Selective targeting of IRAK1 attenuates low molecular weight hyaluronic acid-induced stemness and non-canonical STAT3 activation in epithelial ovarian cancer

Abstract

Advanced epithelial ovarian cancer (EOC) survival rates are dishearteningly low, with ~25% surviving beyond 5 years. Evidence suggests that cancer stem cells contribute to acquired chemoresistance and tumor recurrence. Here, we show that IRAK1 is upregulated in EOC tissues, and enhanced expression correlates with poorer overall survival. Moreover, low molecular weight hyaluronic acid, which is abundant in malignant ascites from patients with advanced EOC, induced IRAK1 phosphorylation leading to STAT3 activation and enhanced spheroid formation. Knockdown of IRAK1 impaired tumor growth in peritoneal disease models, and impaired HA-induced spheroid growth and STAT3 phosphorylation. Finally, we determined that TCS2210, a known inducer of neuronal differentiation in mesenchymal stem cells, is a selective inhibitor of IRAK1. TCS2210 significantly inhibited EOC growth in vitro and in vivo both as monotherapy, and in combination with cisplatin. Collectively, these data demonstrate IRAK1 as a druggable target for EOC.

Fig. 1. TIR signaling is activated in CSCs and cisplatin-resistant EOC.

A Viability assay of A2780 and C30 cells treated with cisplatin. B Quantification of spheroid formation for A2780 and C30 cells. C Hierarchal clustering heatmap of RNA sequencing data from A2780 and C30 cells grown as 2D and 3D cultures. D Volcano plot comparing log2 fold enrichment of genes highly expressed in C30 and A2780 cells grown in 2D. E Volcano plot comparing gene enrichment of highly expressed in A2780 cells grown in 3D and 2D. F List of IPA-activated pathways in C30 cells. Red bars indicate commonly activated pathways. *Indicates TIR-related signaling pathways. G List of IPA-activated pathways in A2780 cells grown in 3D. Red bars indicate commonly activated pathways as in (F). H Venn diagram comparing IPA-activated pathways from (F, G).

Fig. 2. IRAK1 is upregulated in HGSOC.

A Heatmap of mRNA expression of TIR signaling genes from CBioPortal TCGA, Firehose Legacy serous cystadenocarcinoma dataset. B Copy number alterations of TIR signaling genes from CBioPortal TCGA, Firehose Legacy serous cystadenocarcinoma dataset. C Oncoprint of TCGA, Firehose Legacy serous cystadenocarcinoma dataset. D Boxplot of IRAK1 mRNA expression in normal tissue (NT), PT, and RT from TCGA ovarian cancer dataset. E Kaplan–Meier survival curve of TCGA ovarian cancer dataset (high, n = 201, low, n = 172, p = 0.037). F Diagnosis Age versus IRAK1 mRNA expression of TCGA, Firehose Legacy ovarian serous cystadenocarcinoma dataset. Spearman correlation = −0.23, p < 0.0001). G Representative immunohistochemistry for IRAK1 staining of the normal fallopian tube and PT from two patients from HGSOC TMA. H The histological staining score of IRAK1 in TMA, containing matched normal fallopian tube (ctrl), PT, and Met from 100 patients with stage 3/4 HGSOC. ns not significant, ***p < 0.001, ****p < 0.0001.

Fig. 3. LMW HA activates non-canonical IRAK1 signaling and stemness.

A Western blot for IRAK1 across EOC cell lines. B Western blot time course for pIRAK1 (T209) and total IRAK1 in A1847 and OVCAR8 cells following stimulation with LMW HA (200 ng/mL). GAPDH served as an internal control. C Representative images of spheroid formation assay in A1847 and OVCAR8 cells either unstimulated or stimulated with LMW HA for 14 days. D Quantification of (C). Data is represented as mean from three independent experiments ± SEM. E (Top) Representative image of low and high exposures of immunoassay-based kinase array following stimulation with or without LMW HA for 15 min. 1. pSTAT5; 2. pp38; 3. β-Catenin; and 4. pERK. (Bottom) Quantification of pixel density of relative to untreated control. F Western blot of A1847 and OVCAR8 time course assessing phosphorylated and total expression of p38 and STAT3 following stimulation with or without LMW HA. The STAT3 target gene, MYC, was also assessed. *p < 0.05, ****p < 0.0001.

Fig. 4. IRAK1 is critical for HGSOC growth and LMW HA-induced stemness.

A Western blot of IRAK1 in IRAK1 KD and IRAK1 Scr cells and phosphorylated and total p65 and p38. B Heatmap of RNA-sequencing comparing IRAK1 KD and Scr gene expression. C Volcano plot of RNA-seq data from (B). Genes associated with stemness are labeled. D Western blot KD and Scr for pSTAT3, total STAT3, and cleaved Notch1 and Notch3. E Representative image of CF comparing KD and Scr. F Quantification of (E) for (i) colony number and (ii) colony size. Data are represented as mean from three independent experiments ± SEM. G Representative images of Scr and KD spheroid formation. H Quantification of spheroid formation from (G). Data are represented as box plots, min to max, all data points from three independent experiments. I Representative image of NSG mice injected IP with either IRAK1 Scr or KD cells. J Quantification of malignant ascites volume from IRAK1 Scr and KD groups (n = 5 mice per group). Data represented as mean ± SEM showing all data points. K Quantification of tumor weight following microdissection from IRAK1 Scr and KD groups (n = 5 mice per group). Data represented as mean ± SEM showing all data points. L Western blot time course of IRAK1 Scr and IRAK1 KD cells stimulated with LMW HA for phosphorylation and total STAT3. GAPDH served as internal control. M representative images of spheroid formation for IRAK1 Scr and KD cells stimulated with or without LMW HA for 14 days. N Quantification of spheroid number from (M). Data represented as mean ± SEM from three independent experiments. *p < 0.05, **p < 0.01, ***p < 0.001, ns not significant.

Fig. 5. TCS2210 is a selective inhibitor of IRAK1.

A Chemical structure of TCS2210. B In silico docking of TCS2210 with IRAK1: (i) space-filled model (ii) zoomed-in space-filled model with predicted hydrogen bonding interactions. C Kinome tree interaction map of Eurofins ScanMAX analysis with TCS2210. D Western blot of CETSA assay for IRAK1 following incubation with TCS2210 or DMSO vehicle control. E SPR analysis of TCS2210 with active recombinant IRAK1 enzyme. Data are shown for 1:1 kinetic model and 1:1 affinity model. F Western blot of A2780 cells for pIRAK1 (T209) and total IRAK1 following preincubation with or without TCS2210, and stimulation with or without LMW HA.

Fig. 6. TCS2210 induces apoptosis of EOC cells.

A Hexosaminidase viability assay in A1847, A2780, OVCAR3, and OVCAR8 EOC cell lines. Data are represented as means from three independent experiments ± SEM. B Representative images of CF in A18437 and OVCAR8 cells treated with TCS2210. C Quantification of CF number and size from (B). D Left: representative image of scatterplot for Annexin V/PI apoptosis assay in A1847 and OVCAR8 cells treated with TCS2210. Right: quantification of the percentage of apoptotic cells following treatment with TCS2210. Data are represented as mean from three independent experiments ± SEM. E Representative images of A1847 and OVCAR8 spheroid formation following treatment with TCS2210. F Quantification of spheroid formation from (E). Data are represented as mean from three independent experiments ± SEM. G Western blot of A1847 and OVCAR8 cells treated with TCS2210 for 24 h and 48 h. Immunoblots were performed for IRAK1, pSTAT3, STAT3, and MYC. GAPDH served as internal control. H Left, heatmap of % viability for A1847 cells treated with various concentrations of TCS2210 and cisplatin. Right, synergy plot generated using SynergyFinder2.0 of A1847 cells treated in combination with TCS2210 and cisplatin. Data are represented as the mean from three independent experiments. I Top, the experimental design of xenotransplant study. Bottom, quantification of tumor volume of micro-dissected tumors from mice (n = 10 per group) injected IP with A2780 EOC cells and treated with either cisplatin (Cis), TCS2210 (TCS), or combination. Data represented as mean ± SEM showing all data points.