Various tumor cell lines express FGF2 in vitro (Presta et al., 1986; Moscatelli et al., 1986; Halaban et al., 1993, Okumura et al., 1989; Nakano et al., 1992). In situ hybridization and immunolocalization experiments have shown the presence of FGF2 mRNA and/or protein in neoplastic cells, endothelial cells, and infiltrating cells within human tumors of different origin (Schulze-Osthoff et al., 1990; Zagzag et al., 1990; Ohtani et al., 1993; Takahashi et al., 1990; Statuto et al., 1993). Antisense-FGF2 and FGF receptor-1 cDNAs inhibit neovascularization and growth of human melanomas in nude mice (Wang and Becker, 1997). Also, a significant correlation between the presence of FGF2 in cancer cells and advanced tumor stage has been reported (Yamanaka et al., 1993).

In the last few years, various angiogenic factors other than FGF2 have been identified. Among them, VEGF appears to play an important role in tumor neovascularization (Martiny and Marmè, 1995). Indeed, VEGF antagonists, including neutralizing antibodies (Kim et al., 1993), antisense-VEGF cDNA (Saleh et al., 1996), and dominant-negative VEGF receptor mutant (Millauer et al., 1994), can inhibit tumor growth in different experimental models. Also, VEGF levels in tumor biopsies correlate with blood vessel density of the neoplastic tissue and may be of prognostic significance (Samoto et al., 1994; Takahashi et al., 1996).

At variance with VEGF, FGF2 lacks a hydrophobic signal sequence needed to enter the secretory pathway and is usually poorly secreted by producing cells. On the other hand FGF2 may be released by an alternative secretion pathway Mignatti et al., 1991b; Mignatti et al., 1992) and accumulates in the extracellular matrix (ECM), from where it is mobilized by ECM-degrading enzymes (Yeoman 1993) FGF2 is detectable in urine of patients with a wide spectrum of cancers (Chodak et al., 1988; Nguyen et al., 1994) and in cerebrospinal fluid of children with brain tumors (Li et al., 1994). Interestingly, the appearance of an angiogenic phenotype correlates with the export of FGF2 during the development of fibrosarcoma in a transgenic mouse model (Kandell et al., 1991). These data suggest that FGF2 release may occur in vivo and may influence solid tumor growth and neovascularization by autocrine and paracrine modes of action. Accordingly, neutralizing anti-FGF2 antibodies affect tumor growth under defined experimental conditions (Czubayko et al., 1997; Rak and Kerbel, 1997).

Relevant to this point is the recent observation that a secreted FGF-binding protein that mobilizes stored extracellular FGF2 can serve as an angiogenic switch for different tumor cell lines, including squamous cell carcinoma and colon cancer cells (Czubayko et al., 1997). Interestingly, targeting of FGF-binding protein with specific ribozymes reduces significantly the growth and vascularization of xenografted tumors in mice (Czubayko et al., 1997) despite the high levels of VEGF produced by these cells (see Rak and Kerbel, 1997 for a further discussion). These data suggest that modulation of FGF2 expression, release, and mobilization may allow a fine tuning of the angiogenesis process even in the presence of significant levels of VEGF. This hypothesis is supported by the capacity of the two factors to act synergistically in stimulating angiogenesis in vitro and in vivo (Goto et al., 1993; Asahara et al., 1995).

Recently we have described an export-dependent mechanism of action for FGF2 in the human endometrial adenocarcinoma HEC-1B cell line (Coltrini et al., 1995). After transfection with an expression vector harboring a human FGF2 cDNA, different FGF2-expressing clones were obtained and one of them (FGF2-B9 clone) showed the capacity to secrete significant amounts of the growth factor. This clone stimulates angiogenesis in the avascular rabbit cornea and forms highly vascularized tumors growing faster than the non-FGF2 releasing clones (FGF2-A8 and FGF2-B8 clones) when transplanted s.c. into nude mice line (Coltrini et al., 1995). FGF2-transfected HEC-1-B clones may therefore represent an useful experimental model to study the effects on the microvascular architecture of the neoplastic tissue consequent to modifications of tumor microenvironment due to FGF2 expression and release.

FGF2 export decides the biological behavior of FGF2-transfected HEC-1-B cells. A8 and B9 cells overexpress FGF2 at similar levels but only the latter clone releases significant amounts of the growth factor with a consequent increase in fibrinolytic and angiogenic potential. This results in an increased rate of growth in vivo limited to FGF2-B9 tumors.

We studied the microvascular pattern of tumors grown in nude mice originated from HEC-1B cell clones expressing and secreting different levels of FGF2 by means of the microvascular corrosion casting technique. Microvascular corrosion casting allows for qualitative as well as quantitative insights into the tumor vascular system. All relevant parameters defining the microvascular network, such as interbranching distances, intervascular distances, branching angles, and vessel diameters can be determined using 3D stereo pairs of the casted tumor vascularity (Malkusch et al., 1995). We used this powerful method to assess whether differences in FGF2 expression and secretion may differently affect the microvascular architecture of the tumor in our experimental model. When transplanted at the same concentration, FGF2-B9 cells grew faster in nude mice compared to FGF2-A8 and FGF2-B8 clones. The total amount of new vessel formation was far higher in FGF2-B9 tumors than in FGF2-B8 or FGF2-A8 tumors. Also, vessel courses were more irregular and blind ending vessels and evasates were more frequent in FGF2-B9 tumors.
 
 

Moreover, FGF2-B9 tumor microvasculature was characterized by a wider average vascular diameter and by an extreme variability of the diameter of each individual vessel along its course between two ramifications. No statistical differences were instead observed when the distribution curves of the values of intervascular distances, interbranching distances, and branching angles of the microvessel network were compared among the different experimental groups. The distinctive features of the microvasculature of FGF2-B9 tumors were retained, at least in part, in the smaller lesions produced by injection of a limited number of cells.

The data indicate that FGF2 production and release confer to FGF2-B9 cells the ability to stimulate the formation of new blood vessels with distinctive morphological features. Neovascularization of FGF2-B9 lesions parallels the faster rate of growth of the neoplastic parenchyma. This does not affect the overall architecture of the microvessel network that appears to be primed by characteristics of the HEC-1-B tumor cell line and/or by the microenvironment of the host.