Human malignant melanoma (WM239) and colorectal carcinoma (SW480) cell lines were obtained from the American Type Culture Collection (Manassas, VA, USA). Cells were cultured in DMEM (Sigma-Aldrich, Oakville, ON, Canada) containing 10% fetal bovine serum (FBS) (Invitrogen, Burlington, ON, Canada), 1% sodium pyruvate (Sigma-Aldrich), and 0.5% gentamicin (Invitrogen) and maintained in a humidified atmosphere at 37°C and 5% CO₂.

All animal studies were performed according to regulations of the Canadian Council on Animal Care as supervised by the local Animal Care Committee of the University of Guelph. Xenografts were established in male and female RAG1^-/- mice by injecting 2 × 10⁶ WM239 or SW480 cells in 100 μL of 0.5% BSA/PBS subcutaneously into the right flank. Tumor size was monitored through caliper measurements taken every third day, and volume calculated using the formula (length × width²)/2. Tumors were allowed to grow to approximately 200 mm³, and then mice were randomized into treatment and control groups. Treatment consisted of 20 mg/kg/day of cyclophosphamide monohydrate (Sigma-Aldrich) in their drinking water[16]. Water for all mice was changed twice weekly for a period of up to 18 days, mice were euthanized by CO₂ asphyxiation and cervical dislocation, and tissue samples were collected for analysis.

Formalin-fixed paraffin embedded sections of treated and control SW480 and WM239 xenografts, and tissue microarrays of human colorectal carcinoma (Tissue Array Research Program, Centre for Cancer Research, NCI, Frederick, MD, USA), and melanoma (Imgenex, San Diego, CA, USA) were de-paraffinized and subjected to a 10 mM sodium citrate buffer, pH 6.0 for antigen retrieval. Sections were then incubated in DAKO Protein Block (DAKO, Mississauga, ON, Canada) for 1 hour. Antigen detection was performed sequentially by first using goat polyclonal anti-CD31 primary antibody (1:100; Santa Cruz Biotechnology, Santa Cruz, CA, USA) then donkey anti-goat FITC-conjugated secondary antibody (1:200; Santa Cruz Biotechnology) for 30 minutes. Sections were washed and then incubated with rabbit anti-VEGFR2 (1:100; Cell Signaling, Boston, MA, USA) overnight at 4°C, followed by goat anti-rabbit Cy3 (1:200; Jackson ImmunoResearch, West Grove, PA, USA) for 30 minutes. Antigen detection was performed sequentially (as described above). For mural cell quantification, antigen retrieval was in Tris-EDTA buffer solution (pH 9.0), followed by DAKO Protein Block and 5% donkey serum each for 30 minutes. Sections were incubated with goat polyclonal anti-CD31 (1:100; Santa Cruz Biotechnology) for 1 hour then donkey anti-goat FITC-conjugated (1:200; Santa Cruz Biotechnology) for 30 minutes, followed by anti-smooth muscle actin, directly conjugated to Cy3 (1:400; Sigma-Aldrich) for 30 minutes.

Five random fields of view at 200X magnification for each xenograft specimen were captured in a blinded fashion using QCapture software calibrated to a Leica DMLB microscope with an attached Q imaging QICAM fast1394 digital camera. One image/spot at 200X magnification was captured from each specimen in human tissue microarrays. Images were overlaid using Adobe Photoshop 7.0 (Adobe, Toronto, ON, Canada). VEGFR2 positive and negative, anti-smooth muscle actin positive and negative, and total blood vessels per field were quantified from overlaid images as previously reported[15]. Briefly, positive blood vessels were characterized by staining for both VEGFR2 or anti-smooth muscle actin antibodies, and the pan-endothelial cell marker, CD31. Vessel profiles with discontinuous CD31 staining were counted as separate vessels. Branched vessels were counted as a single vessel as long as staining for CD31 and/or VEGFR2 was continuous for the extent of the visible blood vessel profile. Microvessel density was calculated by dividing the total number of blood vessels per field area to determine the number of blood vessels per mm².

Formalin-fixed paraffin embedded sections were de-paraffinized, rehydrated and incubated in 3% H₂O₂ (Fisher, Ottawa, ON, Canada) for 15 minutes, followed by antigen retrieval with acidic citrate buffer as described above. Sections were incubated in DAKO Protein Block for 20 minutes followed by 5% normal goat serum (Vector, Burlington, ON, Canada) for 45 minutes and rabbit anti-carbonic anhydrase-IX (1:500; Abcam, Cambridge, MA, USA) overnight at 4°C. Slides were washed, incubated in biotinylated goat anti-rabbit (1:500; Vector) for 30 minutes then treated with RTU Vectastain Elite, ABC reagent (Vector) for 30 minutes followed by AEC chromogen (Vector) for 5 minutes. Sections were counterstained with Mayer's hematoxylin (Sigma-Aldrich). Relative hypoxic tissue (as CA-IX positive regions) to total section surface area was calculated using the software ImageScope (Aperio, Vista, CA, USA). Sections were evaluated in a blinded fashion.

Tissues were lysed with Cell Lysis Buffer (Cell Signaling Technology) with added aprotinin, PMSF and Phosphatase Inhibitor Cocktail II (all from Sigma-Aldrich). SDS-PAGE was performed with 90 micrograms of total protein loaded into 7.5% polyacrylamide gels and proteins were transferred to PVDF membranes (Roche) and blocked in 5% milk/TBST. Membranes were probed with the following primary antibodies overnight at 4ºC: rabbit anti-VEGFR2 (1:1000; Cell Signaling Technology), mouse anti-TSP-1 (1:500; Lab Vision), mouse anti-tubulin (1:200,000; Sigma-Aldrich), mouse anti-HIF1α (1:400, R&D Systems, Minneapolis, MN), rabbit anti-desmin (1:1500; Abcam), goat anti-CD31 (1:800; Santa Cruz Biotechnology), followed by secondary antibodies goat anti-rabbit POD (1:5000 - 1:20,000), goat anti-mouse POD (1:20,000) or rabbit anti-goat (1:20,000, all from Sigma-Aldrich) for 30 minutes at room temperature. Protein bands of interest were detected with Chemiluminescence Blotting Substrate (Roche), and densitometry performed by normalization to tubulin signal. For desmin blots, signal was normalized to CD31 for each sample.

Briefly, 1 × 10⁴ cells were plated into each well in a 96 well plate in complete medium for 48 hrs then serum starved for 24 hrs. Plates were then exposed to 0-100,000 ng/ml 4-HC (4-hydroperoxycyclophosphamide; HCNiomech/IIT GmbH, Bielefeld, DE; the precursor of 4-hydroxycyclophosphamide, the metabolically active metabolite of cyclophosphamide) and incubated for 72 hours. MTT cytotoxicity assay was performed with a Cell Growth Determination Kit using the manufacturer's protocol (Sigma-Aldrich, Oakville, ON) and absorbance was read at 570 nm with an EL_X800 Universal Microplate Reader (BIO-TEK Instruments Inc, Winooski, VT). Three independent replicates were preformed with each dose done in triplicate.

WM239 and SW480 cells were seeded into 6-well plates at a density of 5.0 × 10⁵ cells. The following day, media was changed to DMEM and 2% FBS. 24 hours later, media was replaced to DMEM and 2% FBS ± CoCl₂ (100 μM), ± 4-HC (100 nM) or a combination of both. 24 hours after treatment, condition media was collected and cell free supernatant was used to measure VEGF-A levels with ELISA using a commercially available ELISA kit (R & D Systems). Cells were counted to standardize VEGF levels. Two independent experiments were performed, and each sample analyzed as duplicates. Results were presented as pg VEGF per 10,000 cells.

Apoptosis was detected in paraffin embedded tissues with the In Situ Cell Death Detection Kit, POD (Roche), according to manufacturer's instructions. TUNEL-positive nuclei were identified with DAB substrate chromogen system (DAKO) and counterstained with Mayer's hematoxylin (Sigma-Aldrich). Quantification was performed in a blinded fashion by scoring the number of positive nuclei/10 high power fields of non-necrotic tissue. Five sections per group were analyzed and averaged.

Comparison of means was performed using either the Student's t-test, or ANOVA followed by post-hoc analysis using the Bonferroni and Tukey's LSD methods. Differences were considered significant when p ≤ 0.05. Data are expressed as the mean and standard error of the mean (SEM).

Dual immunofluorescence staining for VEGFR2 and CD31 was performed on formalin-fixed paraffin embedded tissue microarrays to determine the heterogeneous nature of VEGFR2 expression within the vasculature of clinically relevant tumors. Colorectal carcinoma (n = 36; on average approximately 20 blood vessels/tissue core), and metastatic malignant melanoma (n = 14; approximately 15 blood vessels/tissue core) were evaluated. Heterogeneous expression of VEGFR2 was observed in both tumor types (Figure 1; Table 1). Colorectal carcinoma had a significantly lower average percent VEGFR2 positive blood vessels (60% VEGFR2 positive blood vessels) than malignant melanoma (93% VEGFR2 positive blood vessels) (p < 0.05).

The VEGFR2 status of vessels in colorectal carcinoma and melanoma was compared between human clinical samples and mice xenografted with human cancer cells of the same type, and there were no significant differences in vascular phenotype between xenografts and clinical cases within a cancer type (Table 1) (p < 0.05). We consistently found that colorectal carcinoma xenograft tumors had significantly fewer VEGFR2 positive blood vessels than malignant melanoma xenografts.

To study the influence of this VEGFR2 heterogeneity on response to low dose metronomic cyclophosphamide therapy (CTX), we evaluated xenografts from treated mice. No significant differences in tumor volume between treated and control groups were observed for either SW480 or WM239 xenografts (Figure 2A). Cancer cell toxicity to CTX was evaluated in vitro (Figure 2C), and the approximate LC₅₀ of 3000 ng/ml is well above the expected tissue levels of CTX under this metronomic schedule[16]. Consistent with this we found no significant difference in the incidence of cancer cell apoptosis as detected by TUNEL positive nuclei (Figure 2D). Sections of tumors immunostained for CD31 were then scored for microvessel density (MVD). CTX therapy induced a significant reduction in MVD in both SW480 and WM239 xenografts (p < 0.05; Figure 3A). CTX treatment also induced a significant increase in the levels of endogenous anti-angiogenic protein thrombospondin 1 (TSP-1) in SW480 xenografts, but a significant decrease in TSP-1 levels was seen in WM239 tumors treated with CTX compared to control (p < 0.05; Figure 3C).

The effect of low dose metronomic cyclophosphamide therapy on VEGFR2 expression was also characterized (Figure 4A-D). There was a significant decrease in mean density of VEGFR2 positive blood vessels between control and treated tumors for both SW480 and WM239 xenografts (p < 0.05; Figure 4E). Western blotting of tumor proteins demonstrated a significant reduction in total VEGFR2 between SW480 control and CTX tumors (p < 0.05), but no significant changes in total VEGFR2 in WM239 (p < 0.05; Figure 4G).

We expected to see that successful angiogenic blockade of SW480 and WM239 xenografts would result in increased tumor hypoxia. However, we found no significant differences in HIF1-α levels between control and CTX for either tumor type (Figure 5A), although the average hypoxic area as demonstrated by CA-IX immunostaining was significantly reduced in WM239 CTX compared to control (p < 0.05; Figure 5C-G). While in vitro induction of HIF1α in WM239 and SW480 cells (by CoCl₂ exposure) lead to expected increases in VEGF-A production, no significant difference were observed between cells co-incubated with or without CTX (Figure 5H) hence difference in response to CTX between cell lines is not due to differential regulation of VEGF-A expression.

Our finding of reduced MVD but no increase in tumor hypoxia suggests that tumor vessel functionality (i.e. perfusion) may have been improved in CTX treated tumors. We found significant increases in the proportion of α-SMA positive (i.e. "mature") vessels in SW480 tumors (p < 0.05; Figure 6A). There was also evidence that the relative expression of the mural cell protein desmin was increased in SW480 tumors treated with CTX compared to control but not in WM239 xenografts (Figure 6).