EDI3 knockdown in ER-HER2+ breast cancer cells reduces tumor burden and improves survival in two mouse models of experimental metastasis

Abstract

Background: Despite progress understanding the mechanisms underlying tumor spread, metastasis remains a clinical challenge. We identified the choline-producing glycerophosphodiesterase, EDI3 and reported its association with metastasis-free survival in endometrial cancer. We also observed that silencing EDI3 slowed cell migration and other cancer-relevant phenotypes in vitro. Recent work demonstrated high EDI3 expression in ER-HER2+ breast cancer compared to the other molecular subtypes. Silencing EDI3 in ER-HER2+ cells significantly reduced cell survival in vitro and decreased tumor growth in vivo. However, a role for EDI3 in tumor metastasis in this breast cancer subtype was not explored. Therefore, in the present work we investigate whether silencing EDI3 in ER-HER2+ breast cancer cell lines alters phenotypes linked to metastasis in vitro, and metastasis formation in vivo using mouse models of experimental metastasis.

Methods: To inducibly silence EDI3, luciferase-expressing HCC1954 cells were transduced with lentiviral particles containing shRNA oligos targeting EDI3 under the control of doxycycline. The effect on cell migration, adhesion, colony formation and anoikis was determined in vitro, and significant findings were confirmed in a second ER-HER2+ cell line, SUM190PT. Doxycycline-induced HCC1954-luc shEDI3 cells were injected into the tail vein or peritoneum of immunodeficient mice to generate lung and peritoneal metastases, respectively and monitored using non-invasive bioluminescence imaging. Metabolite levels in cells and tumor tissue were analyzed using targeted mass spectrometry and MALDI mass spectrometry imaging (MALDI-MSI), respectively.

Results: Inducibly silencing EDI3 reduced cell adhesion and colony formation, as well as increased susceptibility to anoikis in HCC1954-luc cells, which was confirmed in SUM190PT cells. No influence on cell migration was observed. Reduced luminescence was seen in lungs and peritoneum of mice injected with cells expressing less EDI3 after tail vein and intraperitoneal injection, respectively, indicative of reduced metastasis. Importantly, mice injected with EDI3-silenced cells survived longer. Closer analysis of the peritoneal organs revealed that silencing EDI3 had no effect on metastatic organotropism but instead reduced metastatic burden. Finally, metabolic analyses revealed significant changes in choline and glycerophospholipid metabolites in cells and in pancreatic metastases in vivo.

Conclusions: Reduced metastasis upon silencing supports EDI3's potential as a treatment target in metastasizing ER-HER2+ breast cancer.

Keywords: Anoikis; Breast cancer; Choline metabolism; GPCPD1; Glycerophospholipid metabolism; HER2 positive breast cancer; Metastasis.

Fig. 1

EDI3 silencing alters glycerophospholipid metabolism in ER-HER2+ breast cancer cells. A Schematic of the inducible lentiviral shRNA vector (SMARTvector™, Dharmacon) for doxycycline-dependent expression of shRNA oligos. B EDI3 mRNA expression and C representative Western blot showing EDI3 protein expression in HCC1954-luc shNEG and HCC1954-luc shEDI3 (oligos shEDI3 #1 and #2) cells after 72 h treatment with 0.01 or 0.1 µg/ml doxycycline. D Overview of the metabolic pathways downstream of EDI3. By hydrolyzing GPC to choline and G3P, EDI3 plays a role in choline metabolism via the Kennedy pathway as well as in the synthesis of signaling and structural glycerophospholipids. E Intracellular GPC/PCho ratio, and intracellular levels of F choline, G G3P, H 16:0 LPA and I 16:0 LPC measured using LC–MS/MS in HCC1954-luc shNEG and shEDI3 after 72 h treatment with 0.01 or 0.1 µg/ml doxycycline (left panel) as well as in SUM190PT (middle panel) and SKBR3 cells (right panel) after silencing EDI3 using two different siRNA oligos. Metabolite levels were determined by calculating the ratios of the integrated peaks of the endogenous metabolites and the internal standards. Quantities of metabolites were normalized to cell number and presented relative to untreated control cells (shNEG or siNEG). Data represent mean ± SD from at least five technical replicates (wells from a six-well plate) from one or two independent experiments (*p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001). GPC, glycerophosphocholine; PCho, phosphocholine; LPA, lysophosphatidic acid; PA, phosphatidic acid; LPC, lysophosphatidylcholine; LPS, lysophosphatidylserine; LPG, lysophosphatidylglycerol; DAG, diacylglycerol; PtdCho, phosphatidylcholine; CDP-Cho, cytidine diphosphate choline; PS, phosphatidylserine; G3P, glycerol-3-phosphate; CDP-DAG, cytidine diphosphate diacylglycerol; PGP, phosphatidylglycerol phosphate; PG, phosphatidylglycerol

Fig. 2

Doxycycline-induced EDI3 silencing influences colony formation, cell adhesion and anoikis resistance. A Representative images (left panel) and corresponding quantification (right panel) of the wound healing assay with HCC1954-luc shNEG and shEDI3 cells treated with 0.1 µg/ml doxycycline. B Representative images (upper panel) and corresponding quantification (lower panel) of colonies (number and size) formed by HCC1954-luc shEDI3 cells treated with 0.01 or 0.1 µg/ml doxycycline at the time of (plated in ± Dox) or after (Dox added 24 h after plating) plating. C Representative pictures (left) of HCC1954-luc shNEG and shEDI3 cells stained with crystal violet after 30 min of adhesion to a fibronectin matrix and quantification (right panel) shown as RFU after 20 and 30 min measured at 570 nm after destaining. D Viability (RFU) of HCC1954-luc shNEG and shEDI3 cells induced with 0.1 µg/ml doxycycline relative to non-induced control cells (0 μg/ml doxycycline) measured 24 h after plating on a poly-HEMA matrix as a measure of anoikis resistance. Scale bars represent 100 μm. Values in graphs represent mean ± SD from at least three independent experiments (*p < 0.05; **p < 0.01). RFU, relative fluorescence units

Fig. 3

Doxycycline-induced EDI3 knockdown in HCC1954 results in reduced formation of lung metastasis in mice. A Schematic illustration of the experimental plan. Luciferase-expressing HCC1954 shNEG and shEDI3 cells were induced with doxycycline for 72 h, followed by tail vein injection into doxycycline pre-treated and untreated CD1 nude mice, respectively. Doxycycline was administered to the mice by a 625 mg/kg doxycycline containing diet (Ssniff) ad libitum. Luminescence signal was measured over six weeks. B Western blot confirming EDI3 knockdown in the doxycycline-treated shEDI3 cells at time of injection. C Representative luminescence images of mice (left panel) showing signal intensity and corresponding quantitative analysis of luminescence signal for HCC1954-luc shNEG (upper right panel) and shEDI3 (lower right panel) normalized to T0. Data represent results of five mice per condition. Box plots: horizontal line, median; box, 25th–75th percentiles; whiskers, min to max (*p < 0.05; **p < 0.01)

Fig. 4

Silencing EDI3 in HCC1954 reduces formation of peritoneal metastasis in mice and improves survival. A Schematic illustration of the experimental plan. Luciferase-expressing HCC1954 shNEG and shEDI3 cells were induced with doxycycline for 72 h. EDI3 mRNA expression and corresponding Western blot showing reduced EDI3 expression in the HCC1954-luc shEDI3 cells at time of injection. Doxycycline-treated (induced) and non-treated (non-induced) cells were injected into the peritoneum of doxycycline pre-treated and untreated CD1 nude mice, respectively. Doxycycline was administered to the mice by a 625 mg/kg doxycycline containing diet (Ssniff) ad libitum. Luminescence signal was measured over ten weeks and the time of survival was recorded. Quantitative analysis of luminescence signal for B shEDI3 and C shNEG normalized to T0. D Corresponding representative luminescence images of mice. Survival of mice was observed for 15 weeks after intraperitoneal injection with E shEDI3 and F shNEG cells. Data in B and C represent seven mice per condition. Data in E and F represent 12 (shEDI3) or seven (shNEG) mice per condition. Box plots: horizontal line, median; box, 25th–75th percentiles; whiskers, min to max. Kaplan–Meier curves: p values were determined by log-rank test (*p < 0.05; **p < 0.01)

Fig. 5

EDI3 knockdown in HCC1954-luc cells reduces metastatic burden. A Ex vivo luminescence imaging of organs and 100 μl ascites fluid (if present) with corresponding quantitative analysis B six and C eight weeks after intraperitoneal injection of doxycycline-induced and non-induced HCC1954-luc shEDI3 cells. Quantitative analysis of luminescence signal analyzed per organ after D six and E eight weeks. F Tables representing the ratio of mice positive for tumors in a specific organ compared to all mice six (left panel) and eight (right panel) weeks after injection. Data represent results of seven mice per condition. Box plots: horizontal line, median; box, 25th–75th percentiles; whiskers, min to max (*p < 0.05; **p < 0.01). l, liver; d, diaphragm; k, kidney; s/s/p, complex of spleen, stomach, and pancreas; a, ascites fluid; g, gonadal white adipose tissue (WAT)

Fig. 6

MALDI-MSI reveals altered GPC and PCho levels in tumors with silenced EDI3 expression. Representative images of HER2 staining and MALDI mass spectrometry imaging of tumors dissected from pancreas of mice eight weeks after intraperitoneal injection of non-induced (-Dox) and doxycycline-induced (+ Dox) HCC1954-luc shEDI3 with corresponding quantification of glycerophosphocholine (GPC), phosphocholine (PCho) and choline levels within the defined tumor regions (black lines), with the GPC/PCho ratio also provided. Scale bars represent 1 mm and 200 µm (zoomed). Values represent mean ± SD for six tumors per condition (*p < 0.05)

Fig. 7

EDI3 knockdown in HCC1954-luc cells reduces ascites formation. A Representative photographs of mice with (left panel) and without (right panel) ascites. Swelling caused by ascites is indicated using arrows. Scale bar represents 1 cm. B Table representing the ratio of mice positive for ascites compared to all mice. C Ascites volume measured in mice. D Quantitative analysis of luminescence signal detected in 100 μl ascites fluid by ex vivo imaging. Data represent results of seven mice per condition. ‘Non-induced’ are mice injected intraperitoneally with HCC1954-luc shEDI3 cells not treated with doxycycline; ‘induced’ are mice injected with doxycycline-treated HCC1954-luc shEDI3 cells and fed a 625 mg/kg doxycycline containing diet (Ssniff) ad libitum for the duration of the experiment. Box plots: horizontal line, median; box, 25th–75th percentiles; whiskers, min to max (*p < 0.05)