FGF signals for cell proliferation and migration through different pathways.

FGFs are pleiotropic growth factors that control cell proliferation, migration and differentiation. However, FGF transduction studies have so far focused primarily on the mitogenic effect of this growth factor family and it has been difficult to assess if the described intracellular signaling pathways are dedicated solely to cell proliferation, or whether they are equally important for the migratory activity often seen in responsive cells. We review here papers in which the migratory effects of this growth factor family were clearly discriminated from proliferative effects. In toto, these studies suggest that cells use different signaling pathways for migration, such as Src and p38 MAP kinase, from those for proliferation, which tend to upregulate the ERKs. Which signaling pathway a cell uses for proliferation or migration appears to depend on many factors, including the structure and the quantity of available FGF trapped in the basal lamina by heparan sulfate co-factors, the disposition of cognate high affinity receptors and the general environment of the cell. Thus the density of the cell population, the state of the cell cycle, the presence of other factors or receptors will modulate the migratory response of cells to FGF.

Expression of nerve growth factor receptors and their prognostic value in human breast cancer.

Nerve growth factor (NGF) has been shown recently to be mitogenic for human breast cancer cells. In the present study, we have assayed the expression of NGF receptors (NGFRs: TrkA and p75) mRNAs in 363 human primary breast cancers, using real-time quantitative reverse transcription-PCR. NGFRs were found in all of the tumor biopsies. TrkA and p75 were positively correlated and were respectively associated with the histoprognostic grading and the tumor type. NGFRs were both related to progesterone receptors. In univariate analyses, TrkA (>upper quartile) was associated with longer overall survival. Histoprognostic grading, tumor size, node involvement, and steroid receptors were also prognostic factors. In Cox multivariate analyses, TrkA was not a prognostic parameter. This study demonstrates the expression of NGFRs in breast cancer and points out that patients with high levels of TrkA have a more favorable overall survival prognosis.

Proteomic analysis reveals that 14-3-3sigma is down-regulated in human breast cancer cells.

The class of molecular chaperones known as 14-3-3 is involved in the control of cellular growth by virtue of its apparent regulation of various signaling pathways, including the Raf/mitogen-activated protein kinase pathway. In breast cancer cells, the sigma form of 14-3-3 has been shown to interact with cyclin-dependent kinases and to control the rate of entry into mitosis. To test for a direct role for 14-3-3 in breast epithelial cell neoplasia, we have quantitated 14-3-3 protein levels using a proteomic approach based on two-dimensional electrophoresis and matrix-assisted laser desorption/ionization mass spectrometry (MALDI-TOF). We show here that 14-3-3sigma protein is strongly down-regulated in the prototypic breast cancer cell lines MCF-7 and MDA-MB-231 and in primary breast carcinomas as compared with normal breast epithelial cells. In contrast, levels of the alpha, beta, delta, or zeta isoforms of 14-3-3 were the same in both normal and transformed cells. The data support the idea that 14-3-3sigma is involved in the neoplastic transition of breast epithelial cells by virtue of its role as a tumor suppressor; as such, it may constitute a robust marker with clinical efficacy for this pathology.

Steroid hormones modulate H19 gene expression in both mammary gland and uterus.

H19 is an imprinted and developmentally regulated gene whose product remains apparently untranslated. In a previous study on breast adenocarcinomas, we reported that overexpression of the H19 gene was significantly correlated with the presence of steroid receptors, suggesting the putative role of hormones in H19 transcription. To determine the mode of steroid action, we have detected levels of H19 RNA synthesis during mammary gland development by in situ hybridization (ISH): two peaks of H19 transcription occur during puberty and pregnancy. Furthermore, we demonstrated by ISH that in the uterus H19 RNA synthesis is high during estrus and metestrus phases. To test steroid control of H19 transcription, ovariectomized and adrenalectomized mice were supplemented, 1 week after surgery, with 17-beta-estradiol (E2, 20 microg/kg/day), progesterone (P, 1 mg/kg/day) or corticosterone (B, 0.3 mg/ kg/day) for 2 weeks. According to ISH data, E2 and to a lesser extent B stimulated H19 transcription in the uterus, whereas P inhibited it. To confirm the in vivo results, in vitro experiments were performed using cultures of MCF-7 cells (a hormone-sensitive mammary cell line). E2 stimulated the endogenous H19 gene of this cell line and tamoxifen inhibited this effect. Furthermore, we performed transient cotransfections in MCF-7, in HBL-100 (another hormone-sensitive mammary cell line) and in BT-20 (a hormone-insensitive mammary cell line) with various constructs of ERalpha (WT or mutated) and PR-A, in presence or absence of steroid hormones. We demonstrated that ERalpha up-regulated the H19 promoter in MCF-7 and in HBL-100, whereas PR-A did not have any effect per se. Moreover, in MCF-7, PR-A antagonized clearly the ERalpha-mediated promoter enhancement, but in HBL-100 this counteracting effect on the ERalpha up-regulation was not found. Interestingly, the same experiments performed in BT-20 cell line provided very similar results as those obtained in MCF-7 cells, with a clear down-regulation mediated by PR-A on the H19 promoter. All these in vitro data are in agreement with in vivo results. In addition, data obtained with ERalpha mutants indicate that H19 promoter activation is both ligand-dependent and ligand-independent. We have thus demonstrated that H19 gene expression is controlled by steroid hormones; furthermore, this gene is highly expressed in hormone-sensitive organs when the hormonal stimulation is accompanied with a morphological repair.

Ductal cyst formation in collagen-embedded adult human islet preparations. A means to the reproduction of nesidioblastosis in vitro.

Neogenesis of endocrine islets from ductal epithelium termed nesidioblastosis has been described in vivo after various experimental conditions (90% pancreatectomy or pancreas wrapping in the rodent) and in clinical pathologies. In the adult regenerating pancreas, a proliferation and organization of ductal epithelium into tubular structures precedes its differentiation into endocrine cells. Reproduction of nesidioblastosis in vitro may provide a novel approach to human islet propagation in vitro. With this aim, adult human islet preparations were cultured in diverse three-dimensional (3D) gels in the presence of serum. After 3-5 days in rat tail collagen gels, proliferating (bromodeoxyuridine-positive) cystic structures appeared associated with islets and as isolated spheres. Percentage labeling indexes of the cysts were 4.1, 18.7, 15.4, and 13.3% after 3, 5, 7, and 10 days of culture, respectively. Immunohistochemistry confirmed the ductal (carbohydrate antigen 19-9) and epithelial (keratin-1) nature of the cysts. No cysts were formed in agarose gels or Vitrogen 100, whereas the cyst number was increased by the quantity of serum (20% > 10%) and gels rich in extracellular matrix components and growth factors (Matrigel). The latter lead to tubular networks. Single endocrine islet cells were observed in the ductal cysts after 7 (2.8%) to 10 (5.6%) days in rat tail collagen. Our observations paralleled the changes characteristic of the regenerating pancreas in vivo. 3D culture may permit the identification of matrix and media constituents promoting the neogenesis of islets and may be the means to increase the mass of endocrine tissue obtained from adult cadaveric pancreases for transplantation.

Cell interactions and regeneration control.

This paper is a review of the main findings of our laboratory on the control of regeneration by cell interactions. These include results related to the role of both cell contact and local soluble factors in regeneration of the legs of insects and newts and of the parapodia and segments of nereis. The pattern of these structures is considered to be defined by positional information distributed as longitudinal and transverse positional value sequences carried by epidermal (insect) or mesenchymal (newt) cells. By associating tissues to create transverse and longitudinal discontinuities in these sequences, single or multiple regenerating structures were obtained. These structures are formed by the intercalation of cells characterized by intermediate positional values which fill the gap between the tissues in contact. Positional information may also be changed during regeneration by the nerve cord in nereis and retinoids in the newts. We describe additional cases where morphogenesis occurs without any overt discontinuity in positional information, such as from a locally injured or non-injured insect trochanter, or after deflection of nerves in nereis and newt. Regeneration following an amputation may be considered as a special case of intercalary regeneration, the first stage being the juxtaposition of normally non-contiguous cells resulting in a longitudinal or/and a transverse gap. We also report studies on local factors produced by nerves and the blastema during newt limb regeneration. The nerve factor is necessary for the division of blastemal cells. After denervation, mesenchyme differentiates in an abnormal way. The mitogenic signal from the nerves is mediated by the PKC pathway. Its production is enhanced by regeneration of cut nerve fibers. The blastema also produces growth factors. We show that the epidermal cap and mesenchyme contain acidic FGF-like factor, and that the proliferating mesenchyme stimulates nerve fibers to regrow into the blastema.

Nerve growth factor induced neurite outgrowth from amphibian neuroepithelial precursor cells is prevented by dipeptides inhibiting ubiquitin-mediated proteolysis.

The effect of dipeptides known to inhibit the ubiquitin-mediated proteolysis has been examined on growth factor induced neurite outgrowth from amphibian neuroepithelial precursor cells in primary culture. Nerve growth factor (NGF) stimulated neuritogenesis from these cells but fibroblast growth factor 2 (FGF-2) only increased the number of melanophores. The neurite outgrowth induced by NGF was inhibited by the dipeptides blocking the ubiquitin mediated proteolysis (Leu-Ala and Leu-Gly) whereas the inactive control dipeptides (Ala-Leu and Ala-His) had no effect. This suggests that ubiquitin-mediated proteolysis involving the ubiquitin ligase E3 is necessary for growth factor induced neuronal differentiation during the development of the central nervous system.

Expression of c-ets-1 and uPA genes is associated with mammary epithelial cell tubulogenesis or neoplastic scattering.

Although the inductive interactions which trigger epithelial morphogenesis have been extensively described, little is known about the transcription factors involved in these processes. During mammary gland morphogenesis, we report the expression of the transcription factor c-ets-1 and one of its target genes uPA in mesenchymal cells during early stages of epithelial invasion, and later in epithelial cells themselves. In vitro studies show that both c-ets-1 and uPA mRNAs can be induced in cultured normal mammary epithelial cells in response to medium conditioned by MRC-5 fibroblasts. In contrast, invasive tumorigenic cell lines from the mammary epithelium express constitutively c-ets-1 and uPA while non-invasive tumorigenic cells do not. In three dimensional co-cultures in collagen gels, a preferential expression of these genes is detected in epithelial cells migrating through the gel either at the tips of normal ducts or in cancerous cells which are scattering. These genes are also expressed in the neighboring fibroblasts. In MRC-5 fibroblasts, conditioned media from tumorigenic epithelial cells induce more efficiently c-ets-1 and uPA mRNA accumulation than do conditioned medium from normal cells. These results suggest that epithelial-mesenchymal interactions trigger c-ets-1 and uPA expression in both compartments during mammary gland morphogenesis. The expression of the genes correlates with invasiveness of epithelial cells irrespective of their being normal or cancerous.

Experimental evidence for FGF-1 control of blastema cell proliferation during limb regeneration of the amphibian Pleurodeles waltl.

During regeneration, blastema cell proliferation depends on several different factors which are, as yet, not fully understood. Previous studies showing the presence of FGF-1 and FGF receptors in the limb blastema make FGF-1 a potentially important molecule for limb regeneration but they do not demonstrate that this factor is active during the process. In the present study, we have first of all confirmed the presence of FGF-1 in limb blastemas of the amphibian Pleurodeles waltl using immunochemistry. Second, we provide evidence in vivo that FGF-1 controls blastema cell proliferation by using different reagents which interfere with FGF activity. Sulfated polysaccharides which bind FGFs, such as heparin, iota-carrageenan and pentosan polysulfate, are able to decrease both 3H-thymidine incorporation and the mitotic index in regeneration blastemas. In addition, suramin which inhibits the binding of growth factors to their receptors, induces the same effect. The presence of receptors in blastema cells is also demonstrated by using the FGF-saporin complex which is known to bind to FGF receptors and to kill cells bearing these receptors. This complex decreases the mitotic index in mesenchyme, while saporin alone did not influence cell proliferation. Finally, results obtained using a neutralizing monoclonal antibody against FGF-1 which was able to specifically reduce blastema cell proliferation, suggests that FGF-1 plays an important function in limb regeneration.

The mitogenic signaling pathway for fibroblast growth factor-2 involves the tyrosine phosphorylation of cyclin D2 in MCF-7 human breast cancer cells.

Fibroblast growth factor-2 (FGF-2) is mitogenic for the human breast cancer cell line MCF-7; here we investigate some of the signaling pathways subserving this activity. FGF-2 stimulation of MCF-7 cells resulted in a global increase of intracellular tyrosine phosphorylation of proteins, particularly FGF receptor substrate-2, the protooncogene product Src and the mitogen-activated protein kinase (MAP kinase) cascade. A major increase in the tyrosine phosphorylation of a 30-kDa protein species was also found. This protein was identified as cyclin D2 by mass spectrometry after trypsin digestion. Immunoprecipitation of cyclin D2 and immunoblotting with anti-phosphotyrosine antibodies confirmed that the tyrosine phosphorylation of cyclin D2 was indeed induced by FGF-2 stimulation. In addition, pharmacological inhibition of Src (with herbimycin A and PP2), and of the MAP kinase cascade (with PD98059), confirmed that Src activity is required for the FGF-2-induced phosphorylation of cyclin D2 whereas MAP kinase activity is not. Thus, tyrosine phosphorylation of cyclin D2 may be a key regulatory target for FGF-2 signaling.

Characterization of Le(x), Le(y) and A Le(y) antigen determinants in KDN-containing O-linked glycan chains from Pleurodeles waltlii jelly coat eggs.

Novel acidic oligosaccharides were isolated in large amounts by reductive alkaline treatment of the jelly coat of Pleurodeles waltlii (Michah) eggs. The oligosaccharides were found to contain the newly described KDN as acidic monosaccharide and possess either the Le(x), Le(y) and A Le(y) antigenic determinants. Occurrence of Le(x) and Le(y) determinants previously recognized as tumor-associated antigen (TAA) demonstrates that mucins of lower animals may represent a rich and easily available source for preparing TAA. Moreover, it reinforces the hypothesis according to which TAA are evolution markers.

Regulation of cell proliferation and urokinase plasminogen activation of human breast epithelial cells by carrageenans.

Carrageenans, a family of polysulphated carbohydrates, are able to inhibit the binding to cells of growth factors such as transforming growth factor 1 (TGF 1), fibroblast growth factor-2 (FGF-2) and platelet-derived growth factor (PDGF) and so modulate cell invasion and proliferation. We studied the effects of carrageenans on the proliferation and on the uPA/PAI-1 system in breast epithelial cells. Carrageenans were able to inhibit the proliferation of both normal breast epithelial cells (NBEC) and breast epithelial cancer cell lines (MDA-MB-231 and MCF-7) but could only inhibit the uPA activity in the MDA-MB-231 cells. Moreover, carrageenans inhibited FGF-2 binding in all three cell types, suggesting that they regulate cell proliferation and uPA/PAI-1 system through two distinct mechanisms. These molecules could be considered as potentially useful anti-cancer agents.

Basic fibroblast growth factor (bFGF): mitogenic activity and binding sites in human breast cancer.

We investigated binding characteristics of basic fibroblast growth factor (bFGF) on membranes prepared from 4 human breast cancer cell lines and 38 primary BC biopsies. Competitive binding experiments were performed and analyzed using the "Ligand" program. Furthermore bFGF mitogenic activity was measured by [3H]thymidine incorporation into DNA from breast cancer cell lines. The presence of high-affinity binding sites was demonstrated in each cell type (MCF-7: Kd = 0.60 nM; T-47D: Kd = 0.55 nM; BT-20: Kd = 0.77 nM; MDA-MB-231: Kd = 0.34 nM). The presence of these high-affinity binding sites was confirmed with saturation experiments. A second class of low-affinity binding sites was detected in the 2 hormone-independent cells (BT-20: Kd = 2.9 nM; MDA-MB-231: Kd = 2.7 nM). bFGF stimulated the proliferation of MCF-7, T-47D, BT-20 but not MDA-MB-231 cell lines. With competition experiments, binding sites were detectable in 36/38 breast cancers; high-affinity binding sites (Kd less than 1 nM) were present in 19/36 cases and low-affinity binding sites (Kd greater than 2 nM) were present in 29/36 cases (the two classes of binding sites were present in 12 breast cancers). No relation between bFGF binding sites and node involvement, histologic type or grading of the tumor was evidenced. There were negative correlations (Spearman test) between total bFGF binding sites and estradiol receptor (P = 0.05) or progesterone receptor (P = 0.009). The demonstration of (1) bFGF specific binding sites in breast cancer membranes, and (2) bFGF growth stimulation of some breast cancer cell lines indicates that this factor may be involved directly in the growth of some breast cancers.

Production in vitro by spinal cord of growth factor(s) acting on newt limb regeneration: influence of regeneration of the nerve fibers.

In order to approach the problem of regulation of growth factor(s) production during limb regeneration in newt, we co-cultivated spinal cord segments and blastemas. First we showed that, like the sensory supply, the spinal cord possesses size-dependent mitogenic capacities for limb blastemas. A 5-mm long spinal segment enhances radiolabelled thymidine incorporation to the same extent as spinal ganglia (1.6-fold). Second, we co-cultivated blastemas with spinal segments, the nerve fibers of which were previously stimulated to regenerate (= stimulated spinal segment) or not (= non-stimulated spinal segment). Only after a 24-h coculture, do stimulated spinal segments enhance thymidine incorporation in blastemas 2-fold more than non-stimulated spinal segments. Our results suggest that during limb regeneration brachial nerves produce more growth factor(s) when regrowing, inducing the proliferation of blastema cells which in return deliver a neuronotrophic factor acting on these nerves.

Embryonic brain-derived heparan sulfate inhibits cellular membrane binding and biological activity of basic fibroblast growth factor.

We have investigated the ability of glycosaminoglycans from embryonic chick brain (15 days old) to interact with basic fibroblast growth factor (bFGF). 35SO4 metabolically labeled glycosaminoglycans were purified and separated on DEAE-cellulose chromatography. Material which eluted between 0.20 and 0.35 M NaCl displaced the binding of [125I]bFGF to brain membrane. This activity was dose-dependent and on the basis to its heparinase sensitivity and chondroitinase insensitivity, has been attributed to heparan sulfate. CL-6B-Sepharose chromatography of this material revealed two glycosaminoglycans of molecular masses of about 15,000 and 65,000. Incubation with [125I]bFGF followed or not by heparinase and chondroitinase treatment of electrotransfert from SDS-PAGE revealed that both of these forms correspond to heparan sulfate chains and bind bFGF. In vitro, embryonic brain-derived heparan sulfate inhibited both bFGF induced [3H]thymidine incorporation in CCL39 cells and neurite outgrowth in PC12 cells. These results suggest that heparan sulfate play an important function in the control of the biological activity of bFGF during brain development.

Stimulation of axon growth from the spinal cord by a regenerating limb blastema in newts.

The effects of limb blastemas of Pleurodeles waltl on axon growth from fragments of spinal cord were studied in vitro. Cultured in a defined medium, spinal cord fragments regenerated sparse, short axons. The culture of spinal fragments in the presence of blastemas greatly enhanced the length, number and survival of axons. Testing separately each of the two components of the blastema showed that only the mesenchyme exerts a neurotropic effect on the spinal fragments. Other tissues such as muscle or skin had a limited neurotrophic effect. Additionally, the neurotrophic activity of blastemas seems to be dependent of its proliferation status. Compared with blastemas of regenerating limbs from young animals, irradiated blastemas (devoid of mitotic activity) and blastemas of regenerating limbs from old animals or differentiated blastemas (both characterized by a low mitotic activity), exhibited a weaker neurotrophic influence. The blastema neurotrophic factor is not an attachment molecule but a soluble one and cannot be nerve growth factor (NGF) or fibroblast growth factor (FGF). It has a relatively low molecular weight (less than 15 kDa) and its protein nature was ascertained by its sensitivity to heating and proteases. As the production of this mesenchyme-derived neurotrophic factor depends upon mesenchymal cell proliferation of the blastema, we suggest that there is loop of positive regulation between spinal nerves and blastema. Blastema tissues may stimulate nerve regeneration allowing the stimulation of proliferation of blastema cells by regenerating nerve fibers. Alternatively, blastema cells may produce a neurotrophic factor whose secretion might be dependent on cell proliferation.

Developmental changes of acidic fibroblast growth factor (aFGF) transcription and expression in mouse brain.

In order to increase our knowledge of the in vivo role of acidic fibroblast growth factor (aFGF) in the central nervous system, we have examined aFGF levels during mouse brain development. Using a specific polyclonal antibody raised against aFGF, we measured levels of aFGF-immunoreactive material (IRMaFGF) in extract of total mouse brain taken at different days of development. We found that the level of measurable IRMaFGF remained low and without significant variation during fetal brain development (0.2 ng/mg of extracted proteins). During the first 11 days postnatal (P0 to P11), IRMaFGF increased from 0.5 to 1.5 ng/mg. Between P11 and P14 IRMaFGF levels went up more rapidly, reaching 5 ng/mg. From P30 to adulthood a constant value of 2.5 ng/mg was measured, aFGF content in the different brain extracts was further characterized by its affinity for heparin-Sepharose, its elution at 1 M NaCl from this column and its capacity to induce thymidine incorporation in quiescent fibroblasts. These results were confirmed at the mRNA level. Northern blot analyses of poly A+ mRNA from brains with a specific riboprobe for bovine aFGF, revealed a major 4.5-Kb transcript and a minor 2.7-Kb transcript detectable only in postnatal brains. A similar pattern to that observed for IRMaFGF was seen with these mRNA transcripts, indicating that these aFGFmRNA are translated in the mouse brain. Our results suggest that aFGF may act in the postnatal phases of brain maturation.

High and low affinity membrane binding sites for fibroblast growth factors in the developing chick brain.

Acidic and basic fibroblast growth factors (aFGF and bFGF), two mitogenic, neurotrophic and angiogenic molecules, are present in the embryonic chick brain but their function remains unclear. In order to approach the biological activity of FGFs during brain development, we have looked for their receptors and studied their regulation through chick brain development. Competitive binding studies realized on brain membranes indicated the presence of two classes of FGF binding sites: high affinity binding sites (dissociation constant, Kd = 100 pM) and low affinity binding sites (Kd = 20 nM). Cross-competition experiments show that these two classes of binding sites both interact with aFGF and bFGF. The number of sites in these two classes of binding sites changes during embryogenesis. On the one hand, the membrane capacity of high affinity sites decreases from E7 (1 +/- 0.2 pmol/mg of protein) to E15 (0.5 +/- 0.2 pmol/mg of protein); on the other hand, the membrane capacity of low affinity sites increases from E15 (25 +/- 4 pmol/mg of protein) to P1 (75 +/- 20 pmol/mg of protein). Cross-linking experiments revealed the presence of two putative receptor forms of molecular masses of about 130 and 95 kDa. These results suggest that the biological activity of aFGF and bFGF during brain embryogenesis could be regulated by the expression of high and low affinity binding sites for these growth factors.

Differential responsiveness of human breast cancer cells to basic fibroblast growth factor: a cell kinetics study.

The effects of basic fibroblast growth factor (bFGF) on breast cancer cells are still contradictory and not fully understood. We have studied the effect of bFGF on the cell cycle kinetics of two breast cancer cell lines (MCF-7 and MDA-MB-231) and an immortalized cell line (HBL-100). The methodology included use of microscopic image analysis with cell numeration, Feulgen staining, Proliferating Cell Nuclear Antigen/Ki-67 immunodetection and bromodeoxyuridine incorporation. We show that bFGF is mitogenic for MCF-7 cells via a mechanism of recruitment of G0 phase cells to reenter into the cell cycle and by decreasing the G1 phase length. No effect of bFGF on cell cycle parameters has been found with either highly metastatic MDA-MB-231 cells or immortalized HBL-100 cells. These results reveal differences in bFGF responsiveness of breast epithelial cells.

[Involvement of sulfated proteoglycans in the control of proliferation of MCF-7 breast cancer cells]. / Implication des protéoglycanes sulfatés dans le contrôle de la prolifération des cellules cancéreuses mammaires MCF-7.

The MCF-7 breast cancer cells exhibit remarkable growth enhancement in response to basic fibroblast growth factor (FGF-2) stimulation in a dose dependent manner. To investigate the involvement of proteoglycans on control of FGF-2 induced proliferation, polysaccharide chains were degraded by specific enzymes. Our results showed that MCF-7 cells were unsensitive to FGF-2 after enzymatic degradation of heparin sulfate proteoglycans (HSPG) by heparinase. After metabolic inhibition of sulphation by sodium chloride, radiolabelled proteoglycans were purified and quantified by ion exchange chromatography. Sodium chlorate treatment reduced by 70% sulfation of proteoglycans. This decrease of sulphation totally inhibited FGF-2-mediated proliferation. The sulphated glycosaminoglycans which were critical in FGF-2-induced proliferation were strictly HSPG, as an addition of heparin in cell culture medium can restore FGF-2 mitogenic activity. In contrast, other glycosaminoglycans (chondroitin sulfate/hyaluronic acid) did not show any effect. These results provide clear evidence for the critical role of HSPG in FGF-2-induced proliferation on MCF-7 breast cancer cells.