Cell line and patient-derived xenograft models reveal elevated CDCP1 as a target in high-grade serous ovarian cancer

Abstract

Background: Development of targeted therapies for high-grade serous ovarian cancer (HGSC) remains challenging, as contributing molecular pathways are poorly defined or expressed heterogeneously. CUB-domain containing protein 1 (CDCP1) is a cell-surface protein elevated in lung, colorectal, pancreas, renal and clear cell ovarian cancer.

Methods: CUB-domain containing protein 1 was examined by immunohistochemistry in HGSC and fallopian tube. The impact of targeting CDCP1 on cell growth and migration in vitro, and intraperitoneal xenograft growth in mice was examined. Three patient-derived xenograft (PDX) mouse models were developed and characterised for CDCP1 expression. The effect of a monoclonal anti-CDCP1 antibody on PDX growth was examined. Src activation was assessed by western blot analysis.

Results: Elevated CDCP1 was observed in 77% of HGSC cases. Silencing of CDCP1 reduced migration and non-adherent cell growth in vitro and tumour burden in vivo. Expression of CDCP1 in patient samples was maintained in PDX models. Antibody blockade of CDCP1 significantly reduced growth of an HGSC PDX. The CDCP1-mediated activation of Src was observed in cultured cells and mouse xenografts.

Conclusions: CUB-domain containing protein 1 is over-expressed by the majority of HGSCs. In vitro and mouse model data indicate that CDCP1 has a role in HGSC and that it can be targeted to inhibit progression of this cancer.

Figure 1

Immunohistochemical analysis of CDCP1 in HGSC, and normal, benign and malignant fallopian tube. (A) Negative staining of a primary HGSC tumour. (B) Weak staining of an HGSC appendiceal metastasis. (C) Moderate staining of a primary HGSC tumour. (D) Strong staining of an HGSC lymph-node metastasis. (E) Negative staining in normal fallopian tube. (F) Weak to moderate CDCP1 expression in the benign fallopian tube epithelium of a patient with HGSC in the ovary. (G) Moderate to strong CDCP1 expression in a fallopian tube showing HGSC involvement. Staining was apparent in the tumour (T) and absent in the adjacent non-malignant fallopian tube (N). Magnifications: (A–D) are × 40 with insets × 10; (E–G) are × 40 with insets × 20. Scale bar is 50 μm.

Figure 2

SKOV3 cells display HGSC features when xenografted intraperitoneally in mice and grown in vitro. (A) H&E staining showing histology representative of a SKOV3 cell xenograft. (B) Anti-p53 immunohistochemical analysis of a SKOV3 cell xenograft showing the absence of staining within tumour cells (left). Western blot analysis of OVCAR3 and SKOV3 cell lysates for p53 (right). OVCAR3 cells carry missense-mutated TP53 that encodes p53-R248W. (C) Representative images of immunohistochemical staining of a SKOV3 cell xenograft for WT1, CA125, cytokeratin 7 and cytokeratin 20. Magnification, × 40. Scale bar is 50 μm.

Figure 3

Silencing of CDCP1 reduces migration and non-adherent growth but not proliferation of ovarian cancer cell lines in vitro. (A) Anti-CDCP1 and -GAPDH western blot analysis of OV90, HEY and SKOV3 lysates of cells stably transduced with lentivirus carrying the CDCP1 targeting sequence V1 or scramble control shRNA. Densitometric analysis of the CDCP1 signal, determined from three independent western blot analyses, is shown at the bottom of the panel. (B) Adherent cell growth. Cells (2000 per well) were seeded and at the indicated time points absorbance was read at 490 nm. Absorbance was measured each day for 4 days. (C) Comparison of migration of OV90, HEY and SKOV3 cells stably transduced with a scramble control or a CDCP1 shRNA. Cells (50000 per well) seeded in serum-free media migrated towards a 10% FCS gradient. (D) Non-adherent cell growth. Cells (8000 per well) were seeded in serum-free media in wells of an ultra-low attachment plate and after 72 h cell viability was assessed. ***P<0.001. Data points represent mean and standard error of the mean from three independent experiments, each with triplicate wells for each time point.

Figure 4

Silencing of CDCP1 reduces intraperitoneal tumour formation of SKOV3 cells in mice. Female NSG mice were injected with SKOV3-shScramble (n=5) or SKOV3-shCDCP1 (n=5) cells (5 × 10⁶). (A) Left, Bioluminescent images of mice after 5 weeks of tumour growth. Right, Graph of the bioluminescent signal (total flux; photons per seconds) obtained from each mouse. (B) Left, Representative images of the peritoneal cavity of mice at the time of killing at 5 weeks after injection of SKOV3-shScramble or SKOV3-shCDCP1 cells. Arrows indicate tumour nodules. Right, Graph of the number of peritoneal tumour nodules present in mice injected with either SKOV3-shScramble or SKOV3-shCDCP1 cells. Data represent mean and standard deviation of each group. (C) Western blot analysis of lysates from three randomly selected SKOV3-shScramble and SKOV3-shCDCP1 xenograft tumours recovered from mice for CDCP1, p-Src-Y416 (pSrc), Src and GAPDH.

Figure 5

Targeting CDCP1 with monoclonal antibody 10D7 reduces in vitro migration and non-adherent, but not adherent growth, of ovarian cancer cell lines. (A) Adherent cell growth. Cells (2000 per well) were seeded in media containing antibody 10D7 or control IgG (50 μg ml⁻¹) and at the indicated times absorbance was read at 490 nm. Absorbance was measured each day for 4 days. (B) Non-adherent cell growth. Cells (8000 per well) were seeded in serum-free media containing 10D7 or control IgG (50 μg ml⁻¹) in wells of an ultra-low attachment plate and after 72 h cell viability was assessed. *P<0.05; **P<0.01; ***P<0.001. (C) Comparison of migration of OV90, HEY and SKOV3 cells (50 000 per well) seeded in serum-free media containing antibody 10D7 or control IgG (50 μg ml⁻¹) migrated towards 10% FCS containing 10D7 or control IgG (50 μg ml⁻¹). Data points represent mean and standard error of the mean from three independent experiments, each with triplicate wells for each time point.

Figure 6

Monoclonal antibody targeting of CDCP1 impedes HGSC PDX growth. (A) H&E and anti-CDCP1 (brown) staining of three patient tumours and PDXs developed from these tumours. Magnification, × 40. CDCP1 is expressed by malignant cells, detected mainly in the membrane with cytosolic staining also apparent. Scale bar is 50 μm. (B–D) Mice were injected with cell slurry from the PDX of patient 28 along with antibody 10D7 or isotype matched control IgG (100 μg). 10D7 and IgG treatments continued weekly at 25 mg kg⁻¹ per mouse (n=4 mice per treatment). (B) Representative images, at the time of killing of mice at week 7, of the peritoneal cavity of mice treated with IgG (left) or 10D7 (right). (C) Graphical representation of the total number of tumours (left), the weight of the largest tumour (middle) and the weight of all tumours combined (right) of groups of mice treated with IgG or 10D7. Data represent mean and standard deviation of each group. (D) Western blot analysis of lysates from four randomly selected IgG- and 10D7-treated PDX tumours recovered from mice. Lysates were examined for CDCP1, p-Src-Y416 (pSrc), Src and GAPDH.