Immunohistochemical analysis

The growth of s.c. tumors in mice is known to be angiogenesis-dependent (413). Thus, in the next step of the experiment we tested, if the differences in tumor growth are in line with the effects of drugs on vascular parameters and the expression of target angiogenic receptors of the RTKIs. Therefore, we stained tissue sections for these parameters.

The expression level of antiangiogenic RTKs may significantly influence treatment response, and in turn, successful therapy can also regulate target receptor localization and function. In the C26 model expression of PDGFRα, -β and FGFR1 was observed not only on mural cells, but also on tumor cells. Expression patterns of PDGFRα, -β and FGFR-1 did not change in response to treatment with any of the compounds (p=0.8265 0.1261 and 0.2983, respectively) in the C26 model (Figure 20-22.).

The receptor distribution of the C38 model showed a different pattern compared to the C26 tumors. C38 tumor cells did not express the aforementioned receptors, but a definite cell population expressing PDGFRα, -β and FGFR1 was detectable on the mural cells. Similarly to the C26 model, no change in the expression of PDGFRα, -β and FGFR1 (p=0.7601, 0.7497

and 0.7178, respectively) was seen (Figure 20-22.).

62

Figure 20. Low power views of C26 and C38 tumor sections stained for PDGFRα (green). Microvessels are labeled with anti-CD31 (red). Nuclei are stained with Hoechst 33342 (blue).

63

Figure 21. Low power views of C26 and C38 tumor sections stained for PDGFRβ (green). Microvessels are labeled with anti-CD31 (red). Nuclei are stained with Hoechst 33342 (blue).

64

Figure 22. Low power views of C26 and C38 tumor sections stained for FGFR1 (green). Microvessels are labeled with anti-CD31 (red). Nuclei are stained with Hoechst 33342 (blue).

65

VEGFR2 expression was detected both on tumor and endothelial cells in the C26 model. Significant differences were shown both when counting the VEGFR2 signal and when measuring the area of VEGFR2+ cells in C26 tumors (p=0.0296 and 0.022 respectively; Figure 23, 25). Post-hoc test showed that VEGFR2 expression was altered only in the sunitinib treated group.

Figure 23. Low power views of C26 tumors stained for VEGFR2 (green). Microvessels are labeled with anti-CD31 (red). Nuclei are stained with Hoechst 33342 (blue).

66

C38 tumors were characterized by a weak VEGFR2 expression of the CD31+

endothelial layer, but no signal of the receptor on tumor cells was observed. No difference in the expression of VEGFR2 in the treated vs. non treated groups (p=0.6857 and 0.4857 for VEGFR2 density and area respectively) was detected (Figure 24-25.).

Figure 24. (A.) Low power views of C38 tumors stained for VEGFR2 (green). Microvessels are labeled with anti-CD31 (red). Nuclei are stained with Hoechst 33342 (blue). (B.) The same images without counterstain with anti-CD31 and Hoechst 33342.

67

Figure 25. VEGFR2 (A.) densities and (B.) areas of C26 and C38 tumors. Data are shown as box (first and third quartiles) and whisker (maximum to minimum) plots with the mean (horizontal bar) from 6 animals per group. VEGFR2 densities were counted in ten viable intratumoral regions. VEGFR2 areas were calculated by counting the number of VEGFR2-positive pixels in ten viable intratumoral regions.

p=0.0296 and 0.022for VEGFR2 density and VEGFR2 area respectively in the C26 model as shown by the Kruskal-Wallis test, p=0.6857 and 0.4857for VEGFR2 density and VEGFR2 area respectively in the C38 model as shown by the Mann Whitney U test.

68

We found a significant difference between the MVD of the C26 groups, p<0.0001. The post-hoc test showed a suppressed MVD by sunitinib, motesanib and minimally vatalanib in that model. Microvessel area was also decreased in the sunitinib, motesanib and less intensively in the sorafenib treated group, p<0.0001 (Figure 26-27.).

In the C38 model however, no difference in the vessel density (p=0.235), but in vessel area was detected, p=0.0341 (Figure 26-27.). It is also important to mention, that major differences in the vasculature of the two groups were observed. While C26 tumors had lots of small vessels, C38 tumors were characterized by only a few, but large and complex vascular structures.

69

Figure 26. Microvessel density (A.) and microvessel area (B.) data of C26 and C38 tumors. Data are shown as box (first and third quartiles) and whisker (maximum to minimum) plots with the mean (horizontal bar) from 6 animals per group. Microvessel densities were counted in ten viable intratumoral regions. Microvessel areas were calculated by counting the number of CD31-positive pixels in ten viable intratumoral regions. p<0.0001 both for MVD and vessel area in the C26 model as shown by the one-way Anova test, p=0.235 and 0.0341 for vessel density and area respectively as shown by unpaired t-test in the C38 model.

70

Figure 27. Microvessel density and area of C26 and C38 tumors. Tumors were labeled for the endothelial cell marker CD31 (red) and nuclei are stained with Hoechst 33342 (blue).

71

MVD clearly correlated with tumor oxygenation. Hypoxia was located in the less vascularized areas of the tumor. Accordingly, a significant increase (p=0.0152) in the intratumoral hypoxic areas was observed in the sunitinib treated group (Figure 28-29.).

The ratio of hypoxic areas did not differ (p=0.9143) in the C38 model (Figure 28-29.).

Figure 28. Hypoxic areas of C26 and C38 tumors. Data are shown as box (first and third quartiles) and whisker (maximum to minimum) plots with the mean (horizontal bar) from 6 animals per group. Hypoxic areas are shown in the percentage of the total tumor sections. p=0.0152 as shown by the Kruskal-Wallis test in the C26 model and 0.9143 as shown by the Mann-Whitney U test in the C38 model.

72

Figure 29. Representative images of hypoxic areas in the C26 and C38 tumors. Green: anti-pimonidazole staining for hypoxia; blue: nuclear staining with Hoechst 33342.

73

Beside the inhibition of endothelial cell proliferation, multi-target antiangiogenic RTKIs also influence PDGFR and FGFR positive pericyte and VSMC recruitment to tumor blood vessels. Therefore, the inhibition of these receptors may result in not only decreased MVD and consequently lower blood flow rate of tumors, but could also facilitate cancer cell metastatization. To observe the structural changes of the vasculature in response to treatment, we examined the expression of laminin, desmin and αSMA of tumor sections. While all vessels remained underlaid with a definite layer of laminin and covered with αSMA, desmin expression has decreased in response to sunitinib and motesanib treatment in the C26 model; p=0.0135 (Figure 30-32.).

Both laminin and αSMA expression were definite and did not change in response to treatment in the C38 model (Figure 30-31.), but unlike in C26, no difference in desmin expression was observed either p=0.9143 (Figure 32-33.).

74

Figure 30. Low power views of C26 and C38 tumor sections stained for the capillary basement membrane component laminin (green) and CD31 (red). Nuclei are stained with Hoechst 33342 (blue).

Note that tumor cells are also weakly positive for laminin.

75

Figure 31. Low power views of C26 and C38 tumor sections stained for αSMA (green) and CD31 (red).

Nuclei are stained with Hoechst 33342 (blue).

76

Figure 32. Desmin expression in C26 and C38 tumors. Tumor sections are immunolabeled for pericyte desmin (green) and CD31 (red). Nuclei are counterstained with Hoechst 33342 (blue).

77

Figure 33. Desmin expression of C26 and C38 tumors. Data are shown as box (first and third quartiles) and whisker (maximum to minimum) plots with the mean (horizontal bar) from 6 animals per group.

Desmin expression is expressed in the % of the microvessels covered by desmin expressing pericytes.

p=0.0135 for the C26 model and 0.9143 for the C38 model as shown by the one-way Anova and the Mann-Whitney U tests respectively.

6.3 Mass spectrometric analysis