Fast growth associated with aberrant vasculature and hypoxia in fibroblast growth factor 8b (FGF8b) over-expressing PC-3 prostate tumour xenografts

Abstract

Background: Prostate tumours are commonly poorly oxygenated which is associated with tumour progression and development of resistance to chemotherapeutic drugs and radiotherapy. Fibroblast growth factor 8b (FGF8b) is a mitogenic and angiogenic factor, which is expressed at an increased level in human prostate tumours and is associated with a poor prognosis. We studied the effect of FGF8b on tumour oxygenation and growth parameters in xenografts in comparison with vascular endothelial growth factor (VEGF)-expressing xenografts, representing another fast growing and angiogenic tumour model.

Methods: Subcutaneous tumours of PC-3 cells transfected with FGF8b, VEGF or empty (mock) vectors were produced and studied for vascularity, cell proliferation, glucose metabolism and oxygenation. Tumours were evaluated by immunohistochemistry (IHC), flow cytometry, use of radiolabelled markers of energy metabolism ([18F]FDG) and hypoxia ([18F]EF5), and intratumoral polarographic measurements of pO2.

Results: Both FGF8b and VEGF tumours grew rapidly in nude mice and showed highly vascularised morphology. Perfusion studies, pO2 measurements, [18F]EF5 and [18F]FDG uptake as well as IHC staining for glucose transport protein (GLUT1) and hypoxia inducible factor (HIF) 1 showed that VEGF xenografts were well-perfused and oxygenised, as expected, whereas FGF8b tumours were as hypoxic as mock tumours. These results suggest that FGF8b-induced tumour capillaries are defective. Nevertheless, the growth rate of hypoxic FGF8b tumours was highly increased, as that of well-oxygenised VEGF tumours, when compared with hypoxic mock tumour controls.

Conclusion: FGF8b is able to induce fast growth in strongly hypoxic tumour microenvironment whereas VEGF-stimulated growth advantage is associated with improved perfusion and oxygenation of prostate tumour xenografts.

Figure 1

Expression of VEGF after transfection of PC-3 cells with VEGF or mock expression vectors. Western blot analysis of conditioned medium of PC-3/VEGF cell clones revealed several clones positive for 23 kD VEGF protein. The larger bands at approximately 60 kD may represent cleaved formes of larger splice variants in these cells. The expression level in mock cells was very low. One PC-3/VEGF clone (PC-3/VEGF3) and one PC-3/mock clone (PC-3/mock2) were selected for in vivo studies.

Figure 2

Growth and morphology of FGF8b, VEGF and mock tumours. A, Growth of subcutaneous FGF8b, VEGF and mock tumours (n = 30, n = 22 and n = 14, respectively). Tumour diameter in 2 perpendicular dimensions was measured once a week and tumour volume was calculated according to the formula V = (π/6)(d₁ × d₂)^3/2 and presented as a function of time (mean ± SEM). Differences in tumour volumes between the groups were significant at all time points between 2 and 4 weeks (p < 0.001). B, H&E staining of representative FGF8b, VEGF and mock tumours (Bar 200 μm). C-D, The density of CD31-positive blood capillaries (μm/mm²) was counted in a blinded manner from 3 fields of the FGF8b, VEGF and mock tumours (51 ± 27 μm/mm², n = 18, 97 ± 4 μm/mm², n = 72, and 36 ± 21 μm/mm², n = 49, respectively), p* < 0.05, p*** < 0.001 (Bar 200 μm). E, The relative area of necrosis was counted in a blinded manner from 3 fields in FGF8b (n = 6), VEGF (n = 6) and mock (n = 6) tumours, p* < 0.05, p** < 0.01.

Figure 3

The effect of ectopic FGF8b and VEGF on proliferation. Ki67 immunostaining of FGF8b (n = 40), VEGF (n = 5) and mock (n = 23) tumours (Bar 200 μm). The relative number of Ki67-positive cells in tumours was counted in a blinded manner in 3 fields per tumour and the results were expressed as percentage of positive cells per mm². There were significantly more proliferative cells (p < 0.001) in both FGF8b and VEGF tumours compared with mock tumours.

Figure 4

Perfusion and oxygenation status of tumours. A, Labelling with the perfusion marker Hoechst 33342 showed that the flow in blood vessels was better in the VEGF tumours compared with the FGF8b and mock tumours, where Hoechst 33342 labelling was detected only in the tumour periphery (Bar 200 μm). Examples of the intratumoral distribution of [¹⁸F]EF5 in FGF8b, VEGF and mock tumour sections are shown in B. FGF8b and mock tumours showed mainly peripherally located uptake of [¹⁸F]EF5, whereas uptake into VEGF tumours was more uniform throughout the tumours. Uptake intensity is not comparable between these images, since they were not corrected for injected dose or cross-calibrated between separate studies. C, The tumour-to-blood (T/B) uptake ratio of [¹⁸F]EF5 (T/B ratio) in FGF8b (n = 11), VEGF (n = 12) and mock (n = 29) tumours is expressed as mean ± SD. Accumulation of [¹⁸F]EF5 was significantly lower (p < 0.05) in VEGF tumours compared with mock tumours, whereas no significant difference in T/B ratio was seen between FGF8b and mock tumours. Partial pressures of oxygen (pO₂) in VEGF (n = 3), FGF8b (n = 3) and mock (n = 3) tumours are shown in D. The mean values of pO₂ measurements are shown as a curve. VEGF tumours were relatively well oxygenated in comparison with FGF8b and mock tumours.

Figure 5

Relationship between tumour hypoxia and proliferation. Localisation of areas positively stained for HIF1α and Ki67 was visually detected from microscope images in FGF8b and mock tumours. The arrows show positive HIF1α staining and negative Ki67 staining in A and B, and C and D. C and D are captions from the indicated areas in A and B, respectively (Bar 200 μm).

Figure 6

Glucose metabolism. A, Intratumoral distribution of [¹⁸F]FDG-derived radioactivity in tumour sections (FGF8b n = 6, VEGF n = 6 and mock n = 5). Uptake intensity is not comparable between these images, since they were not corrected for injected dose or cross-calibrated between separate studies. B, Tumour-to-blood (T/B) uptake ratio of [¹⁸F]FDG (T/B ratio) in FGF8b (n = 6), VEGF (n = 6) and mock (n = 5) tumours is expressed as mean ± SD. The accumulation of [¹⁸F]FDG was significantly lower (p < 0.05) in FGF8b tumours compared with VEGF and mock tumours. Number of cells/cm³ was determined using TrueCount tubes and flow cytometry. On the right, we show relative uptake of [¹⁸F]FDG after balancing the uptake against cell number. C, Immunostaining of the glucose transporter GLUT1 showed decreased staining in VEGF tumours compared with mock and FGF8b tumours (FGF8b n = 5, VEGF n = 4 and mock n = 7 tumours) (p < 0.05 VEGF vs. mock, Bar 200 μm). Analysis was carried out in a blinded manner in three representative non-overlapping fields of the tumours, and the data presented as mean ± SD.