FOXK2 promotes ovarian cancer stemness by regulating the unfolded protein response pathway
Yaqi Zhang, … , Mazhar Adli, Daniela Matei
Yaqi Zhang, … , Mazhar Adli, Daniela Matei
Published March 29, 2022
Citation Information: J Clin Invest. 2022;132(10):e151591. https://doi.org/10.1172/JCI151591.
View: Text | PDF
Research Article Cell biology Oncology

FOXK2 promotes ovarian cancer stemness by regulating the unfolded protein response pathway

Abstract

Understanding the regulatory programs enabling cancer stem cells (CSCs) to self-renew and drive tumorigenicity could identify new treatments. Through comparative chromatin-state and gene expression analyses in ovarian CSCs versus non-CSCs, we identified FOXK2 as a highly expressed stemness-specific transcription factor in ovarian cancer. Its genetic depletion diminished stemness features and reduced tumor initiation capacity. Our mechanistic studies highlight that FOXK2 directly regulated IRE1α (encoded by ERN1) expression, a key sensor for the unfolded protein response (UPR). Chromatin immunoprecipitation and sequencing revealed that FOXK2 bound to an intronic regulatory element of ERN1. Blocking FOXK2 from binding to this enhancer by using a catalytically inactive CRISPR/Cas9 (dCas9) diminished IRE1α transcription. At the molecular level, FOXK2-driven upregulation of IRE1α led to alternative XBP1 splicing and activation of stemness pathways, while genetic or pharmacological blockade of this sensor of the UPR inhibited ovarian CSCs. Collectively, these data establish what we believe is a new function for FOXK2 as a key transcriptional regulator of CSCs and a mediator of the UPR, providing insight into potentially targetable new pathways in CSCs.

Authors

Yaqi Zhang, Yinu Wang, Guangyuan Zhao, Edward J. Tanner, Mazhar Adli, Daniela Matei

×

Figure 3

FOXK2 directly regulates IRE1α expression in OC cells.

Options: View larger image (or click on image) Download as PowerPoint
FOXK2 directly regulates IRE1α expression in OC cells. (A) Density plots...
(A) Density plots (upper) and heatmaps (lower) of normalized FOXK2 and H3K27Ac ChIP-seq reads at regions differentially bound by FOXK2 in OVCAR5. (B) Venn diagram shows numbers of overlapping genes FOXK2 peaks in ChIP-seq (FDR < 0.05) and DEGs in RNA-seq (log2[fold change] > 2, FDR < 0.05) in OVCAR5 transduced with shFOXK2-2 (shFOXK2) versus shCtrl. (C) Volcano plot of overlapping genes described in B. (D) Integrative Genomics Viewer (IGV, https://software.broadinstitute.org/software/igv/) tracks of the FOXK2 binding peak in the ERN1 gene, and ERN1 mRNA by RNA-seq in shCtrl and shFOXK2 cells. The FOXK2 binding motif is indicated along with the position of gRNA sequences used (1, 2, 3). (E) ChIP-qPCR shows binding of FOXK2 to the ERN1 gene in HGSOC tumors (n = 3). Amplification of a sequence 1 kb downstream was used as a control. (F) ChIP-qPCR measured enrichment of H3K27Ac in the FOXK2 binding site of the ERN1 gene (n = 3). Amplification of a sequence 1 kb downstream was used as a control. (G) ERN1 mRNA levels in OVCAR5 transduced with nontargeting dCas9-gRNA (dCas9-NT) or dCas9-sgRNA targeting the FOXK2 binding motif on ERN1 (dCas9-ERN1-1 through -3) (n = 3). The position of target sequences for gRNAs is indicated in D. (H) ChIP-qPCR using the same primers and control as in E shows binding of FOXK2 to the ERN1 gene in OVCAR5 transduced with dCas9-NT, dCas9-ERN1-1, or dCas9-ERN1-2. (I and J) ERN1 mRNA levels in shCtrl- and shFOXK2-transduced OC cells and HGSOC tumors (n = 3) (I) and in EV- or FOXK2-OE–transfected OC cells (n = 3) (J). (K) Scatter plot shows the correlation between mRNA levels of FOXK2 and ERN1 in ovarian tumors profiled by TCGA (n = 427). *P < 0.05; **P < 0.01; ***P < 0.005; ****P < 0.0001, by unpaired, 2-tailed Student’s t test.