Histology and immunohistochemistry

Paraffin sections (5 μm thickness) were placed on Superfrost+ slides for histology and immunohistochemistry (IHC). Slides were deparaffinized in xylene and subsequently rehydrated in a graded isopropanol series. For immunohistochemistry, antigen retrieval was performed in sodium citrate buffer (pH 6.0) in a 95°C water bath for 15 min. Endogenous peroxidase activity was quenched with 3% hydrogen peroxide for 30 min and non‐specific binding sites were blocked with a 1:1 ratio of 5% normal goat serum (Life Technologies, Burlington, ON, Canada) and DAKO block (Dako, Carpinteria, CA, USA) for 2 h at room temperature followed by overnight incubation at 4°C with the anti‐VEGFR2 (1:1600); anti‐phospho‐VEGFR2 (1:125); anti‐NRP‐1 (1:11 600) or anti‐c‐CBL (1:200), followed by washing and incubation in biotinylated anti‐rabbit IgG (1:200; Vector Laboratories, Burlington, ON, Canada) for 1 h, at room temperature. Peroxidase‐streptavidin complex (1:200; Cedarlane, Burlington, ON, Canada) was then applied for 1 h at room temperature, and peroxidase activity visualized with 3,3′‐diaminobenzidine tetrahydrochloride (DAB; Vector Laboratories). Negative controls consisted of slides incubated with no primary antibody and slides incubated with an irrelevant primary antibody (rabbit polyclonal anti‐pan‐cytokeratin; Fig. S1). Positive controls consisted of vascular endothelial cells for VEGFR2, pVEGFR2 and NRP‐1 (internal control) and canine testis for c‐CBL (Fig. S1). Slides were counterstained with Harris haematoxylin, dehydrated and coverslipped. Immunohistochemical labelling for VEGF‐A was conducted by the Michigan State University Diagnostic Center for Animal and Population Health (Lansing, MI, USA). Additional slides were stained with haematoxylin and eosin, and used to assess mitotic count, numbers of multinucleate cells and tumour growth pattern as previously reported (Thompson et al. 2011b).