Pituitary follicular cells secrete a novel heparin-binding growth factor specific for vascular endothelial cells

A growth factor for vascular endothelial cells was identified in the media conditioned by bovine pituitary follicular cells and purified to homogeneity by a combination of ammonium sulfate precipitation, heparin-sepharose affinity chromatography and two reversed phase HPLC steps. The growth factor was a cationic, heat stable and relatively acid stable protein and had a molecular weight, as assessed by silver-stained SDS-PAGE gel, of approximately 45,000 under non reducing conditions and approximately 23,000 under reducing conditions. The purified growth factor had a maximal mitogenic effect on adrenal cortex-derived capillary endothelial cells at the concentration of 1-1.2 ng/ml (22-26 pM). Further characterization of the bioactivity of the growth factor reveals that it exerts mitogenic effects also on vascular endothelial cells isolated from several districts but not on adrenal cortex cells, lens epithelial cells, corneal endothelial cells, keratynocytes or BHK-21 fibroblasts, indicating that its target cells specificity is unlike that of any previously characterized growth factor. Microsequencing reveals a unique N-terminal amino acid sequence. On the basis of its apparent target cell selectivity, we propose to name this factor vascular endothelial growth factor (VEGF).     

Homocysteine is positively associated with cytokine IL-18 plasma levels in coronary artery bypass surgery patients

Homocysteine, cytokines (IL-18, IL-6, IL-8) are involved in vascular inflammation and coronary artery disease. Homocysteine influences endothelial IL-6 and IL-8 cytokine expression and release, however, an association between homocysteine and IL-18 has not been previously investigated in endothelial/smooth muscle cells and or in coronary artery disease. We report in 9 coronary artery bypass surgery (CABG) patients a positive correlation r = 0.86 between homocysteine and IL-18 plasma levels (p < 0.05). Plasma IL-18 levels are significantly higher in those patients with elevated homocysteine compared to those with normal levels (p < 0.02; 153 +/- 19 pg/ml versus 116 +/- 14 pg/ml respectively). Our in vitro cell culture studies suggest that the source of IL-18 in CABG patients with elevated homocysteine is not from vascular smooth muscle or endothelial cells.     

Isolation and characterization of a newly identified endothelial cell mitogen produced by AtT-20 cells.

An endothelial cell growth factor with unique specificity for vascular endothelial cells has been purified from the conditioned medium of the AtT-20 pituitary cell line. This growth factor, which has been characterized as a homodimer composed of two subunits with mol. wts of 23 kd is a potent mitogen for vascular endothelial cells in vitro with activity detectable at 50 pg/ml and saturation at 1 ng/ml. It was also angiogenic in vivo. In contrast with other endothelial mitogens of the fibroblast growth factor family, it has a unique target cell specificity. It did not stimulate the growth of other cell types of the vascular system such as vascular smooth muscle cells or that of mesoderm and neuroectoderm derived cells. Microsequencing revealed an amino-terminal sequence with no homology to any known protein. The release of this novel endothelial cell growth factor by pituitary derived cells and its unique target cell specificity suggest that it could play an important role in the angiogenic process.     

Characterization of a hepatic proliferation inhibitor (HPI): effect of HPI on the growth of normal liver cells--comparison with transforming growth factor beta

Improvements in the purification of a hepatic proliferation inhibitor (HPI) from adult rat liver have yielded a product that has an inhibitory activity 1,000-fold greater than previously reported. The growth inhibitory activity, which could be eluted from SDS-PAGE at 17-19 kilodaltons (kD), was compared to that of transforming growth factor beta (TGF-beta). The ID50 of the HPI preparation in Fischer rat liver epithelial cells was 50 pg/ml (2.5 pM) compared to a value of 260 pg/ml (10.4 pM) obtained for pure human TGF-beta. Both inhibitors also modulated the stimulation of DNA synthesis in primary hepatocytes by either epidermal growth factor or a growth stimulatory activity prepared from serum of hepatectomized rats. The ID50s of HPI and TGF-beta in these cells were 250 pg/ml and 40 pg/ml, respectively. In contrast to TGF-beta the growth inhibitory activity of HPI was unaltered in the presence of an antibody raised against TGF-beta. The possible mechanism of action of HPI is discussed.     

Specific binding of vasculotropin to bovine brain capillary endothelial cells

Recently, a new growth factor was purified to homogeneity; its biological activity appeared to be restricted to vascular endothelial derived cells. As it was also angiogenic in vivo, it was provisionally named vasculotropin. An iodination procedure used to label vasculotropin did not damage the molecule; it was thus possible to undertake binding studies. The binding of iodinated vasculotropin to bovine brain capillary endothelial cells reached saturation at 7 ng/ml and half maximal binding occurred at 1.5 ng/ml. Scatchard analysis of the data demonstrated 2 classes of binding sites with apparent dissociation constants of 4 and 41 pM and 600 and 4,100 sites per cell respectively. The interaction was specific since an excess of unlabelled vasculotropin, but no Fibroblast Growth Factor or Transforming Growth Factor Beta almost totally abolished the binding of the tracer. A sensitive radioreceptor-assay convenient for measuring vasculotropin in biological samples is described.     

Control of proliferation of bovine vascular endothelial cells.

The effects of Fibroblast Growth Factor (FGF) and Epidermal Growth Factor (EGF) on the proliferation of bovine vascular endothelial cells has been examined. FGF induces the initiation of DNA synthesis and cell proliferation in cloned endothelial cells of fetal and adult origin at concentrations as low as 1 ng/ml and is saturating at 50 ng/ml. EGF had no effect over the same range of concentrations. The mitogenic effect of FGF is blocked by a crude extract of cartilage. Platelet extract is also mitogenic for vascular endothelial cells although to a lesser extent than the purified FGF. In contrast to vascular endothelial cells, both EGF and FGF are mitogenic for vascular smooth muscle cells although EGF is less mitogenic than FGF at 100 ng/ml. The mitogenic effect of EGF and FGF on vascular smooth muscle is not blocked by the addition of a crude extract of cartilage, thus demonstrating the specificity of the chalone like effect of cartilage crude extract for endothelial cells.     

A possible involvement of ion transporter in tumor necrosis factor alpha and cycloheximide-induced apoptosis of endothelial cells

We examined the tumor necrosis factor alpha (TNFalpha)-induced apoptosis of vascular endothelial cells from the standpoint of ion channels. Cultured vascular endothelial cells from bovine carotid artery were used. Apoptosis was determined by a propidium iodide assay. Treatment of the endothelial cells with TNFalpha and cycloheximide for 6 h induced nuclear fragmentation in a TNFalpha dose-dependent manner (1-10 ng/ml). Concomitant treatment of endothelial cells with TNFalpha at a dose of 10 ng/ml and cycloheximide at a dose of 10 microg/ml elicited endothelial cell apoptosis as high as 23.4+/-4.1% at 6 h after administration. However, 10 ng/ml TNFalpha alone elicited a little apoptosis at 6 h after its administration (% apoptosis=4.1+/-0.8%). Cycloheximide (10 microg/ml) did not induce apoptosis at all. Concomitant treatment of endothelial cells with 1 mmol/l of 4,4-diisothiocyanatostilbene-2,2-disulfonic acid, which is a chloride bicarbonate exchanger blocker, partially inhibited the TNFalpha and cycloheximide-induced endothelial cell apoptosis. On the other hand, endothelial cell apoptosis due to TNFalpha and cycloheximide was completely inhibited by benzyloxycarbonyl-Asp-CH2OC(O)-2,6-dichlorobenzene (50 micromol/l), an inhibitor of caspase. Moreover, pyrrolidine dithiocarbanate, an inhibitor of nuclear factor kappa B (NF-kappaB), also suppressed endothelial cell apoptosis induced by TNFalpha and cycloheximide completely. These findings suggest that the endothelial cell apoptosis induced by TNFalpha and cycloheximide is closely related to not only chloride ions, but also both NF-kappaB and caspase activation. That is to say, there is a possibility that chloride ions or bicarbonate (pH) may play an important role in signal transduction such as NF-kappaB and caspase activation in the apoptosis induced by TNFalpha and cycloheximide.     

Bovine brain and pituitary fibroblast growth factors: comparison of their abilities to support the proliferation of human and bovine vascular endothelial cells

The mitogenic effects of brain and pituitary fibroblast growth factors (FGF) on vascular endothelial cells derived from either human umbilical vein or bovine aortic arch have been compared. Both brain and pituitary FGF are mitogenic for low density human umbilical endothelial (HUE) cell cultures maintained on either fibronectin- or laminin-coated dishes or on biomatrices produced by cultured cells such as bovine corneal endothelial cells or the teratocarcinoma cell line PF-HR-9. Pituitary FGF triggered the proliferation of HUE cells at concentrations as low as 0.25 ng/ml, with a half-maximal response at 0.55 ng/ml and optimal effect at 2.5 to 5 ng/ml. It was 50,000-fold more potent than commercial preparations of endothelial cell growth factor and 40 times more potent than commercial preparations of pituitary FGF. Similar results were observed when the effect of pituitary FGF was tested on low density cultures of adult bovine aortic endothelial cells. When the activity of brain and pituitary FGF on low density HUE cell cultures was compared, both mitogens were active. To confirm the presence in brain extract of both acidic and neutral, as well as of basic mitogen, for HUE cells, brain tissues were extracted at acidic (4.5), neutral (7.2), and basic (8.5) pH. The three types of extracts were equally potent in supporting the proliferation of either HUE or adult bovine aortic endothelial cells. When the various extracts were absorbed at pH 6.0 on a carboxymethyl Sephadex C-50 column, the neutral and basic extracts had an activity after adsorption similar to that of unadsorbed extracts. In contrast, extracts prepared at pH 4.5 lost 90-95% of their activity which was recovered in the adsorbed fraction containing FGF.     

Specific binding of vascular permeability factor to endothelial cells

Vascular permeability factor (VPF), also known as vascular endothelial cell growth factor, has recently been purified from guinea pig, human, and bovine sources. We show that various fetal or adult endothelial cell strains originating from either capillary or large vessels possess specific high affinity and saturable binding sites for guinea pig tumor-derived [125I]VPF. Two classes of sites with KDs of approximately 10 pM and 1 nM were detected for all endothelial cell types examined. Guinea pig [125I]VPF binding to endothelial cells was inhibited by human VPF (ID50 = 0.8 ng/ml) and by suramin (ID50 = 75 micrograms/ml) but not by heparin. Cross-linking experiments revealed specific [125I]VPF-receptor complexes of two types. Most of the complexes migrated very slowing in SDS-PAGE, indicating that they were of very high molecular weight and probably highly cross-linked. A portion of the molecules migrated as 270 kDa complexes, indicating that the molecular weight of the endothelial cell VPF receptor is about 230 kDa.     

Isoform of Vascular Endothelial Cell Growth Inhibitor （VEGI72-251） Increases Interleukin-2 Production by Activation of T Lymphocytes

The objective of this study is to investigate the characteristics of the recombinant variant ofhuman vascular endothelial cell growth inhibitor,VEGI_(72-251),and compare its biological activities with that ofits prototype VEGI_(24-174),The recombinant plasmid containing the variant VEGI_(72-251) gene was constructedand expressed in Escherichia coli.The effects of the expressed VEGI_(72-251) on cell proliferations were checkedin the human umbilical vein endothelial cell line and certain tumor cell lines (ECV304 and B 16).The inhibitionof VEGI_(72-251) on angiogenesis was detected in the chorioallantoic membrane of chick embryos.In comparisonwith VEGI_(24-174),the recombinant human VEGI_(72-251) seems to have no effect on the proliferation of endothelialcells and the angiogenesis of the chorioallantoic membrane in vitro.An enzyme-linked immunosorbent assay-based method was used for the measurement of interleukin-2 (IL-2) production by peripheral blood monocytes(PBMCs) treated with VEGI_(72-251).PBMCs were pretreated with VEGI_(72-251) (1.25-12.50μg/ml) for 24 h invitro,and the IL-2 concentration in PBMC medium was increased from 354 pg/ml to 1256 pg/ml.It can beconcluded that VEGI_(72-251) is able to increase the level of human IL-2 production by the activation of Tlymphocytes.Differing from VEGI_(24-174) on anti-angiogenesis,VEGI_(72-251) may serve as an anti-cancer factorthrough its activation of T lymphocytes.     

The characterization of radioimmunoassay for rat pancreatic polypeptide in serum.

A radioimmunoassay for the measurement of rat pancreatic polypeptide (RPP) in serum or plasma has been developed and characterized using a new guinea-pig anti-rat-PP antibody. The assay provides a high degree of sensitivity and lacks cross-reactivity (CR less than 0.01%) to neuropeptide Y and peptide YY. It also does not interact with PPs of other species or peptide hormones namely, amylin, glucagon, human insulin, human-PP, human-proinsulin, rat C-peptide and rat insulin. The assay employs synthetic rat PP as standards from concentrations of 21-2100 pg/ml (i.e., 5-500 pM) and produces a sensitivity limit of 19 pg/ml (4.5 pM) PP at +/- 3 S.D. The intra- and interassay % coefficient of variations are 6.4% and 5.9%, respectively. The % recovery of RPP added to rat serum samples ranges from 98% to 103%. Assay of serum volumes ranging from 25 microliters to 100 microliters does not significantly alter the expected RPP level. The migration patterns of rat serum PP and that of a synthetic RPP are identical by Sephadex G-50 chromatographic analysis. The mean values of fasting and a 2 h post-feeding plasma RPP levels in normal rats are 40 +/- 2 and 80 +/- 10 pg/ml (9.5 pM and 19.0 pM), respectively. Rat-PP release during insulin induced hypoglycemia in conscious rats rises from 38 +/- 5 pg/ml to 261 +/- 34 pg/ml (9.0 to 62.1 pM, P less than 0.005) by 30 min. Additionally, the antibody used in this study cross-reacts well with mouse-PP as determined by linear serum dilution curves, thus making it useful in the measurement of murine-PP. In conclusion, we have developed and validated a sensitive and specific rat-PP assay. This assay provides a new tool for the reliable measurement of PP in physiologic studies using rat and mouse animal models.     

Bradykinin-induced release of prostacyclin and thromboxanes from bovine pulmonary artery endothelial cells. Studies with lower homologs and calcium antagonists

Bovine pulmonary artery endothelial cells, in serum-free culture medium, release small quantities of prostacyclin and thromboxane A2 (3-10 and 0.1-0.3 ng/ml; measured as immunoreactive 6-ketoprostaglandin F1 alpha and thromboxane B2, respectively). The release of these substances is stimulated by up to 20-fold during a 3 min incubation with the vasodilator, bradykinin (Arg1-Pro2-Pro3-Gly4-Phe5-Ser6-Pro7-Phe8-Arg9). Endothelial cells incubated with [3H]arachidonic acid for 24 h and then exposed to bradykinin for 3 min release 3H into the medium, approximately 65% of which co-chromatographs with 6-ketoprostaglandin F1 alpha and 3% with thromboxane B2. The effects of bradykinin are dose-related and are often discernible when the hormone is used at concentrations believed to occur physiologically (10 pg/ml; approximately 10 pM). Furthermore, the bradykinin molecule must be intact: none of its lower homologs affects the release of prostacyclin, thromboxane A2, or 3H unless used at concentrations (1 microM or higher) unlikely to be achieved in vivo. The release appears to involve calcium uptake and calmodulin: it is abolished by EGTA (5 mM) and inhibited by the 'slow channel' calcium antagonists, verapamil and nifedipine (10-100 microM), and by the calmodulin inhibitor, trifluoperazine (3-30 microM). Our findings suggest that bradykinin exerts some of its hormonal effects by acting on specific receptors possessed by vascular endothelial cells; receptor activation is associated with calcium transport, arachidonate mobilization, and a selective synthesis of prostacyclin, a vasodilator in its own right.     

Determination of immunoreactive endothelin in medium from cultured endothelial cells and human plasma

We have developed a sensitive and selective radioimmunoassay for porcine/human endothelin (ET1). The assay has a detection limit of 0.62 pg/tube and exhibits no cross-reactivity to atrial natriuretic peptide, arginine vasopressin, or angiotensin II. Procedures were developed for extraction of endothelin from human plasma samples and samples of buffer from endothelial cell incubations using C18 Sep-Pak extraction cartridges. The mean recovery following extraction was approximately 80%. Both bovine and porcine aortic endothelial cells were found to produce immunoreactive endothelin (IR-ET) with porcine cells producing 4.7 +/- 1.1 ng of IR-ET/mg cell protein after 6 hours. Human plasma samples were extracted, assayed and found to contain a mean concentration of 2.0 +/- 0.4 pg/ml of IR-ET.     

Responses of vascular endothelial oxidant metabolism to lipopolysaccharide and tumor necrosis factor-alpha.

Quantification of intracellular and extracellular levels and production rates of reactive oxygen species is crucial to understanding their contribution to tissue pathophysiology. We measured basal rates of oxidant production and the activity of xanthine oxidase, proposed to be a key source of O2- and H2O2, in endothelial cells. Then we examined the influence of tumor necrosis factor-alpha and lipopolysaccharide on endothelial cell oxidant metabolism, in response to the proposal that these inflammatory mediators initiate vascular injury in part by stimulating endothelial xanthine oxidase-mediated production of O2- and H2O2. We determined a basal intracellular H2O2 concentration of 32.8 +/- 10.7 pM in cultured bovine aortic endothelial cells by kinetic analysis of aminotriazole-mediated inactivation of endogenous catalase. Catalase activity was 5.72 +/- 1.61 U/mg cell protein and glutathione peroxidase activity was much lower, 8.13 +/- 3.79 mU/mg protein. Only 0.48 +/- 0.18% of total glucose metabolism occurred via the pentose phosphate pathway. The rate of extracellular H2O2 release was 75 +/- 12 pmol.min-1.mg cell protein-1. Intracellular xanthine dehydrogenase/oxidase activity determined by pterin oxidation was 2.32 +/- 0.75 microU/mg with 47.1 +/- 11.7% in the oxidase form. Intracellular purine levels of 1.19 +/- 1.04 nmol hypoxanthine/mg protein, 0.13 +/- 0.17 nmol xanthine/mg protein, and undetectable uric acid were consistent with a low activity of xanthine dehydrogenase/oxidase. Exposure of endothelial cells to 1000 U/ml tumor necrosis factor (TNF) or 1 microgram/ml lipopolysaccharide (LPS) for 1-12 h did not alter basal endothelial cell oxidant production or xanthine dehydrogenase/oxidase activity. These results do not support a casual role for H2O2 in the direct endothelial toxicity of TNF and LPS.     

Endotoxin stimulates endothelin-release in vivo and in vitro as determined by radioimmunoassay

A marked increase in immunoreactive endothelin was observed in rat serum collected within 10-15 min after infusion of endotoxin. Endothelin level was 117 +/- 11.5 pg/ml (mean +/- S.E., N = 4) in rats exposed to endotoxin as compared with undetectable levels (less than 2 pg/ml, N = 4) in controls. We have also observed a significant stimulation of endothelin-release by endotoxin from cultured bovine transformed thoractic aortic endothelial cells at concentrations of endotoxin ranging between 0.1 and 10.0 micrograms/ml. Serum was indispensable for the stimulating effect of endotoxin, although serum itself did not show any effect at the concentration used (1%). These results suggest that endothelin plays an important role in mediation of pathophysiological responses caused by endotoxin. The levels of endothelin were measured by radioimmunoassay with high sensitivity.     

Plasma concentrations of endothelin in patients with abnormal vascular reactivity. Effects of ergometric exercise and acute saline loading

We measured circulating concentrations of endothelin, a recently discovered vasoconstrictor peptide produced by vascular endothelial cells, in healthy subjects and in patients with abnormal vascular reactivity. Endothelin concentrations were determined by radio-immunoassay after extraction of plasma using Sep-Pak C-18 cartridges in healthy subjects (n = 20), in patients with diabetes mellitus type I (n = 10), in patients with mild to moderate essential hypertension (n = 12) and in non-dialyzed patients with stable chronic renal failure (n = 12). Plasma concentrations were similar in healthy controls, in diabetics and in hypertensive patients averaging 5.0 +/- 0.6 pg/ml, 4.7 +/- 0.2 pg/ml and 6.5 +/- 1.0 pg/ml, respectively. In contrast, plasma concentrations of endothelin were markedly elevated in patients with chronic renal failure averaging 16.6 +/- 2.9 pg/ml (p less than 0.005). No correlations were observed between serum creatinine concentrations ranging from 124 to 850 mumol/l or blood pressure and plasma concentrations of endothelin. Bicycle ergometric exercise in six healthy subjects and an acute modest i.v. saline load of 1,000 ml of 0.45% NaCl administered within 60 min in six patients with mild essential hypertension did not affect plasma concentrations of endothelin. Thus, it is unlikely that vascular synthesis of endothelin is related to acute physiological changes in systemic hemodynamics or to the circulatory and renal responses to acute extracellular fluid volume (ECFV) expansion. A potential role of endothelin, however, in the control of regional blood flow cannot be excluded. Elevated plasma concentrations of endothelin observed in patients with chronic renal failure require further investigations.     

Characterization of the receptor to vasculotropin on bovine adrenal cortex-derived capillary endothelial cells

Recently a new growth factor was purified to homogeneity, and its bioactivity seemed to be restricted to vascular endothelial derived cells. As it was also angiogenic in vivo, it was provisionally named vasculotropin (VAS). As an iodination procedure used to label VAS did not damage the molecule, it was possible to undertake binding studies. The binding of iodinated vasculotropin to bovine adrenal cortex-derived capillary endothelial cells was saturable at 250 pM, and half-maximal binding occurred at 47 pM. Scatchard's analysis of the data demonstrated two apparent classes of binding sites with apparent dissociation constants of 2 and 82 pM displaying 280 and 3400 binding sites, respectively. The binding was specific; half-displacement was observed with a 2-fold excess of unlabeled VAS. The structurally related platelet-derived growth factor did not compete in a radioreceptor assay. 125I-VAS was displaced by suramin and not by heparin. 125I-VAS was covalently cross-linked to its cell surface receptor on intact bovine adrenal cortex-derived capillary endothelial cells using the homobifunctional agents ethylene glycol bis(succinimidyl succinate) or disuccinimidyl tartarate. A major macromolecular species with an apparent molecular mass of 230,000 Da was labeled under reducing and nonreducing conditions. These data demonstrate the existence of a specific binding protein for VAS and an estimation of the size at 185,000 Da.     

Interleukin-2 enhances the release of endothelin-1 from the rat mesenteric artery

We measured the ET-1 concentration in plasma and in the perfusate of the mesenteric arteries of rats treated with a therapeutic dose of IL-2 for 7 days (100000 U/Kg, iv.). The plasma ET-1 concentration in rats given IL-2 was 14.2 +/- 3.2 pg/ml which was significantly greater than that in the controls (2.5 +/- 0.4 pg/ml, P less than 0.05). The mesenteric arteries also released a significantly greater amount of ET-1 (29.5 +/- 1.6 pg/h) than that in controls (16.8 +/- 2.3 pg/h, P less than 0.01). The arterial blood pressure was significantly lower after IL-2 treatment than the pre-dosing level (P less than 0.05). It is concluded that IL-2 induces ET-1 release from the vascular wall, possibly as a result of reversible endothelial dysfunction caused by IL-2.     

Cell density dependent effects of TGF-beta demonstrated by a plasminogen activator-based assay for TGF-beta.

Transforming growth factor-beta 1 (TGF-beta 1) induces a decrease in plasminogen activator (PA) expression in confluent cultures of bovine aortic endothelial (BAE) cells. We describe an assay using the suppression of PA expression in confluent BAE cells by TGF-beta 1 which detects concentrations of the growth factor ranging from 5 to 200 pg/ml and has an ED50 of 15-20 pg/ml. The assay can be performed in 96-well plates and requires a minimum of 35 ul of solution per sample, thereby limiting the amount of reagents required and allowing many samples to be tested in a single assay. Here we demonstrate that the effect of TGF-beta 1 on PA expression in BAE cells depends on the length of time the cells are exposed to the growth factor and the density at which the cells are plated. In cells plated at a high density (3.5 x 10(5) cells/cm2), both 4 h and 24 h exposures to TGF-beta 1 suppress PA expression. However, with cells plated sparsely (3.5 x 10(4) cells/cm2), a 4 h exposure to TGF-beta 1 increases PA expression 2-fold, whereas a 24 h exposure results in an 85% inhibition of basal PA expression. The paradoxical stimulation of PA expression in cells at a sparse density upon 4 h exposure to TGF-beta 1 occurs in a dose-dependent manner with an ED50 of 15-20 pg/ml. This bifunctional response of PA production in cells exposed to TGF-beta 1 may have implications with regard to the role of TGF-beta 1 in angiogenesis.     

Brain-derived growth factor is a chemoattractant for fibroblasts and astroglial cells

Bovine brain-derived growth factor (BDGF), a 16-17 kDa protein with biochemical properties resembling brain-derived acidic fibroblast growth factor (acidic FGF) and endothelial cell growth factor, was found to have potent chemotactic activity for bovine ligament fibroblasts, human skin fibroblasts and rat astroglial cells, maximal at 100-200 pg/ml. The chemotactic activity was completely blocked by protamine sulfate (5 ug/ml), an inhibitor of receptor-binding and mitogenic activity of BDGF. BDGF did not stimulate migration of human monocytes. These results indicate that the effects of BDGF 'in vivo' might extend to mesenchymal cell recruitment.