Loss of primary cilia promotes EphA2-mediated endothelial-to-mesenchymal transition in the ovarian tumor microenvironment

Abstract

Endothelial-to-mesenchymal transition (EndMT) is closely associated with tumor progression. Endothelial cells (ECs) in the tumor microenvironment (TME) use EndMT programs to facilitate tumor progression; however, the underlying mechanisms in ovarian cancer are poorly understood. Here, we describe the involvement of primary cilia in EndMT of the ovarian TME. We showed that ECs from human ovarian tumors displayed robust EndMT and impaired cilia formation, as was also observed in ECs in response to ovarian cancer cell culture-conditioned media (OV-CM). Notably, ECs lacking primary cilia exhibited increased OV-CM-induced EndMT. Vascular abnormalities, such as enhanced cell migration and vessel permeability, were observed in vitro. Furthermore, in vivo experiments using endothelial-specific kinesin family member 3A (Kif3a)-knockout mice showed enhanced EndMT in the ovarian TME. Mechanistically, we identified ephrin type-A receptor 2 (EphA2) as a key regulator of EndMT. Upon OV-CM treatment, EphA2 expression increased, and depletion of EphA2 in ECs decreased OV-CM-induced EndMT and vascular abnormalities. These results highlight that the loss of primary cilia and the consequent EphA2 activation are key mechanisms by which EndMT programs induce the acquisition of cancer-associated fibroblast-like cells in the ovarian TME, thereby promoting ovarian cancer progression.

Keywords: EPH receptor A2 (EphA2); endothelial‐to‐mesenchymal transition (EndMT); kinesin family protein 3a (Kif3a); ovarian cancer; primary cilia.

Fig. 1

Robust EndMT and impaired primary cilia formation in Ovarian TAEs. (A) Protein expression of the endothelial markers and the mesenchymal markers in ovarian TAEs. TAEs were isolated from cell suspensions using magnetically activated cell sorting using CD31 antibody‐based magnetic cell sorting. Cell lysates were analyzed by immunoblot and GAPDH was used as the loading control. TAEs were isolated from three independent human ovarian tumor samples (n = 3). (B) Immunofluorescence images of primary cilia in normal ECs and ovarian TAEs. Primary cilia on the cells were immunostained with ARL13B (green), and acetylated‐α‐tubulin (red). Nuclei were stained with DAPI. The arrows indicate the primary cilia. Values represent the mean ± SD from three independent experiments. The percentage of ciliated cells was evaluated (n > 100 cells per condition); **P < 0.01, *P < 0.05, by one‐way ANOVA with Dunnett's test. n = 3 (Scale bar = 20 μm). (ECs, Endothelial cells; EndMT, Endothelial‐to‐mesenchymal transition; FN, Fibronectin; HMVEC, Human microvascular endothelial cells; HUVEC, Human umbilical vein endothelial cells; SM22α, smooth muscle protein 22‐alpha; TAE, Tumor‐associated endothelial cell; α‐SMA, alpha‐smooth muscle actin).

Fig. 2

Loss of primary cilia shows increased EndMT. (A, B) HUVECs were treated with CM or OV‐CM for 24 h. (A) Protein expression of the endothelial markers and the mesenchymal markers. Cell lysates were analyzed by immunoblot and GAPDH was used as the loading control. Representative results from three independent experiments are shown. (B) Immunofluorescence images of primary cilia. Primary cilia on the cells were immunostained with ARL13B (green) and acetylated‐α‐tubulin (red). Nuclei were stained with DAPI. The arrows indicate the primary cilia. Values represent the mean ± SD from four independent experiments. The percentage of ciliated cells was determined (n > 100 cells per condition); **P < 0.01, by 2‐tailed student's t test. n = 4 (Scale bar = 20 μm). (C) Quantification of Kif3a expression in HUVECs. HUVECs were transfected with siKif3A 1 ng·L⁻¹ for 48 h, and siGFP was used as a negative control. The quantification was examined by RT‐qPCR (left) and immunoblot (right). mRNA and protein levels were normalized by β‐Actin; **P < 0.01, by two‐tailed student's t test. n = 3 (D) Immunofluorescence images of primary cilia in HUVECs transfected with 1 ng·L⁻¹ siKif3A. Primary cilia on the cells were immunostained with ARL13B (red) and acetylated‐α‐tubulin (green). Nuclei were stained with DAPI. Values represent the mean ± SD from three independent experiments. The arrows indicate the primary cilia; *P < 0.05, by two‐tailed student's t test. n = 3 (Scale bar = 20 μm) (E) Protein expression of the endothelial markers and the mesenchymal markers. HUVECs were transfected with siKif3A 1 ng·L⁻¹ for 48 h and treated with OV‐CM for 24 h. Cell lysates were analyzed by Immunoblot. Values represent the mean ± SD from three independent experiments. *P < 0.05, **P < 0.01, ****P < 0.0001, by 2‐tailed student's t test. (F) Immunofluorescence analysis of ID8 tumor sections. Tumors were sectioned at 6 μm and immunostained with CD31 (green) and α‐SMA (red). Nuclei were stained with DAPI. Values represent the mean ± SD from three independent experiments. Representative data are shown from four mice; *P < 0.05, by two‐tailed student's t test. (Scale bar = 20 μm). (CM, control media; EndMT, Endothelial‐to‐mesenchymal transition; FN, Fibronectin; HUVEC, Human umbilical vein endothelial cells; Kif3a, Kinesin Family Member 3A; OV‐CM, Ovarian cancer cell culture‐conditioned media; SM22α, smooth muscle protein 22‐alpha; α‐SMA, alpha‐smooth muscle Actin).

Fig. 3

EndMT induces vascular abnormality in vitro. (A) Migration assay in HUVECs treated with OV‐CM. The migration of HUVECs was measured using a wound‐healing assay. Images were taken every 12 h after scratch for 24 h in the presence of OV‐CM. Values represent the mean ± SD from three independent experiments. Representative images of n = 3 were shown; *P < 0.05, by two‐tailed student's t test. (Scale bar = 200 μm) (B) Cell permeability assay in HUVECs. The permeability of an EC monolayer was investigated by measuring the fluorescence of FITC‐dextran diffusing over a transwell membrane. Values represent the mean ± SD from three independent experiments. **P < 0.01, by two‐tailed student's t test. n = 3 (C) Dil‐Ac‐LDL uptake assay in HUVECs. HUVECs were incubated with Dil‐Ac‐LDL 10 μg·mL⁻¹. Nuclei were stained with DAPI (Scale bar = 100 μm). Representative images of n = 3 were shown. (CM, control media; EndMT, Endothelial‐to‐mesenchymal transition; HUVEC, Human umbilical vein endothelial cells; OV‐CM, Ovarian cancer cell culture‐conditioned media).

Fig. 4

EphA2 Is a key regulator of EndMT. (A, B) HUVECs were treated with CM or OV‐CM (A) RTK assay in HUVECs. HUVECs were treated for 72 h and cell lysates were analyzed using a phosphoreceptor tyrosine kinase array. Values represent the mean ± SD from three independent experiments. ****P < 0.0001, two‐way ANOVA with Bonferroni's test. (B) Protein expression of EphA2 and p‐EphA2 at Tyr⁷⁷². HUVECs were treated with OV‐CM or CM for 24 h and cell lysates were analyzed by Immunoblot. Values represent the mean ± SD from three independent experiments. *P < 0.05, ***P < 0.001, by 2‐tailed student's t test. (C, D) HUVECs transfected with siEphA2 were treated with OV‐CM for 24 h. (C) Immunofluorescence analysis in cells with an anti–VE‐cadherin antibody. Representative images of n = 3 were shown (Scale bar = 40 μm). (D) Protein expression of the endothelial markers and the mesenchymal markers. Cell lysates were analyzed by immunoblot and β‐Actin was used as the loading control. Values represent the mean ± SD from three independent experiments. *P < 0.05, ***P < 0.001, ****P < 0.0001, by 2‐tailed student's t test. (E) Migration assay in HUVECs transfected with siEphA2. The migration of EphA2 depleted HUVECs was measured using a wound‐healing assay. Images were taken after 24 h of scratch in the presence of OV‐CM. Values represent the mean ± SD from three independent experiments. Representative images of n = 3 were shown; **P < 0.001, ***P < 0.0001, by two‐way ANOVA with Tukey's test. (Scale bar = 200 μm) (F) Dil‐Ac‐LDL uptake assay in HUVECs. HUVECs were transfected with siEphA2 20 nm for 48 h and treated with OV‐CM for 24 h. EphA2 depleted HUVECs were incubated with Dil‐Ac‐LDL 10 μg·mL⁻¹. Nuclei were stained with DAPI (blue). Representative images of n = 3 were shown (Scale bar = 100 μm). (G) Expression of EphA2 in Kif3a‐depleted HUVECs treated with OV‐CM. HUVECs transfected with siEphA2 were treated with OV‐CM for 24 h. Cell lysates were analyzed by Immunoblot and β‐Actin was used as the loading control. Values represent the mean ± SD from three independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001, by 2‐tailed student's t test. (H) Schematic summarizing the relationship between EphA2 and primary cilia in EndMT. (CM, control media; EndMT, Endothelial‐to‐mesenchymal transition; EphA2, EPH Receptor A2; FN, Fibronectin; HUVEC, Human umbilical vein endothelial cells; Kif3a, Kinesin Family Member 3A; OV‐CM, Ovarian cancer cell culture‐conditioned media; SM22α, smooth muscle protein 22‐alpha; α‐SMA, alpha‐smooth muscle Actin).