. Elizabeth D Lightbody3,
  5. Juliette Wilson-Sanchez1,2,
  6. Ali Hamade1,2,
  7. Stephen Chenard1,2,
  8. Natasha A. Jawa4,
  9. Brian J. Laight2,5,
  10. Afrakoma Afriyie-Asante6,
  11. Kathrin Tyryshkin7,
  12. Martin Koebel8 and
  13. Madhuri Koti1,2
  1. 1Biomedical and Molecular Sciences, Queen's University, Kingston, Ontario, Canada
  2. 2Queen's Cancer Research Institute, Queen's University, Kingston, Ontario, Canada
  3. 3Dana-Farber Cancer Institute, Harvard University, Cambridge, Massachusetts, USA
  4. 4Centre for Neuroscience Studies & School of Medicine, Queen's University, Kingston, Ontario, Canada
  5. 5Pathology and Molecular Medicine, Queen's University Cancer Research Institute, Kingston, Ontario, Canada
  6. 6Biochemistry, Microbiology and Immunology, University of Ottawa, Ottawa, Ontario, Canada
  7. 7Pathology and Molecular Medicine, Queen's University, Kingston, Ontario, Canada
  8. 8Pathology and Laboratory Medicine, University of Calgary, Calgary, Alberta, Canada
  1. Correspondence to Dr Madhuri Koti; kotim{at}


Background High-grade serous ovarian carcinoma (HGSC) is the most lethal gynecologic malignancy characterized by resistance to chemotherapy and high rates of recurrence. HGSC tumors display a high prevalence of tumor suppressor gene loss. Given the type 1 interferon regulatory function of BRCA1 and PTENgenes and their associated contrasting T-cell infiltrated and non-infiltrated tumor immune microenvironment (TIME) states, respectively, in this study we investigated the potential of stimulator of interferon genes (STING) pathway activation in improving overall survival via enhancing chemotherapy response, specifically in tumors with PTEN deficiency.

Methods Expression of PTEN protein was evaluated in tissue microarrays generated using pretreatment tumors collected from a cohort of 110 patients with HGSC. Multiplex immunofluorescence staining was performed to determine spatial profiles and density of selected lymphoid and myeloid cells. In vivo studies using the syngeneic murine HGSC cell lines, ID8-Trp53–/–; Pten–/– and ID8-Trp53–/–; Brca1–/–, were conducted to characterize the TIME and response to carboplatin chemotherapy in combination with exogenous STING activation therapy.

Results Patient tumors with absence of PTEN protein exhibited a significantly decreased disease specific survival and intraepithelial CD68+ macrophage infiltration as compared with intact PTEN expression. In vivo studies demonstrated that Pten-deficient ovarian cancer cells establish an immunosuppressed TIME characterized by increased proportions of M2-like macrophages, GR1+MDSCs in the ascites, and reduced effector CD8+ cytotoxic T-cell function compared with Brca1-deficient cells; further, tumors from mice injected with Pten-deficient ID8 cells exhibited an aggressive behavior due to suppressive macrophage dominance in the malignant ascites. In combination with chemotherapy, exogenous STING activation resulted in longer overall survival in mice injected with Pten-deficient ID8 cells, reprogrammed intraperitoneal M2-like macrophages derived from Pten-deficient ascites to M1-like phenotype and rescued CD8+ cytotoxic T-cell activation.

Conclusions This study reveals the importance of considering the influence of cancer cell intrinsic genetic alterations on the TIME for therapeutic selection. We establish the rationale for the optimal incorporation of interferon activating therapies as a novel combination strategy in PTEN-deficient HGSC.

  • Immunotherapy
  • Interferon Inducers
  • Tumor Microenvironment
  • Immunomodulation
  • Genital Neoplasms, Female

Data availability statement

Data are available upon reasonable request.

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


  • Studies to-date have shown high-grade serous ovarian cancer (HGSC) is both genetically and immunologically heterogenous and demonstrates poor response to immune checkpoint blockade therapies.


  • We report that cancer cell genotype influences the tumor immune microenvironment (TIME) and disease progression, where cancer cell intrinsic loss of BRCA1 is associated with an active and immune infiltrated TIME in contrast to tumors with cancer cell intrinsic loss of PTEN . PTEN deficiency correlates with increased macrophage recruitment to the TIME in both humans and mice. Ascites fluid from mice injected with PTEN-deficient ID8 ovarian cancer cells show a remarkable increase in M2-like macrophages. This is further supported by in vitro experiments that demonstrate the influence of PTEN-deficient cell-derived factors in inducing macrophage polarization. Although PTEN deficiency results in decreased sensitivity to carboplatin alone, we reveal exogenous stimulator of interferon genes pathway activation potentiates response to carboplatin chemotherapy, prolongs overall survival, repolarizes M2-like suppressive macrophages and rescues T-cell activation in the PTEN-deficient TIME.


  • We establish the rationale for optimal incorporation of interferon activating therapies in combination strategies to treat PTEN-deficient HGSC.


High-grade serous ovarian carcinoma (HGSC) is the most common and aggressive histological subtype of epithelial ovarian cancer. Patients with HGSC often present with advanced disease, generally managed with debulking surgery followed by platinum and taxane-based combination chemotherapy.1 Although most patients show initial sensitivity to chemotherapy, over 80% relapse and subsequently develop resistance to platinum.1 Despite the success of poly-ADP ribose polymerase (PARP) inhibitors in a subpopulation of patients with HGSC (homologous recombination deficient and BRCA-mutated), newer treatment options such as anti-angiogenic agents and immune checkpoint inhibitors have demonstrated modest or no improvements in HGSC.2 3 While significant advancements have been made to delineate pretreatment tumor immune states that associate with differential treatment outcomes, their therapeutic vulnerabilities remain to be fully exploited. Our previous reports on the pretreatment tumor immune microenvironment (TIME) states in patients with HGSC have demonstrated that higher expression of type I interferon genes (IFN-1) associates with increased response to chemotherapy.4 5

Recent evidence further suggests that the evolution of distinct TIME states and downstream responses can be driven by cancer cell intrinsic alterations in tumor suppressor genes that regulate cellular IFN-1 pathways.6 While mutations in the tumor suppressor gene, TP53, are universal and a characteristic feature of HGSC, additional mutations in genes associated with DNA damage repair (DDR) pathways are present in approximately 50% of HGSC cases.7 Among the DDR genes, mutations in breast cancer type I susceptibility protein 1 and 2 (BRCA1/BRCA2) have been widely reported to correlate with increased chemosensitivity.8 9 Interestingly, tumors with DDR deficiency also exhibit high infiltration of CD8+ T cells.10 In contrast to the chemosensitive behavior of tumors with BRCA1 mutations, those with mutations in the phosphatase and tensin homolog (PTEN) gene (present in ~10% of HGSC cases) frequently exhibit a chemoresistant profile.7 In other solid tumors, PTEN deficiency via mutations, copy number loss, or epigenetic silencing is also associated with a decreased infiltration of immune cells in the tumor microenvironment.11 This can potentially be attributed to PTEN deficiency associated cytokine signaling that stimulates an immunosuppressive microenvironment.11 Importantly, both BRCA and PTEN genes have a regulatory function in activation of IFN-1 pathways,12 13 wherein BRCA loss confers constitutive activation of cyclic GMP–AMP synthase (cGAS) stimulating an IFN-1 response through the stimulator of interferon genes (STING) pathway and PTEN controls the import of interferon regulatory factor 3 (IRF3), a master transcription factor responsible for IFN-ß production. While therapies such as PARP inhibitors have shown promise in BRCA1/2 and other homologous recombination (HR) deficient immune infiltrated HGSC tumors,14 15 there remains an unmet need for those patients who exhibit an underactive immune state that often accompanies a chemoresistant tumor phenotype. In the context of chemoresistance associated genetic alterations, reports in melanoma and glioma have shown that PTEN loss associated PI3K/AKT pathway activation plays a crucial role via altering macrophage activation within the TIME towards a tumor promoting functional state.16 17

We have previously shown that therapeutic activation of STING pathway leads to production of chemokines such as CXCL10 and CCL5 which recruit cytotoxic CD8+T cells to the TIME, enhances response to carboplatin chemotherapy, sensitizes to programmed cell death protein-1 (PD-1) immune checkpoint blockade and overall survival (OS) in the ID8 murine model of HGSC.18 Here, we aimed to investigate whether cancer cell intrinsic loss of Pten or Brca1 influence the TIME, response to chemotherapy, and survival outcomes. We further provide evidence supporting the benefit of exogenously activating the STING pathway in the context of BRCA1 or PTEN protein deficiency. Findings from this study demonstrate that although therapeutic STING pathway activation leads to improved response to chemotherapy and improved OS, PTEN-deficient tumors with a suppressed pretreatment TIME state exhibit a more pronounced benefit through reinvigoration of the TIME.


Patient specimens

Archival tumor specimens from 110 patients with HGSC were accessed under the institutional health research ethics board approval (University of Calgary, Alberta, Canada). Tissue microarrays (TMAs) were constructed from formalin-fixed paraffin-embedded tumors following histopathological confirmation of HGSC by a pathologist (MKoe). A total of 110 chemo naïve tumors were used for subsequent analysis. The clinicopathological parameters of the cohort are provided in online supplemental table 1. Duplicate or triplicate 0.6 mm cores were extracted from the areas of interest of each tumor specimen and embedded into a recipient block of paraffin for TMA construction.

Supplemental material

Immunostaining of HGSC TMA

Four μm TMA sections were stained with PTEN rabbit monoclonal antibody (1/50 dilution; clone D4.3 XP; Cell Signaling Technologies, Massachusetts, USA). Multiplex immunofluorescence staining of TMA sections was performed using antibodies specific to CD8, CD68, and pan-cytokeratin (Molecular and Cellular Immunology Core, BC Cancer Research, British Columbia, Canada). Slides were scanned using Vectra Polaris (MOTiF) Multispectral Imaging System (Akoya Biosciences, Massachusetts, USA). Individual tumor cores on the scanned TMA cores were segmented and annotated for stromal versus epithelial compartments using the HALO Link image analysis Software (Indica Labs, New Mexico, USA). Algorithms to detect markers of interest were generated on HALO Link outputting both a total cell number using 4',6-Diamidino-2-Phenylindole, Dihydrochloride (DAPI) staining and total cells positively stained with markers of interest within tissue compartments for PTEN-deficient and PTEN-intact tumors. Patient clinical data from samples within TMAs were accessed for association with clinical parameters.

Murine ovarian cancer cell lines

The syngeneic murine ovarian cancer cell lines ID8-Trp53–/–; Pten–/– and ID8-Trp53–/–; Brca1–/–, were kindly provided by Dr Ian McNeish (Imperial College, UK; online supplemental figure 2C). Both cell lines were maintained in Dulbecco’s Modified Eagle’s Medium (Sigma-Aldrich, Ontario, Canada) supplemented with 4% heat-inactivated fetal bovine serum (FBS) (Sigma-Aldrich, Ontario, Canada), 1% penicillin–streptomycin (100 µg/mL), and 1% insulin transferrin sodium selenite liquid media supplement (Sigma-Aldrich, Ontario, Canada) and incubated at 37°C/5% CO2. Cells for bioluminescent imaging were fluorescently tagged with firefly luciferase by viral transduction using previously established protocols.19

Macrophage migration assay

IC21 murine peritoneal macrophage cell line was obtained from American Type Culture Collection (Manassas, Virginia, USA) and cultured in RPMI 1640 (Sigma-Aldrich, Ontario, Canada) supplemented with 10% FBS and 1% penicillin/streptomycin. ID8 Trp53–/–; Pten–/– and ID8 Trp53–/–; Brca1–/– cells were resuspended at a density of 5×105 in 4 mL complete growth media and seeded in 6-well plates. Conditioned media from both lines were collected 72 hours post seeding. Migration assays were performed in uncoated 24-well transwell plates with 8 µm pore inserts and 6.5 mm in diameter (Corning, New York, USA). IC21 cells were added at a cell concentration of 5×105 in 100 µL of serum-free media into the upper chamber and allowed to migrate through the insert membrane for 16 hours in a 37°C/5% CO2 atmosphere. Conditioned media (600 µL) from either ID8 derivative cell line was placed into the lower chambers. Recombinant murine macrophage chemoattractant protein 1 (MCP-1) (20 ng/mL; PeproTech, New Jersey, USA) was used as a positive control. Cells which migrated onto the transwell inserts were fixed with 4% paraformaldehyde (PFA), stained with DAPI (1 µg/mL; Sigma-Aldrich, Ontario, Canada) and mounted onto a positively charged slide. Inserts were imaged using EVOS Cell Imaging System and quantified using ImageJ software.

Immunofluorescence staining to detect cytosolic DNA in ID8 cells

Immunofluorescence staining was performed to assess the constitutive cytosolic double stranded DNA (dsDNA) expression in untreated cells. ID8-Trp53–/–; Pten–/– or ID8-Trp53–/–; Brca1–/– cells were seeded at a density of 5×105 on coverslips placed in a 6-well plate and left to adhere overnight in complete growth media at 37°C/5% CO2. Coverslips were washed with phosphate-buffered saline (PBS) and fixed for 10 min with 4% PFA. Following permeabilization of cells (0.2% Triton X-100 in 1× PBS) and blocking (0.1% Triton X-100, 1% bovine serum albumin in 1× PBS) at room temperature (RT), cells were incubated in anti-dsDNA monoclonal antibody (1:100; EMD Millipore, Ontario, Canada) overnight at 4°C. Coverslips were washed with blocking solution three times at RT and incubated in blocking buffer containing Alexa Fluor 488 anti-mouse IgG antibody (1:300; Invitrogen, Massachusetts, USA) for 1 hour at RT in the dark. After counterstaining with DAPI for 10 min, coverslips were rinsed with PBS and mounted with antifading fluorescence medium (Invitrogen, Massachusetts, USA) onto a slide. Slides were imaged using EVOS Cell Imaging System and quantified using ImageJ software.

In vivo studies

All in vivo procedures performed were approved by the University Animal Care Committee at Queen’s University (ID number: 2018–1871). A total of 5–6×106 ID8-Trp53–/–; Pten–/– or ID8-Trp53–/–; Brca1–/– cells in PBS were injected via intraperitoneal (i.p.) route in 8–10 week old female C57BL/6 mice (Charles River Laboratories, Quebec, Canada). All mice were maintained under specific pathogen-free conditions. Treatments were initiated 3 weeks post-cancer cell injections. Mice were randomized into three treatment groups (n=10–15 per group): vehicle (PBS), carboplatin, or carboplatin+STING agonist (2’3’=c-di-AM(PS)2 (Rp, Rp); InvivoGen, California, USA). Carboplatin was used at a dose of 10 mg/kg two times per week for four consecutive weeks and STING agonist was used at a dose of 4 mg/kg one time per week for three doses. To determine the effect of cancer cell intrinsic versus immune cell intrinsic STING activation in vivo, 5–6×106 ID8-Trp53–/–; Pten–/– or ID8-Trp53–/–; Brca1–/– cells in PBS were injected via i.p. route in 8–10 week old female C57BL/6J-Sting1gt/J (STING-KO) mice (Jackson Laboratories International, Connecticut, USA) and profiled at endpoint (when abdominal diameter of mice reached ~34 mm).

Multiplex cytokine analysis of post-treatment plasma and ascites fluid

Ascites fluid and plasma samples collected 24 h following initial STING agonist dose or post carboplatin chemotherapy alone from each treatment group were subjected to multiplex cytokine analysis using the mouse Cytokine Array/Chemokine Array 31-Plex Discovery Assay (includes eotaxin, granulocyte colony-stimulating factor (G-CSF), granulocyte macrophage-colony-stimulating factor (GM-CSF), IFN-γ, interleukin (IL)-1α, IL-1β, IL-2, IL-3, IL-4, IL-5, IL-6, IL-7, IL-9, IL-10, IL-12 (p40), IL-12 (p70), IL-13, IL-15, IL-17A, IP-10 (CXCL10), CXCL1, LIF, LIX, MCP-1 (CCL2), macrophage colony-stimulating factor (M-CSF), MIG (CXCL9), MIP-1α, MIP-1β, MIP-2, RANTES (CCL5), tumor necrosis factor-α, and VEGF) at Eve Technologies (Alberta, Canada). All samples were analyzed in biological triplicates. The standard curve regression was used to calculate the concentration of each target cytokine.

Local and systemic immune profiling using polychromatic flow cytometry

Measurement of immune cell proportions within splenocytes and ascites cells was conducted using flow cytometry to characterize systemic and local immune profiles, respectively. Splenocytes from all three treatment groups (control, carboplatin, carboplatin+STING agonist) were collected 24 h post initial carboplatin or carboplatin+STING agonist dose. Following the sacrifice of mice, ascites were aspirated using an 18-gage needle. Single-cell suspensions of splenocytes were prepared by mechanical dissociation of the tissue followed by passing dissociated tissue through a 40 µm cell strainer. A 1× red blood cell (RBC) lysis buffer was added to ascites cells and splenocytes. Cells were digested in RPMI-1640 media containing 20 µg/mL DNase (Roche, Basel, Switzerland) and 1 mg/mL collagenase IV (STEMCELL Technologies, British Columbia, Canada) for 30 min at 37°C/5% CO2. Cells were washed in PBS, counted and subjected to flow cytometry analysis of target immune populations (CD45+ cells) using a lymphoid panel (CD3 T cells, CD4 helper T cells, CD8 cytotoxic T cells, CD19 B cells, NK1.1 natural killer cells, CD62L/CD69 T-cell activation, PD-1 checkpoint; BioLegend, California, USA) and myeloid panel (CD11b myeloid cells, CD11c dendritic cells, F4/80 macrophages, CD80 M1-like macrophages, CD206 M2-like macrophages, GR1 myeloid derived suppressor cells, programmed death ligand-1 (PD-L1) checkpoint), on the CytoFLEX S Flow Cytometer (Beckman Coulter, Ontario, Canada). Single color positive controls, as well as unstained and fluorescence-minus-one negative controls were used for each antibody and their respective panel to determine gates. Gating and analysis were conducted using FlowJo V.10 Software (Beckton Dickinson Biosciences, Ontario, Canada).

NanoString-based tumor immune transcriptomic profiling

Immune gene expression profiling was performed using the pre-built nCounter Mouse PanCancer Immune Profiling panel which includes genes associated with immune function, various cancer-related pathways, and housekeeping genes (NanoString Technologies, Washington, USA) to determine both the baseline differences in the TIME generated from ID8 cells of different genotypes and to measure the effect of STING agonist treatment on the TIME. Total RNA of tumors collected from mice injected with ID8-Trp53–/–; Pten–/– or ID8-Trp53–/–; Brca1–/– cells with and without treatment was isolated at endpoint using the total RNA Purification Kit (Norgen Biotek, Ontario, Canada) as per the manufacturer’s instructions. The purity and concentration of isolated RNA was estimated spectrophotometrically using the NanoDrop ND-100 spectrophotometer (NanoDrop Technologies, Delaware, USA). Total RNA (150 ng) was used as a template for digital multiplexed profiling at the Queen’s Molecular Pathology Laboratory’s as per previously established protocols.18 Raw nCounter NanoString counts were normalized using nSolver software V.3.0 (NanoString Technologies, Washington, USA), using NanoString’s built-in positive controls. Messenger RNA content normalization was performed using housekeeping genes and overall assay efficiency was calculated using the geometric mean of each control. Differential gene expression between comparison groups was computed on nSolver, and Benjamini-Hochberg method was used to adjust for the false discovery rate. Heatmaps were generated on nSolver software (NanoString Technologies, Washington, USA.

Macrophage polarization assay

IC21 cells were seeded at a density of 5×105/well in a 6-well plate and incubated overnight at 37°C in complete media. Ascites generated from untreated mice injected with either ID8-Trp53–/–; Pten–/– or ID8-Trp53–/–; Brca1–/– cells were centrifuged to remove cellular debris and fluid fraction was collected separately. Ascites fluid was passed through Amicon centrifugal filter concentrators (EMD Millipore, Ontario, Canada). Concentrated ascites fluid protein levels were quantified using a Bradford assay. IC21 cells were incubated in media containing equal concentrations of ascites fluid in media from either genotype (1:100 ratio) for 24 hours. Expression of M2-like macrophage associated markers (F4/80, CD206 and PD-L1) was measured using flow cytometry.

Macrophage and T-cell co-culture assay

Splenocytes were collected from a healthy female mouse spleen, mechanically dissociated, passed through a 40 µm strainer and followed by RBC lysis. Single cell suspensions were then incubated in RPMI-1640 supplemented with 20 µg/mL DNase and 1 mg/mL collagenase IV for 30 min at 37°C/5% CO2. Splenic T cells were enriched using a magnetic-based commercial CD8 T-cell negative selection kit (STEMCELL Technologies, British Columbia, Canada; online supplemental figure 3A) as per the manufacturer’s protocol. Ascites from mice (n=5 for each group) injected with either ID8-Trp53–/–; Pten–/– or ID8-Trp53–/–; Brca1–/– cells were collected, passed through a 70 µm strainer, lysed using 1× RBC lysis buffer and incubated in RPMI-1640 supplemented with 20 µg/mL DNase and 1 mg/mL collagenase IV for 30 min at 37°C/5% CO2. Macrophages were enriched using a magnetic-based commercial F4/80 macrophage positive selection kit (Miltenyi Biotec, Maryland, USA; online supplemental figure 3B) as per manufacturer’s instructions. A 96-well plate was coated with anti-CD3e (1 mg/mL; Thermo Fisher, California, USA) and anti-CD28 (5 mg/mL; Thermo Fisher, Ontario, Canada) antibodies for 4 hours at 37°C/5% CO2. CD8+T cell-to-macrophage ratio was seeded at 1:1 (100,000 of each cell type) into triplicate wells for each mouse in 200 µL of IL-2 (5 IU) treated media containing 1% gentamycin and 10% FBS and incubated for 48 hours at 37°C/5% CO2. Four hours prior to the 48-hour time point, brefeldin A (Cayman, Michigan, USA) was added to each well (10 µg/mL). Supernatant containing T cells in suspension was aspirated and placed into a new 96-well plate for flow cytometric analysis of activation state stained with anti-CD45, CD3, CD8, F4/80, PD-1 and IFN-γ antibodies (BioLegend, California, USA). 1× EDTA was used to lift adherent macrophages off of wells and cells were placed into a new 96-well plate for staining of macrophage state. The expression of CD45, F4/80, CD80, CD206, CD8 and PD-L1 (BioLegend, California, USA) was measured using polychromatic flow cytometry (Beckman Coulter). Intracellular levels of IFN-γ in CD8+T cells treated with macrophages from different genotypes were measured to determine the influence of macrophages on T-cell activation.

Statistical analysis

Statistical analyses were performed using GraphPad Prism V.9.0 software (GraphPad prism, version 9.5.1, Software, California, USA) as described in the results. Statistics for Kaplan-Meier survival analysis on patient cohort was performed using the survminer survival package in R (V.3.5.2, R Studio, Massachusetts, USA). All analyses used Mann-Whitney non-parametric test (for data that deviates from normality) to compare two conditions unless otherwise indicated. Results are expressed as a mean±SD. A p value<0.05 was considered statistically significant.


PTEN-deficient human HGSC tumors associate with decreased OS and exhibit a distinct macrophage infiltration profile

Pre-chemotherapy treated tumors from 110 patients with HGSC (online supplemental table 1) were evaluated for PTEN protein expression using a pre-established PTEN scoring system.20 Loss of PTEN protein expression was observed in 9.1% of cases within this cohort. Tumors with either complete absence of PTEN protein expression (n=10) in epithelial compartments or with normal expression (n=57) were subjected to further analyses. Kaplan-Meier survival analysis demonstrated a statistically significant shorter disease specific survival (log-rank test, p=0.0376; HR=0.45; figure 1A) in tumors with absence of PTEN expression compared with those with normal PTEN expression. Immunofluorescence staining revealed decreased CD8+ cytotoxic T cells and CD68+ macrophages in tumors with complete absence of PTEN expression (figure 1B) compared with tumors with PTEN presence (figure 1C), however, these differences were not statistically significant potentially due to intratumoral heterogeneity. In this cohort, CD68+ macrophages were only observed to heavily infiltrate the stromal compartment of PTEN-deficient tumors and displayed low densities in the epithelial compartments of these tumors (figure 1D; p=0.0005 between compartments). In contrast, PTEN-intact tumors displayed intraepithelial CD68+ macrophages revealing macrophage patterns specific to the absence of PTEN (p>0.05 between compartments). This suggests a potential relationship between HGSC PTEN alterations driving variations in intratumoral macrophage infiltration patterns.

Figure 1

PTEN deficiency in the tumor immune microenvironment of patients with HGSC leads to shorter overall survival and decreased immune infiltration. (A) Kaplan-Meier survival analysis using clinical data for patients with PTEN intact (n=10) had significantly higher disease specific survival than patients with PTEN absent (n=57; p=0.0376). H&E stained and multiplex immunofluorescence stained images showing CD8+ cytotoxic T cells and CD68+ macrophages within tumors of patients with; (B) complete absence of PTEN (PTEN absent; n=9) and (C) presence of PTEN (PTEN intact; n=51) in stromal and epithelial compartments of patient tumors using immunohistochemistry. (D) Infiltration patterns of CD68+ macrophages between different compartments in tumors with PTEN absence and PTEN intact were compared, revealing a significantly higher density of CD68+ macrophages in the stroma compared with the epithelium in tumors with PTEN absence (p=0.0005). Average of duplicate or triplicate cores for each sample was taken and Mann-Whitney non-parametric test was used to determine statistical significance of differences between immune cell infiltration patterns. Log-rank test was applied to determine statistical significance of Kaplan-Meier survival analysis, using R statistical software. **p<0.005, ***p<0.001 . P value<0.05 was considered statistically significant. HGSC, high-grade serous ovarian cancer; ns, not significant; PTEN, phosphatase and tensin homolog.

Cancer cell genotype influences the TIME and disease progression in the ID8 murine model of HGSC

We first evaluated whether the cancer cell intrinsic loss of Pten or Brca1 led to survival differences in the ID8 syngeneic murine model of HGSC. Mice were injected with ID8-Trp53–/–, ID8-Trp53–/–; Pten–/–, or ID8-Trp53–/–; Brca1–/– cells (henceforth denoted as ‘Trp53-deficient’; ‘Pten-deficient’ or ‘Brca1-deficient’ cells, respectively). Mice injected with Pten-deficient cells had a statistically significant shorter median OS of 39 days compared with those injected with Brca1-deficient cells, which displayed a median OS of 45 days (p=0.0002), both genotypes displaying a shorter OS than mice injected with Trp53 deficient ID8 cells (56 days; figure 2A).

Figure 2

PTEN-deficient HGSC tumor immune microenvironment and disease progression is distinct from the BRCA1-deficient HGSC. (A) Kaplan-Meier survival analysis of mice injected with either ID8-Trp53−/−, ID8-Trp53−/−; Pten−/− and ID8-Trp53−/−; Brca1−/− cells. (n=10–15) for each treatment group. (B) In vivo bioluminescent imaging of mice injected with ID8-Trp53−/−; Pten−/− and ID8-Trp53−/−; Brca1−/− luciferase-tagged cells tracking tumor progression one time per week over 4 weeks. Total RNA from untreated tumors of different genotypes were subjected to NanoString gene expression profiling using the PanCancer immune gene panel displayed as a (C) heatmap showing differential expression pattern of groups of genes involved in various immune functions, (D) two-fold differentially expressed genes and (E) 0.5-fold differentially expressed genes. Proportion of cells derived from ascites of mice injected with ID8-Trp53−/−; Pten−/− and ID8-Trp53−/−; Brca1−/− cells expressing (F) CD45+CD3+CD8+CD69+, (G) CD45+CD11b+F4/80+CD206+, and (H) CD45+CD11b+GR1+. (I) IL-6, (K) IL-10 and (G) CXCL-10 cytokine levels derived from the ascites fluid of untreated mice injected with ID8-Trp53−/−; Pten−/− and ID8-Trp53−/−; Brca1−/− cells. Log-rank (Mantel-Cox) test was applied to derive significant differences in (A). Mann-Whitney non-parametric test was used for (F–L) *p<0.05, **p<0.005, ***p<0.001, ****p<0.0001. Gene expression data analysis was performed using nSolver Advanced Analysis Software. Mean±SD. HGSC, high-grade serous ovarian cancer; IL, interleukin; MHC, major histocompatibility complex; NK, natural killer; PTEN, phosphatase and tensin homolog; TNF, tumor necrosis factor; UT, untreated.

Given our research question, all subsequent analyses were performed using the Pten-deficient and Brca1-deficient cell lines to investigate tumor immune associated disease states. In vivo bioluminescence imaging revealed a more rapid tumor progression in mice implanted with Pten-deficient cells compared with Brca1-deficient cells by week 4 (figure 2B). To investigate the effects of Pten or Brca1 loss on the associated TIME, we characterized baseline immune transcriptomic profiles of tumors generated from either Pten-deficient or Brca1-deficient cells (figure 2C,E; online supplemental figure 1A,B). A total of 119 genes were significantly differentially expressed between the two groups. NanoString nSolver-based analysis revealed an enrichment in genes associated with innate and adaptive immunity, such as antigen processing and T-cell function in tumors generated from Brca1-deficient cells compared with those from Pten-deficient cells (figure 2C,D). Contrastingly, expression of genes associated with cancer progression, angiogenesis, and extracellular matrix stiffening (such as Angpt2, Col3a1 and Col1a1) were significantly increased in Pten-deficient tumors (figure 2C,E). A twofold lower expression of IFN activated genes Cxcl9 and Cxcl10 was observed within Pten-deficient tumors compared with the Brca1-deficient tumors (figure 2D). Pten-deficient tumors also showed decreased expression of genes involved in several cytotoxic immune cells, however, genes associated with an exhausted T-cell phenotype such as Cd279 (PD-1) and Lag3 were overexpressed (online supplemental figure 2A).

Polychromatic flow cytometry-based analysis of both myeloid and lymphoid cells within the cellular fraction of ascites generated from Pten-deficient cells showed significantly decreased proportions of CD8+ cytotoxic T cells (online supplemental figure 1C) and activated CD8+CD69+ cytotoxic T cells compared with Brca1-deficient ascites cells (figure 2F). Significantly increased proportions of F4/80+ macrophages within the ascites generated from mice injected with Pten-deficient cells was also observed (figure 3B). A significant increase in suppressive immune populations, such as CD206+M2 macrophages and GR1+MDSCs was also observed in the ascites generated from Pten-deficient cells (figure 2G,H). Corresponding ascites cytokine levels showed elevated levels of MCP-1 (figure 3C), IL-6 and IL-10 (figure 2I; p=0.017 and figure 2K; p=0.029). Compared with ascites from mice injected with Brca1-deficient cells, CXCL10, a chemokine critically involved in the recruitment of cytotoxic immune cells, level was significantly decreased in the ascites fluid from mice injected with Pten-deficient cells (figure 2L; p=0.0314). Quantification of secreted chemokines from Pten-deficient cells in supernatants derived from in vitro propagated cells revealed a similar trend with significantly decreased levels of both CXCL10 and CCL5 as compared with Brca1-deficient cells (online supplemental figure 1E,F; p=0.0022).

Figure 3

Pten-deficient ovarian cancer cells alter their tumor immune microenvironment towards a suppressive state via polarizing macrophages to M2-like phenotype. Total RNA from untreated tumors of different genotypes were subjected to NanoString gene expression profiling using the PanCancer immune gene panel displayed as a (A) volcano plot showing differential expression pattern of genes associated with macrophage function. (B) Proportion of cells derived from ascites of mice injected with ID8-Trp53−/−; Pten−/− and ID8-Trp53−/−; Brca1−/− cells expressing CD45+CD11b+F4/80+cells. (C) MCP-1 chemokine levels derived from the ascites fluid of untreated mice injected with ID8-Trp53−/−; Pten−/− and ID8-Trp53−/−; Brca1−/− cells. (D) Transwell migration assay of IC21 peritoneal macrophage cells stimulated (from left to right) with media control (Dulbecco’s Modified Eagle’s Medium), positive control (20 ng/mL MCP-1), or 72 hours conditioned media of ID8-Trp53−/−; Pten−/− and ID8-Trp53−/−; Brca1−/− cells. Stimulated for 16 hours; images 20x magnification. Flow cytometric analysis of IC21 macrophage proportions expressing (E) CD45+F4/80+CD206+ post-stimulation with concentrated ascites derived from mice injected with different genotypes of HGSC cells. US, unstimulated IC21 cells; M1 positive control represented by IC21 cells stimulated with IFN-ɣ (50 ng/mL) and LPS (100 ng/mL) in complete media; M2 positive control represented by IC21 cells stimulated with IL-4 (10 ng/mL) and IL-10 (20 ng/mL) in complete media. Averages of triplicate wells/experiment for three repeated experiments displayed. Mann-Whitney non-parametric test was used for (B–C) *p<0.05 **p<0.005, ***p<0.001, ****p<0.0001. One-way analysis of variance applied for D and E. Mean±SD. HGSC, high-grade serous ovarian cancer; IFN, interferon; IL, interleukin; MCP-1, macrophage chemoattractant protein 1; ns, not significant; PTEN, phosphatase and tensin homolog.

Pten-deficient ovarian cancer cells recruit and polarize macrophages into M2-like phenotype

We next evaluated the expression profiles of genes associated with innate and adaptive immune cells in tumors from mice injected with Pten-deficient or Brca1-deficient ID8 cells. NanoString-based immune transcriptome analysis revealed increased expression of macrophage associated genes in Brca1-deficient tumors compared with Pten-deficient tumors (online supplemental figure 1G). While the expression of genes associated with macrophage phenotypes and function was higher in Brca1-deficient tumors, significantly increased expression of genes such as colony-stimulating factor 1 (p<0.05), responsible for the differentiation of hematopoietic stem cells into macrophages, was observed in Pten-deficient tumors (figure 3A).

While macrophage-associated intratumoral gene expression was relatively low in Pten-deficient tumors compared with Brca1-deficient tumors, flow cytometric analysis of ascites revealed significantly increased proportions of F4/80+ macrophages within the ascites generated from mice injected with Pten-deficient cells (figure 3B). Further, corresponding ascites cytokine levels showed elevated MCP-1 levels in mice injected with Pten-deficient cells (figure 3C). Transwell migration assays of genetically distinct ID8 cells demonstrated a significantly increased migration of IC21 peritoneal macrophage cells incubated with conditioned media from Pten-deficient cells compared with that from Brca1-deficient cells (figure 3D).

We next investigated whether cancer cell intrinsic loss of tumor suppressor genes influences the polarization of macrophages. Treatment of IC21 peritoneal macrophage cells with concentrated ascitic fluid derived from mice injected with Pten- or Brca1-deficient cells was used to assess polarization of IC21 macrophages into either an antitumor M1 or protumor M2-like phenotype. A higher proportion of F4/80+CD206+ cells, indicating polarization to an M2-like phenotype, was observed on treatment with ascites that was generated from Pten-deficient mice (p=0.0079; figure 3E). Of note, macrophages derived directly from the local ascites environment of mice injected with Pten-deficient ID8 cells displayed a higher proportion of PD-L1 immune checkpoint expressing macrophages (online supplemental figure 1D) compared with those from the Brca1-deficient ascites. These results demonstrate that the suppressed tumor immune state associated with the loss of Pten in HGSC may be driven by increased polarization of tumor infiltrating macrophages into an M2-like phenotype.

Host STING pathway is critical in antitumor immunity against PTEN-deficient tumors

Given the established IFN-1 regulatory functions of Brca1 and Pten, we next determined the cancer cell intrinsic activation of the STING pathway. Immunofluorescence staining revealed increased cytosolic dsDNA in Brca1-deficient cells compared with Pten-deficient cells under unstimulated conditions (figure 4A). On stimulation of Pten-deficient or Brca1-deficient cells with STING agonist, we also observed that STING protein levels in Brca1-deficient cells peak at 3 hours post-treatment as compared with 6 hours for Pten-deficient cells (data not shown). Moreover, expression levels of genes related to STING pathway, including Tmem173 (STING; p>0.05), Irf3 (p=0.0355), NFκB (p=0.0013), chemokines Cxcl10 (p>0.05) and Ccl5 (p>0.05), and the receptors Ifnar1, Ifnar2, and Ifngr1, were decreased in Pten-deficient tumors compared with Brca1-deficient tumors (figure 4B; online supplemental figure 2A).

Figure 4

Host STING pathway activation is critical to improved overall survival outcome. (A) Confocal immunofluorescence images showing dsDNA (FITC-green) and DAPI (nucleus, blue) in unstimulated ID8 cells of varying genotypes. Magnification, ×40; scale bar 100 µm. (B) Isolated RNA from untreated tumors from ID8-Trp53−/−; Pten−/− and ID8-Trp53−/−; Brca1−/− cells were subjected to NanoString-based gene expression profiling to identify variation in genes associated with the cGAS-STING pathway. (C) Kaplan-Meier survival analysis of mice implanted with ID8-Trp53−/−; Pten−/− and ID8-Trp53−/−; Brca1−/− cells in wild-type (WT) C57BL6 mice or Sting1 deficient mice (STING-KO) was reported. (D) Kaplan-Meier survival analysis of untreated control (UT) or STING agonist treated STING-KO mice injected with either ID8-Trp53−/−; Pten−/− and ID8-Trp53−/−; Brca1−/− cells. Log-rank test was performed to assess statistical significance between groups. (n=5–15) for each treatment group. Mann-Whitney non-parametric test was applied for (B). *p<0.05 **p<0.005, ***p<0.001, ****p<0.0001. Data analysis was performed using nSolver Advanced Analysis Software. IRF3, interferon regulatory factor 3; mRNA, messenger RNA; ns, not significant; PTEN, phosphatase and tensin homolog; STING, stimulator of interferon genes.

We then investigated whether response to combination chemoimmunotherapy is affected by cancer cell intrinsic loss of Pten or Brca1. STING-KO mice injected with Pten-deficient ID8 cells revealed significantly shorter median OS (26 days) compared with wild-type (WT) mice (median OS of 39 days; figure 4C). Similar differences in survival were also observed in STING-KO mice injected with Brca1-deficient ID8 cell compared with WT mice (median OS of 34 and 44 days, respectively; figure 4C). This highlights the importance of the host STING pathway in antitumor immunity independent of cancer cell intrinsic differences in STING pathway activation.

Further, the addition of exogenous activation of the STING pathway using a STING agonist resulted in no significant survival benefit in STING-KO mice injected with Pten-deficient cells compared with the untreated mice (figure 4D). However, treatment of STING-KO mice injected with Brca1-deficient cells, with STING agonist significantly prolonged survival compared with the control group, suggestive of a role for pre-existing DDR deficiency-mediated IFN-1 response (figure 4D).

Exogenous activation of STING increases the response of PTEN-deficient tumors to carboplatin chemotherapy

Tumors exhibiting chemoresistant malignant cells, such as those with PTEN deficiency, require improved response outcomes, which may be achieved by direct STING pathway activation. In vitro carboplatin chemosensitivity assay confirmed a higher IC50 in Pten-deficient cells (3.755 µM) compared with Brca1-deficient cells (0.6974 µM; online supplemental figure 2B). Response to carboplatin chemotherapy was significantly increased in mice injected with Pten-deficient ID8 cells following addition of STING agonist in the treatment regimen compared with those treated with carboplatin alone (median OS 64 and 51 days, respectively; figure 5B). Kaplan-Meier survival analysis and log-rank (Mantel-Cox) test showed STING agonist monotherapy does not impart significant survival benefit in Pten-deficient mice (median OS 38.5 days) compared with the vehicle group (median OS 39 days, p=0.0516; figure 5B). Similar treatment response patterns were observed in mice injected with Brca1-deficient cells (figure 5C); however, the effects of prolonged survival were more pronounced in mice injected with Pten-deficient ID8 cells (figure 5B,C). Gene expression profiling of Pten-deficient tumors generated post-treatment with either carboplatin or carboplatin+STING agonist revealed decreased expression of cytotoxic T-lymphocytes-associated protein 4, IL-10 and CD163 macrophages and increased expression of granzyme B and signal transducer and activator of transcription 1 (STAT1) genes on combination treatment (data not shown). Similar patterns of increased CD3e, CD8a, CD8b, and decreased CD163 macrophage gene expression was observed in Brca1-deficient tumors treated with the combination of carboplatin and STING agonist compared with carboplatin alone. Plasma CXCL10 and CCL5 cytokine levels were increased on combination carboplatin and STING agonist treatment (figure 5D,E), where levels of CXCL10 were higher in plasma of mice with untreated Pten-deficient tumors compared with untreated Brca1-deficient tumors.

Figure 5

Treatment with STING agonist post carboplatin chemotherapy increases the response of Pten-deficient ovarian tumors. (A) Schematic demonstrating the timeline of cell injection and subsequent treatment course and dosage of carboplatin and STING agonist. Kaplan-Meier survival analysis of mice injected with either (B) ID8-Trp53−/−; Pten−/− or (C) ID8-Trp53−/−; Brca1−/− cells for different treatment groups (n=8–15 each). Cytokine profiles of (D) CXCL10 and (E) CCL5 from the plasma of untreated control, carboplatin treated, or combination carboplatin + STING agonist treated mice of either genotype. Log-rank (Mantel-Cox) test was applied to derive significant differences between treatment groups. *p<0.05 **p<0.005, ***p<0.001, ****p<0.0001. Two-way analysis of variance was applied. Mean±SD. i.p., intraperitoneal; PTEN, phosphatase and tensin homolog; STING, stimulator of interferon genes.

Exogenous STING pathway activation repolarized M2-like macrophages and increased T-cell activation within the Pten deficient TIME

Based on the finding that STING pathway activation increased immune cell recruitment and imparted a survival benefit in Pten-deficient tumors, we further investigated the mechanism by which this genotype-specific immune suppression occurs and the ability of STING activation to reinvigorate antitumor T-cell activation. In vitro activated CD8+T cells derived from age matched healthy mice were treated with macrophages isolated from the ascites of mice injected with either Pten-deficient or Brca1-deficient ID8 cells to assess for influence on T-cell activation levels. Results revealed a significantly decreased intracellular T-cell IFN-γ expression on co-culture with macrophages derived from ascites of mice injected with ID8-Trp53−/−; Pten−/− cells compared with those from ascites of mice injected with ID8-Trp53−/−; Brca1−/− cells (p=0.008; figure 6A). To further explore this, ascites from untreated, carboplatin treated, and combination carboplatin+STING agonist treated mice were isolated and the proportion of key immune populations was analyzed. Using FlowJo plug-ins which allow for the concatenation of profiles from multiple samples into representative t-Distributed Stochastic Neighbor Embedding (t-SNE) plots, a decrease in myeloid proportions and a shift away from F4/80+ macrophages with a CD206+M2 like phenotype was observed in ascites from mice injected with ID8-Trp53−/−; Pten−/− cells (figure 6B). Additionally, proportions of CD69+CD8+ activated T cells in the ascites cellular fraction increased on treatment with combination therapy compared with that of the vehicle-treated mice (figure 6C).

Figure 6

Exogenous STING pathway activation induces antitumor phenotype of immune cells within the Pten-deficient tumor immune microenvironment. (A) Co-culture of ascites-derived macrophages of mice injected with ID8-Trp53−/−; Pten−/− or ID8-Trp53−/−; Brca1−/− cells. Baseline proportions of CD80+M1 macrophages and CD206+M2 macrophages added to healthy T cells displayed. Histogram showing representative average of IFN-ɣ producing CD8+T cells from triplicate wells 48 hours post-stimulation with macrophages derived from ascites of either cell genotype (using n=5 mice for each genotype). Averages of triplicate wells used within this assay were displayed as a bar graph. (B) t-SNE plots generated from a concatenation of myeloid markers expressed on ascites derived from ID8-Trp53−/−; Pten−/− cells using FlowJo FlowSOM and PhenoGraph plug-in features (n=5 for each treatment group: untreated, carboplatin, and carboplatin+STING agonist). Corresponding cell proportions for each marker within ascites from individual mice displayed as bar graphs below. (C) Proportion of Live+CD45+CD3+CD8+CD69+ activated T cells within the ascites generated from untreated, carboplatin or carboplatin+STING agonist treated ID8-Trp53−/−; Pten−/− or ID8-Trp53−/−; Brca1−/− cells; analyzed using one-way analysis of variance. Mann-Whitney non-parametric test was used for (A). *p<0.05 **p<0.005, ***p<0.001, ****p<0.0001. Mean±SD. IFN, interferon; ns, not significant; PTEN, phosphatase and tensin homolog; STING, stimulator of interferon genes.


Cancer cell intrinsic genetic alterations such as loss of tumor suppressor gene function via mutations, copy number alterations, or epigenetic modifications are key determinants of the pretreatment TIME and clinical outcomes.21 The HGSC TIME exhibits a spectrum of immune active and underactive states that associate with differential treatment response.22 In the current study, we investigated this phenomenon in the BRCA1 and PTEN loss associated TIME to mimic the polarized HGSC disease states that exhibit contrasting immune landscapes. In both patient tumors and a syngeneic murine model of HGSC, we demonstrate how cancer cell intrinsic genetic alterations contribute to variable survival outcomes and tumor immune phenotypes. The BRCA1 gene loss associated activation of STING pathway and consequent immune infiltration via induction of IFN-1 associated chemokines has been reported in ovarian cancer.23 However, the PTEN loss associated underactive TIME, chemoresistance and poor prognosis in HGSC remains understudied.24 25 Findings from our study validated the previous reports21 23 24 26 on BRCA1 deficiency associated cytosolic abundance of dsDNA in EOC cells and downstream constitutive activation of the cGAS-STING pathway.23 PTEN, however, functions to dephosphorylate IRF3 and stimulates its translocation into the nucleus to begin transcription of IFN-1 genes.13 Thus, its absence results in decreased cGAS-STING pathway activation and subsequent decreased IFN-1 response.

While we previously demonstrated that patients with chemoresistant HGSC tumors display a low density of CD8+tumor infiltrating lymphocytes (TILs) and lower expression of IFN-1 genes,4 the difficulty in treating HGSC lies primarily in our inability to precisely attribute the specific tumor immune states associated with the vast genetic heterogeneity in this cancer. Indeed, widespread prevalence of TP53 mutations in HGSC results in baseline DDR deficiency likely affecting IFN-1 response and innate immune activity via the cGAS-STING pathway.27 The cell lines used within this study contain a knockout of either Brca1 or Pten on a Trp53 deficient backbone, thus baseline STING pathway activation is observed in all derivatives of the ID8 cell lines. Supporting this notion, recent in vitro studies in breast and pancreatic cell lines have demonstrated that TP53 loss of function mutations interfere with the cytoplasmic DNA sensing machinery required for cGAS-STING pathway activity.28 Furthermore, mutations in TP53 also correlate with increased expression of immune cell recruiting chemokines CXCL9, CXCL10 and CXCL11, and an inflammatory TIME compared with WT TP53 in HGSC specifically.28 29 Here, we demonstrate that the effect of Trp53 gene loss is further altered with co-occurring mutations in DDR genes such as Brca1 and Pten. Survival analysis in mice revealed that irrespective of Trp53 loss of function imparting constitutive activation of cellular IFN-1 pathways, deletion of another tumor suppressor gene results in a more aggressive disease state decreasing OS independent of which additional gene was deleted. Importantly, in this study we validated the relatively enhanced OS in mice injected with Trp53 deficient ID8 cells alone, compared with those with additional loss of either Pten or Brca1, shedding light on the importance of determining the influence of paralleled loss of DDR genes on the TIME, tumor progression, and treatment responses.

To begin, in a cohort of adjuvant treated HGSC tumors, stratification of patients based on complete absence or presence of PTEN within both epithelial and stromal compartments revealed significantly decreased OS in patients with PTEN deficiency. Despite trends of decreased CD8+T cells and CD68+ macrophages in the absence of PTEN, a high degree of variability was observed in this cohort, confirming the challenge faced with cohort-specific differences in biomarker translation within HGSC. Interestingly, PTEN-deficient tumors displayed a significantly greater density of CD68+ macrophages within the stromal compartment compared with the epithelial compartment. Such differences were not observed in PTEN-intact tumors, suggesting a potential role for the loss of PTEN in the inhibition of immune migration into the tumor epithelial compartment. It is plausible that in tumors with PTEN loss, reduced infiltration of macrophages also impedes the process of tumor antigen cross presentation to CD8+T cells and may undergo metabolic reprogramming.30 31 Their decreased chemokine expression in the TIME may thus further contribute to lower infiltration of immune cells contrasting tumors with BRCA1 loss. Other mechanisms compounding this effect such as Pten-loss associated hypoxia contribute to an ‘immune excluded’ phenotype observed in other solid tumor cancers, may also be relevant within HGSC tumors.32 A comprehensive understanding of intratumoral localization of macrophages and the associated environmental cues within the different tumor compartments on their resultant phenotype and function as M1 versus M2 is critical.

In mice injected with Pten-deficient cells, both cellular and secreted profiles of malignant ascites and tumors revealed a suppressed and underactive tumor microenvironment, reflected by significantly decreased proportions of activated CD8+T cells and increased GR1+MDSCs. Interestingly, ascites from mice injected with Pten-deficient cells showed significantly increased proportions of CD206+M2 macrophages. In vitro analysis revealed Pten-deficient cancer cell-secreted factors led to migration of significantly more IC21 peritoneal macrophages as compared with those from Brca1-deficient cells. This suggests cancer cell intrinsic PTEN mutations specifically promote an immune microenvironment with enhanced polarization of macrophages toward an M2-like suppressive behavior. Such a polarization could also result from elevated levels of the cytokine IL-10,33 which was significantly higher in ascites from mice injected with Pten-deficient cells. Another key finding reflecting an immunosuppressive microenvironment driven by Pten deficiency in tumors was the reduced expression of intracellular IFN-γ in CD8+T cells co-cultured with macrophages derived from ascites. Similar findings have been previously reported in colorectal cancer.34 Thus, during Pten-deficient tumor progression and peritoneal metastasis, the protumoral environment generated impairs the effector function of critical cytotoxic immune populations relative to other genotypes.

While there are no approved agents directly targeting STING pathway in HGSC, there are multiple clinical trials currently evaluating STING agonists in combination with immune checkpoint inhibitors (NCT04609579, NCT04144140 and NCT05070247) in advanced solid tumors based on preclinical findings.35 Administration of intraperitoneal STING agonist was used primarily as a means to stimulate a potent IFN-1 response that would subsequently rescue or stimulate the function of pre-existing immune cells and increase the recruitment of additional antitumor immune cells via secretion of CXCR3 receptor binding chemokines. Presence of STING and IFN-1s are well-established in their ability to impart robust antitumor immune effects via activation of antigen presenting cell priming of tumor antigens to T cells, subsequently leading to activation of cytotoxic T cells.36 37 We previously established that exogenous STING pathway activation increased IFN-1 signaling by elevating levels of the immune cell recruiting chemokines CXCL10 and CCL518, demonstrating increased CD8+ cytotoxic T-cell recruitment in mice injected with either Pten-deficient or Brca1-deficient cells. Addition of STING agonist rescued the effector functions of immune cells within Pten-deficient HGSC tumors by increasing cytotoxic T-cell activation, enhancing response to carboplatin, and resulted in a more pronounced increase in their OS. As such, HGSC tumors harboring mutations resulting in underactive immune states or growth patterns, similar to that observed with PTEN deficiency, may be targeted for therapies that directly activate the STING pathway. A delayed activation of STING pathway in Pten-deficient cells compared with Brca1-deficient cells is also suggestive of additional IFN regulatory mechanisms leading to an ultimate effect of reduced chemokine secretion, which are beyond the scope of this study and offer avenues for further investigation. In this study we demonstrate the benefit of cancer cell intrinsic STING activation within Pten-deficient cells, however, STING-KO (Sting1–/–) mice injected with Pten-deficient cells resulted in decreased OS relative to WT C57BL/6 mice. While the emphasis of this work was on cancer cell intrinsic genetic alterations and STING pathway activity, the importance of host IFN-1 activation in cancer progression is evident (figure 4), independent of genotype; further studies are required to gain an understanding on the role of STING pathway activation within immune cells, specifically myeloid cells in the context of Pten deficiency.

PTEN deficiency in tumors was associated with an abundance of immunosuppressive M2-like macrophages in the malignant ascites. Several mechanisms within the complex TIME of HGSC may indeed contribute to an M2-dominated state.30 31 Most relevant to our study is the reduced feed-forward loop activation of STAT1 and CXCL10 that may act in parallel to other cancer cell intrinsic mechanisms.38 39 In BRCA1-deficient tumors, a significantly active STAT1/CXCL10 axis in both the cancer and tumor infiltrating immune cells has been widely reported.23 40 41 Furthermore, the decreased expression of CXCL10 by M2-like macrophages may be additive to the overall aggressive behavior of Pten-deficient cancer cells.42 Finally, M2-like tumor-associated macrophages (TAMs) are known to promote HGSC survival and angiogenesis.43 In mice treated with STING agonist, we observed a shift from M2 to M1 macrophages. This finding is in concordance with similar recent studies in both colorectal cancer and breast cancer models where repolarization of M2 macrophages to an M1 phenotype was observed on exogenous STING pathway activation.34 44 This suggests that STING pathway activation in Pten-deficient HGSC displays the benefits associated with decreased M2s, such as decreased angiogenesis, secretion of immunosuppressive cytokines such as IL-10 and transforming growth factor-β, while providing rationale for the combined use of STING agonists and immune checkpoint blockade therapy, particularly PD-L1 antagonism. Although previous reports have revealed mechanistic links between Pten loss and macrophage driven cancers such as glioma,16 this remains to be investigated in HGSC. As such, tumors which exhibit inactive immune states resulting from genetic alterations such as PTEN loss may see greater benefit from direct STING pathway and IFN-1 activating therapies (ie, STING agonists following an immunogenic cell death inducing chemotherapy, oncolytic viruses).45 Indeed, as recently reported, PARP inhibitors and other DDR kinase inhibitors have also been shown to indirectly activate the STING pathway although at a lesser magnitude.46 47

Our study is not without limitations, particularly, our investigation focuses on the influence of individual and isolated genetic alterations. Human HGSC cells can express several co-occurring alterations7 and future work should incorporate these findings to further our understanding on the effects of the interactions and relationships of the several common gene losses reported in HGSC. Herein, we explore the effect of PTEN and BRCA1 loss in a metastatic syngeneic ID8 murine model of HGSC, however, investigating the impact of cancer cell intrinsic genotype on the resultant TIME within other models is critical to our understanding.48 To conclude, activation of the STING pathway is largely influenced by pre-existing IFN regulatory genetic alterations such as DDR gene deficiency, prevalent in HGSC tumors. Exogenous activation of the STING pathway overcomes the immunosuppressive local microenvironment driven by Pten loss in cancer cells. As further research considers the cancer cell intrinsic genetic alterations,49 integrating genomic correlates of the TIME in patients with HGSC will provide an improved platform for rapid evaluation of optimal therapeutic combinations.

Data availability statement

Data are available upon reasonable request.

Ethics statements

Patient consent for publication


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.


  • Twitter @NoorShakfa, @juliettew_s, @JawaTasha, @BrianLaight, @CanImmun

  • Contributors MKoti designed the overall study, provided insights on experimental design and interpretation of results. NS and MKoti wrote the manuscript. NS designed and performed all in vivo and in vitro experiments and analyzed the data. DL and EDL assisted with in vivo work. GC performed IF assays. DL, EDL, JW-S, AH, SC, and BL assisted with in vitro and flow cytometry experiments. NJ performed all analyses of patient clinical data (survival analyses and cox proportional hazards analyses) using the statistical software R and provided statistical support for survival subgroup analyses. MKoebel performed the histopathological evaluation of patient tumor specimens and designed the tissue microarray (TMA). KT provided guidance and assistance with NanoString data analysis and overall statistical analysis of patient TMA based studies.

  • Funding This work was supported by research and infrastructure grants from the Canadian Institutes of Health Research (grant number 159497), Ontario Ministry of Research Innovation and Science; Early Research Award (award number ER17-13-133), Queen’s University Research Initiation Grant and Canada Foundation for Innovation (grant number 37798) to MKot. Additional support was provided to NS by the Franklin Bracken Fellowship, Dean’s Doctoral Award through Queen’s University, and the Ontario Graduate Scholarship program. We thank Shakeel Virk at the Queen’s Laboratory for Molecular Pathology (QLMP) for his assistance with TMA imaging and Halo, Katy Milne at BC Cancer’s Molecular and Cellular Immunology Core (MCIC) for immunofluorescence staining of TMAs.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.