Coexistent ARID1A-PIK3CA mutations promote ovarian clear-cell tumorigenesis through pro-tumorigenic inflammatory cytokine signalling

Abstract

Ovarian clear-cell carcinoma (OCCC) is an aggressive form of ovarian cancer with high ARID1A mutation rates. Here we present a mutant mouse model of OCCC. We find that ARID1A inactivation is not sufficient for tumour formation, but requires concurrent activation of the phosphoinositide 3-kinase catalytic subunit, PIK3CA. Remarkably, the mice develop highly penetrant tumours with OCCC-like histopathology, culminating in haemorrhagic ascites and a median survival period of 7.5 weeks. Therapeutic treatment with the pan-PI3K inhibitor, BKM120, prolongs mouse survival by inhibiting the tumour cell growth. Cross-species gene expression comparisons support a role for IL-6 inflammatory cytokine signalling in OCCC pathogenesis. We further show that ARID1A and PIK3CA mutations cooperate to promote tumour growth through sustained IL-6 overproduction. Our findings establish an epistatic relationship between SWI/SNF chromatin remodelling and PI3K pathway mutations in OCCC and demonstrate that these pathways converge on pro-tumorigenic cytokine signalling. We propose that ARID1A protects against inflammation-driven tumorigenesis.

Figure 1. A new Arid1a conditional allele to explore ARID1A tumor suppressor function in vivo

(A) Targeting scheme to insert two loxP sites flanking exon 5 and 6 of mouse ARID1A. (B) Representative Southern blot showing properly targeted ES cell clones. Insertion of loxP sites generates 5.6 kb EcoRV and 8.5 kb ApaI genomic fragments. (C) PCR genotyping and quantitative, dual-color fluorescent western blot of ARID1A (red) and βACTIN (green) protein expression in whole-embryo lysates from four Arid1a^fl/+ and four Arid1a^fl/+;Sox2Cre^0/+ embryos. (D) Graph of quantitative western blot results depicting normalized protein levels. Significant differences based on the average normalized protein expression ±SD of four independent embryos were calculated using a two-tailed Student’s t test (p-values <0.05 were considered significant). (E,F) Representative LacZ-stained, control (uninjected) or intrabursal AdCRE-injected (Gt)Rosa26lacZ ovary showing patterns of CRE-mediated recombination in the ovarian surface epithelium. (F, inset) Histological section of LacZ-stained AdCRE-injected (Gt)Rosa26lacZ ovary. (G-I) Live-luminescence overlaid whole-mount images of an intrabursal AdCRE injected, (Gt)Rosa26luciferase female mouse. (I) Excised reproductive tract. Asterisks in H indicate wound staple.

Figure 2. Concurrent ARID1A loss and PIK3CA activation leads to ovarian tumorigenesis in the mouse

(A) Plot of ARID1A and PIK3CA alterations across available TCGA datasets. Mutation or co-mutation frequencies are expressed as a percentage of all tumor samples for each cancer. A Fisher’s exact test was used to calculate the significance of association between ARID1A and PIK3CA mutations (*p-value <0.05). (B) Genotyping scheme to detect CRE-deleted (Arid1aΔ) allele in tumor genomic DNA samples. RT-PCR was used to distinguish PIK3CA and transgenic (Gt)Rosa26Pik3ca^*H1047R transcripts in tumor RNA samples. The H1047R mutation in (Gt)Rosa26Pik3ca^*H1047R protects against MslI digest of amplified tumor cDNA. (C) ARID1A loss or (Gt)Rosa26Pik3ca^*H1047R transcripts were detected in tumor RNA samples by RT-PCR. Significant differences based on the average normalized mRNA expression ±SD between replicates of a control liver sample and replicates of three independent tumor samples were calculated using a two-tailed Student’s t test (p-values <0.05 were considered significant). (D,E) ARID1A expression is observed in cells throughout the normal uterus and ovary by IHC. (E) ARID1A is expressed in the OSE of normal ovaries (arrowhead). (F) ARID1A expression is not observed in the tumors. (E) P-AKT S473 levels are low in the normal uterus. (H, I) P-AKT S473 in the normal ovary is highest in the OSE (E, arrowhead) and these levels are greatly increased in ovarian tumors (F, arrowhead). Asterisk in E denotes an oocyte. All sections processed for IHC were lightly counter-stained with methyl green. (J,K) Morbid Arid1a^fl/fl;(Gt)Rosa26Pik3ca^*H1047R mouse at sacrifice with hemorrhagic ascites (inset), primary ovarian tumor of moderate size, and bilateral tumor metastases (arrowheads). (L,M) Morbid Arid1a^fl/fl;(Gt)Rosa26Pik3ca^*H1047R mouse at sacrifice with hemorrhagic ascites (inset), large primary ovarian tumor, and no visible metastases. The mice shown in J-M were sacrificed at 7 and 9 weeks post-AdCRE, respectively, because of visible ascitic fluid burden. (N,O) Arid1a^fl/+;(Gt)Rosa26Pik3ca^*H1047R mice at 11-weeks post-AdCRE showing no evidence for tumor formation. In K and M, dashed circles indicate primary ovarian tumor on injected ovary. In N, arrows denote the AdCRE injected ovary. In K, M, and O, asterisks denote the uninjected, control ovary. UCEC, uterine corpus endometrial carcinoma; UCS, uterine carcinosarcoma; STAD, stomach adenocarcinoma; CESC, cervical squamous cell carcinoma and endocervical adenocarcinoma; HNSC, head and neck squamous cell carcinoma; COAD/ READ, colon and rectum adenocarcinoma; LUSC, lung squamous cell carcinoma; BRCA, breast invasive carcinoma; LUAD, lung adenocarcinoma; LGG, brain lower grade glioma; OV, ovarian serous cystadenocarcinoma; KIRC, kidney renal clear-cell carcinoma; LAML, acute myeloid leukemia; ACC, adrenocortical carcinoma; BLCA, bladder urothelial carcinoma; GBM, glioblastoma multiforme; KICH, kidney chromophobe; KIRP, kidney renal papillary cell carcinoma; PAAD, pancreatic adenocarcinoma; PRAD, prostate adenocarcinoma; SKCM, skin cutaneous melanoma; THCA, thyroid carcinoma.

Figure 3. Therapeutic PIK3CA inhibition promotes animal survival by reducing tumor cell viability

(A) Primary ascites-derived tumor cell isolation scheme. (B) Representative western blots to highlight the primary tumor cell validation scheme and demonstrate that ARID1A loss and heightened P-AKT S473 is observed in primary Arid1a^fl/fl;(Gt)Rosa26Pik3ca^*H1047R ascitic tumor cells versus normal ovarian surface epithelial (OSE) cells. (C) Western blots of P-AKT S473 and total AKT levels in Arid1a^fl/fl;(Gt)Rosa26Pik3ca^*H1047R ascitic tumor cells grown in the presence 10, 1, 0.001, or 0 µM BKM120 or equivalent dilutions (v/v) of DMSO vehicle control for 48 hrs. (D) BKM120 treatment reduces tumor cell viability with a half-maximal inhibitory concentration (IC₅₀) of 0.96 +/− 0.25 µM. Significant differences based on the average percent inhibition ±SD of three independent tumor cell lines treated with BKM120- versus vehicle-treated cells were calculated using a two-tailed Student’s t test (*p-value <0.05). (E,F) Administration of chow-fed BKM120 reduces PI3K pathway activity, as shown by reduced P-S6 Ser235/236 levels, and extends the Arid1a^fl/fl;(Gt)Rosa26Pik3ca^*H1047R median survival latency to 11 weeks. BKM120 treatment commenced 4 weeks after AdCRE injection. P-S6 Ser235/236 IHC was performed on tumor histological sections from three independent BKM120 treated or untreated mice. (F) Untreated Arid1a^fl/fl;(Gt)Rosa26Pik3ca^*H1047R mice succumb to cancer with a median (µ_1/2) latency of period of 7.5 weeks. Statistical differences in survival (BKM120 chow-fed versus untreated) were calculated using a log rank test (p-values <0.05 were considered significant).

Figure 4. Arid1a^fl/fl;(Gt)Rosa26Pik3ca^*H1047R ovarian tumors manifest clear-cell-like histological features

(A–E) Hematoxylin and Eosin (H&E) stained sections of tumor architectural patterns observed in intrabursal AdCRE-injected Arid1a^fl/fl;(Gt)Rosa26Pik3ca^*H1047R mice. Whole-mount tumor images are shown inset in A and E. Asterisks denote the uterus. Dotted line A’ demarcates the ovarian surface-tumor boundary. Arrowhead in D’ denotes a hobnail cell. (F–H) H&E stained sections of Arid1a^fl/fl;(Gt)Rosa26Pik3ca^*H1047R tumor metastases found on the surface of gastrointestinal tract (F), kidney (G), and uninjected, control ovary (H). Panels A’-H’ are high magnification images of regions demarcated by dashed boxes in A-H. (I) H&E stained exfoliated tumor cell aggregate. (K) H&E stained example of hobnail cells on the luminal surface a tumor. Arrowheads in D’, F’, I, and K denote hobnail cells on tumor surface. (J) Dot plot of primary tumor mass compared to uninjected control ovary. Significant differences based on the average mass ±SD between tumors and matched uninjected ovaries were calculated using a two-tailed Student’s t test (p-values <0.05 were considered significant). (L–N) H&E stained high magnification images of the ovarian surface from Arid1a^fl/+;(Gt)Rosa26Pik3ca^*H1047R, (Gt)Rosa26Pik3ca^*H1047R, and Arid1a^flfl mice. (M,N) OSE hyperplasia (arrowheads) is observed in (Gt)Rosa26Pik3ca^*H1047R and Arid1a^fl/+;(Gt)Rosa26Pik3ca^*H1047R mice. OV, ovary; Oo, oocyte; T, tumor; IE, intestinal epithelium; KD, kidney; OV, ovary; UT, uterus.

Figure 5. Arid1a^fl/fl;(Gt)Rosa26Pik3ca^*H1047R ovarian tumors and human OCCC share molecular characteristics

(A) H&E stained image of Arid1a^fl/fl;(Gt)Rosa26Pik3ca^*H1047R tumor histology. (B) Near-adjacent histological section showing HNF1β immunoreactivity by IHC. (A’,B’) High magnification images of regions demarcated by dashed boxes in A and B. (C) Box plots of normalized expression values of human OCCC discriminant genes (N=159) for all mouse tumor and human EOC subtype samples. Significant differences between the mouse tumors and each human histological EOC subtype were calculated using a Wilcoxon test (*significant p-value of <0.05; #not significant p-value=0.6). NS, not significant.

Figure 6. Tumor-derived IL-6 promotes OCCC tumor cell growth and survival

(A) Top OCCC-specific genes in common between mouse and human tumors. Top six IPA and MSigDB (GSEA) predictions of the top upregulated genes are included. (B) RT-PCR validation of IL-6 expression in primary tumors and peritoneal metastases. Significant differences based on average normalized mRNA expression ±SD between peritoneal metastases or primary tumors and matched uninjected ovaries were calculated using a two-tailed Student’s t test. (C,D) Mouse IL-6 levels in the serum and ascitic fluid Arid1a^fl/fl;(Gt)Rosa26Pik3ca^*H1047R mice, as measured by anti-IL-6 ELISAs. Significant differences based on the average protein concentration ±SD between wild-type versus Arid1a^fl/fl;(Gt)Rosa26Pik3ca^*H1047R mice were calculated using a two-tailed Student’s t test. (E) IL-6 expression in an Arid1a^fl/fl;(Gt)Rosa26Pik3ca^*H1047R tumor by IHC. (F) MTT absorbance values plotted with log antibody concentration (µg mL⁻¹) for non-specific rat IgG-treated (control) or rat anti-mouse IL-6 treated Arid1a^fl/fl;(Gt)Rosa26Pik3ca^*H1047R ascitic tumor cells after 96 hrs. of treatment. Plot represents the average absorbance value ±SD for treatment performed on three independent cell lines. Significant differences between control- and anti-IL-6-treated cells were calculated using a two-tailed Student’s t test (*significant p-value <0.05). (G) Primary ascitic tumor cells were treated 10, 1, 0.01, or 0 µg mL⁻¹ rat anti-mouse IL-6 or non-specific Rat IgG-treated (control). (H,I) IL-6 expression in Arid1a^fl/fl;(Gt)Rosa26Pik3ca^*H1047R ascitic tumor cells stably expressing control shRNAs or IL-6 shRNAs by ELISA and western blot. Each replicate represents in stable pool of IL-6 shRNA expressing cells from three independently isolated tumor cell lines. Significant differences based on the average protein concentration ±SD between control- versus IL-6 shRNA were calculated using a two-tailed Student’s t test. (J) MTT absorbance values plotted over time for control shRNA-, IL-6 shRNA-, or IL-6 shRNA-expressing cells supplemented with 10 ng/mL IL-6. Cells were plated at 2×10e4 cells per mL at hour 0, then MTT measurements were taken every 24 hours for a total of 96 hours. Significant differences between control- and IL-6-shRNA expressing cells were calculated using a two-tailed Student’s t test (*significant p-value < 0.05). (K,L) Whole-mount images of control- and IL-6-shRNA tumorgrafts (arrowheads) on the right flank of nude mice and images of corresponding H&E-stained tumorgraft sections. (M) Plot of growth rates for control- and IL-6-shRNA expressing tumorgrafts over 21 days of measurement. Significant differences between control- and IL-6-shRNA tumorgraft growth rates were calculated using a two-tailed Student’s t test. Only p-values <0.05 were considered significant.

Figure 7. Coexistent ARID1A-PIK3CA mutations induce IL-6 expression in a cooperative manner

(A) IL-6 mRNA expression in AdCRE-infected primary OSE cells carrying all allele combinations. (B) Normalized IL-6 ELISA measurements of AdCRE-infected primary OSE cells carrying all allele combinations. Significant differences based on the average normalized expression value ±SD among mutant allele combinations versus wildtype were calculated using a two-tailed Student’s t test. (C) Western blot showing ARID1A, P-AKT S473, total AKT, P-STAT3 Y705, and total STAT3 levels in AdCRE-infected primary OSE cells carrying all allele combinations. AdCRE-infected wildtype cells served as controls. (D) MTT cell viability assay on primary Arid1a^fl/fl;(Gt)Rosa26Pik3ca^*H1047R OSE cells treated with AdControl, AdCRE, or AdCRE supplemented with 5 µg mL⁻¹ anti-IL-6 antibodies. (E) Treatment of normal (wild type) primary OSE cells with vehicle (v/v), 10 ng mL⁻¹ IL-6, or 10 ng mL⁻¹ IL-6 supplemented with 5 µg mL⁻¹ anti-IL-6 antibodies. Significant differences based on the average MTT absorbance value ±SD between three independent replicates of AdControl- versus AdCRE-infected or vehicle versus IL-6-treated cells were calculated using a two-tailed Student’s t test. (F) Western blot of normal primary OSE cells showing dose-dependent increases in P-STAT3 Y705 levels following IL-6 treatment. Co-treatment with 5 µg mL⁻¹ anti-IL-6 antibodies blocked P-STAT3 Y705 induction. (G) ARID1A occupancy at the IL6 locus was detected by chromatin immunoprecipitation (ChIP) using anti-ARID1A antibodies (denoted as IP) on crosslinked chromatin from AdControl- or AdCRE-infected Arid1a^fl/fl;(Gt)Rosa26Pik3ca^*H1047R primary OSE cells. Non-specific isotype matched antibodies (denoted as NS) were used in control ChIPs. ARID1A occupancy at the IL6 promoter (site or primer pair B) is reduced in ARID1A depleted cells treated with AdCRE. Average percent ChIP input ±SD represent data from experiments performed using three primary OSE cell isolations. Significant differences were calculated using a two-tailed Student’s t test. (H) Proposed model of IL-6 regulation by ARID1A and PIK3CA mutations in OCCC. Only p-values <0.05 were considered significant.