Dynamic modulation of phosphoprotein expression in ovarian cancer xenograft models

The platinum sensitive OV1002 ovarian cancer xenograft model was derived from a high grade serous adenocarcinoma, while the HOX424 xenograft model was of clear cell/endometrioid origin and had reduced responsiveness to platinum. Both cell lines were established at the Institute of Genetics and Molecular Medicine, University of Edinburgh as described previously [31]. Ovarian tumor fragments were implanted subcutaneously into adult (12 week old) female CD-1 nu/nu mice and were allowed to grow to 4–6 mm in diameter in negative pressure isolators (La Calhene, Cambridge, UK). Mice were then stratified into treatment groups such that mean tumor volumes were equivalent across the groups. Tumor size was monitored twice weekly and relative tumor volumes (RTV) were calculated for each individual tumor by dividing the tumor volume on day t (Vt) by the tumor volume on day 0 (V0). Mice were grouped into three groups: untreated controls, carboplatin treated (50 mg/kg) and carboplatin (50 mg/kg) + paclitaxel treated (10 mg/kg). Treatment was administered as a single intraperitoneal dose on day 0. Groups of both treated and untreated control mice were sacrificed on Day 1, 2, 4, 7 and 14 after treatment. Tumors were placed into formalin, fixed and paraffin embedded. Tissue sections were cut and stained with a range of antibodies as described below against 49 protein targets (Table 1). The xenograft dataset consisted of 99 OV1002 and 67 HOX424 xenograft samples. Phenotypic and raw data are shown in Additional file 1. There were 3–8 biological replicates at each time point and treatment, except at day 2 for HOX424, where there were 2 and 1 replicates for each treatment. Agreement between sample expression levels was good as measured by the correlation coefficients (r) among replicates of each condition (mean r 0.94, 95 % confidence interval 0.934–0.955, Additional file 2, column E). The xenograft studies were undertaken under a UK Home Office Project Licence in accordance with the Animals (Scientific Procedures) Act 1986 and studies were approved by the University of Edinburgh Animal Ethics Committee.

Protein expression was measured by immunofluorescence using methods previously described [25]. Tissue microarrays were prepared from xenograft material that had been formalin-fixed at the time of collection and had been processed into paraffin blocks. Microarray slides were deparaffinized and antigen-retrieved by pressure-cooking. Endogenous peroxidases were blocked with 2.5 % hydrogen peroxide for 15 min and non-specific binding blocked with serum-free protein block for 15 min. Slides were then incubated with primary antibodies (Table 2) with AE1/AE3 mouse monoclonal cytokeratin antibody at room temperature for 1 h. After washing, sections were incubated for 1 h at room temperature with secondary antibodies, which included an Alexa 555-conjugated goat anti-mouse antibody diluted 1:100, and prediluted goat anti-rabbit antibody conjugated to a horseradish peroxidase decorated dextran-polymer backbone (EnVision, Dako). Slides were then incubated for 10 min with Cy5-tyramide, which is activated by horseradish peroxidase, to visualise target protein expression. 4', 6-Diamidino-2-phenylindole (DAPI; Molecular Probes, Eugene, Ore) was used to stain the nuclear compartment. For analysis, pan-cytokeratin antibody was used to identify tumor cells and normal epithelial cells, DAPI counterstain to identify nuclei, and Cy-5-tyramide detection for target protein in compartmentalised (tissue and subcellular) analysis of tissue sections. Monochromatic images of each 0.6 mm TMA core were captured at 20× objective using an Olympus AX-51 epifluorescence microscope, and high-resolution digital images analysed by the AQUAnalysis software. AQUA scores, which represent the sum of the Cy5-tyramide score for the target protein divided by the area of the cellular compartment (cytoplasm or nucleus), were then generated for each sample. A detailed description of the AQUA analysis has been reported elsewhere [32].

AQUA scores were log-transformed averaged from replicate cores. Means of differentially expressed proteins log fold change values for each time point were hierarchically clustered using Cluster 3.0 in order to identify significant proteins with similar temporal expression profiles. Heatmaps were visualized using TreeView [33]. Bioconductor package limma [34] was used for differential expression calculations. Significant genes had FDR adjusted p-values below 0.05. Treated samples were contrasted to pooled untreated control from all time points in each xenograft. Differentially expressed proteins were classified as expressed early (Days 1–4) or late (Days 7, 14), and transient (expressed significantly in 1 time point) or continuous (expressed significantly at least 2 continuous time points).