Suramin inhibition of growth factor receptor binding and mitogenicity in AKR-2B cells

Suramin, a polyanionic compound, has previously been shown to dissociate platelet-derived growth factor (PDGF) from its receptor. In the present study suramin was found to inhibit the growth of sparse cultures of AKR-2B cells in fetal bovine serum (FBS)-supplemented medium in a dose-dependent, reversible fashion. Suramin also inhibited the ability of FBS, transforming growth factor beta (TGF beta), heparin-binding growth factor type-2 (HBGF-2), and epidermal growth factor (EGF) to stimulate DNA synthesis in density-arrested cultures of AKR-2B cells. The inhibition of growth factor-stimulated mitogenicity was directly correlated to the dose of suramin required to inhibit the binding of 125I-labeled TGF beta, HBGF-2, and EGF to their cell surface receptors. Suramin affected TGF beta and HBGF-2-related events at a 10-15-fold lower dose than that required for EGF-related events. It was also noted that suramin inhibited TGF beta-stimulated soft agar colony formation of AKR-2B (clone 84A) cells as well as the spontaneous colony formation of AKR-MCA cells, a chemically transformed derivative of AKR-2B cells. This demonstrates that suramin's spectrum of action for growth factors and their receptors should be extended to include TGF beta, HBGF-2, and EGF as well as PDGF. The data further suggest that the spontaneous growth of AKR-MCA cells in soft agar is dependent on growth factor binding to cell surface receptors.     

Regulation of rat proximal tubule epithelial cell growth by fibroblast growth factors, insulin-like growth factor-1 and transforming growth factor-beta, and analysis of fibroblast growth factors in rat kidney.

Growth factors may play an important role in regulating the growth of the proximal tubule epithelium. To determine which growth factors could be involved, we have investigated the mitogenicity of various purified factors in rat kidney proximal tubule epithelial (RPTE) cells cultured in defined medium. Fibroblast growth factors, aFGF (acidic FGF) and bFGF (basic FGF), stimulate DNA synthesis in a dose-dependent manner, with ED50 values of 4.5 and 3.2 ng/ml, respectively; their effects are not additive. With cholera toxin in the medium, both aFGF and bFGF can replace insulin or epidermal growth factor (EGF) to attain the maximum level of cell growth, but they cannot replace cholera toxin. Cholera toxin specifically potentiates the effects of FGFs on DNA synthesis. At high cell density, both insulin and insulin-like growth factor 1 (IGF-1) induce DNA synthesis more effectively than EGF, FGFs and cholera toxin. The high concentration (0.2-1.0 microgram/ml) of insulin required for cell growth can be replaced by a low concentration of IGF-1 (10-20 ng/ml), indicating that insulin probably acts through a low affinity interaction with the IGF-1 receptor. Transforming growth factor-beta 1 (TGF-beta 1) inhibits DNA synthesis induced by individual factors and combinations of factors in a concentration-dependent manner. Northern blot analysis shows that mRNA for TGF-beta 1, IGF-1, and aFGF, but not bFGF are present in rat kidney. Western blot analysis and bioassay data confirmed that the majority of FGF-like protein in rat kidney is aFGF. The data suggest that in addition to EGF, IGFs, and TGF-beta, FGFs may also be important kidney-derived regulators of proximal tubule epithelial cell growth in vivo and in vitro.     

Effect of suramin on growth of androgen-responsive mouse tumor (Shionogi carcinoma 115) and its autonomous subline (Chiba subline 2)

Suramin has been shown to inhibit the binding of various growth factors to their receptors. Shionogi Carcinoma 115 cells (SC 115 cells) and Chiba Subline 2 cells (CS 2 cells) are clones of an androgen-responsive mouse tumor cell and its autonomous subline, respectively. Since the growth of SC 115 and CS 2 cells are assumed to be regulated by their own fibroblast growth factor (FGF)-like growth factors, the present study was undertaken to examine the effect of suramin on these cells. Suramin inhibited the growth of SC 115 and CS 2 cells in a dose dependent manner. The inhibition of suramin was reversible up to 50 micrograms/ml. Suramin reversibly changed the shape of these cells from fibroblast-like to polygonal and epithelial-like ones, and inhibited 3H-thymidine incorporation into these cells which was evoked by acidic and basic FGFs, and conditioned medium obtained from CS 2 cells. The binding of 125I-basic FGF to SC 115 and CS 2 cells was inhibited by suramin. However, suramin had no effect on growth factor production and the hst-1 gene expression on CS 2 cells. In conclusion, suramin inhibited the autocrine and paracrine growth of SC 115 and CS 2 cells by blocking the binding of autocrine growth factors to their receptors.     

Basic and acidic fibroblast growth factors modulate the epidermal growth factor receptor by a protein kinase C-independent pathway

Human acidic and basic fibroblast growth factors (aFGF and bFGF) inhibit epidermal growth factor (EGF) receptor binding in mouse Swiss 3T3 cells. Scatchard analysis indicates that aFGF and bFGF cause a decrease in the high affinity EGF receptor population, similar to that observed for activators of protein kinase C such as phorbol esters, platelet-derived growth factor (PDGF) and bombesin. However, unlike phorbol esters, aFGF and bFGF inhibit EGF binding in protein kinase C-deficient cells. The time course and dose response of inhibition of EGF binding by both aFGF and bFGF are very similar, with an ID50 of approximately 0.10 ng/ml. In contrast to bombesin but like PDGF, neither aFGF nor bFGF act on the EGF receptor through a pertussis toxin-sensitive G protein. These results indicate that both acidic and basic FGF depress high affinity EGF binding in Swiss 3T3 cells with similar potency through a protein kinase C/Gi-independent pathway.     

Neural retina of chick embryo in organ culture: effects of blockade of growth factors by suramin

The neural retina is a highly organized organ whose final histoarchitecture depends on the presence of diverse growth factors and on their interactions with extracellular matrix components. However, the role of growth factors on retinal development is not fully understood. Suramin has been shown to produce diverse cellular effects via the simultaneous block of the action of several growth factors. We have therefore studied the effects of suramin on organotypic culture of chick embryo neural retina in order to gain further insights into the participation of growth factors in neural retinal development. Neural retina was incubated for 24 h with suramin at 50-200 microM and then processed to determine cell proliferation, nuclear morphology, and actin distribution. Suramin provoked extensive morphological changes revealed by a decrease in BrdU incorporation, alterations in cellular organization, and disruption of the outer limiting membrane, with the emergence of cellular elements through it. All of these effects were dose-dependent and markedly attenuated by the simultaneous presence of suramin and fibroblast growth factor 2 (FGF-2) in the culture medium. These findings indicate that suramin induces pleiotropic effects on the histoarchitecture of the chicken neural retina in organ culture and suggest that FGF-2 is one of the biological modulators involved in the maintenance of the structural organization of the chicken neural retina.     

Understanding the mechanism of the antimitogenic activity of suramin

Fibroblast growth factors (FGFs) play crucial roles in the regulation of key cellular processes such as angiogenesis, differentiation, and tumor growth. Suramin, a polysulfonated naphthylurea, is known to be a potent inhibitor of FGF-induced angiogenesis. Using isothermal titration calorimetry, we demonstrate that human acidic fibroblast growth factor (hFGF-1) binds to suramin with high affinity in the nanomolar range. The suramin:hFGF-1 binding stoichiometry is estimated to be 2:1. Size-exclusion chromatography data reveal that suramin oligomerizes hFGF-1 to form a stable tetramer. Thermal unfolding experiments monitored by steady state fluorescence, and limited trypsin digestion analysis data suggest that suramin-induced oligomerization of hFGF-1 occurs in two steps. The first step involves the binding of suramin at specific sites on the protein. Two molecules of suramin appear to bind simultaneously to one molecule of hFGF-1. Binding of suramin possibly involves formation of solvent-exposed nonpolar surfaces in hFGF-1. In the second step, FGF appears to oligomerize through coalescence of the solvent-accessible nonpolar surfaces. Results of the NMR experiments reveal that suramin binds to residues in the heparin binding pocket as well as to residues involved in FGF receptor binding. On the basis of the results of this study, we propose a model to explain the molecular mechanism(s) underlying the antimitogenic activity of suramin. To our knowledge, this is the first study in which suramin interaction sites on FGF have been characterized.     

Modulating the Structure of EGFR with UV Light: New Possibilities in Cancer Therapy

The epidermal growth factor receptor (EGFR) is a member of the ErbB family of receptor tyrosine kinases. EGFR is activated upon binding to e.g. epidermal growth factor (EGF), leading to cell survival, proliferation and migration. EGFR overactivation is associated with tumor progression. We have previously shown that low dose UVB illumination of cancer cells overexpressing EGFR prior to adding EGF halted the EGFR signaling pathway. We here show that UVB illumination of the extracellular domain of EGFR (sEGFR) induces protein conformational changes, disulphide bridge breakage and formation of tryptophan and tyrosine photoproducts such as dityrosine, N-formylkynurenine and kynurenine. Fluorescence spectroscopy, circular dichroism and thermal studies confirm the occurrence of conformational changes. An immunoassay has confirmed that UVB light induces structural changes in the EGF binding site. A monoclonal antibody which competes with EGF for binding sEGFR was used. We report clear evidence that UVB light induces structural changes in EGFR that impairs the correct binding of an EGFR specific antibody that competes with EGF for binding EGFR, confirming that the 3D structure of the EGFR binding domain suffered conformational changes upon UV illumination. The irradiance used is in the same order of magnitude as the integrated intensity in the solar UVB range. The new photonic technology disables a key receptor and is most likely applicable to the treatment of various types of cancer, alone or in combination with other therapies.     

Heparin modulation of the neurotropic effects of acidic and basic fibroblast growth factors and nerve growth factor on PC12 cells

Nerve growth factor (NGF) and acidic or basic fibroblast growth factor (aFGF and bFGF, respectively) induce neurite outgrowth from the rat pheochromocytoma cell line, PC12. The neurites induced by these three factors are stable for up to a month in cell culture in the continued presence of any of the above growth factors. bFGF (ED50 = 30 pg/ml) is 800 fold more potent in stimulating neurite outgrowth than aFGF (ED50 = 25 ng/ml) and 260 fold more potent than NGF (ED50 = 8 ng/ml). While the neurotropic activities of aFGF and NGF are potentiated by heparin, that of bFGF is both partially inhibited or stimulated, depending upon the concentration of bFGF. Radioreceptor binding experiments show that aFGF and bFGF bind to a common binding site on the PC12 cell surface. Affinity labeling studies demonstrate a single receptor with an apparent molecular weight of 145,000 daltons, which corresponds to the high molecular weight receptor identified in BHK-21 cells. NGF does not appear to compete with aFGF or bFGF for binding to the receptor. Heparin blocked the binding of bFGF to the receptor but had only a small inhibitory effect on the binding of aFGF to the receptor. Thus, it appears that heparin inhibition of the neurotropic effects of bFGF occurs, at least in part, by impairing the interaction of bFGF with the receptor, while having little effect on that of aFGF. The stimulatory effects of heparin on the neurotropic activity of aFGF, bFGF, and NGF may occur through a site not associated with the respective cellular receptor for the growth factors.     

The Beneficial Effect of Suramin on Monocrotaline-Induced Pulmonary Hypertension in Rats

Background

Pulmonary hypertension (PH) is a progressive disorder characterized by an increase in pulmonary artery pressure and structural changes in the pulmonary vasculature. Several observations indicate that growth factors play a key role in PH by modulating pulmonary artery smooth muscle cell (PA-SMC) function. In rats, established monocrotaline-induced PH (MCT-PH) can be reversed by blocking platelet-derived growth factor receptors (PDGF-R), epidermal growth factor receptors (EGF-R), or fibroblast growth factor receptors (FGF-R). All these receptors belong to the receptor tyrosine kinase (RTK) family.

Methods and Results

We evaluated whether RTK blockade by the nonspecific growth factor inhibitor, suramin, reversed advanced MCT-PH in rats via its effects on growth-factor signaling pathways. We found that suramin inhibited RTK and ERK1/2 phosphorylation in cultured human PA-SMCs. Suramin inhibited PA-SMC proliferation induced by serum, PDGF, FGF2, or EGF in vitro and ex vivo. Treatment with suramin from day 1 to day 21 after monocrotaline injection attenuated PH development, as shown by lower values for pulmonary artery pressure, right ventricular hypertrophy, and distal vessel muscularization on day 21 compared to control rats. Treatment with suramin from day 21 to day 42 after monocrotaline injection reversed established PH, thereby normalizing the pulmonary artery pressure values and vessel structure. Suramin treatment suppressed PA-SMC proliferation and attenuated both the inflammatory response and the deposition of collagen.

Conclusions

RTK blockade by suramin can prevent MCT-PH and reverse established MCT-PH in rats. This study suggests that an anti-RTK strategy that targets multiple RTKs could be useful in the treatment of pulmonary hypertension.     

Mechanisms of 4-hydroxytamoxifen anti-growth factor activity in breast cancer cells: alterations of growth factor receptor binding sites and tyrosine kinase activity

We previously demonstrated that antiestrogen 4-hydroxytamoxifen (OH-Tam) blocks the mitogenic activity of growth factors in breast cancer. We now investigate this mechanism by evaluating how OH-Tam affects growth factor binding and receptor tyrosine kinase activity. We show here that OH-Tam has an opposite effect on epidermal growth factor (EGF) and insulin-like growth factor-1 (IGF-1) binding in estrogen receptor (ER) positive cells. A decrease in IGF-1 binding sites may explain the reduced IGF-I mitogenic effect, whereas an increase in high affinity EGF binding associated with a decrease in in vitro receptor autophosphorylation rather favors the possibility of an alteration in EGF receptor tyrosine kinase activity. We conclude that OH-Tam may prevent growth factor action in ER+ cells both by modulating the concentration of growth factor binding sites and by altering growth factor receptor functionality.     

Effects of acidic fibroblast growth factor and epidermal growth factor on subconfluent fetal rat calvaria cell cultures: DNA synthesis and alkaline phosphatase activity

The effects of acidic fibroblast growth factor (aFGF) and epidermal growth factor (EGF) were examined in subconfluent fetal rat calvaria cell cultures, in the presence of 2% serum. Maximal effect of aFGF and EGF on DNA synthesis measured by [³H]thymidine incorporation was observed after 18 h. aFGF stimulated DNA synthesis by 3.5-fold with an ED₅₀ of 0.75 ng/ml while a 2.3-fold EGF stimulation was recorded with an ED₅₀ of 0.067 ng/ml. 5-Bromo-2-deoxyuridine staining showed a higher stimulation of proliferation in the scattered cells than in the cell clusters. An 18 h aFGF or EGF treatment decreased alkaline phosphatase (ALP) activity by 40 and 23%, respectively, as compared with control cultures. This inhibition was more pronounced after 48 h in the presence of the effectors but no modification of the ALP electrophoretic mobility was observed. These data suggest that aFGF is a less potent mitogen than EGF and a higher inhibitor of ALP activity in fetal rat calvaria cell culture.     

Comparison of the ability of basic and acidic fibroblast growth factor to stimulate the proliferation of an established keratinocyte cell line: modulation of their biological effects by heparin, transforming growth factor beta (TGF beta), and epidermal growth factor (EGF)

The bioactivity of both bFGF and aFGF in the BALB/MK-1 cell line has been compared to that of EGF. Our results indicate that, for that cell type, aFGF was far more potent than bFGF in inducing cell proliferation. In the presence of heparin, aFGF was as potent as EGF. In addition, excess bFGF has an inhibitory effect on the proliferation of MK cells exposed to a saturating concentration of aFGF, therefore acting as a partial agonist of aFGF. Surprisingly, bFGF, although it had low biological activity, was capable of synergizing the effect of EGF. In its presence, cultures exposed to saturating concentration of EGF have a final cell density 3- to 4-fold higher than that of counterpart cultures exposed to EGF alone. TGF beta, which in previous studies has been shown to inhibit the growth of keratinocytes, also inhibited the growth of BALB/MK-1 cells in response to either bFGF or aFGF. These studies suggest a role for FGF in regulating BALB/MK proliferation. aFGF provides positive growth signals which can be negatively modulated by excess bFGF or TGF beta, while bFGF, although a poor mitogen, could act by potentiating the effect of subsaturating concentrations of EGF.     

Design of GFB-111, a platelet-derived growth factor binding molecule with antiangiogenic and anticancer activity against human tumors in mice

We have designed a molecule, GFB-111, that binds to platelet-derived growth factor (PDGF), prevents it from binding to its receptor tyrosine kinase, and blocks PDGF-induced receptor autophosphorylation, activation of Erk1 and Erk2 kinases, and DNA synthesis. GFB-111 is highly potent (IC50 = 250 nM) and selective for PDGF over EGF, IGF-1, aFGF, bFGF, and HRGbeta (IC50 values > 100 microM), but inhibits VEGF-induced Flk-1 tyrosine phosphorylation and Erk1/Erk2 activation with an IC50 of 10 microM. GFB-111 treatment of nude mice bearing human tumors resulted in significant inhibition of tumor growth and angiogenesis. The results demonstrate the feasibility of designing novel growth factor-binding molecules with potent anticancer and antiangiogenic activity.     

Real-time kinetic studies on the interaction of transforming growth factor alpha with the epidermal growth factor receptor extracellular domain reveal a conformational change model

Transforming growth factor alpha (TGF-alpha), epidermal growth factor (EGF), and related factors mediate their biological effects by binding to the extracellular domain of the EGF receptor, which leads to activation of the receptor's cytoplasmic tyrosine kinase activity. Much remains to be determined, however, about the detailed molecular mechanism involved in this ligand-induced receptor activation. The determination of the binding mechanism and the related thermodynamic and kinetic parameters are of prime importance. To do so, we have used a surface plasmon resonance-based biosensor (the BIAcore) that allows the real-time recording of the interaction between TGF-alpha and the extracellular domain of the EGF receptor. By immobilizing different biotinylated derivatives of TGF-alpha on the sensor chip surface, we demonstrated that the N-terminus of TGF-alpha is not directly involved in receptor binding. By optimizing experimental conditions and interpreting the biosensor results by several data analysis methods, we were able to show that the data do not fit a simple binding model. Through global analysis of the data using a numerical integration method, we tested several binding mechanisms for the TGF-alpha/EGF receptor interaction and found that a conformational change model best fits the biosensor data. Our results, combined with other analyses, strongly support a receptor activation mechanism in which ligand binding results in a conformation-driven exposure of a dimerization site on the receptor.     

Structural and thermodynamic analysis of thrombin:suramin interaction in solution and crystal phases

Suramin is a hexasulfonated naphthylurea which has been recently characterized as a non-competitive inhibitor of human alpha-thrombin activity over fibrinogen, although its binding site and mode of interaction with the enzyme remain elusive. Here, we determined two X-ray structure of the thrombin:suramin complex, refined at 2.4 Å resolution. While a single thrombin:suramin complex was found in the asymmetric unit cell of the crystal, some of the crystallographic contacts with symmetrically related molecules are mediated by both the enzyme and the ligand. Molecular dynamics simulations with the 1:1 complex demonstrate a large rearrangement of suramin in the complex, but with the protein scaffold and the more extensive protein–ligand regions keep unchanged. Small-angle X-ray scattering measurements at high micromolar concentration demonstrate a suramin-induced dimerization of the enzyme. These data indicating a dissimilar binding mode in the monomeric and oligomeric states, with a monomeric, 1:1 complex to be more likely to exist at the thrombin physiological, nanomolar concentration range. Collectively, close understanding on the structural basis for interaction is given which might establish a basis for design of suramin analogues targeting thrombin.     

Acidic fibroblast growth factor (HBGF-1) stimulates DNA synthesis in primary rat hepatocyte cultures.

Acidic fibroblast growth factor (aFGF) stimulated DNA synthesis in primary rat hepatocyte cultures in a dose-dependent manner with maximal effect at 10-50 ng ml-1. This activity was dependent on the presence of heparin at a concentration of 10-50 micrograms.ml-1. Insulin interacted synergistically with aFGF, as it did with epidermal growth factor (EGF). The response to aFGF was only 50% that found with EGF. The disparity was not due to different kinetics of DNA synthesis, since the peak response for both growth factors occurred at 36-72 hr after plating of the hepatocytes. The potential relevance of this novel hepatocyte mitogen to normal and pathological liver growth is discussed.     

Suramin inhibits death receptor-induced apoptosis in vitro and fulminant apoptotic liver damage in mice

Suramin is a polysulfonated derivative of urea and has been widely used both to treat infections and as a chemotherapeutic drug. Suramin has been shown to inhibit growth factor signaling pathways; however, its effect on apoptosis is unknown. Here we show that suramin inhibits apoptosis induced through death receptors in hepatoma and lymphoma cells. It also inhibits the proapoptotic effect of chemotherapeutic drugs. The antiapoptotic mechanism is specific to cell type and is caused by reduced activation, but not altered composition, of the death-inducing signaling complex (DISC), and by inhibition of the initiator caspases 8, 9 and 10. Suramin also shows similar effects in in vivo models: apoptotic liver damage induced by CD95 stimulation and endotoxic shock mediated by tumor-necrosis factor (TNF) are inhibited in mice, but necrotic liver damage is not inhibited in a rat model of liver transplantation. Thus, the antiapoptotic property of suramin in the liver may be therapeutically exploited.     

Suramin Induces Phosphorylation of the High-Affinity Nerve Growth Factor Receptor in PC12 Cells and Dorsal Root Ganglion Neurons

Abstract: Suramin is a polysulfonated naphthylurea with demonstrated antineoplastic activity. Toxicity includes adrenal insufficiency and peripheral neuropathy. Although the mechanism of antitumor activity is unknown, inhibition of binding of growth factors to their receptors has been suggested. Growth factors inhibited by suramin include platelet-derived growth factor, fibroblast growth factor, transforming growth factor, epidermal growth factor, insulin-like growth factor, and nerve growth factor (NGF). In these studies, suramin was shown to be cytotoxic to PC12 cells in a dose-dependent manner. At lower doses and in surviving cells, we observed the induction of neurite outgrowth. To determine the mechanism of suramin-induced neurite outgrowth, PC12 cells were exposed to suramin and/or NGF for various time periods and treated cells were analyzed, by western blot analysis, for expression of tyrosine phosphoproteins. There was a similarity in the pattern of tyrosine-phosphorylated proteins in PC12 cells stimulated with suramin or NGF. Of particular interest was the rapid phosphorylation (by 1 min) of the high-affinity NGF (TrkA) receptor. Activation of other members of the signal-transduction cascade (Shc, p21 ^ras , Raf-1, ERK-1) revealed similar phosphorylation levels induced by suramin and NGF. Parallel studies were performed in rat dorsal root ganglion cultures; suramin potentiated neurite outgrowth and activated the NGF receptor on these cells. This finding of specific patterns of tyrosine phosphorylation of cellular proteins in response to suramin treatment demonstrated that suramin is a partial agonist for the NGF receptor in both PC12 cells and dorsal root ganglion neurons.     

Roles of insulinlike growth factor 1 (IGF-1) and the IGF-1 receptor in epidermal growth factor-stimulated growth of 3T3 cells.

BALB/c3T3 cells are exquisitely growth regulated and require platelet-derived growth factor, epidermal growth factor (EGF), and insulinlike growth factor 1 (IGF-1) for growth. When BALB/c3T3 cells are transfected with plasmids constitutively expressing both EGF and the human IGF-1 receptor mRNAs, the cells are capable of growing in serum-free medium without the addition of any exogenous growth factor. These cells, called p5 cells, can grow for prolonged periods in serum-free medium. BALB/c3T3 cells transfected with only the IGF-1 receptor expression plasmid (p6 cells) do not grow in serum-free medium but do grow if IGF-1 (or insulin in supraphysiological concentrations) is added. p6 cells also grow in response to EGF, confirming that the combination of EGF and an overexpressed IGF-1 receptor is sufficient for the growth of 3T3 cells. We have found that in EGF-stimulated p6 cells there is an increase in the expression of IGF-1 mRNA, that IGF-1 is secreted into the medium, and that the growth of p5 cells and EGF-stimulated p6 cells is inhibited by exposure to antisense oligodeoxynucleotides to IGF-1 receptor RNA. Finally, while cells constitutively expressing both EGF and EGF receptor RNAs grow, albeit modestly, in serum-free medium, their growth is also inhibited by an antisense oligodeoxynucleotide to IGF-1 receptor RNA. In contrast, in cells overexpressing the IGF-1 receptor, IGF-1-mediated cell growth occurs independently of the platelet-derived growth factor and EGF receptors (Z. Pietrzkowski, R. Lammers, G. Carpenter, A. M. Soderquist, M. Limardo, P. D. Phillips, A. Ullrich, and R. Baserga, Cell Growth Differ. 3:199-205, 1992, and this paper). These data indicate that an important role for EGF is participation in the activation of an autocrine loop based on the IGF-1-IGF-1 receptor interaction, which is obligatory for the proliferation of 3T3 cells.