A CD147-targeted small-molecule inhibitor potentiates gemcitabine efficacy by triggering ferroptosis in pancreatic ductal adenocarcinoma

Abstract

CD147 has emerged as a promising tumor-specific therapeutic target. Identifying small-molecule inhibitors that promote its proteolysis represents a critical step toward advancing clinical translation, while elucidating its mechanisms of action could further accelerate this process. In this study, we identify dracorhodin perchlorate (DP) as a potent CD147 inhibitor that induces autophagy-dependent degradation. DP significantly inhibits cell proliferation and enhances sensitivity to gemcitabine in pancreatic cancer cells. Mechanistically, CD147 inhibition upregulates acyl-CoA synthetase long-chain family member 4 (ACSL4) expression through H3K9 lactylation and suppresses the sterol regulatory element-binding protein 1 (SREBP1)/stearoyl-CoA desaturase-1 (SCD1) signaling pathway, collectively disrupting the balance of polyunsaturated and monounsaturated fatty acids, ultimately triggering ferroptosis. The combination of DP and gemcitabine demonstrates remarkable synergistic anti-tumor effects in orthotopic xenograft models, spontaneous KPC mouse models, and patient-derived organoid (PDO) and xenograft (PDX) models. In conclusion, this study reveals a mechanism by which CD147 regulates ferroptosis and supports combining DP with gemcitabine as a therapeutic strategy to improve patient outcomes in pancreatic ductal adenocarcinoma.

Keywords: ACSL4; CD147; H3K9la; dracorhodin; ferroptosis; gemcitabine; pancreatic ductal adenocarcinoma.

Graphical abstract

Figure 1

Identification and characterization of DP as a CD147 inhibitor (A) Immunohistochemical staining of CD147 in PDAC tumor and matched adjacent non-tumor tissues. Scale bars, 50 μm. (B) Immunoblotting of CD147 in 10 paired PDAC tumor (T) and non-tumor (NT) tissues. (C) CD147 staining in a PDAC tissue microarray. Scale bars, 50 μm. (D) Kaplan-Meier analysis of overall survival in patients with PDAC stratified by CD147 expression level in the tissue microarray. p values were calculated using the log rank test. (E) CD147 expression in HPDE and PDAC cell lines by immunoblotting. (F) 3D scatterplot of virtual screening results of 1,880 natural compounds, based on docking score, molecular weight, and the number of hydrogen bond donors (HBDs). The top 100 candidates are highlighted in red. (G) Molecular docking model showing predicted interactions between DP and CD147. (H) Bio-layer interferometry (BLI) analysis showing concentration-dependent binding of DP to CD147. (I) Pull-down of CD147 using biotin-conjugated DP and streptavidin beads, followed by immunoblotting. (J) Cellular thermal shift assay (CETSA) showing DP-mediated stabilization of CD147. (K) Isothermal dose-response fingerprint (ITDRF-CETSA) of CD147 at 46°C. (L) CD147 levels in PDAC cells treated with increasing concentrations of DP for 24 h. (M) Cycloheximide chase assay in PANC-1 cells treated with 1 μM cycloheximide ±30 μM DP. (N and O) CD147 levels in PANC-1 cells pretreated with MG132 (5 μM), bafilomycin A1 (1 μM), or chloroquine (20 μM) for 2 h before DP treatment (30 μM, 24 h). (P) CD147 expression in PANC-1 cells treated with DP (20 μM) ± rapamycin (1 μM). See also Figure S1.

Figure 2

DP suppresses proliferation and enhances chemosensitization via targeting CD147 in PDAC (A) IC₅₀ values of DP in HPDE and PDAC cell lines at 24 and 48 h. (B and C) Dose-response curves (B) and bright-field (BF) images (C) of PANC-1, AsPC-1, and HPDE cells treated with DP for 48 h. Scale bars, 200 μm. (D) Validation of CD147 knockdown by immunoblotting. (E and F) Cell viability after 48 h DP treatment in control and CD147-knockdown PDAC cells. (G) GEM sensitivity in BxPC3-GR cells ± CD147 knockdown. (H) Pearson correlation analysis between GEM response and CD147 expression in PDOs. (I) CD147 knockdown efficiency in PDO#4 validated by immunoblotting. (J) CD147 expression by IHC in PDOs and matched tumor tissues; BF images of GEM-treated PDOs. Scale bars, 100 μm. (K) GEM response in PDO#4 with or without CD147 knockdown. (L) Synergy between DP and GEM assessed by combination index (CI < 1 indicates synergy). (M) CD147 levels in PANC-1 and BxPC3-GR cells treated with DP, GEM, or both. (N) BF images of PDO#2, PDO#3, and PDO#4 with or without CD147 knockdown following treatment with DMSO, GEM (0.5 μM), DP (20 μM), or their combination for 48 h. Scale bars, 100 μm. (O) Schematic of xenograft study design. (P and Q) Bioluminescent imaging (P) and quantification (Q) of tumor burden in C57BL/6J mice bearing orthotopic pancreatic tumors. n = 5 mice per group. (R and S) Images and weights of tumors harvested from C57BL/6J mice bearing orthotopic pancreatic. n = 5 tumors per group. (T and U) Quantification of CD147 (T) and Ki67-positive (U) cells. n = 3 tumors. Statistical analyses were performed by one-way ANOVA followed by Tukey’s multiple comparison test (Q and S–U). Data are presented as mean ± SEM (B, E–G, K, Q, and S–U). ∗∗p < 0.01, ∗∗∗p < 0.001. See also Figures S2 and S3.

Figure 3

DP induces ferroptosis through alterations in PUFA/MUFA composition (A) Viability of PDAC cells treated with DP (20 or 40 μM) for 48 h ± inhibitors of necroptosis (necrostatin-1, 20 μM), ferroptosis (ferrostatin-1, 0.75 μM; deferoxamine, 7.5 μM), pyroptosis (disulfiram, 25 μM), or apoptosis (Z-VAD-FMK, 20 μM). (B) Transmission electron microscopy of PANC-1 cells treated with DMSO, DP (30 μM), or RSL3 (0.5 μM) for 24 h. Scale bars, 1 μm. (C) Confocal imaging of PANC-1 cells treated with DP (30 μM) for 24 h, stained with the lipid peroxidation probe C11-BODIPY 581/591. Scale bars, 20 μm. (D) Flow cytometric analysis of lipid peroxidation in PANC-1 cells treated with increasing concentrations of DP for 24 h using the C11-BODIPY 581/591 probe. (E) Total ROS levels in PDAC cells after 24 h DP treatment, assessed by flow cytometry using the general ROS probe 2',7'-dichlorodihydrofluorescein diacetate (DCFH-DA). (F and G) Lipid ROS levels in PDAC cells treated with DP (30 μM) ± ferrostatin-1 (0.75 μM) or deferoxamine (7.5 μM) (F), or in combination with RSL3 (0.5 μM) (G). (H) Quantification of malondialdehyde (MDA) levels after 24 h DP treatment. (I) Immunoblotting of SLC7A11 and GPX4 in DP-treated PDAC cells. (J) Intracellular GSH levels expressed as GSH/GSSG ratio. (K) Kyoto Encyclopedia of Genes and Genome pathway enrichment of differentially expressed genes identified by RNA-seq in PANC-1 cells treated with DP (30 μM). (L and M) Principal component analysis (PCA) (L) and bubble plot (M) illustrating global metabolic reprogramming in DP-treated PANC-1 cells, with a specific focus on alterations in polyunsaturated fatty acids (PUFAs), monounsaturated fatty acids (MUFAs), and their phospholipid-bound forms (PUFA-PLs and MUFA-PLs). Statistical analyses were performed using one-way ANOVA followed by Dunnett’s multiple comparison test (A, D, E, H, and J) and Tukey’s multiple comparison test (F and G). Data are presented as mean ± SEM (A, D–H, and J). ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. See also Figure S4.

Figure 4

DP sensitizes GEM by targeting CD147-mediated ferroptosis (A) Quantification of lipid ROS levels in shCtrl and shCD147 PANC-1 cells treated with DP (30 μM) for 24 h by flow cytometry using the C11-BODIPY 581/591 probe. (B and C) Quantification of malondialdehyde (MDA) levels (B) and GSH ratio (GSH/GSSG) (C) in shCtrl and shCD147 PANC-1 cells after 24 h DP treatment. (D) Transmission electron microscopy of CD147-knockdown PANC-1 cells showing ferroptosis-related morphological changes. Scale bars, 1 μm. (E) Quantification of cell proliferation in shCtrl and shCD147 PANC-1 cells treated with or without ferrostatin-1 (0.75 μM). (F) Quantification of cell viability in shCtrl and shCD147 PANC-1 cells treated with RSL3 (0.5 μM) for 48 h. (G and H) Quantification of lipid ROS levels (G) and MDA levels (H) in PANC-1 cells treated with GEM (1 μM) with or without CD147 knockdown for 24 h. (I) Quantification of cell viability in shCtrl and shCD147 cells treated with GEM (1 μM) ± ferrostatin-1 (0.75 μM) for 48 h. (J and K) Quantification of lipid ROS levels (J) and MDA levels (K) in GEM-treated BxPC3-GR cells with or without CD147 knockdown for 24 h. (L) Quantification of cell viability in BxPC3-GR cells treated with GEM (1 μM) ± ferrostatin-1 (0.75 μM) for 48 h. (M and N) Quantification of cell viability in PANC-1 (M) and BxPC3-GR (N) cells treated with GEM (1 μM), DP (20 μM), or their combination with or without ferrostatin-1 (0.75 μM) for 48 h. (O) Confocal imaging of PDO#4 treated with DMSO, GEM (0.5 μM), DP (20 μM), or their combination for 24 h, stained with C11-BODIPY 581/591. Scale bars, 100 μm. (P and Q) Images and weights of tumors harvested from nude mice bearing xenograft treated with DP (40 mg/kg, intraperitoneally [i.p.], every other day) with or without ferrostatin-1 (3 mg/kg/day, i.p.). n = 5 tumors per group. (R) Quantification of lipid ROS levels in xenograft tissues by flow cytometry using the C11-BODIPY 581/591 probe. Statistical analyses were performed by one-way ANOVA followed by Tukey’s multiple-comparison test (A–C, G–N, Q, and R). Data are presented as mean ± SEM (A–C, E–N, Q, and R). ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. See also Figure S4.

Figure 5

DP enhances ACSL4 expression via H3K9 lactylation to promote ferroptosis (A) Immunoblotting of MCT1 and MCT4 in PDAC cells after 24 h DP treatment. (B) Quantification of intracellular lactate levels in PDAC cells treated with DP for 24 h. (C and D) Immunoblotting of pan-lysine lactylation (Pan-Kla) and histone H3 lactylation marks (H3K9la, H3K14la, H3K18la, and H3K56la) in DP-treated PANC-1 cells. (E) Time course of H3K9la induction in PANC-1 cells after DP treatment. (F and G) Quantification of intracellular lactate levels (F) and histone H3 lactylation marks (G) in cells with or without CD147 knockdown. (H) ChIP-seq analysis showing genome-wide enrichment of H3K9la near transcription start sites. (I) Overlap analysis between H3K9la-enriched genes and ferroptosis/lipid metabolism-related gene sets. (J) Visualization of H3K9la ChIP-seq peaks at the ACSL4 genomic locus using IGV. (K) Immunoblotting of H3K9la levels in PDAC cells treated with oxamate (50 mM), alone or in combination with DP (30 μM) for 24 h. (L) ChIP-qPCR analysis of H3K9la enrichment at the ACSL4 promoter in PANC-1 cells treated with DP (30 μM), with CD147 knockdown, or in combination with oxamate (50 mM). (M–P) Quantification of ACSL4 mRNA (M) and protein levels (N–P) in cells treated with DP or CD147 knockdown. (Q) IHC staining of ACSL4 in PDO#4 treated with DMSO or DP (20 μM) for 24 h. Scale bars, 50 μm. (R) Immunoblotting of ACSL4 in PDAC cells treated with DP (30 μM) alone or in combination with oxamate (50 mM) or 2-DG (10 mM). (S) Quantification of lipid ROS levels in PDAC cells treated with DP (30 μM) alone or in combination with ACSL4 siRNA (0.2 μM). (T) Quantification of cell viability in PANC-1 cells treated with DP (30 μM) with or without ACSL4 siRNA (0.2 μM) for 48 h. (U) Quantification of cell viability in shCtrl and shCD147 cells treated with RSL3 (0.5 μM) ± PRGL493 (10 μM). Statistical analyses were performed using one-way ANOVA followed by Dunnett’s multiple comparison test (B and F) and Tukey’s multiple comparison test (L, S, and U), and unpaired Student’s t test (M). Data are presented as mean ± SEM (B, F, L, M, and S–U). ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. See also Figure S5.

Figure 6

Targeting CD147 suppresses SCD1 expression via the AMPK-SREBP1 pathway (A–C) Quantification of SCD1 mRNA (A) and protein levels (B–C) in PDAC cells treated with DP (30 μM) for 24 h or with CD147 knockdown. (D and E) Quantification of intracellular pH in PANC-1 cells by flow cytometry using BCECF-AM after 24 h DP treatment or CD147 knockdown. (F) Assessment of mitochondrial membrane potential by JC-1 staining in PANC-1 cells treated with DP (30 μM) or with CD147 knockdown. Scale bars, 100 μm. (G and H) Quantification of intracellular ATP (G) and AMP (H) levels in PDAC cells treated with DP (30 μM) for 24 h or with CD147 knockdown. (I and J) Immunoblotting of phosphorylated AMPK (p-AMPK), total AMPK, and SREBP1 in PANC-1 cells treated with DP or CD147 knockdown. (K) IHC staining of SCD1 in PDOs treated with or without DP (20 μM) for 24 h. Scale bars, 50 μm. (L) Immunoblotting of SCD1 in PDAC cells with or without SCD1 overexpression. (M and N) Quantification of cell viability in PDAC cells with or without SCD1 overexpression after DP (30 μM) treatment for 48 h. (O and P) Quantification of lipid ROS levels (N) and malondialdehyde (MDA) concentrations (O) in PDAC cells with or without SCD1 overexpression, treated with DP (50 μM) alone or in combination with ACSL4 siRNA (0.2 μM). Statistical analyses were performed by Student’s t test (A, E, G, and H) and one-way ANOVA followed by Tukey’s multiple-comparison test (O and P). Data are presented as mean ± SEM (A, E, G, H, and M–P). ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. See also Figure S6.

Figure 7

Combined DP and GEM therapy suppresses PDAC tumor growth in preclinical models (A) Schematic diagram of the treatment protocol in KPC mice. (B) Representative PET-CT images and quantification of standardized uptake values in KPC mice after 4 weeks of treatment. Tumor margins are outlined. (C) Gross anatomy of the abdominal cavity in a KPC mouse showing tumor (T), liver (L), and spleen (S). (D) Images and weights of tumors harvested from KPC mice. n = 3 tumors per group. (E) Kaplan-Meier survival analysis of KPC mice across treatment groups. n = 8 mice per group. p values were calculated using the log rank test. (F) Schematic illustration of PDX model establishment and treatment schedule. (G and H) Quantification of tumor volume over time (G) and Kaplan-Meier survival curves (H) in PDX-bearing mice across treatment groups. n = 8 mice per group. p values were calculated using the log rank test. (I) Images and weights of tumors harvested from PDX models. n = 5 tumors per group. (J) IHC staining of Ki67, CD147, ACSL4, and SCD1 in PDX tumor tissues. Scale bars, 100 μm. (K and L) Quantification of IHC signals for CD147, ACSL4, SCD1, and Ki67 in PDX tumors. Statistical analyses were performed by one-way ANOVA followed by Tukey’s multiple-comparison test (B, D, G, I, K, and L). Data are presented as mean ± SEM (B, D, G, I, K, and L). ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. See also Figure S7.