Organdependent Expression Of Basic Fibroblast Growth Factor By Tumor Cells

This laboratory has shown that expression of basic FGF (bFGF) by tumor cells is dependent on the site of implantation. When HRCC (SN12) cells were implanted in different organ microenvironments in nude mice, the expression of bFGF was 10-20 times higher in those tumors implanted in the kidney than those implanted in the subcutaneous tissues (51). The kidney tumors were more highly vascularized than tumors implanted in the subcutis. In sharp contrast, the expression of IFN- p was high in epithelial cells and fibroblasts surrounding the subcutaneous tumors, but no IFN-P was found in or around HRCC tumors growing in the kidney. The parental cell line (SN12) and metastatic clone also differed in bFGF expression. The alteration in bFGF level by the site of implantation was caused by adaptation to the organ microenvironment, as was demonstrated when the cells were reestablished in culture, and the levels of bFGF returned to the previously in vitro concentration after 4 wk (51).

Expression of bFGF in HRCC (SN12) is cell-density-dependent. By in situ mRNA hybridization (ISH) and Northern blot analysis, an inverse correlation was found between increasing cell density and bFGF expression (62). Fluorescence-activated cell sorting (FACS), immunohistochemistry, and ELISA confirmed this finding at the protein level. Tumor cells harvested from dense cultures (low bFGF expression), and then plated under sparse conditions, expressed high levels of bFGF. S imilar data was obtained using endothelial cells. The effect was not mediated by soluble factors released into the culture medium.

The in vitro conditions of sparse vs confluent can be argued to be irrelevant under in vivo conditions, in which almost all cells are in contact with neighboring cells. The in vivo manifestation of cell-density-dependent regulation may be interpreted as the difference in gene expression in the center of a tumor vs the periphery or leading edge. This laboratory investigated the expression of multiple metastasis- and angiogenesis-related genes, including bFGF, in patients with various stages of HCC. By ISH, the level of bFGF was significantly higher in patients with Dukes' stage C (node-positive tumors without further evidence of metastasis) and Dukes' stage D (distant metastases), than in those with Dukes' stage B disease (node-negative tumors without evidence of metastasis) (63). Of particular interest was the fact that Northern blot hybridization did not detect mRNA transcripts for bFGF. Careful analysis by ISH revealed that bFGF was expressed at the highest levels in subpopulations of cells at the periphery of the tumor (invasive edge), which probably represents the most active portion of the invading tumor. A follow-up study utilized the same ISH technique to predict the disease recurrence in patients with colon cancer. The authors were able to identify patients who appeared to be free of metastasis at the time ofinitial surgery (Dukes' stage B), yet developed distant metastases at a later date. These patients had relatively high bFGF expression (along w ith increased expression of other metastasis-related genes) (64). These studies demonstrate that host cells surrounding the tumor may affect angiogenesis-related gene expression, possibly by direct cell-to-cell contact or by secretion of paracrine factors.

Recent clinical observations have noted antiangiogenic effects in vascular tumors, including hemangioma (65-70), Kaposi's sarcoma (71-74), melanoma (75), basal cell and squamous cell carcinoma (76), and bladder carcinoma (77), using recombinant interferons. These tumors have also been documented as producing high levels of bFGF, often detectable in the urine or serum of these patients (78,79). These findings, along with in vivo observations, prompted investigation of whether IFNs could modulate the expression of the angiogenic molecule, bFGF. The authors found that IFN-a and IFN-p, but not IFN-y, downregulated the expression of bFGF mRNA and protein in HRCC, as well as in human bladder, prostate, colon, and breast carcinoma cells (80). The inhibitory effect of IFN-a and -P on bFGF expression was cell-density dependent and independent of the antiproliferative effects of IFNs (80,81). The authors recently confirmed that IFN can inhibit bFGF production in an in vivo model system. Systemic administration of human IFN-a decreased the in vivo expression of bFGF, decreased blood vessel density, and inhibited tumor growth of a human bladder carcinoma implanted orthotopically in nude mice (Dinney et al., submitted for publication). Both in vitro and in vivo, the downregu-lation of bFGF required a long exposure of cells to low concentrations of IFNs. In addition, when IFN was withdrawn, cells resumed production of bFGF. These observations are consistent with the findings from Ezekowitz, Mulliken, and Folkman (65), in which complete regression of fatal hemangioma required daily subcutaneous injections of low level IFN-a-2a over a course of 7-8 mo.

0 0

Post a comment