LKB1 and STRADα Promote Epithelial Ovarian Cancer Spheroid Cell Invasion

Abstract

Late-stage epithelial ovarian cancer (EOC) involves the widespread dissemination of malignant disease throughout the peritoneal cavity, often accompanied by ascites. EOC metastasis relies on the formation of multicellular aggregates, called spheroids. Given that Liver Kinase B1 (LKB1) is required for EOC spheroid viability and LKB1 loss in EOC cells decreases tumor burden in mice, we investigated whether the LKB1 complex controls the invasive properties of human EOC spheroids. LKB1 signalling was antagonized through the CRISPR/Cas9 genetic knockout of LKB1 and/or the RNAi-dependent targeting of STE20-related kinase adaptor protein (STRAD, an LKB1 activator). EOC spheroids expressing nuclear GFP (green) or mKate2 (red) constructs were embedded in Matrigel for real-time live-cell invasion monitoring. Migration and invasion were also assessed in spheroid culture using Transwell chambers, spheroid reattachment, and mesothelial clearance assays. The loss of LKB1 and STRAD signalling decreased cell invasion through Matrigel and Transwell membranes, as well as mesothelial cell clearance. In the absence of LKB1, zymographic assays identified a loss of matrix metalloproteinase (MMP) activity, whereas spheroid reattachment assays found that coating plates with fibronectin restored their invasive potential. A three-dimensional EOC organoid model demonstrated that organoid area was greatly reduced by LKB1 loss. Overall, our data indicated that LKB1 and STRAD facilitated EOC metastasis by promoting MMP activity and fibronectin expression. Given that LKB1 and STRAD are crucial for EOC metastasis, targeting LKB1 and/or STRAD could disrupt the dissemination of EOC, making inhibitors of the LKB1 pathway an alternative therapeutic strategy for EOC patients.

Keywords: STE20-related kinase adaptor protein (STRAD); epithelial ovarian cancer (EOC); liver kinase B1 (LKB1); mesothelial clearance; metastasis; organoid; spheroid.

Figure 1

LKB1 and STRADα regulate EOC spheroid size and viability. (A) HeyA8, HeyA8 STK11 KO, OVCAR8, and OVCAR8 STK11 KO cells were transfected with 10 nM siNT or 10 nM siSTRADA for 72 h. OVCAR3 and OVCAR4 cells were transfected with 10 nM siNT, 10 nM siSTRADA, or 10 nM siSTK11 for 72 h. (B) FT190 and FT190 STK11 KO cells were transfected with 10 nM siNT or 10 nM siSTRADA for 72 h. For each experimental condition, 2000 cells were seeded in a 96-well ULA cluster plate and imaged every 24 h using an IncuCyte S3 Live-Cell Analysis System. Representative brightfield images of each experimental condition were captured 48–168 h post-seeding for all cell lines. CellTiter-Glo assessed spheroid cell viability at the 168 h time point, and representative graphs for relative viability are found to the left of the representative brightfield images. Spheroid area was measured using the Sartorius spheroid module analysis tool, and spheroid doubling time was calculated using the rate of change of the spheroid area. Representative graphs for spheroid area and spheroid doubling time are located to the right of the representative brightfield images. The quantification represents three to five independent experiments (mean ± SD). Significance is indicated as * = p < 0.05, ** = p < 0.01, *** = p < 0.001, and **** p < 0.0001. Scale bars = 0.5 mm.

Figure 2

EOC spheroid cell dispersion and mesothelial clearance are dependent on LKB1 and STRADα. HeyA8, HeyA8 STK11 KO, OVCAR8, and OVCAR8 STK11 KO cells were transfected with 10 nM siNT or 10 nM siSTRADA for 72 h. OVCAR3 and OVCAR4 cells were transfected with 10 nM siNT, 10 nM siSTRADA, or 10 nM siSTK11 for 72 h. For all treatments, 2000 cells were seeded in 96-well ULA plates to form spheroids for 24 h. Spheroids were reattached using standard tissue-culture-treated 96-well plates and imaged after 24 h (HeyA8 and OVCAR8) or 48 h (OVCAR3 and OVCAR4) using a Leica DMI 4000B inverted microscope. ImageJ was used to quantify initial spheroid size and cell migration post-reattachment relative to siNT. Graphs from three independent experiments (mean ± SD) are located to the right of the representative images. Based on each cell line’s response to the downregulation of STK11 or STRADA, (A) reattached HeyA8 and OVCAR4 spheroids were paired and (B) reattached OVCAR8 and OVCAR3 spheroids were paired. Black scale bars = 1 mm and red bars represent one of four quadrants measured to quantify relative migration post-spheroid reattachment. (C) A schematic representation of mesothelial clearance assay. ZT-GFP human mesothelial cells (in green) are displaced by re-attaching EOC spheroids (in red) expressing mKate2 conjugated to a nuclear localization signal (NLS). (D) HeyA8, HeyA8 STK11 KO, OVCAR8, and OVCAR8 STK11 KO cells stably expressing mKate2-NLS (red) were transfected with 10 nM siNT or 10 nM siSTRADA for 72 h. For all experimental conditions, 2000 cells were seeded in 96-well ULA plates and grown for 24 h. EOC spheroids were reattached to a confluent monolayer of ZT-GFP mesothelial cells cultured on collagen-coated 24-well plates. Images of mesothelial clearance at 24 h post-reattachment were captured using a Leica DMI 4000B inverted microscope. ImageJ was used to quantify EOC spheroid area and mesothelial clearance area. Relative mesothelial clearance for each treatment was standardized to original spheroid size. Scale bars = 0.5 mm. Significance is indicated as * = p < 0.05, ** = p < 0.01, *** = p < 0.001, and **** = p < 0.0001.

Figure 3

Downregulating LKB1 and STRADα disrupts invasion, as established through 3D in vitro models of mesothelial layers. (A) A schematic representation of the mesothelial invasion assay. In the Transwell chamber, the upper chamber was filled with serum-free medium and the lower chamber was filled with serum-rich medium, separated by a porous Transwell membrane. The Transwell membranes were coated with rat-tail collagen followed by a confluent monolayer of ZT-GFP mesothelial cells (green). mKate2-NLS-expressing EOC spheroid cells (red) were reattached to the mesothelial layer: EOC spheroid cells cleared the mesothelial layer, reorganized the underlying collagen, and invaded across the membrane to the lower chamber. (B) HeyA8, HeyA8 STK11 KO, OVCAR8, and OVCAR8 STK11 KO cells stably expressing mKate2-NLS (red) were transfected with 10 nM siNT or 10 nM siSTRADA for 72 h. For all experimental conditions, 2000 EOC cells were seeded in 96-well ULA plates and grown for 24 h. EOC spheroids were reattached to a confluent monolayer of ZT-GFP mesothelial cells cultured on rat-tail collagen in the top chamber of a Transwell membrane. A Leica DMI 4000B inverted microscope was used to image mesothelial clearance 24 h post-reattachment. ImageJ was used to quantify EOC spheroid area and mesothelial clearance area. The mesothelial clearance for each treatment was standardized to the original spheroid size and siNT was set to 1. Scale bars = 0.5 mm. (C) Spheroid cell invasion was assessed 48 h post-seeding. The Transwell membranes were fixed, stained with DAPI, and mounted on microscope slides. Images of DAPI (blue), mKate2-NLS (red), and GFP (green) were acquired using a Leica DMI 4000B inverted microscope, and DAPI-stained nuclei overlapping with mKate2-NLS were counted using ImageJ. The percentage of invading cells for each experimental condition was standardized to the siNT control. Scale bars = 40 μm. The quantification represents three independent experiments (mean ± SD). Significance is indicated as * = p < 0.05, ** = p < 0.01, *** = p < 0.001, and **** = p < 0.0001.

Figure 4

Downregulating LKB1 and STRADα disrupts EOC spheroid invasion through Matrigel. (A) HeyA8, HeyA8 STK11 KO, OVCAR8, and OVCAR8 STK11 KO cell lines expressing mKate2-NLS (red) were transfected with 10 nM siNT or 10 nM siSTRADA for 72 h. After seeding 2000 cells per well in a 96-well ULA plate for 24 h, spheroids were embedded in Matrigel at a final concentration of 50%. After 72 h, spheroid invasion area was quantified using ImageJ and normalized to the siNT controls from three independent experiments. Graphs of relative Matrigel invasion are located below representative brightfield and mKate2-NLS images. (B) HeyA8 or OVCAR8 wild-type and STK11 KO cell lines expressing nuclear-localized mKate2 (red) or GFP (green) were mixed in a 1:1 ratio. For each cell line, 1000 wild-type and 1000 STK11 KO cells were seeded per well in a 96-well ULA plate to generate mixed spheroids. After 24 h, spheroids were embedded in a final concentration of 50% Matrigel. An IncuCyte S3 Live-Cell Analysis System imaged the spheroids 72 h post-Matrigel embedding, and ImageJ was used to quantify the number of invading cells. Graphs from three independent experiments (mean ± SD) are shown and significance is indicated as * = p < 0.05, ** = p < 0.01, *** = p < 0.001, and **** = p < 0.0001. Scale bars = 1 mm.

Figure 5

The impact of LKB1 and STRADα on matrix metalloproteinase (MMP) 2/9 activity. HeyA8, HeyA8 STK11 KO (A), OVCAR8, and OVCAR8 STK11 KO (B) cells were transfected with 10 nM siNT or 10 nM siSTRADA for 72 h. All experimental conditions were reseeded in a standard 6-well tissue-culture-treated plate (monolayer condition) or in a 96-well ULA plate (spheroid condition). Both monolayer cells and spheroids were serum-starved for 48 h, and the conditioned media were concentrated prior to zymographic assays. Gels for zymography contained gelatin and were incubated in MMP-reactivating buffers for 16 h. After MMP reactivation, the gels were stained with Coomassie Brilliant Blue G-250 prior to imaging and densitometry. Quantification represents relative MMP9 and MMP2 activities relative to the siNT control in monolayer cells, as determined from three independent experiments (mean ± SD). (C) HeyA8 and OVCAR8 cell monolayers and reattached spheroids responsible for generating the conditioned media utilized in zymography were lysed and subjected to SDS-PAGE and immunoblotting using antibodies specific for LKB1, STRADα, and vinculin. (D) HeyA8, HeyA8 STK11 KO, OVCAR8, and OVCAR8 STK11 KO cells stably expressing mKate2-NLS (red) were transfected with 10 nM siNT or 10 nM siSTRADA for 72 h. For all experimental conditions, 2000 cells were seeded in 96-well ULA plates and grown for 24 h. The EOC spheroids were reattached to a confluent monolayer of ZT-GFP mesothelial cells cultured on rat-tail collagen. After 72 h of mesothelial clearance, EOC spheroids and ZT cell co-cultures were serum-starved for 48 h to generate conditioned media for zymographic assays. Quantifications of MMP9 activity relative to ZT-GFP mesothelial cells from three independent experiments (mean ± SD) are graphed to the right of the representative zymographic blots. (E) EOC spheroids were reattached to a confluent monolayer of ZT-GFP cells 24 or 72 h prior to zymography. Red boxes in upper zymograms are re-displayed below for better visualization of MMP9 activity. Quantifications of MMP9 activity relative to siNT controls from three independent experiments (mean ± SD) are graphed below representative zymographic blots. Significance is indicated as * = p < 0.05, ** = p < 0.01, *** = p < 0.001, and **** = p < 0.0001.

Figure 6

Fibronectin restores STK11 KO cell migration post-spheroid reattachment. (A) Global transcriptome analysis of OVCAR8 wild-type (WT) and OVCAR8 STK11 KO cultured as 24 h spheroids using the Affymetrix Human Clariom S microarray. The fold changes of mRNA transcripts expressed in OVCAR8/OVCAR8 STK11 KO spheroids were graphed, and fibronectin (FN1) mRNA expression is indicated (red dot). Only mRNAs with a greater expression in OVCAR8 spheroids are represented in the graph. (B) RT-qPCR was performed to assess FN1 mRNA expression in HeyA8, HeyA8 STK11 KO, OVCAR8, and OVCAR8 STK11 KO spheroids transfected with 10 nM siNT or 10 nM siSTRADA for 72 h. FN1 expression was normalized to siNT controls from three independent experiments (mean ± SD). (C) Proteomic analysis of OVCAR8 wild-type (WT) and OVCAR8 STK11 KO spheroids grown for 24 h. Spheroids were lysed and subjected to tandem mass tag isobaric labelling and mass spectrometry analysis. All detectable peptides of the kinome were graphed as OVCAR8-OVCAR8 STK11 KO, and fibronectin protein abundance is indicated (red dot). Only proteins with a greater abundance in OVCAR8 cells are represented in the graph. (D) HeyA8, HeyA8 STK11 KO, OVCAR8, and OVCAR8 STK11 KO cells were transfected with 10 nM siNT or 10 nM siSTRADA for 72 h. Cells were reseeded into 6-well ULA plates to generate spheroids, and after 72 h, spheroids were lysed for SDS-PAGE and immunoblotting using antibodies specific for fibronectin, LKB1, STRADα, and actin. LKB1 and STRADα verified knockout and knockdown, respectively, whereas total fibronectin was quantified, standardized to actin, and graphed below representative immunoblots relative to siNT control from three independent experiments (mean ± SD). (E) HeyA8, HeyA8 STK11 KO, OVCAR8, and OVCAR8 STK11 KO cells were transfected with 10 nM siNT or 10 nM siSTRADA for 72 h. The conditioned media (CM) were isolated and prepared for immunoblotting and the cells that produced the CM were lysed and subjected to SDS-PAGE and immunoblotting using antibodies specific for fibronectin, LKB1, and STRADα. The blots were stripped and stained with Coomassie Brilliant Blue G250, which functioned as a loading control. Total fibronectin was quantified for cell lysates and CM, standardized to Coomassie Brilliant Blue G250, and graphed. The graphs are shown below representative immunoblots relative to siNT control from three independent experiments (mean ± SD). (F) HeyA8, HeyA8 STK11 KO, OVCAR8, and OVCAR8 STK11 KO cells were transfected with 10 nM siNT or 10 nM siSTRADA for 72 h. Spheroids grown for 24 h in 96-well ULA plates were reattached for 48 h to standard tissue-culture-treated plates or 5 µg/cm² fibronectin-coated plates. Cell viability was assessed by an alamarBlue assay followed by fixation and cell staining using Hema3. The relative migration and cell viability was quantified, and graphs from three independent experiments (mean ± SD) are shown to the right of the representative micrographs. Significance is indicated as * = p < 0.05, ** = p < 0.01, *** = p < 0.001, and **** = p < 0.0001.

Figure 7

LKB1 inactivation disrupts EOC organoid growth. (A) HeyA8, HeyA8 STK11 KO, OVCAR8, and OVCAR8 STK11 KO cell lines were seeded in Cultrex BME with modified organoid growth medium and imaged with an IncuCyte S3 Live-Cell Analysis System every 12 h for 14 days (HeyA8) or 21 days (OVCAR8). The IncuCytes organoid analysis software calculated total organoid area (µm²) and organoid count per image. Graphs of average organoid size (i.e., organoid area/count) are located to the right of the representative micrographs. (B) Brightfield images of the EOC organoids were acquired using a Leica DMI 4000B inverted microscope after 14 days (HeyA8) or 21 days (OVCAR8) of growth. Graphs from three independent experiments (mean ± SD) are located below the representative images and significance is indicated as * = p < 0.05 and ** = p < 0.01. Scale bars = 200 μm. (C) HeyA8 and OVCAR8 organoids and spheroids cultured for 3 or 21 days were lysed and subjected to SDS-PAGE and immunoblotting using antibodies specific for LKB1 and vinculin.

Figure 8

Proposed model of LKB1 pathway activity during ovarian cancer metastasis. EOC cells disseminate from the primary ovarian tumor and form multicellular spheroids in the peritoneal cavity. EOC spheroids colonize distant sites by invading across the mesothelial lining of the peritoneal organs, reorganizing the underlying extracellular matrix, and establishing secondary tumors. Our in vitro models of EOC metastasis indicated that intact LKB1 signalling is important for EOC spheroid formation and viability in suspension, mesothelial clearance after reattachment, extracellular matrix remodeling, and new secondary tumor growth. The ablation of LKB1 expression decreased each of these key metastatic steps, which may be due to reduced viability, fibronectin production, or MMP9 activity.