ULK1 promotes metastatic progression in experimental models of epithelial ovarian cancer

Abstract

Epithelial ovarian cancer (EOC) is a leading cause of gynecological cancer mortality, driven largely by late diagnosis and chemoresistance. While autophagy is critical for EOC spheroid survival during metastasis, the role of ULK1, a key regulator of autophagy, in EOC progression remains unclear. To investigate this, we utilized CRISPR/Cas9 technology to delete ULK1 in EOC cell lines OVCAR8, HEYA8, ES2 and the fallopian tube epithelial cell line FT190. ULK1 loss and autophagy disruption were confirmed in EOC spheroids, with reduced Beclin-1 phosphorylation, impaired LC3 processing, and p62 accumulation. ULK1 knockout decreased EOC spheroid cell viability via increased apoptosis, and impaired matrix-bound organoid growth, offering new insights into ULK1 activity in affecting EOC tumor growth and spread. These findings were supported by in vivo xenograft models, in which ULK1 loss significantly reduced tumor burden and metastatic potential. ULK1 requirement during metastasis was supported by diminished invasive capacity of ULK1 knockout spheroid cells in mesothelial clearance assays. To investigate ULK1 mechanisms contributing to EOC tumor progression and metastasis, we conducted proteomic analyses of OVCAR8 spheroids, which revealed ULK1 loss disrupted critical pathways, including MEK-MAPK, PI3K-AKT-mTOR, and apoptosis regulation. Although ULK1 knockout failed to synergize with standard-of-care chemotherapeutics, it significantly enhanced sensitivity to MEK and mTOR inhibition. Analysis of ovarian cancer datasets demonstrates that high ULK1 mRNA correlates with a poorer 10-year overall and progression-free survival; in fact, its expression is further elevated in metastases as compared with primary tumors and normal tissue. Treatment of metastatic patient-derived organoids with the clinical ULK1 inhibitor DCC-3116, MEK inhibitor trametinib, or mTORC1/2 inhibitor AZD-8055 reduced viability in a subset of these samples, reflecting inter-patient heterogeneity and need for biomarker-guided selection. Overall, this study highlights ULK1 as a critical regulator of multiple steps of EOC disease progression, underscoring its potential as a therapeutic target in advanced ovarian cancer.

Background

Epithelial ovarian cancer (EOC) has the fifth-highest death-to-incidence ratio for all cancers in women and is the leading cause of death from gynecologic cancers due to its late-stage diagnosis and lack of effective strategies for treating chemoresistant disease [1]. Patients with EOC are typically treated with aggressive surgical debulking and cytotoxic carboplatin/paclitaxel combination chemotherapy; however, nearly 80% of these patients relapse within five years [2]. Investigations into the mechanisms that support cell survival during metastasis and regrowth of refractory EOC cells after treatment are of critical importance and are an area of active research. EOC spreads by tumor cells disseminating directly into the peritoneal space, often suspended in ascites, and then attach to the serosal surfaces of the abdominal cavity to form secondary deposits [3, 4]. Clusters of metastatic EOC cells known as spheroids accumulate in the malignant fluid of patients with advanced disease [5]. Spheroids are known to promote metastasis with increased cell survival in the face of chemotherapy and possess enhanced adhesive and invasive capabilities [6, 7]. Additionally, our lab and others have clearly demonstrated that spheroid cells undergo numerous phenotypic changes, including cellular quiescence [8], epithelial-mesenchymal transition [9], activated stress metabolism [10, 11], and autophagy [12], all of which contribute to the tumor cell dormancy phenotype of residual disease, and that may drive chemo-resistant recurrence.

Autophagy is an evolutionarily conserved and tightly controlled metabolic degradation process in which proteins and organelles are broken down in the lysosomes [13]. This degradative process yields metabolic substrates from lysosomal activity, thereby providing vital nutrients for essential cellular functions during nutrient scarcity or energy stress [14]. Autophagy typically operates at basal levels to fulfill fundamental homeostatic functions but can quickly escalate under stress [15]. Autophagy induction is controlled by the Unc-51-like kinase (ULK1) complex, consisting of ULK1/2, autophagy-related gene 13 (ATG13), and focal adhesion kinase-interacting protein (FIP200) [16]. ULK1 is a serine-threonine kinase that responds to upstream signals of nutrient and energy availability to trigger autophagy through the initiation of phagophore formation. Under nutrient-abundant conditions, mechanistic target of rapamycin complex 1 (mTORC1) phosphorylates ULK1, thereby inhibiting its activity and initiating autophagy. Conversely, the absence of nutrients leads to mTORC1 deactivation, while AMP-activated protein kinase (AMPK) phosphorylates to activate ULK1 [17]. Thus, ULK1 and autophagy are considered an essential regulatory hub to control energy supplies in an equilibrium with cellular demands.

The role of ULK1 in autophagy initiation has been reviewed in [18], and its involvement in tumorigenesis—promoting tumor growth, invasiveness, and survival—has been documented across various cancer types [19–23]. However, few studies have investigated the role of ULK1-mediated autophagy in EOC with a specific focus on tumor growth and metastatic progression. We have previously demonstrated that AMPK activation [24] and AKT-mTORC1 downregulation [8] work in coordination to induce autophagy in high-grade serous ovarian cancer (HGSOC) spheroids. We also showed that EOC spheroids had increased ULK1 expression, which parallels autophagy induction, and that transient knockdown and inhibitor treatments in vitro blocked autophagy and reduced cell viability [25]. Building on our prior work on autophagy in EOC spheroids and utilizing CRISPR/Cas9 technology for ULK1 ablation, we aimed to elucidate ULK1’s role in tumorigenesis by examining its contributions to spheroid survival, metastatic dissemination, and tumor growth to establish its potential as a therapeutic target in EOC metastasis.

Methods

Cultured cell lines

The cell lines OVCAR8, OVCAR8-ULK1KO, HEYA8, and HEYA8-ULK1KO were grown in RPMI-1640 medium (Wisent), whereas ES2, ES2-ULK1KO, FT190, and FT190-ULK1KO were grown in DMEM/F12 medium (Life Technologies). All growth media were supplemented with 10% fetal bovine serum. OVCAR8, HEYA8, and ES2 cells were procured from the American Type Culture Collection. Adherent cells were sustained on tissue culture-treated polystyrene (Sarstedt, Newton, NC, USA) and spheroids were maintained in Ultra-Low Attachment (ULA) cluster plates (Corning, NY, USA). The immortalized human fallopian tube secretory epithelial cell line FT190 [26] was generously provided by R. Drapkin from the University of Pennsylvania, Philadelphia, PA, USA. The human lung mesothelial ZT-GFP cell line [7] was generously provided by Marcin Iwanicki from the Stevens Institute of Technology All cell lines were authenticated through short tandem repeat analysis by the Center for Applied Genomics (The Hospital for Sick Children, Toronto, ON, Canada) and routinely examined them for mycoplasma using a Universal Mycoplasma Detection Kit (30–1012 K; ATCC).

Generation of ULK1KO cell lines

CRISPR/Cas9 (sc-400516-KO-2 Lot# C3016, Santa Cruz Biotechnology) was used to ablate ULK1 in OVCAR8, HEYA8, ES2, and FT190 cells. Briefly, cells were seeded at 1–1.5 × 10⁵ per well in a 6-well plate and transfected the following day. Cells were trypsinized four days post-transfection and sorted using fluorescence-activated cell sorting (FACS) in 96-well plates. The clones were left to grow for a minimum of two weeks, after which colony formation was observed. Colonies were harvested and plated into 6-well plates and then 10 cm plates upon reaching confluency. Cells were harvested for protein lysates, screened for ULK1 loss via western blotting and passaged for continued culture and subsequent clone pooling.

Generation of Nuclight GFP and RFP cell lines

Cells were transduced with Incucyte® Nuclight Green Lentivirus (EF1a, Puro) (Sartorius, #4624) or Incucyte® Nuclight Red Lentivirus (EF1a, Puro) (Sartorius, #4476) following the manufacturer’s instructions. After transduction, cells were cultured in complete media supplemented with puromycin (1 μg/mL; BioShop, #5E38885) to select successfully transduced cells. The isolated GFP+ and RFP+ cells were expanded and used for further analysis and experimentation.

Generation of mCherry-eGFP-LC3 cell lines

Parental and ULK1KO cells were transfected with mCherry-eGFP-LC3B autophagy reporter plasmid [pBABE-puro mCherry-EGFP-LC3B was a gift from Jayanta Debnath (Addgene plasmid # 22418; http://n2t.net/addgene:22418;RRID:Addgene_22418] [27]. After transfection, cells were grown in complete media containing G-418 (400 μg/mL, Wisent, #450-130-QL) for two weeks to select those with successful reporter plasmid integration. Following the selection phase, the cells were cultured in complete medium without G-418 for another four weeks, allowing for growth and recovery. Cells were sorted using FACS to identify double-positive cells (GFP+, mCherry+).

Generation of luc2tdTomato cell lines

Cells were transduced with pCDH-EF1-Luc2-P2A-tdTomato, according to the manufacturer’s instructions [pCDH-EF1-Luc2-P2A-tdTomato was a gift from Kazuhiro Oka (Addgene plasmid # 72486; http://n2t.net/addgene:72486;RRID:Addgene_72486)]. After transduction, the cells were subjected to FACS to isolate the tdTomato+ cells. The cells were cultured in a complete medium for growth and recovery. The cells were then subjected to another round of FACS to gate and select for populations of cells with similar tdTomato expression intensities. Cell populations were seeded in a serial dilution for bioluminescent imaging (BLI) analysis using D-luciferin and the IVIS Lumina S5 system (PerkinElmer) to determine optimal populations (i.e., similar reporter gene expression among lines) to use for subsequent in vivo xenograft assays.

Antibodies and reagents

Antibodies against ULK1 (#8054S), p62 (#5114S), LC3B (#2775S), Beclin1 S30 (#5410S), Beclin1 (#3738S), p21 (#2947), p27 (3686), AKT S473 (#9271), AKT (#9272S), MEK1/2 S217/221 (#9154S), MEK (#8727), ERK1/2 Thr202/Tyr204 (#9101), ERK1/2 (#9102), cleaved-PARP (#9541S), P70S6K Thr389 (#9234S), P70S6K (#2708S), p38 MAPK Thr180/Y182 (#4511S), 4EBP1 T37/46 (#2855S), 4EBP1 (#9452S) were purchased from Cell Signaling Technology. Anti-ULK2 antibody (AB97695), ATG16L1 (AB187671), ATG16L1 S30 (AB19016), and mCherry (AB167453; 1:500) were purchased from Abcam. Anti-actin antibody (A2066; 1:25000) was purchased from Millipore. Antibodies against tubulin (T5168; 1:40000) and vinculin (V9264; 1:25000) were purchased from Sigma-Aldrich. Horseradish peroxidase (HRP)-conjugated antibodies against mouse IgG (NA931; 1:10000) and rabbit IgG (NA934; 1:10000) were purchased from GE Healthcare. Antibodies were diluted in tris-buffered saline-Tween 20 containing either 5% bovine serum albumin or non-fat milk 1:1000 or as stated otherwise. Adenovirus expressing green fluorescent protein (Ad-GFP) was a kind gift from Dr. B. C. Vanderhyden (Ottawa Health Research Institute) and prepared as described previously [11]. Rat-tail collagen was purchased from Gibco (#963791) and Matrigel was purchased from Corning (CLS356231). Paclitaxel was purchased from Cayman Chemical Company (#10461) and stored at −20 °C as 5 mM in DMSO stocks. Carboplatin was received from the London Regional Cancer Program and stored at 4 °C as 27 mM in saline stocks. Olaparib, (#HY-10162), ralimetinib (#HK-13241), and DCC-3116 (#HY-160699) were purchased from MedChemExpress, trametinib (#7709) was purchased from Tocris Bioscience, and AKT inhibitor VIII (Akti-1/2) was purchased from EMD/Calbiochem (#12408).

Immunoblot analysis

The Bio-Rad Mini-PROTEAN II Electrophoresis System was used for immunoblotting following the manufacturer’s guidelines, utilizing in-house prepared gels (30% acrylamide/bis solution 37.5:1, catalog number 1610158; Bio-Rad). Densitometric analysis was conducted using Image Lab 6.05 software suite (Bio-Rad).

Preparation of whole-cell lysates

For assessment of all proteins, 4-, 8-, 24-, 48, and 72-hour whole-cell lysates were generated from adherent cells cultured at a density of 0.75–1× 10⁶ cells in 10 mL medium (10 cm dish), or spheroid cells cultured at a density of 1–3× 10⁶ cells in 15 mL medium (35 mm ULA well). Cell seeding numbers were chosen to obtain acceptable protein yields for each cell line.

Whole-cell lysates

Adherent cells cultivated on tissue culture-treated plates or dishes were collected by removing the medium, washing twice with cold PBS, and scraping into a modified radioimmunoprecipitation (RIPA) buffer (50 mM HEPES (pH 7.4), 150 mM NaCl, 10% glycerol, 1.5 mM MgCl₂, 1 mM EGTA, 1% Triton X-100, 0.1% SDS, 1 mM Na₃VO₄, 10 mM NaF, 1 mM PMSF, 1 × SIGMAFAST protease inhibitor cocktail (cat. S8820; Sigma), 10 mM beta-glycerophosphate). Spheroids (minimum of 1.5× 10⁶ cells per sample) were collected by transferring the cell suspension into an ice-cold conical tube, followed by centrifugation using a swinging bucket rotor (800 × g at 4 °C for 4 min) to form a pellet. The medium was aspirated, and the pellet was resuspended in cold PBS (at least 10 mL of cold PBS. This process was repeated by resuspending the pellet in cold PBS, followed by centrifugation and aspiration of the PBS. The resulting cell pellets were lysed using modified RIPA buffer, vortexed, exposed to one freeze-thaw cycle, and clarified by centrifugation (maximum × g at 4 °C for 20 min).

Spheroid viability assays

Bulk spheroids

Cells were placed in 24-well ultra-low attachment (ULA) cluster plates at a density of 0.5–1× 10⁵ cells per well in 1 mL of medium. The spheroids were then collected into chilled microcentrifuge tubes and centrifuged at 500 × g for 3 min to form pellets. After aspirating the medium, the pellets were washed once with 500 μL of PBS, centrifuged once more, and resuspended in 50–250 μL of trypsin/EDTA. The suspension was incubated at 37 °C with gentle agitation every 10 min until no aggregates were visible (10–30 min). Trypsin was inactivated by adding an equal volume of FBS, followed by the addition of Trypan Blue dye (Gibco™ 15250061) at 1:1 ratio and gentle mixing via pipetting. Cell counting was performed using the TC20 Automated Cell Counter (Bio-Rad Laboratories).

Individual spheroids

Cells were seeded in a 96-well round-bottom ULA plate at a density of 2000 cells per well in 100 μL of medium. For alamarBlue assays, cells were incubated at a final dilution of 1:10 (alamarBlue to media) for 4, 24, or 48 h and fluorescence was measured using an Agilent Biotek Synergy H1 plate reader. For Caspase-Glo 3/7 (Promega, G8092) assays, 100 μL of reagent was added to each well, and the plate was frozen at −80 °C. After 24 h, the plates were thawed at room temperature in the dark for 60 min on a plate rocker. The contents of wells were transferred to individual wells of a 96-well opaque white plate, and luminescence was measured on the Agilent Biotek Synergy H1 plate reader.

Cell proliferation

Cells expressing Nuclight GFP were placed in 96-well standard-well ultra-low attachment (ULA) cluster plates at a density of 0.5–1× 10⁵ cells per well in 200 µL of medium. Fluorescent images were captured at 4 h intervals in the Incucyte® S3 System (Sartorius). Growth curves and doubling times were calculated using GraphPad Prism 10. Adherent and spheroid doubling time calculations were performed using green and green mean intensity features of the Incucyte® S3 system, respectively.

Organoid culture

Cell line organoid culture

Cells were seeded at a density of 5000 cells/well as droplets in 50 µL of Cultrex Basement Membrane Extract (BME) PathClear Type 2 (Cedarlane, Burlington, ON, Canada) on a 24-well standard tissue culture plate (Corning). Droplets were overlaid with an EOC organoid specific media containing Advanced DMEM/F-12 (Invitrogen) supplemented with B-27™ (Invitrogen), Forskolin (Sigma), GlutaMAX™ (Invitrogen), HEPES (Wisent), Human EGF (Peprotech Inc.), Human FGF-10 (Peprotech Inc.), Nicotinamide (Sigma), N-Acetyl-L-cysteine (Sigma), Recombinant Human Noggin (R&D Systems) and ROCK-inhibitor (Y27632 dihydrochloride, Sigma). Brightfield images were captured every 12 h with the Incucyte® S3 System. The total organoid area (µm²) and number of organoids per well were quantified using the Organoid Analysis Software via the Incucyte® S3 System.

Patient-derived organoid culture

Patient consent for the clinical specimens from which PDO cultures were derived was obtained according to our institution’s research ethics board-approved protocol (#115904). Cells were seeded at a density of 25,000 cells/well as droplets in 50 µL of Matrigel (Corning) on a 96-well standard tissue culture plate (Corning). Droplets were overlaid with an patient-derived organoid specific media containing Advanced DMEM/F-12 (Invitrogen) supplemented with Antibiotic-Antimycotic (WISENT), B-27™ Supplement (Invitrogen), GlutaMAX™ Supplement (Invitrogen), Recombinant Human FGF basic/FGF2/bFGF Protein (Novus Biologicals), Human FGF-10 (Peprotech Inc.), ROCK-inhibitor (Y27632 dihydrochloride, Sigma), Human EGF (Peprotech Inc.), Forskolin (Sigma), N-Acetyl-L-cysteine (Sigma), Recombinant Human Noggin (R&D Systems), Nicotinamide (Sigma), β-Estradiol (Sigma), HEPES (WISENT), N-2 Supplement (Invitrogen), SB431542 hydrate (Sigma), and Human Recombinant BMP-2 (Sigma).

Scratch wound closure migration assay

Confluent cell monolayers were scratched with a pipette tip and immediately imaged (0 h time point). Images were acquired up to 36 h post-scratch and ImageJ [28] was used to measure scratch width and calculate scratch area.

Mesothelial clearance assay

ZT-GFP mesothelial cell line with red fluorescence-positive spheroids

Human ZT-GFP mesothelial cells (1–1.5× 10⁵ cells in 1 mL) were seeded into each well of a 24-well tissue culture plate and incubated for 24 h. An empty well containing 1 mL of media served as a control. Cells expressing luc2tdTomato (OVCAR8 and HEYA8) or Nuclight RFP (ES2) (2000 cells per well) were seeded into 96-well ULA plates and incubated for 24 h then individual spheroids were transferred using a P200 and placed onto mesothelial cell monolayers or control wells, with at least 6 spheroids per well. Green and red fluorescent images were captured 24 h later using a Leica DMI4000B inverted microscope and spheroid displacement was quantified using Fiji (Fiji is just ImageJ).

Primary human mesothelial cells with mouse ascites-derived spheroids

Prior to cell culture, collagen (50 μg) was dissolved in 70% ethanol and added to each well of a 24-well tissue culture plate and incubated at room temperature for 2 h. The wells were then aspirated, washed with PBS, and re-aspirated before adding mesothelial cells. To generate fluorescing primary mesothelial cells, 3.5 μL of recombinant human Ad5-green fluorescent protein (Ad-GFP) vector stock per 100,000 cells was added to the cell suspension, which was then seeded to the collagen layer at a density of 1–1.5× 10⁵ cells and incubated for 24 h. Mouse xenograft ascites-derived spheroids were generated by seeding cells into 96-well ULA plates at 2000 cells per well in 100 μL and incubated for 24 h. The following day, mesothelial cells were washed twice with PBS to remove Ad-GFP-containing media, and 1 mL of fresh media was added to each well, including control wells without mesothelial cells. Spheroids were transferred to a 24-well plate at 6 per well using a P200. Green and red fluorescent images were captured 24 h later and spheroid displacement was quantified as described above.

Xenotransplantation assays

Eight NOD/SCID female mice (8–10 weeks old; Charles River Laboratories) were injected intraperitoneally with luc2tdTomato cells with the following cell numbers in 150 μL PBS: OVCAR8/OVCAR8-ULK1KO, 2× 10⁶; HEYA8/HEA8A8-ULK1KO, 1× 10⁶). Mice were randomly assigned to receive either parental or ULK1KO cells. For survival analyses, mice were monitored daily after intraperitoneal injection and euthanized using established criteria for humane endpoints (lethargy, hunched posture, impaired breathing, weight loss, and excessive ascites) as per protocol guidelines. Animals that died prematurely without evidence of tumor burden (i.e., in the absence of jaundice, lethargy, abdominal distension, or a palpable mass) were excluded from the survival analysis. Mice received weekly injections of D-luciferin (Perkin Elmer, #122799) at 75 mg/kg in 100 µL PBS to monitor tumor progression via BLI using the IVIS Lumina S5 system (PerkinElmer). Tumor locations and evidence of ascites for each mouse was assessed and recorded at necropsy. The mice were provided chow (Envigo, #2919) and water ad libitum throughout the study. All animal experiments were approved by the Institutional Animal Care and Use Committee of the University of Western Ontario (London, Ontario, Canada) and were performed in accordance with approved guidelines.

Immunohistochemistry

Sectioning and staining of tumor specimens were performed by the Molecular Pathology Core Facility at Robarts Research Institute (London, Ontario, Canada). Images of stained tumor sections were captured using an Aperio ScanScope slide scanner (Leica). IHC analysis was performed using the Fiji distribution in ImageJ software [29]. Ki67-positive nuclei were masked using the Trainable Weka Segmentation plugin [30], and masked regions were counted using a minimum particle area of 120 pixels. Cleaved caspase-3 staining in xenograft tumor sections was evaluated using the “IHC Profiler” plugin for ImageJ, as described previously [31]. Positive caspase-3 staining represents the combined “high-positive” and “positive” scores as defined by IHC profiler.

Carboplatin and paclitaxel dose-response curves

To determine carboplatin and paclitaxel half-maximal inhibitory concentration (IC₅₀) values, 2000 cells in 100 µL media were seeded in a standard 96-well plate for adherent culture, allowed to attach for 24 h, then treated with carboplatin or paclitaxel over a 12-point concentration gradient. After 72 h of treatment, cell viability was determined using the alamarBlue Cell Viability Reagent (Invitrogen CAT# DAL 1025) according to the manufacturer’s instructions. To determine carboplatin and paclitaxel IC₅₀ values of spheroids, 2000 cells in 100 µL media were seeded in a 96-well ULA plate. After 72 h, cells were treated individually over a 12-point concentration gradient for an additional 72 h. Following treatment, viability was determined by alamarBlue viability assay. Viability was assessed at 4 h and 48 h post alamarBlue incubation for carboplatin and paclitaxel, respectively, and IC₅₀ values were calculated using GraphPad Prism 10.

Proteomic mass spectrometry

Protein extraction and mass spectrometer analysis were performed on OVCAR8 wild-type and OVCAR8-ULK1KO 24-hour spheroids. Spheroids were lysed using three 5-s pulse rounds of sonication at 35% amplitude lysed OVCAR8 wild-type and OVCAR8-ULK1KO spheroids at 24 h post seeding in lysis buffer containing 200 mM 4-(2-hydroxyethyl)-1-piperazinepropanesulphonic acid (EPPS; pH 8.6), 6 M guanidine, 1 mM PMSF, 100 mM NaF, and phosphatase inhibitor cocktail (2 mM NaF, 2 mM imidazole, 1.15 mM Na₂MoO₄, 1 mM Na₄P₂O₇, 4 mM Na₂C₄H₄O₆, 2 mM Na₃VO₄, and 1 mM β-glycerophosphate). Lysates were incubated in the dark with 5 mM Tris (2-carboxyethyl) phosphine and 15 mM indole-3-acetic acid for 30 and 45 min, respectively, and quenched with 5 mM dithiothreitol. Sera-Mag^TM SpeedBeads (GE Healthcare, Little Chalfont, UK; 65152105050250) were added to the lysates, followed by equal volumes of 100% ethanol. The resulting mixture was incubated on a shaker for 10 min. The supernatant was removed from the mixture, and the beads were washed and resuspended in 50 mM EPPS buffer (pH 8.5). After beads and EPPS buffer were subjected to a 2-h LysC digestion at 1 mAu per 100 μg of protein, trypsin was added at a 1:50 ratio for overnight digestion. The beads were washed with water and 30% acetonitrile the following day to elute the peptides, which were stored at −80 °C.

For mass spectrometry, the peptides were analyzed using a Q Exactive Plus mass spectrometer coupled with an EASYLCn-1000 system (Thermo Fisher Scientific). The peptides were loaded onto an Easy-LCn-1000 and separated on an EASY-Spray 75 μm × 500 mm at 45 °C using an ES803A analytical column (Thermo Fisher Scientific) at a flow rate of 300 nl/min. Raw mass spectrometry data were processed using FragPipe (version 20.0) and Rstudio with the Tidyverse R package for data manipulation, mice R package for imputing missing data, and LIMMA R package for differential expression analysis. Pathway analysis was performed using Kegg [32] (http://bioinformatics.sdstate.edu/go/) and Reactome databases [33] (https://reactome.org).

Spheroid drug treatments and reattachment

Cells were placed in 24-well ULA cluster plates at a density of 5× 10⁴ cells per well in 1 mL of medium. After 24 h, spheroids were treated with individually with carboplatin (100 µM), paclitaxel (50 nM), olaparib (20 µM), AKTi 1/2 (5 µM), trametinib (10 nM), or ralimetinib (15 µM). Spheroid cell viability was performed using Trypan Blue Exclusion assay as described above at 96 h for all drug treatments, except for Olaparib, which was performed at 192 h. For spheroid reattachment assays, inhibitor-treated spheroids were reattached to standard tissue culture plates for 48 h and viability was assessed using alamarBlue at 1:10 dilution in media. Fluorescence was measured using the Agilent Biotek Synergy H1 plate reader.

Statistical analysis

Statistical analyses were performed using GraphPad Prism 10 (GraphPad Software) and the details for specific statistical tests are described in each figure legend.

Results

Differential requirement for ULK1 between EOC and noncancer precursor spheroids

Autophagy induction is controlled by the ULK complex, notably by ULK1 kinase activity [16]. Our previous studies showed that EOC spheroids exhibit elevated ULK1 expression, which is associated with increased autophagy activation, and that targeted ULK1 knockdown or inhibition effectively disrupted autophagy activation [25]. To further elucidate the role of ULK1 in autophagy and tumorigenesis, we ablated ULK1 in OVCAR8, HEYA8, and ES2 EOC cells using CRISPR/Cas9 and pooled multiple independent clones to generate a population of ULK1 knockout (ULK1KO) cells. To validate the effect of ULK1 loss on autophagy, we examined proteins involved in the autophagic pathway. We detected a substantial reduction in LC3II:I across all three EOC ULK1KO spheroids, while significant increases in p62 expression were observed in OVCAR8 and HEYA8 ULK1KO spheroids compared to their parental cell lines (Fig. 1A, B). This is important, as p62 accumulation serves as an indicator of autophagy inhibition, whereas its decrease suggests autophagy induction [34]. Additionally, the LC3II:I ratio, derived from the processing of LC3I to LC3II, a marker of autophagosome membranes, reflects the activation of autophagy [35]. We observed a significant decrease in the LC3II:I ratio observed as early as 4 and 24 h within OVCAR8 and HEYA8-ULK1KO spheroids, respectively (Supplementary Fig. 1A, B). Interestingly, ULK1 loss resulted in an elevation of basal LC3I (Fig. 1A, B) and a significant reduction in p-ATG16L1 (S278) (Supplementary Fig. 2A), supporting the established role of ULK1 in the early processing of LC3 [36]. Additionally, we verified the abrogation of ULK1 activity through loss p-Beclin-1 (S30) levels (Fig. 1A, B), as this is a known substrate for ULK1 activity [37].

Fig. 1. Differential requirement for ULK1 between EOC and noncancer precursor spheroids.

A OVCAR8, HEYA8, and ES2 EOC cells, and FT190 precursor cells with or without intact ULK1 were seeded as adherent and spheroid cultures. Protein lysates were harvested 72 h after seeding for western blot analysis of protein markers of autophagy and ULK1 activity as indicated. B Densitometric analysis of autophagy markers and ULK1 activity in EOC and FT190 cells relative to their expression in parental cells in adherent conditions. Data are displayed as mean ± SEM (N = 3–4); Two-way ANOVA followed by Šidák’s multiple comparisons test, *P < 0.01, **P < 0.001, ****P < 0.0001.

High-grade serous ovarian cancer (HGSOC) is the most prevalent EOC histotype that arises from the preneoplastic lesions in the secretory epithelium of the distal fallopian tube [38]. Therefore, we ablated ULK1 in FT190 cells, an immortalized human fallopian tube secretory epithelial cell line [26]. Although ULK1 loss resulted in reduced phosphorylated Beclin-1 (S30) levels as seen in EOC cells, no significant differences in the LC3II:I ratio were observed between FT190 parental and FT190 ULK1KO spheroids (Fig. 1A, B; Supplementary Fig. 1C), suggesting that ULK1 activity is not essential for autophagy activation in these precursor cells. To address the potential compensation due to ULK1 loss in FT190 cells, we investigated ULK2 protein expression, a homolog of ULK1 that is believed to be redundant in autophagy activation [39]. Although no changes were observed in OVCAR8 and HEYA8 cells with ULK1 loss, we observed a small yet significant increase in ULK2 in ES2 and FT190 ULK1KO adherent cells (Fig. 1A, B). These results suggest that ULK1 is vital for autophagy activation in EOC spheroids, whereas its role may be redundant in HGSOC precursor cells.

Cell viability is significantly impaired in EOC ULK1KO spheroids

Spheroids enhance EOC cell viability during metastasis by protecting them from anoikis and chemotherapy-induced damage [40]. To understand whether ULK1 and autophagy activities contribute to this property, ULK1KO cells were grown in suspension culture to assay for differences in spheroid morphology, density, integrity, and viable cell number. We found that HEYA8-ULK1KO spheroids showed obvious differences in morphology with decreased density and cell number, and impaired integrity compared to parental spheroids, whereas OVCAR8 and ES2-ULK1KO spheroids retained overall morphology but also displayed reduced viable cell number (Supplementary Fig. 3A). A significant reduction in viable cells was observed by Trypan Blue exclusion cell counting on day 7 and 10 across all EOC ULK1KO spheroids, whereas ES2-ULK1KO significantly decreased viable cell number as early as day 3 (Fig. 2A). Although all EOC spheroids increased in cell number up to day 3, expansion plateaued in OVCAR8 spheroids after this time; HEYA8 and ES2 parental spheroids continued to expand in cell number, yet OVCAR8 and HEYA8-ULK1KO spheroids failed to do so (Fig. 2A). Interestingly, loss of ULK1 in FT190 spheroids resulted in a further reduction in viable cells as compared with parental spheroids (Fig. 2A). EOC parental and ULK1KO cells expressing nuclear-localized GFP were generated to facilitate fluorescence imaging as an indirect indicator of spheroid growth. However, no differences in growth rate were observed by fluorescence imaging between EOC parental and ULK1KO spheroids (Supplementary Fig. 3B). To determine whether reduced viable cell number in ULK1KO spheroids occurs via cytostasis or cell death, we assessed markers for proliferation and apoptosis. Since EOC spheroids display features of quiescent cells compared to adherent cells [8], we evaluated the expression of tumor suppressor proteins p21 and p27, which are established markers of tumor cell dormancy [41]. In line with this, we have shown previously that cellular quiescence in EOC spheroids is associated with increased p27 [8]. Herein, we observed significant reductions in p21 and p27 expression in HEYA8 and ES2-ULK1KO spheroids, yet in contrast these markers exhibited significant increases in OVCAR8-ULK1KO spheroids (Fig. 2B). We measured apoptosis activity in spheroids at multiple time points using the Caspase 3/7 Glo assay. We observed a significant surge in apoptosis within 24–48 h of spheroid formation, with elevated activity persisting for up to 72 h in HEYA8, OVCAR8 and FT190 spheroids due to ULK1 ablation (Fig. 2C). Interestingly, we observed decreased apoptotic activity in ES2-ULK1KO spheroids across all time points (Fig. 2C). These findings demonstrate that one primary mechanism whereby ULK1 loss impairs spheroid cell viability may be through apoptosis induction.

Fig. 2. Viability is significantly impaired in EOC ULK1KO spheroids.

A The number of viable OVCAR8, HEYA8, ES2 and FT190 cells were counted by Trypan Blue Exclusion Assay. Data displayed as mean ± SEM (N = 3, with n = 3 technical replicates per experiment); Two-way ANOVA followed by Šidák’s multiple comparisons test, *P < 0.01, **P < 0.001, ****P < 0.0001. B OVCAR8, HEYA8, and ES2 parental and ULK1KO cells were seeded for spheroid culture and protein lysates were harvested at 72 h for western blot analysis. Densitometric analysis of p27 and p21 in EOC spheroids relative to expression in parental spheroid conditions. Data displayed as mean ± SEM (N = 3–4); Student’s t test, *P < 0.05, **P < 0.001. C Apoptosis activity was measured using Caspase-Glo 3/7 luminescence assays. Cells were seeded at 2000 cells per well in a 96-well ULA plate. Data reflects luminescence relative to parental spheroids at each time point. Data displayed as mean ± SEM (N = 3–4, with n = 5 technical replicates per experiment); Student’s t test, *P < 0.05, **P < 0.001, ***P < 0.0001.

ULK1 is required for EOC tumor growth and spread in xenograft models

Given our findings that ULK1 is critical for autophagy activation and spheroid viability, we sought to investigate its role in EOC tumor formation and metastasis directly, areas that have not been explored previously. To initiate these studies in cell culture, we grew parental and ULK1KO cells as matrix-embedded organoids for up to 18 days and assessed expansion over time. Although total organoid number was similar between parental and ULK1KO lines, there was a significant reduction organoid area over time in OVCAR8 and HEYA8-ULK1KO lines, while no differences were observed in ES2-ULK1KO cells (Fig. 3A; Supplementary Fig. 4A). To investigate ULK1 function in spheroid attachment, invasion and migration, we used the mesothelial clearance assay, an experimental model that mimics the early steps of EOC metastasis [42]. We transferred pre-formed red fluorescent EOC spheroids onto either standard tissue culture plastic or GFP-expressing ZT human mesothelial cells. This allowed us to separately quantify spheroid attachment and dispersion from mesothelial cell displacement properties. Interestingly, OVCAR8-ULK1KO cells displayed significantly reduced ability to disperse on tissue culture plastic, ES2-ULK1KO cells had a significantly enhanced ability to displace on tissue culture plastic, while HEYA8-ULK1KO cells were unchanged relative to parental cells (Supplementary Fig. 4B). However, all EOC ULK1KO spheroids displayed a significantly reduced ability to displace mesothelial cells (Fig. 3B). These findings appear to be autophagy-dependent, as treatment of parental cell line spheroids with chloroquine at the time of reattachment inhibited mesothelial clearance to a similar degree to ULK1 loss alone (Supplementary Fig. 4C, D). These findings of altered cell motility properties were recapitulated using a scratch wound closure assay, where ULK1 loss significantly decreased wound closure rate in OVCAR8 cells but not in HEYA8 cells (Supplementary Fig. 5). These cell culture-based results suggest ULK1 may have additional functions to promote EOC spheroid cell invasiveness and metastatic capacity in vivo.

Fig. 3. ULK1 loss impairs EOC organoid growth and spheroid invasion.

A EOC parental and ULK1KO cells were grown as organoids and images were captured for up to 18 days using the IncuCyte S3 Live-Cell Analysis System. Scale bar represents 800 μm and 200 μm, respectively. Organoid area was quantified using the IncuCyte S3 Live-Cell Analysis System. Data reflects the total organoid area and displayed as mean ± SEM (N = 3, with n = 4 technical replicates per experiment); Two-way ANOVA followed by Šidák’s multiple comparisons test; *P < 0.01, **P < 0.001. B The mesothelial clearance assay was utilized to evaluate EOC spheroid invasion. Spheroids expressing tdTomato (OVCAR8 and HEYA8) or Nuclight Red (ES2) were seeded onto ZT-GFP cells and imaged 24 h later. Displacement was quantified using Fiji. Data reflects area of ZT-GFP cells displaced normalized to the original spheroid size and displayed as mean ± SEM (N = 3, with n = 6 technical replicates per experiment); Student’s t test, *P < 0.05, **P < 0.001. Scale bar represents 300 μm.

Fig. 4. ULK1 is required for EOC tumor growth and spread in xenograft models.

A Representative bioluminescent images of mice injected i.p. with OVCAR8 parental and ULK1KO cells. The BLI signal for each image is normalized independently based on the optimal imaging parameters for that specific mouse, including binning, radiance, and exposure values. Scale bar indicates luminescence intensity, with red representing higher luminescence values to blue representing lower luminescence values. B Total flux (photons/sec) was used as a measure of tumor burden as assessed weekly via bioluminescence imaging. Data are displayed as mean ± SEM; Student’s t test, *P < 0.05, **P < 0.01. C Petal plots representing the proportion of mice displaying tumors at peritoneal sites or presence of ascites, as indicated. Radial gridlines represent 10% gradations (OVCAR8, n = 6; OVCAR8-ULK1KO, n = 7). D Representative bioluminescent images of mice injected i.p. with HEYA8 parental and ULK1KO cells. E Total flux (photons/sec) was as a measure of tumor burden assessed weekly via bioluminescence imaging. Data are displayed as mean ± SEM; Student’s t test, *P < 0.05. F Petal plots representing the proportion of mice displaying tumors at peritoneal sites or presence of ascites, as indicated. Radial gridlines represent 10% gradations (HEYA8, n = 7; HEYA8-ULK1KO, n = 6).

To model EOC metastasis, whereby malignant cells from the primary tumor are shed directly into the peritoneal cavity [2, 3], we injected luciferase/tdTomato-expressing cells intraperitoneally into female NOD/SCID mice and monitored tumor progression over time via BLI (Fig. 4A, D). Due to the encouraging results from in vitro organoid experiments and the availability of luciferase-expressing cell lines, these experiments were performed with OVCAR8- and HEYA8-ULK1KO cells and their respective controls. Mice injected with OVCAR8-ULK1KO cells showed reduced tumor burden at all time points, with significant decreases observed during the mid-to-late stage of disease progression (Fig. 4B). In contrast, HEYA8-ULK1KO cells had a significant decrease only at very early stages of disease progression (Fig. 4E), which was lost at later time points. At experimental endpoint, EOC cells with ULK1 loss resulted in fewer tumor lesions observed at several metastatic sites, with notable decreases in ascites formation and omental metastasis (Fig. 4C, F), two canonical features of metastatic EOC. Despite reduced tumor growth and metastatic spread, no significant differences were observed in survival rates (Supplementary Fig. 6A). In addition, no differences in either Ki67- or Caspase-3-positive IHC staining on tumor samples were seen (Supplementary Fig. 6B). Taken together, our findings suggest that ULK1 impacts EOC progression by affecting intrinsic tumor cell growth, and spheroid adhesion and invasion at later steps of metastasis.

ULK1 knockout does not sensitize EOC spheroids to standard-of-care treatment

Elevated autophagy levels have been linked to poor prognostic outcomes in cancer patients due to cytotoxic drug resistance [43]. Since we observed reduced tumor burden and metastasis due to ULK1 ablation in injected EOC cells, yet this did not alter overall survival, we tested whether ULK1 loss and autophagy disruption would sensitize EOC spheroids to chemotherapy. Using our in vitro spheroid model system, we assessed cell viability by treating spheroids with either carboplatin or paclitaxel as single agents. Carboplatin treatment resulted in a significant increase in viable cells in OVCAR8-ULK1KO spheroids and no difference in HEYA8-ULK1KO spheroids. No significant differences in viability were observed between parental and ULK1KO spheroids under paclitaxel treatment (Fig. 5A, B). We also treated spheroids with Olaparib, a poly (ADP-ribose) polymerase (PARP) inhibitor used as maintenance therapy in select HGSOC patients [44]. Both parental and ULK1KO spheroids showed significant sensitivity to Olaparib, yet ULK1 loss did not alter this effect (Fig. 5C). These results suggest that while ULK1 knockout reduces spheroid cell viability, its combination with standard-of-care therapies elicited no further improvement, underscoring our subsequent studies to identify alternative strategies to improve efficacy.

Fig. 5. ULK1 loss does not sensitize EOC spheroids to standard-of-care therapeutics.

Spheroids of each indicated cell line were cultured for 24 h before treating with either A carboplatin (100 µM) or B paclitaxel (50 nM) for 72 h, or C olaparib (20 µM) for 168 h. Viable cell number was measured by Trypan Blue Exclusion Assay. Data displayed as mean ± SEM (N = 3, with n = 3 technical replicates per experiment); Two-way ANOVA followed by Šidák’s multiple comparisons test, *P < 0.01, **P < 0.001, ****P < 0.0001.

ULK1 loss disrupts key cell survival pathways in EOC spheroids

In addition to its well-established role as a primary regulator of autophagy, ULK1 plays critical roles in energy metabolism, mitochondrial homeostasis, and vesicular trafficking [16]. However, its non-canonical functions in EOC remain largely unexplored yet may be implicated in driving EOC progression and thus serve as new therapeutic targets to improve the limited efficacy of standard chemotherapies as seen in our findings. To this end, we performed proteomic mass spectrometry and bioinformatic analyses on OVCAR8 and OVCAR8-ULK1KO spheroids to identify potential ULK1-regulated pathways in our experimental system (Supplementary Table S1). Through KEGG and Reactome pathway analyses of our resultant dataset, we found significant changes in biological pathways related to cell survival, including apoptosis, and PI3K-AKT-mTOR and MAPK signaling that were shared between both analyses (Fig. 6A; Supplementary Tables S2, S3). Since we had already observed enhanced apoptosis in ULK1KO spheroids (Fig. 2C), we sought to validate members of PI3K-AKT-mTOR and MAPK signaling pathways, given their critical roles in regulating tumor progression [45, 46]. We observed significant reductions in p-MEK (S217/221) and p-p38 (Thr180/Tyr182) levels in OVCAR8 and HEY8-ULK1KO spheroids, while expression of p-MEK (S217/221) was unchanged and p-p38 (Thr180/Tyr182) was increased in ES2-ULK1KO spheroids. No changes in downstream p-ERK (Thr202/Tyr204) were detected in EOC ULK1KO spheroids (Fig. 6B, C; Supplementary Fig. 7A). This trend extended to the PI3K-AKT-mTOR pathway, with a decrease in p-AKT (S473) in OVCAR8 and HEYA8-ULK1KO spheroids and yet an increase in ES2-ULK1KO spheroids. Further downstream analysis of AKT revealed contrasting effects, with increased P70S6K phosphorylation (T389) in OVCAR8 and HEYA8-ULK1KO spheroids, while no differences were observed in ES2-ULK1KO spheroids. Interestingly, we observed decreased 4EBP1 phosphorylation (Thr37/46) in OVCAR8 and HEYA8-ULK1KO spheroids (Supplementary Fig. 7B).

Fig. 6. ULK1 loss disrupts cell survival pathways in EOC spheroids.

A Differentially expressed proteins identified by label-free proteomics of OVCAR8 parental and ULK1KO spheroids were applied to KEGG and Reactome pathway enrichment analysis using STRING and Cytoscape applications. Proteins increased or decreased in OVCAR8- ULK1KO spheroids versus parental controls identified major pathways with the respective gene counts and false discovery rates as indicated. B EOC parental and ULK1KO cells were seeded into adherent and spheroid cultures. Protein lysates were harvested 72 h after seeding for western blot analysis of markers representing MEK-MAPK and PI3K-AKT-mTOR signaling pathways. C Densitometric analysis of western blots was performed relative to parental adherent conditions. Phosphorylated proteins were normalized to their respective total protein levels (except p38 Thr180/Y182 that was normalized to vinculin). Data are displayed as mean ± SEM (N = 3–4); Two-way ANOVA followed by Šidák’s multiple comparisons test; *P < 0.01, **P < 0.001, ***P < 0.0001.

Female mice xenografted with OVCAR8 and OVCAR8-ULK1KO cells developed malignant ascites with reduced prevalence due to ULK1 loss (Fig. 4C). These ascites samples were returned to cell culture to study whether their inherent pathobiology had changed during disease progression in mice. OVCAR8-ULK1KO spheroids from ascites-derived lines remained autophagy deficient, as evidenced by a decreased LC3II:I ratio, increased p62, suppressed ULK1 activity with reduced p-Beclin-1 S30, and lower ULK2 expression. Phosphorylated-MEK1/2 (S217/221) was reduced, which was consistent with original OVCAR8-ULK1KO spheroid cells. However, the changes in p-AKT (S473), p-P70S6K (T389), and p-p38 (Thr180/Tyr182) seen in pre-injection OVCAR8-ULK1KO spheroids were not observed in ascites-derived lines, although total AKT protein decreased and total P70S6K increased in ascites-derived lines (Supplementary Fig. 7A, B). Despite the observed changes in MEK-MAPK and PI3K-AKT-mTOR signaling proteins, ascites-derived OVCAR8-ULK1KO spheroids retained impaired metastatic potential in mesothelial clearance and reattachment assays, indicating persistent functional defects post-xenografting (Supplementary Fig. 7C, D). Collectively, these observations suggest that ULK1 disruption leads to reprogramming of key signaling pathways known to impact tumor progression and cancer cell survival.

ULK1 ablation enhances efficacy of MEK and mTOR inhibition

To explore novel treatment strategies that achieve enhanced efficacy in the context of ULK1 loss, we targeted the dysregulated PI3K-AKT-mTOR and MEK-MAPK pathways in ULK1KO spheroids using specific inhibitors. Treatment with AKTi-1/2 (AKT inhibitor) and ralimetinib (p38 inhibitor) resulted in significantly increased cell viability in OVCAR8-ULK1KO spheroids, while no differences were observed in HEYA8-ULK1KO spheroids (Supplementary Fig. 9A). However, treatment with either trametinib (MEK inhibitor) or AZD-8055 (mTORC1/2 inhibitor) resulted in significantly reduced cell viability in ULK1KO spheroids (Fig. 7A), indicating that ULK1 loss may sensitize EOC spheroid cells to MEK and mTOR inhibition. We conducted spheroid reattachment assays to evaluate the effect of these inhibitors on this key step in the metastatic process [47]. Untreated OVCAR8-ULK1KO spheroids had significantly reduced reattachment, which was further decreased due to trametinib and AZD-8055 treatment (Fig. 7B). So too did treatment with trametinib and AZD-8055 significantly reduce HEYA8-ULK1KO spheroid reattachment (Fig. 7B). Interestingly, treatment with AZD-8055 failed to reduce spheroid reattachment in ES2 cells, although it did affect viability (Fig. 7B).

Fig. 7. Genetic and pharmacologic ablation of ULK1 combined with MEK and mTORC1/2 inhibition.

A Spheroids were cultured for 24 h before treating with either trametinib (10 nM) or AZD-8055 (10 nM) for 72 h before performing Trypan Blue Exclusion Assay. Data displayed as mean ± SEM (N = 3, with n = 3 technical replicates per experiment); Two-way ANOVA followed by Šidák’s multiple comparisons test, *P < 0.01, **P < 0.001, ****P < 0.0001. B Inhibitor-treated spheroids were transferred to standard tissue-culture treated plates and cell viability was assessed at 48 h post-reattachment by alamarBlue assay. Data displayed as mean ± SEM (N = 3, with n = 3 technical replicates per experiment); Two-way ANOVA followed by Šidák’s multiple comparisons test, *P < 0.01, **P < 0.001, ****P < 0.0001. C, D EOC cells were seeded at 2000 cells per well in a 96-well ULA plate and treated after 24 h with DCC-3116 (1$\mathrm{\mu }\mathrm{M}$), trametinib (10 nm), or AZD-8055, alone or in combination. Viability was assessed after an additional 72 h using an alamarBlue assay. Data displayed as mean ± SEM (N = 3, with n = 5 technical replicates per experiment); One-way ANOVA followed by a Tukey’s multiple comparisons test, *P < 0.05, **P < 0.001, ****P < 0.0001.

To further understand the therapeutic potential of targeting ULK1 in advanced ovarian cancer, we inhibited ULK1 in EOC spheroids using DCC-3116, a specific ULK1/2 inhibitor that is currently in phase 1/2 clinical trials as monotherapy or in combination with MEK1/2 (trametinib) or EGFR inhibitors (NCT04892017 and NCT05957367). To confirm that DCC-3116 inhibits ULK1 in EOC, we treated OVCAR8 and ES2 spheroids and assessed p-BECN1 S30 as a readout of ULK1 activity. Indeed, we observed a substantial reduction of p-BECN1 S30 at both 1μM and 2μM DCC-3116, with a concordant decrease in spheroid cell viability (Supplementary Fig. 10). We then treated EOC spheroids with 1μM DCC-3116 alone or in combination with trametinib (10 nM) and AZD-8055 (10 nM). Monotherapy of DCC-3116, trametinib, and AZD-8055 significantly decreased EOC spheroid viability, while combination treatment did not further reduce viability (Fig. 7C, D). The majority of reduced spheroid cell viability appeared to be driven by either trametinib or AZD-8055 alone.

Our findings demonstrate PI3K-AKT-mTOR and MEK-MAPK signaling pathways contribute to EOC spheroid viability and metastatic properties and may represent important therapeutic targets particularly when combined with ULK1 ablation and autophagy blockade.

ULK1 expression associates with worse survival, and clinical ULK1 inhibitor decreases patient-derived organoid viability

We previously reported that ULK1 mRNA expression is negatively associated with 5-year overall and progression-free survival [25]. Extending this analysis with updated datasets to 10 years [48], we again found that high ULK1 mRNA correlates significantly with worse overall and progression-free survival (Fig. 8A). To contextualize expression across disease states, TNMplot analysis showed highest ULK1 mRNA in metastasis, significantly higher than both normal and primary tumor, with no significant differences between normal and tumor (Fig. 8B) [49]. Motivated by these observations and our functional data on ULK1 in disease progression, we tested DCC-3116 in ex vivo cultures of patient-derived organoids (PDOs) isolated at debulking surgery from metastatic sites. In PDO 24 and PDO 103, single-agent DCC-3116, trametinib, or AZD-8055 reduced organoid viability, yet combinations of DCC-3116 with trametinib or AZD-8055 did not yield additional benefit. By contrast, PDO 66 and PDO 102 were insensitive to all single agents and combinations (Fig. 8C, D). Collectively, these results underscore inter-patient heterogeneity, indicate that a subset of advanced EOC models are responsive to ULK1 inhibition alone, and support biomarker-guided strategies to identify patients likely to benefit from targeting these pathways.

Fig. 8. Implications of ULK1 expression and activity on ovarian cancer patient metastasis.

A Correlation of low and high ULK1 expression with overall survival (n = 1656) and progression-free survival (n = 1435) among ovarian cancer patients, regardless of histotype and stage, using TCGA and GEO gene expression microarray datasets from the online tool accessed at www.kmplot.com/ovar. Hazard ratios and log rank tests indicate a significantly worse prognosis due to high ULK1 expression in ovarian tumors. B Publicly available gene chip data were obtained using the TNMplot database, which aggregates expression profiles from TCGA, GTEx, and GEO cohorts. ULK1 mRNA levels were compared across normal ovarian tissue, primary ovarian tumors, and metastatic ovarian cancer samples. Expression values were downloaded, re-graphed in GraphPad Prism 10, and analyzed using a one-way ANOVA followed by a Tukey’s multiple comparisons test, ****P < 0.0001. C, D Patient-derived organoids were seeded at 25,000 cells per well in 96-well standard tissue culture plates and allowed to grow for 7 days before treatment with DCC-3116 (1$\mathrm{\mu }\mathrm{M}$), trametinib (10 nM), or AZD-8055 (10 nM), alone or in combination. Viability was assessed after an additional 72 h using an alamarBlue assay. Data displayed as mean ± SEM (n = 6–8 replicates per experiment); One-way ANOVA followed by a Tukey’s multiple comparisons test, *P < 0.05, **P < 0.001, ****P < 0.0001.

Discussion

Epithelial ovarian cancer is a highly lethal gynecologic cancer characterized by late-stage diagnosis, high relapse rates, and the formation of chemo-resistant spheroids that contribute to peritoneal metastasis. To the best of our knowledge, this is the first study to elucidate the role of ULK1 function using the combination of in vitro, ex vivo, and in vivo models of EOC metastasis. Our findings revealed that beyond its role in regulating autophagy, ULK1 deficiency significantly impacted tumor progression, leading to reduced spheroid viability, diminished invasive capacity, and impaired organoid growth. Additionally, our tumor xenograft models demonstrated that ULK1 loss significantly decreases tumor growth and spread, highlighting its critical role in supporting key processes in EOC metastasis and tumor development. Our study is underscored by our proteomic mass spectrometry analysis, which revealed dysregulated PI3K-AKT-mTOR and MAPK signaling and increased sensitivity to MEK and mTOR inhibition in ULK1KO spheroids. The findings provide new insights into potential autophagy-independent functions of ULK1 and highlight novel therapeutic strategies in EOC.

As we expected, ULK1 loss ablated LC3II:I processing and elevated p62 levels in EOC spheroids. Although our previous research suggests that Beclin-1 is not required for autophagy activation in EOC spheroids [50], the absence of p-Beclin-1 (S30) in ULK1KO cells highlights its utility as a specific biomarker for ULK1 activity in EOC, particularly if assessing on-target activity of ULK1 inhibitors. Interestingly, ULK1 loss did not impair autophagy activation in the FT190 HGSOC precursor cell line. Instead, there was a substantial increase in ULK2 protein expression in FT190 ULK1KO cells suggesting a compensatory mechanism for autophagy activation. Furthermore, ULK1 loss in ES2 spheroids failed to ablate LC3 processing and p-BECN1 S30 levels, and like FT190 cells, had elevated ULK2 expression in adherent conditions. Consistent with our findings, a previous study reported that ULK1 inhibition did not alter ULK2 expression in HGSOC cells [51]. Additionally, ULK2 has been shown to compensate for ULK1 loss in mouse embryonic fibroblasts, but not in cerebellar granule neurons [52]. Our observations suggest ULK2 may serve to activate autophagy in precancerous and select malignant EOC cells to compensate for ULK1 loss.

EOC metastasis occurs by direct dissemination of tumor cells into the peritoneal cavity, where spheroid clusters suspended in ascites promote secondary tumor formation through enhanced survival, adhesion, and invasiveness. Using spheroid models that mimic these unique metastasis mechanisms, our results highlight ULK1’s pivotal role in EOC progression. ULK1 loss significantly impaired viability, increased apoptosis, and reduced invasive capacity of EOC spheroids. These reductions in spheroid cell viability appear to be driven by increased apoptosis rather than altered cell growth. Anoikis, a programmed cell death triggered by the loss of cell attachment to the extracellular matrix, serves as a critical barrier to metastasis [53]. Spheroids, however, resist anoikis by activating autophagy [54] and PI3K-AKT-mTOR and MAPK signaling to promote apoptosis resistance [55]. Given our findings of dysregulated PI3K-AKT-mTOR and MAPK signaling in ULK1KO spheroids and the role of autophagy as a defense mechanism preceding apoptosis [56], the disruption of these survival pathways likely compromises the ability of spheroids to resist anoikis, leading to increased apoptosis and reduced viability.

In addition to reduced tumor cell dissemination, we observed fewer metastases and reduced ascites formation. We initially speculated that this reduction in secondary tumors was primarily due to ULK1’s role in regulating cell viability in suspension rather than directly on secondary growth or altered invasive capacities. However, we observed significant reductions in mesothelium invasion of spheroids lacking ULK1. Our findings corroborate a previous study demonstrating that inhibiting autophagy restricts the invasiveness of ovarian cancer cells via the negative regulation of p62 on ERK1/2 activity for invadopodium formation [57]. In addition, ULK1 loss significantly impaired the growth of OVCAR8 and HEYA8 EOC cells in organoid culture, while it did not impair the growth of ES2 organoids. Taken together, these data suggest that metastatic cells reaching secondary sites may exhibit compromised invasiveness and impaired re-initiation of tumor growth.

In xenograft models, ULK1 knockout reduced tumor burden and limited metastatic spread, demonstrating its importance in systemic disease progression. Similarly, evidence from both in vitro and in vivo investigations indicate that ULK1 depletion significantly reduces pancreatic and hepatocellular carcinoma growth [22, 23]. However, ULK1’s role in cancer development might be context-specific among different malignancies. For example, in breast cancer models, ULK1 loss has been associated with an increased likelihood of osseous metastasis [58]. In contrast, EOC rarely metastasizes to distant sites such as the lungs, skin, bones, or brain via hematogenous dissemination, highlighting the underlying mechanisms driving EOC metastasis are biologically different from many other carcinomas [59]. Despite significant reductions in tumor burden, mice injected with ULK1 knockout cells did not exhibit improved survival. These mice developed distended abdomens, jaundice, and significant weight loss, ultimately requiring sacrifice in accordance with approved guidelines. This underscores the lethality of EOC, as tumor cells, even with ULK1 loss, can still disseminate to vital abdominal organs, driving disease progression and mortality.

We reported that autophagy levels and ULK1 mRNA overexpression are correlated with poor survival outcomes in advanced-stage ovarian cancer [25], making it a promising therapeutic target. While combining ULK1 inhibition with standard-of-care chemotherapeutics might seem promising to enhance anti-tumor effects, the results are not uniformly positive. Previous studies have suggested that ULK1 loss can enhance chemotherapy sensitivity in OVCAR8 cells [60]; however, our findings, along with in vitro data [61], indicate that ULK1 inhibition may instead reduce the efficacy of standard first-line chemotherapeutics used in EOC. Previous studies in gastric cancer have shown that elevated p62 expression activates the transcription of Chemokine C-C motif ligand 2 (CCL2), a cytokine associated with drug resistance and contributes to cisplatin resistance [62]. The significant increase in p62 expression we observed following ULK1 loss may similarly underlie the reduced chemotherapeutic efficacy in our system. While p62 accumulation is often used as a marker of impaired autophagy, p62 also functions as a signaling hub that activates NRF2, NF-κB, and mTORC1 pathways to promote tumor survival and stress resistance [63]. In addition, p62 has been linked to impaired DNA damage repair, further contributing to therapy resistance in several cancers [64]. Thus, p62 may not only reflect disrupted autophagy in our models but may also contribute to EOC spheroid viability.

While ULK1 is widely recognized as a critical regulator of autophagy, its autophagy-independent functions, especially in the context of EOC, have been less studied. As such, we performed protein mass spectrometry on OVCAR8 and OVCAR8-ULK1KO spheroids, which verified our findings of increased apoptosis and revealed significant alterations in critical signaling pathways, including PI3K-AKT-mTOR and MAPK signaling. The PI3K-AKT-mTOR pathway is a hallmark cancer promoter that governs essential processes such as cell growth, motility, survival, and metabolism [65]. Similarly, MAPK pathways play a central role in regulating fundamental processes such as cell proliferation, differentiation, and stress responses [46]. We demonstrated a dysregulation in these pathways due to ULK1 loss, highlighted by decreases in p-MEK (S217/22), p-p38 (T180/Thr182), and p-AKT (S473), and an increase in p-p70S6K (T389) in OVCAR8 and HEYA8 EOC ULK1KO spheroids, whereas there were increases in p-p38 (T180/Thr182) and p-AKT (S473) in ES2 ULK1KO spheroids. MAPK signaling is implicated in ovarian cancer, regulating autophagy and apoptosis [66] and, together with AKT, promote invasion and proliferation [67]. Separately, ULK1 links mTOR/AMPK-driven growth control and MAPK crosstalk via p38, and autophagy sustains RTK/endosomal signaling that feeds ERK/AKT [68, 69]. However, to our knowledge, a direct ULK1-dependent dysregulation of MEK-MAPK and PI3K-AKT-mTOR signaling in ovarian cancer models has not been demonstrated.

Therefore, we were intrigued to assess the therapeutic potential of targeting these pathways in our ULK1 deficient system. While AKT and p38 inhibition significantly decreased EOC spheroid viability, ULK1 loss did not increase drug sensitivity and, in some cases, exhibited potential antagonistic effects. However, ULK1 loss significantly enhanced the sensitivity of EOC spheroids to MEK inhibition via trametinib and mTORC1/2 inhibition via AZD-8055. Many preclinical and clinical studies have explored targeting the PI3K-AKT-mTOR pathway, including the use of mTOR inhibitors, in EOC [70]. However, the clinical application of AZD-8055 is limited by its pharmacokinetics, inadequate intratumoral concentrations, and dose-limiting toxicities [71]. Its successor, AZD-2014 (vistusertib), initially showed reduced liver toxicity [71], but recent studies, such as those in meningiomas, reported poor tolerance, with most participants discontinuing the trial [72]. Interestingly, combination therapy with vistusertib and anastrozole in advanced hormone receptor-positive endometrial cancer demonstrated manageable adverse events and improved overall response rates and progression-free survival (NCT02730923) [73]. Trametinib represents a new standard-of-care option for relapsed or persistent low-grade serous ovarian cancer [74], and has shown success in a patient with recurrent HGSOC who had several lines of prior therapy [75]. These findings suggest that blocking autophagy via ULK1 inhibition combined with MEK inhibition via trametinib or mTORC1/2 inhibition could be effective in EOC.

We explored the translational potential of ULK1 blockade using DCC-3116, a selective ULK1 inhibitor being developed for combination with autophagy-activating targeted agents (e.g., MEK/RTK inhibitors such as trametinib) [21, 76]. Across EOC spheroids, DCC-3116, trametinib, and AZD-8055 impaired viability, but PDO responses were heterogeneous: in two of four metastatic PDOs, single agents were effective and adding trametinib or AZD-8055 to DCC-3116 provided no additional benefit. Patient-derived organoids (PDOs) are highly heterogeneous, reflecting diverse patient populations and different stages of disease progression [77]. It is therefore likely that only a subset of patients will benefit from ULK1 inhibitors, either as monotherapy or in combination with cytotoxic or targeted agents. Future work should focus on identifying which patients are most likely to respond, and on determining the optimal doses and combination strategies needed to achieve a therapeutic benefit.

Conclusion

Our comprehensive analysis underscores ULK1’s multifaceted role in EOC, where its influence extends beyond autophagy regulation to impact key cell survival pathways, particularly apoptosis and MAPK and PI3K-AKT-mTOR signaling networks. Although ULK1 loss did not enhance the efficacy of standard-of-care chemotherapeutics, it significantly sensitized EOC spheroids to MEK and mTORC1/2 inhibition. Since higher ULK1 expression portends a poorer prognosis for advanced OC patients, our research provides additional evidence that targeting ULK1 may offer a therapeutic strategy for controlling metastatic tumor growth and spread in select EOC patients.

Author contributions

JDW and TGS conceptualized and designed the study. JDW, EJT, LV, YRV, OH and AB acquired data. TGS supervised and obtained funding for this study. BS, MJB, YRV and SL provided additional resources. JDW wrote the original draft of the manuscript and TGS edited the manuscript. All authors have read and approved the final manuscript.

Funding

We acknowledge the funding support from the Cancer Research Society to TGS and the London Health Sciences Foundation through donations to the Mary and John Knight Translational Ovarian Cancer Research Unit. JDW was supported by an Obstetrics & Gynecology Graduate Scholarship from the Department of Obstetrics and Gynecology at the Western University and a Queen Elizabeth II Graduate Scholarship in Science and Technology (Ontario Government).

Data availability

All data generated or analyzed during this study are included in this published article and its supplementary information files.

Competing interests

The authors declare no competing interests.

Ethics approval and consent to participate

All animal experiments were approved by the Institutional Animal Care and Use Committee of the University of Western Ontario (London, Ontario, Canada) and were performed in accordance with approved guidelines.

Supplementary information

The online version contains supplementary material available at 10.1038/s41388-026-03702-2.