Small-Molecule MMRi62 Induces Ferroptosis and Inhibits Metastasis in Pancreatic Cancer via Degradation of Ferritin Heavy Chain and Mutant p53

This article is featured in Highlights of This Issue, p. 481

Pancreatic ductal adenocarcinoma (PDAC) ranks the fourth leading cause of cancer-related deaths in the Western world (1). The estimated new cases of pancreatic cancer were 60,430 in the United States for 2021 with an estimated 48,220 deaths (1), and 90,100 in China with estimated 79,400 deaths for 2015 (2). The PDAC mortality rate has not been much improved in three decades because of late diagnosis, early metastasis, and limited response to chemotherapy and radiotherapy, and are projected to surpass breast, prostate, and colorectal cancers to become the second leading causes of cancer-related death by 2030 (3). Even for resectable tumors, surgery alone only has limited benefit, as >90% of patients relapse and then die after surgery without additional therapy (4). Surgery followed by adjuvant chemotherapy (gemcitabine plus capecitabine) is the standard of care for resectable tumors and the median survival in these patients is 26 months, with a 5-year survival of 30% (5). FOLFIRINOX regimen (5-fluorouracil/leucovorin with irinotecan and oxaliplatin) and gemcitabine/nab-paclitaxel has been used for advanced PDAC but with severe toxicity that limits patient enrollment (6). Therefore, new therapeutic strategies for PDAC are highly anticipated (5).

TP53 mutation is one of the key drivers of tumorigenesis in most human cancer types (7). Mutant p53 proteins (mut-p53) often lose wild-type p53 (wt-p53) function whereas several hot-spot p53 mutants even acquire novel oncogenic functions, namely gain-of-function driving aberrant cell proliferation, chemoresistance, disruption of tissue architecture, migration, invasion, and metastasis (8). It was established in mouse models that KRAS and TP53 mutations are sufficient to promote PDAC development (9). In humans, four major driver genetic mutations in PDAC include KRAS, CDKN2A, TP53, and SMAD4 (10, 11). Genomic sequencing of primary PDAC pancreatic tumors found 60%–70% bearing TP53 mutations that occur in late-stage PanIN lesions, suggesting critical roles of TP53 mutation in PDAC transformation and progression (12, 13). In addition, mut-p53 drives PDAC metastasis in mouse models (14, 15) and confers PDAC resistance to chemotherapies (16).

Although mutant RAS-targeted therapies are still lacking, screens of RAS-selective lethal compounds identified that RAS-driven pancreatic cancer cells are predisposed to ferroptosis, a mode of cell death involving reactive oxygen species (ROS) production and autophagic degradation of ferritin (17, 18). Although the molecular mechanisms regulating ferroptosis is under active research, identification of novel pro-ferroptotic agents represents a new area of potential PDAC therapeutics. In this report, we describe small-molecule MMRi62 as a ferroptosis inducer in pancreatic cancer cells by mechanisms involving induced degradation of mut-p53 and ferritin heavy chain (FTH1) that may have potential to suppress both PDAC growth and metastasis.

Small-molecule MMRi62 was synthesized in house at Chemler's laboratory. The Betti reaction was for MMRi62 synthesis as previously described in literatures (19, 20). The structure and purity of the synthesized MMRi62 were determined by NMR and the purity was determined by high-resolution mass spectra obtained at SUNY Buffalo's mass spec facility on a ThermoFinnigan MAT XL spectrometer. Bafilomycin A1 was obtained from Sigma (Cat# 19–148) and Carfilzomib was purchased from Cayman Chemical (Cat# 17554). All the compounds were dissolved in DMSO as 10 mmol/L stocks and stored at −20°C.

Panc1, BxPc3, panc10.05, AsPc1, Capan2, and SW1990 pancreatic cancer cells were purchased in December 2017 from the ATCC and cultured as recommended. Cells were regularly authenticated by growth rate and morphologic observation. Panc1, BxPc3, panc10.05, AsPc1, Capan-2, and AsPc1cells were grown in a humidified incubator at 37°C under an atmosphere of 5% CO₂ in air and periodically checked for Mycoplasma contamination. SW1990 cells were grown in a humidified incubator at 37°C without CO₂ in air. Genetic status of KRAS and TP53 were obtained from published studies (21–23). Proliferation assay was carried out with resazurin cell proliferation assay following the manufacturer's instruction (Biotium; Cat# 30025–1). The IC₅₀ values and combination index (CI) were obtained by Chou-Median-Effect Equation using CompuSyn software (24) and dose–effect curves were obtained by GraphPad using affected fractions of compound-treated wells normalized against no-drug control wells using non-liner regression model.

Cells (∼70% confluency) were collected via scraping on ice and lysed in a buffer containing 8 mol/L urea, 0.1 mol/L phosphate, 0.01 mol/L Tris, pH8.0, 15 mmol/L imidazole, 0.2% triton-X100. After vigorous vortex, the lysates were clarified by centrifugation at 12,000 rpm for 10 minutes at room temperature (RT) followed by protein quantification using the Pierce BCA Kit. Equal amounts of protein (20 μg) were separated by 8% or 12% SDS-PAGE followed by transfer to nitrocellulose membranes and blotting with primary antibodies overnight and species-specific horseradish peroxidase–conjugated secondary antibodies and chemiluminescence. The densitometry was performed using ImageJ. Primary antibodies used in this study include rabbit NCOA4 polyclonal antibody (Abclonal, cat# A5695), rabbit Anti-LC3B (Sigma, Cat# L7543), mouse mAb for p53 (DO1), and MDM2 (4B11; Millipore-Sigma Cat#OP43 and cat# OP143), rabbit MDM4 (Proteintech cat# 17914–1-AP), rabbit mAb FTH1 (Cell Signaling #4393S), and rabbit polyclonal GAPDH (FL-335; Santa Cruz Biotechnology, sc-25778). P53 ubiquitination assay was performed as previously described (25).

Cells were plated in wells of 24-well plates with glass cover slips and allowed to attach to cover slips overnight. Following drug exposure, cells were fixed and permeabilized with methanol. Slides were blocked in 3% FBS and then the primary antibody was added overnight. Fluorescent secondary antibodies, goat–anti-mouse IgG (H+L)-Alexa Fluor 594 (Jackson Immunoresearch Labs, cat# 115–585–166) or donkey anti-rabbit IgG (H+L)-Cy3 (Jackson Immunoresearch Laboratories, cat# 711–165–152) were added for 1 hour at RT. Nuclei were counterstained with DAPI (5 mg/mL) at 1:10,000 dilution and coverslips mounted using propyl gallate. Cells were imaged using a Zeiss Axioplan 2 fluorescent microscope mounted with a Hamamatsu ORCA-ER camera.

Cells were plated in 6-cm dishes in approximately 80% confluency and allowed to attach overnight. Cells were then treated with different doses of MMRi62 for 72 hours followed by harvesting and washing with 1X PBS. Cells were stained with Cellular ROS red (0.5 μL/mL assay buffer, Abcam #186027) using phenol red-free media containing 2% FBS. After 30 minutes of incubation in the dark at 37°C, dye-containing media was removed by centrifugation and cells were suspended in phenol red-free media containing 2% FBS. Cells were then analyzed by flow cytometry (LSR IIA; BD Biosciences) using the PE Texas Red channel and results were quantified using the WinList 3D 7.1 software. Data are presented as fold change compared with control.

Cells were infected with retrovirus containing the lentiviral vector pLKO.1-shp53 DNA (addgene #19119; ref. 26). Cells were cultured with puromycin for 2 weeks, and puromycin-resistant clones were selected. Knockdown of p53 was detected by Western blot analysis with anti-p53 antibody.

Two thousand cells per well were played on 6-well plates one day before and the cells were treated with MMRi62 at different concentrations for 24 hours. Then, cells were replenished with drug-free complete medium and cultured for 2 weeks during which the cell culture was replenished with fresh medium every 3 days. Cells were fixed with cold methanol and stained with 0.05% (w/v) crystal violet solution at the end of 2-week period of cell culture.

Dissociated single cells (1,000 cells/well) into ultra-low attachment 96-well plates (Fisher, Cat# 07200603) were cultured in 50 μL medium mixed with 3% Matrigel for 72 hours. The spheroids were treated by adding 100 μL of DMSO or 4 μmol/L MMRi62-containing 3% Matrigel in medium continuously. The plates with spheroids were then scanned on different days and the resulting images were stored as high-resolution (1,200 dpi) grayscale TIFF files. Signal intensity and density were quantified using Image-Pro Plus (Media Cybernetics).

All animal experiments have been conducted in accordance with an IACUC (Institutional Animal Care and Use Committee) protocol approved by IACUC of Roswell Park Comprehensive Cancer Center. Orthotopic tumors were established in female SCID mice at 4–6 weeks of age. While under isoflurane anesthetization, a 1-cm incision was made in the left abdomen of the animal to exteriorize the pancreas. Longitudinal treatment response studies were carried out with an established orthotopic model of pancreatic cancer by implanting 0.5 × 10⁶/50 μL of Panc1luc or BxPc3luc cells in a PBS-Matrigel mixture. After injection into the pancreas, the incision was sutured aseptically. Mouse body weight was measured on a balance and tumor size was measured by bioluminescence imaging (BLI) two to three times per week. Once the tumor became well-established 8 days after tumor implantation, mice were randomized according to tumor size and treatment with MMRi62 was started. The mice were injected with vehicle or MMRi62 at a dose of 25 mg/kg in a formulation of 1.25% MMRi62–10% N-Methyl-2-pyrrolidone-10% Cremphor-EL-78.75% PBS intraperitoneally twice per week. The tumor growth and body weight were measured every week for 31 days.

The Transwell chamber (pore size, 8.0 μmol/L; Millipore) with Matrigel (for invasion assay) coating was inserted into a 24-well culture plate. For the invasion assay, 8 μmol/L pore inserts were coated with Matrigel (BD Biosciences). PDAC cells in log-phase growth were cultured in 6-well plates in medium containing 1% FBS for 24 hours before drug treatment. The PDAC cells (100 μL, 1 × 10⁵) suspended in DMEM containing 1% FBS with or without 4 μmol/L MMRi62 were seeded in the top chamber, and 500 μL of medium containing 20% FBS was placed in the lower chamber as a chemoattractant. The Transwell chamber was incubated for 24 hours. The invaded cells on the bottom surface of filter were fixed in methanol and stained with 0.05% w/v crystal violet solution for 20 minutes following by wash with running water in a bucket. Cell migration and invasion were determined by counting the stained cells under a light microscope in 10 randomly selected fields.

Pancreatic cancer cells were grown to confluence in 24-well plates. The monolayer was then artificially wounded using a sterile 200-μL pipette tip. Cell debris was removed by washing the monolayer with PBS. The cells were then incubated with medium containing 4 μmol/L MMRi62. Wound closure was monitored by images taken at various time points at the same spot with an inverted microscope equipped with a digital camera showing cell migrated into the wound. The extent of healing was calculated as the ratio of the remaining wound areas to the original wound area.

All the IC₅₀ value results were obtained from triplicate experiments and each individual experiment had triplicate at each dose point. The immunoblotting experiments were repeated at least three times and the representative data were presented. Statistical analyses were performed using PRISM6 Software (GraphPad Software, Inc.), ANOVA, and the Student t test. Statistical significances indicated with * were determined by a P value of <0.05.

We previously reported identification of RING-domain inhibitors of MDM2–MDM4 E3 complex in a high-throughput screen. We identified several hits that we called MMRi, which stands for MDM2–MDM4 RING domain inhibitors (27). MMRi6 as one of the primary hits belongs to quinolinol class that is drawing attention in drug development (28, 29). MMRi62 was identified in a cell-based screen using leukemia/lymphoma cell lines as a potent apoptosis inducer among MMRi6 analogs (Fig. 1A top for MMRi62 structure). Mechanism of action studies uncovered that MMRi62 promotes MDM4 degradation and p53-independent apoptosis in leukemia cells (Lama R and colleagues, unpublished observation). To explore whether MMRi62 has antitumor activity in solid tumors, we included MMRi62 in a panel of therapeutics in a Panc1 cell-based screen. We observed that MMRi62 did inhibit Panc1 cell proliferation in addition to gemcitabine. We then tested MMRi62 in multiple PDAC cell lines for which their KRAS and TP53 status could be retrieved from literature. Our result showed that four of six PDAC cell lines were sensitive to MMRi62 with an IC₅₀ value ranging from 0.59 to 1.65 μmol/L in a 72 hours proliferation assay (Fig. 1A, middle) and three lines showed intrinsic resistance to MMRi62 with IC₅₀ value of approximately 10 μmol/L. Although our cell line panel is small, it appeared that transcriptional subtype of PDAC determines MMRi62 sensitivity because three of the four MMRi62-sensitive cell lines fell into quasimesenchymal (QM-PDA) subtypes and the three resistant cell lines fell into classical/epithelial subtype as defined by transcriptome profiling (30, 31). Given that the MMRi62 sensitivity is not dependent of p53 status, we focused on cell lines Panc1 (KRASG12D:TP53R273) and BxPc3 (TP53Y220) bearing mutant p53 because it occurs in 60%–70% PDAC (12, 13). To better understand how MMRi62 inhibits PDAC proliferation, we performed trypan blue exclusion assay and found that MMRi62 treatment for 24 hours induced loss of cell viability with increased trypan blue-positive cells in MMRi62 concentration-dependent manner with significant difference as compared with non-treated cells (Fig. 1B, top, representative images and bottom quantification of live cells). We then performed clonogenic assay with three cell lines, including Panc1, BxPc3, and HPAFII. HPAFII was included because it bears a TP53P151S mutant that confers resistance to cell death by anoikis in head and neck cancer cell lines (32). Pre-treatment with MMRi62 for 24 hours followed by clonogenic growth in the absence of the compound effectively inhibited PDAC cell colony formation in all three cell lines in an MMRi62-concentration dependent manner (Fig. 1C). To investigate whether the pro-death effect of MMRi62 goes beyond 2-dimensional culture system, we performed experiments with PDAC spheroids. The Panc1 and BxPc3 spheroids were treated with solvent DMSO or 4 μmol/L MMRi62-containing medium and spheroid images were taken on days 1, 4, 7, and 10. As shown in Fig. 1D for the representative spheroid images and Fig. 1E for the median diameter of spheroids, MMRi62 significantly inhibited spheroid growth as the diameter of spheroids in DMSO control group increased from 570 μmol/L on day 1 to 720 μmol/L on day 10 whereas the diameter of spheroids decreased from 505 μmol/L on day 1 to 287.5 μmol/L on day 10 in the MMRi62 group (Fig. 1D). The difference between DMSO and the MMRi62-treated group was statistically significant (Fig. 1E) for both Panc1 and BxPc3 cells. Hematoxylin and eosin (H&E) stain of the spheroids collected on 10 days showed that Panc1 formed spheroid structure with irregular edges and BxPC-3 formed compact spheroid structures with vacuole inside the spheroids (Fig. 1D, H&E).

To understand the molecular events involved in MMRi62-induced PDAC cell death, we first investigated how MMRi62 affects the expression of MDM4, MDM2, and p53 because they were the initially identified targets of MMRi62. Although MMRi62 treatment at low μmol/L range for 24 hours did not induce measurable downregulation of MDM4 or MDM2 in Panc1 cells by Western blotting analysis, it did consistently downregulate MDM2 in BxPc3 cells (Fig. 2A). However, MMRi62 does not downregulate MDM2 and MDM4 in PDAC cells as effectively as in leukemia/lymphoma cells. Interestingly, we observed a moderate downregulation of p53R273H in Panc1 and significant downregulation of p53Y220C in BxPc3 after 24 hours treatment (Fig. 2A). Although MMRi62 also downregulates wt-p53 in Capan2 cells (Supplementary Fig. S1), MMRi62 effects on mut-p53 downregulation in Panc1 and BxPc3 cells are significantly stronger. Of note, mut-p53 downregulation by MMRi62 in PDAC cells is of slow kinetics: Extended treatment with MMRi62 for 72 hours caused more significant downregulation than 24 hours treatment (compare, Fig. 2A and B). We then looked at the molecular events involved in apoptosis, that is, increased caspase 3-mediated PARP cleavage (cPARP) in a prolonged treatment of 72 hours. Our results indicated that MMRi62 only slightly increased the levels of cPARP in the two cell lines, indicating that apoptosis was induced but was not likely the leading cause of cell death. Because PDAC cells are predisposed to ferroptosis, a mode of cell death initiated by dysregulation of iron metabolism (17, 18), we then investigated the hallmarks of ferroptosis, increased autophagy activity using LC3-phosphatidylethanolamine conjugate (LC3-II) as a marker (18). We found that MMRi62 significantly increased LC3B-II levels in a dose-dependent manner, which is in sharp contrast with a significant decrease in p53 levels in 72-hour–treated cells (Fig. 2B). Interestingly, phospholipid-unconjugated LC3-I was also increased (Fig. 2B, LC3-I). This increased autophagy activity was not associated with changes in ATG5 expression and occurred even in the presence of a slight decrease in ATG1 in Panc1 cells. Immunofluorescence (IF) microscopy confirmed that MMRi62 treatment downregulated p53 expression and increased LC3 expression in both cell lines (Fig. 2C). Similarly, p53 downregulation was also confirmed by IF in MMRi62-treated spheroids of Panc1 and BxPc3 cells (Supplementary Fig. S2A). Then we measured the production of cellular ROS, another feature of ferroptosis. Our results showed that MMRi62 significantly increased cellular ROS in a dose-dependent manner as compared with non-treated cells (Fig. 2D). Taken together, these results suggest that MMRi62 induces ferroptosis in PDAC cells associated with p53 downregulation.

To confirm that MMRi62 induces ferroptosis, we determined the expression of FTH1 and ferritin receptor NCOA4, two negative regulators of ferroptosis that are downregulated during ferroptosis (33, 34). Our results showed that MMRi62 treatment induced significant decrease in NCOA4 and FTH1 protein expression in both Panc1 and BxPc3 cells (Fig. 3A). We confirmed with IF microscopy that the MMRi62-induced NCOA4 downregulation also occurred in the spheroids of Panc1 and BxPc3 cells (Supplementary Fig. S2B). To understand how MMRi62 induces downregulation of p53 and FTH1 in PDAC cells, we performed rescue expression using proteasomal inhibitor carfilzomib and lysosomal inhibitor bafilomycin A1 (BAF). Our results showed that carfilzomib nearly fully rescued p53, a minor rescue of NCOA4 but not FTH1, whereas BAF1 nearly fully rescued both FTH1 and NCOA4 downregulation in Panc1 cells and partially in BxPc3 cells but showed no effect on mut-p53 downregulation in both cell lines (Fig. 3B). BAF rescue of FTH1 and NCOA4 degradation also suggests that MMRi62-induced increase of LC3-I and LC3-II is due to activation of autophagy rather than inhibition of lysosomal function. The partial rescue of NCOA4 and FTH1 downregulation in BxPc3 cells suggests possible transcriptional mechanisms that might be involved in addition to the lysosomal degradation mechanism. As expected, MMRi62 treatment increased ubiquitinated mut-p53 in Panc1 cells in in vivo ubiquitination assays in which His-ubiquitinated proteins were brought down by nickel beads followed by WB of mut-p53 in Panc1 cells (Fig. 3C). These results suggest that MMRi62 promotes proteasomal degradation of mut-p53 but lysosomal degradation of NCOA4 and FTH1.

To understand whether p53 expression has direct effect on expression of FTH1 and NCOA4 and how mut-p53 is involved in MMRi62 anti-PDAC activity, we performed experiments with p53 knockdown cells. We obtained efficient p53 knockdown in BxPc3 and HPAFII, but not in Panc1 cells. Knockdown of mut-p53 increased expression of NCOA4 and FTH1, the two negative regulators of ferroptosis, suggesting that high expression of mut-p53 predisposes PDAC cells to ferroptosis (Fig. 3D). To understand how mut-p53 knockdown affects MMRi62 sensitivity, we performed proliferation assay in the presence of MMRi62 with the p53 knockdown cells. The IC₅₀ value of MMRi62 was slightly increased in both shp53-BxPc3 cells and shp53-HPAFII cells compared shRNA control cells (Fig. 3E), suggesting that a minor contribution of mut-p53 status to MMRi62 induced anti-growth effect. Because mut-p53 knockdown increases NCOA4 and FTH1, these results suggest that MMRi62 induced downregulation of FTH1/NCOA4 and mut-p53 are independent effects of MMRi62, and MMRi62-induced cell death is independent of p53 downregulation and p53's role in regulating the expression of NCOA4 and FTH1 (Fig. 3D and E).

We next examined MMRi62's in vivo efficacy in orthotopic tumor models with Panc1 and BxPc3 cells stably expressing luciferase for BLI. The experiments were performed by treating the nude mice bearing the Panc1luc or BxPc3luc orthotopic tumors with vehicle or MMRi62 (25 mg/kg/d, 2 d/wk for 2 weeks) by intraperitoneal injection. Mouse body weight was weighed, and tumor growth was measured by BLI on days 9, 18, 28, and 37 after tumor implantation. Our results showed that MMRi62 treatment caused body weight loss at day18 but within 20% range and started to recover by days 28 and 37 (Fig. 4A and B, left middle). MMRi62 significantly suppressed the growth of Panc1luc by 52.71% (P < 0.05) and 60.77% of BxPc31luc orthotopic tumors compared with the tumors in control animals (Fig. 4A and B, middle right). The average wet weights of tumor collected at the end of the experiment were 2.85 g (Panc1luc) and 2.81 g (BxPc3luc), respectively, in the MMRi62 group compared with 3.99 g (Panc1luc) and 4.29 g (BxPc3luc) in the vehicle-treated control group with significant differences (P = 0.0002 and 0.001 for Panc1luc and BxPc3luc, respectively, Fig. 4A and B, bottom left). The average long diameters of the tumors were 1.36 cm (Panc1luc) and 1.48 cm (BxPc3luc), respectively, in the MMRi62 group compared with 2.29 cm (Panc1luc) and 2.47 cm³ (BxPc3luc), respectively, in the control group with significant differences (P = <0.0001 for both Panc1luc and BxPc3luc, Fig. 4A and B, bottom right).

We validated the death mode and changes in p53 and NCOA4 in these orthotopic tumors. Results of H&E staining showed that Panc1luc and BxPc3luc orthotopic tumors manifested necrotic tissue structures in the MMRi62 group (Fig. 4C). MMRi62 treatment reduced the expression of p53 and NCOA4 in both Panc1luc and BxPc3luc orthotopic tumors compared with vehicle groups (Fig. 4D, compare V with M in duplicates). Results from IF microscopy for activated caspase 3 showed increase AC3 staining, suggesting a death mode composed of apoptosis and necrosis (Fig. 4E).

Early metastasis is one of the major reasons for low 5-year survival rate of PDAC patients in which mut-p53 plays a critical role (14, 15). Owing to MMRi62's activity to downregulate mut-p53, we determined whether MMRi62 inhibits cell migration and invasion in PDAC cells. Our results from wound-healing assays showed that non-treated Panc1 cells migrated into the entire wounded area by 72 hours, whereas treatment with 4 μmol/L MMRi62 significantly inhibited Panc1 cell migration (Fig. 5A, left, compare C with M). Similarly, non-treated BxPc3 cell migrated into the entire wounded area by 48 hours, whereas treatment with 4 μmol/L MMRi62 significantly inhibited BxPc3 cell migration (Fig. 5A, right, compare C with M). We used Transwell assay to test the effect of MMRi62 on cell invasion, MMRi62 (4 μmol/L) reduced the invasion of Panc-1 cells (Fig. 5B, left) by 46.5% (Fig. 5B, right) and of BxPc-3 cells (Fig. 5C, left) by 70.0% (Fig. 5C, right) as compared with the non-treated controls. We next examined the effect of MMRi62 on PDAC metastasis in mice bearing Panc1luc and BxPc3luc orthotopic tumors in the experiment described previously in Fig. 4. Necropsy showed that 3 out of 4 of the Panc1luc-bearing mice and 2 out of 4 of BxPc3luc-bearing mice in the vehicle group developed metastatic lesions in the peritoneum, whereas none of the mice treated with MMRi62 had peritoneal dissemination in both Panc1luc and BxPc3luc models (Fig. 5D, small red dot). Liver metastasis occurred in 1 out of 4 mice in the vehicle-treated mice in both Panc1luc and BxPc3luc models. However, none of the MMRi62-treated mice in both models developed liver or lung metastasis (Fig. 5D, red cube and large red dot). These results were confirmed by histopathologic examinations of the liver tissues from mice in the control and treatment groups (Fig. 5E). These data suggest that MMRi62 abrogated PDAC metastasis in these two PDAC models.

We investigated whether combination of MMRi62 and current chemotherapy gemcitabine could synergistically inhibit PDAC proliferation. Our synergism analysis indicated a weak synergism between 30% and 75% inhibition effect levels with CI being about 0.7 to 0.5 (Fig. 6A), suggesting that simultaneous administration of MMRi62 and gemcitabine may have synergism in low effect range but risks antagonism in high effect range. We further tested MMRi62 sensitivity in established gemcitabine-resistant BxPc3 cells (Gem-R-BxPc3 and Gem-R-HPAFII). Our results indicated that Gem-R-BxPc3 and Gem-R-HPAFII were extremely resistant to gemcitabine with IC₅₀ > 100 μmol/L whereas BxPc3 is extremely sensitive to gemcitabine (IC₅₀, 0.047 μmol/L) and HPAFII is relatively resistant to gemcitabine (IC₅₀, 8.06 μmol/L). However, MMRi62 sensitivities of Gem-R-BxPc3 and Gem-R-HPAFII were slightly reduced compared with parental cells: IC₅₀ = 2.38 ± 0.88 and 3.72 ± 0.36 μmol/L, respectively, for BxPc3 and Gem-R-BxPc3 and IC₅₀ = 4.04 ± 0.59 and 6.62 ± 0.52 μmol/L, respectively, for HPAFII and Gem-R-HPAFII (Fig. 6B and C). These results suggest that MMRi62 if used in sequential combination might be a strategy for killing gemcitabine-resistant cells.

PDAC incidence has been on the rise and chemoresistance, both intrinsic and acquired, poses significant challenges to current combination therapies (35). Active autophagy plays an essential role in PDAC development because mice lacking the autophagy genes Atg5 or Atg7 accumulate low-grade, pre-malignant pancreatic intraepithelial neoplasia lesions without progressing to high-grade pancreatic intraepithelial neoplasia and PDAC (36). Elevated autophagy in PDAC provides a surviving mechanism to resist to-die under nutrient limitation of tumor microenvironment or therapeutic treatment (37). Consequently, depletion of autophagic machinery components, or hydroxychloroquine inhibition of autophagy, suppressed PDAC proliferation and decreased tumorigenicity in vivo (38). Autophagy is bulk degradation of proteins and other macromolecules for recycling building blocks and survival of nutrient-scarce conditions. However, paradoxically, ferritinophagy is an autophagic degradation of ferritin triggers cell death called ferroptosis (18). Ferroptosis as a preferred mode of cell death for KRAS mutant PDAC cells was discovered in studies with a RAS-selective lethal compounds like erastin and RSL3 (17). Cellular targets identified with different ferroptosis inducers include voltage-dependent anion channel, cystine/glutamate antiporter system, GPX4 and GSH as cellular targets in ferroptosis induction (39). Ferroptosis is associated with NCOA4-mediated ferritinophagy, increased ROS and deregulation of cellular iron (18). Currently, no ferroptosis inducer agents for PDAC therapy are available in clinic. In this report, we described previously MMRi62 as a new ferroptosis inducer in PDAC cells that inhibited PDAC growth, cell migration in vitro and metastasis in vivo.

MMMRi62 is small-molecule quinolinol initially identified as a RING domain modifier compound that induces preferential MDM4 degradation by MDM2–MDM4 E3 complex and p53-independent apoptosis in leukemia cells (Lama R and colleagues, unpublished observation). Except for targeting MDM2–MDM4, whether MMRi62 has other cellular targets is currently unknown. This study discovers a novel pro-ferroptosis activity of MMRi62 associated with induced mut-p53 and FTH1 degradation. Our data suggest that the anti-PDAC effect of MMRi62 involves a mixture of death modes of apoptosis and ferroptosis, although necrosis might also be involved. The contribution of apoptosis may be limited because caspase 3 activation (Fig. 4) is associated with weak downstream effect of PARP cleavage (Fig. 2). In contrast, ferroptosis appears to be the dominating death mode because MMRi62 treatment caused robust lysosome-dependent downregulation of NCOA4 and FTH1 associated with significantly increased autophagy and ROS production (Figs. 2–4). MMRi62-induced ferroptosis is not strictly dependent on KRAS mutation because it also occurs in wt-KRAS BxPc3 cells. This is in line with literature showing that KRAS mutation is not a prerequisite in ferroptosis because erastin-induced ferroptosis requires KRAS mutation in PDAC but is antagonized by mutant KRAS in RMS13 rhabdomyosarcoma cells (40). Our finding that MMRi62-induced ferroptosis is mutant-KRAS–independent provides a useful scaffold structure for developing new anti-PDAC ferroptosis inducers. NCOA4/FTH1 downregulation in ferritinophagy is a hallmark of ferroptosis reported with erastin (18). Consistent with reports that cancer cells with EMT phenotype are vulnerable to ferroptotic cell death due to their dependence on a lipid peroxidase pathway (41), we found that QM-PDA cells were more sensitive to MMRi62-induced cell death (Fig. 1A). However, 5 μmol/L ferrostatin-1 failed to rescue MMRi62-induced ferroptosis or FTH1 degradation in Panc1 and BxPc3 cells, suggesting that MMRi62-induced ferroptosis does not involve lipid oxidation at which ferrostatin-1 inhibits and rescues erastin-induced ferroptosis (42). We propose that MMRi62-induced FTH1 degradation triggers ferroptosis, because FTH1 has been validated genetically as a key target in ferroptosis in a report showing FTH1-loss causes massive ferroptosis in the developing wing of Drosophila associated with production of ROS (33). Currently, how MMRi62 induces lysosomal degradation of NCOA4 and FTH1 and whether NCOA4 and FTH1 are direct targets of MMRi62 is under investigation. Nevertheless, this study identified MMRi62 as a potent FTH1 degradation inducer useful for FTH1 targeting.

Role of p53 in ferroptosis regulation is cell type and ferroptosis-inducer dependent (43). Wt-p53 positively regulates ferroptosis by transcriptionally repressing SLC7A11, a component of anti-oxidant system Xc⁻ in U2OS and MCF7 cells (44) but it inhibits ferroptosis by inhibiting dipeptidyl-peptidase-4–dependent lipid peroxidation in colon cancer cells (45). Mutant p53 was reported to sensitize cancer cells to ferroptosis involving reduced expression of SLC7A11 (43, 46). We found that mut-p53 might predispose PDAC cells to ferroptosis by reduced expression of FTH1/NCOA4 (Fig. 3D). This projection was confirmed by a slightly decreased MMRi62 sensitivity in shp53 PDAC cell lines (Fig. 3E). Therefore, although MMRi62 promotes mut-p53 degradation (Figs. 2 and 3), MMRi62-induced ferroptosis is likely mediated by FTH1 downregulation. A surprising in vivo result was that MMRi62 completely suppressed PDAC metastasis in two orthotopic mouse models. This effect is likely mediated at least partially by mut-p53 downregulation that occurred in those tumor tissues (Fig. 5D), because mutant p53 is an established metastasis driver in PDAC mouse models (14). Because of limited number of animals in this study, the MMRi62 effect on metastasis warrants further study using a larger cohort of animals.

TP53 mutation occurs in 75% of invasive pancreatic adenocarcinomas (12) and mut-p53 proteins are often stable and accumulate at high levels in cancer cells in contrast with the fast turnover and low expression of wt-p53 protein in normal cells. Strategy to reactivate mut-p53 for cancer therapy (47) may be challenged with limited efficacy because mut-p53 is often found in late-stage cancers in which the p53 downstream pathways are often disrupted together with other oncogenic mutations such as MYC, RAS, and/or PI3K (48). MMRi62 might be an alternative strategy to eliminate mut-p53–mediated oncogenic activities, including metastasis. In this sense, our finding places MMRi62 in a category of mut-p53–targeting agents. Taken together, this study identified MMRi62 as a new class of ferroptosis inducer with distinct mechanisms of action than the reported ferroptosis inducers (39). The dual targeting activity for mut-p53 and FTH1 renders MMRi62 an advantage over other ferroptosis inducers. Further studies on underlying mechanisms of ferroptosis induction by MMRi62 and development of better MMRi62-like small molecules may bring about promising new therapies for recalcitrant cancers such as PDAC.

X. Wang reports grants from National Cancer Institute, NIH, Roswell Park Alliance Foundation, and National Natural Scientific Foundation of China during the conduct of the study; as well as reports a patent 10624881 issued. No disclosures were reported by the other authors.

J. Li: Conceptualization, data curation, formal analysis, investigation, writing–original draft, writing–review and editing. R. Lama: Formal analysis, investigation, writing–review and editing. S.L. Galster: Resources, investigation, writing–review and editing. J.R. Inigo: Investigation. J. Wu: Investigation. D. Chandra: Investigation, writing–review and editing. S.R. Chemler: Resources, investigation, writing–review and editing. X. Wang: Conceptualization, formal analysis, supervision, funding acquisition, writing–original draft.

This work was supported in part by grants from R01CA208352 (to X. Wang), Roswell Park Alliance Foundation (to X. Wang and S.R. Chemler), and by the National Natural Scientific Foundation of China (to J Li; no. 81472246). The Roswell Park Comprehensive Cancer Center Core grant CA16056 from NCI is gratefully acknowledged. The authors also thank Roswell Park Small Molecule Screening Shared Resource for the initial high-throughput screening of MMRi preliminary hits and Roswell Park's Translational Imaging Shared Resource (TISR) for BLI imaging. We thank Yuping Wang for her technical assistance.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.