Rapid rounding of human epidermoid carcinoma cells A-431 induced by epidermal growth factor

Epidermal growth factor (EGF) induces rapid rounding of A-431 human epidermoid carcinoma cells in Ca(++)-free medium. Cell rounding is not induced by a variety of other polypeptide hormones, antiserum to cell membranes, local anesthetics, colchicine, cytochalasin B, or cyclic nucleotides. However, trypsin, like EGF, induces rounding of A- 431 cells in the absence of Ca(++). Both trypsin- and EGF-induced rounding are temperature dependent, appear to be energy dependent, and are inhibited by cytochalasins, suggesting that the active participation of microfilaments in cell rounding. However, a medium transfer experiment suggests that EGF-induced rounding is not attributable to secretion of a protease, and a number of serine protease inhibitors have no effect on the EGF-induced rounding process. Cell rounding is not attributable to the slight stimulation by EGF of the release of Ca(++) that is observed in the Ca(++)-free medium, as stimulation of such release by the ionophore {"type":"entrez-nucleotide","attrs":{"text":"A23187","term_id":"833253","term_text":"A23187"}}A23187 neither induces cell rounding nor blocks EGF-induced rounding. Cells that have rounded up after treatment with EGF or trypsin spread out upon addition of Ca(++) to the medium, even in the continuing presence of EGF or typsin. Like the cell-rounding process, the cell-spreading process is temperature dependent, appears to be energy dependent, and is inhibited by cytochalasin B. Thus, EGF does not destroy the ability of the cell to spread; rather, in the presence of the EGF (or trypsin), cell spreading and the maintenance of the flattened state become dependent on external Ca(++). Because untreated cells remain flattened in the absence of Ca(++), the data suggest that EGF may disrupt Ca(++)-independent mechanisms of adhesion normally present in A-431 cells.     

Regulation of the epidermal growth factor receptor phosphorylation state by sphingosine in A431 human epidermoid carcinoma cells

The regulation of protein phosphorylation by sphingosine in A431 human epidermoid carcinoma cells was examined. Sphingosine is a competitive inhibitor of phorbol ester binding to protein kinase C (Ca2+/phospholipid-dependent enzyme) and potently inhibits phosphotransferase activity in vitro. Addition of sphingosine to intact A431 cells caused an inhibition of the phorbol ester-stimulated phosphorylation of two protein kinase C substrates, epidermal growth factor (EGF) receptor threonine 654 and transferrin receptor serine 24. We conclude that sphingosine inhibits the activity of protein kinase C in intact A431 cells. However, further experiments demonstrated that sphingosine-treatment of A431 cells resulted in the regulation of the EGF receptor by a mechanism that was independent of protein kinase C. First, sphingosine caused an increase in the threonine phosphorylation of the EGF receptor on a unique tryptic peptide. Second, sphingosine caused an increase in the affinity of the EGF receptor in A431 and in Chinese hamster ovary cells expressing wild-type (Thr654) and mutated (Ala654) EGF receptors. Sphingosine was also observed to cause an increase in the number of EGF-binding sites expressed at the surface of A431 cells. Examination of the time course of sphingosine action demonstrated that the effects on EGF binding were rapid (maximal at 2 mins) and were observed prior to the stimulation of receptor phosphorylation (maximal at 20 mins). We conclude that sphingosine is a potently bioactive molecule that modulates cellular functions by: 1) inhibiting protein kinase C; 2) stimulating a protein kinase C-independent pathway of protein phosphorylation; and 3) increasing the affinity and number of cell surface EGF receptors.     

Cell strains of A431 epidermoid carcinoma with altered epidermal growth factor reception

Epidermal growth factor inhibits proliferation of A431 cells when added to the cultural medium. Strains of A431 cells, resistant to EGF (800 ng/ml), were obtained by one-step selection after the treatment of these cells by 1-methyl-3-nitro-1-nitrosoguanidine (MNNG). Two of the obtained strains differ from the initial line in the EGF reception.     

Rapid enhancement of protein phosphorylation in A-431 cell membrane preparations by epidermal growth factor.

Membranes prepared from A-431 human epidermoid carcinoma cells retained the ability to bind 125I-labeled epidermal growth factor (EGF) in a specific manner. In the presence of [gamma-32P]ATP and Mn2+ or Mg2+, this membrane preparation was capable of phosphorylating endogenous membrane components, including membrane-associated proteins; the major phosphorylated amino acid residue detected in partial acid hydrolysates was phosphothreonine. The binding of EGF to these membranes in vitro resulted in a severalfold stimulation of the phosphorylation reaction; again, the major phosphorylated amino acid residue detected in partial acid hydrolysates was phosphothreonine. Membrane-associated dephosphorylation reactions did not appear to be affected by EGF. The phosphorylation reaction was not stimulated by cyclic AMP or cyclic GMP in the absence or presence of EGF. The phosphorylation system of the membrane was able to utilize [gamma-32P]GTP in both the basal and EGF-stimulated reactions. The enhanced membrane phosphorylation was specific for EGF and its derivatives; a wide variety of other peptide hormones were ineffective. The A-431 membrane preparation also was capable of phosphorylating exogenous proteins, such as histone, phosvitin, and ribonuclease, by a process which was stimulated by EGF. These findings suggest that one of the biochemical consequences of the binding of EGF to membranes is a rapid activation of a cyclic AMP-independent phosphorylating system.     

Relation of epidermal growth factor receptor concentration to growth of human epidermoid carcinoma A431 cells

The relation between the concentration of epidermal growth factor (EGF) receptor/kinase and effects of EGF on cell proliferation has been studied using variant A431 cells and antagonist anti-EGF receptor monoclonal antibodies. Clonal A431 cell variants selected for escape from the EGF-mediated growth inhibition of parental A431 cells all have reduced concentrations of EGF receptor/kinase; Harvey sarcoma virus-transformed A431 cells, which have escaped from EGF-mediated growth inhibition, also have reduced EGF receptors. Three clonal variants which have reacquired EGF-mediated growth inhibition have 2- to 4-fold more EGF receptor than their respective parent variant. A biphasic response with stimulation at low and inhibition at high concentrations of EGF was especially evident in revertants of clone 29. Three separate antagonist monoclonal anti-EGF receptor antibodies block the growth inhibitory effects of EGF and uncover EGF-mediated growth stimulation. These studies indicate that in A431 cell variants a continuum of ligand-activated EGF receptors determines proliferative responses from low concentrations of active receptors under basal conditions to intermediate concentrations causing growth stimulation to high concentrations, causing inhibition of cell proliferation.     

Amlodipine,a Ca2+ channel blocker,suppresses phosphorylation of epidermal growth factor receptor in human epidermoid carcinoma A431 cells

AimsAmlodipine, a dihydropyridine Ca²⁺ channel blocker, inhibits the proliferation of human epidermoid carcinoma A431 cells in vitro and in vivo. This study examined the underlying mechanism of this antiproliferative effect in relation to epidermal growth factor receptor (EGFR) signaling.Main methodsThe tyrosine phosphorylated active state of EGFR in A431 cells incubated with the test agents was evaluated by western blot with anti-phosphotyrosine antibody. EGFR phosphorylation levels in A431 xenograft tumors were assessed by immunostaining of matrigel plug sections and western blotting for phosphoEGFR in A431 xenograft tumor homogenates.Key findingsIn vitro treatment of exponentially growing A431 cells with amlodipine decreased the tyrosine phosphorylation states of EGFR. Amlodipine also suppressed the EGF-stimulated phosphorylation of EGFR and a membrane scaffolding protein, caveolin-1, in serum-starved A431 cells. Amlodipine attenuated the EGF-stimulated phosphorylation of EGFR coimmunoprecipitated with caveolin-1 without affecting the EGFR/caveolin-1 interaction. Crosslinking experiments showed that amlodipine also suppressed the EGF-stimulated phosphorylation of EGFR predimers. Addition of cholesterol abolished these inhibitory effects of amlodipine plus its inhibition of cell growth. Furthermore, treatment of mice with amlodipine (10 mg/kg/day × 7 days, i.p.) decreased the levels of phosphorylated EGFR in A431 xenograft tumors.SignificanceThe results indicated that amlodipine inhibits tyrosine phosphorylation of EGFR in vitro and in vivo, possibly via modulating cholesterol-rich, caveolin-1-containing membrane microdomains.     

Neutrophil elastase induces mucin production by ligand-dependent epidermal growth factor receptor activation

Neutrophil products are implicated in hypersecretory airway diseases. To determine the mechanisms linking a proteolytic effect of human neutrophil elastase (HNE) and mucin overproduction, we examined the effects of HNE on MUC5AC mucin production in human airway epithelial (NCI-H292) cells. Stimulation with HNE for 5-30 min induced MUC5AC production 24 h later, which was prevented by HNE serine active site inhibitors, implicating a proteolytic effect of HNE. MUC5AC induction was preceded by epidermal growth factor receptor (EGFR) tyrosine phosphorylation and was prevented by selective EGFR tyrosine kinase inhibitors, implicating EGFR activation. HNE-induced MUC5AC production was inhibited by a neutralizing transforming growth factor-alpha (TGF-alpha, an EGFR ligand) antibody and by a neutralizing EGFR antibody but not by oxygen free radical scavengers, further implicating TGF-alpha and ligand-dependent EGFR activation in the response. HNE decreased pro-TGF-alpha in NCI-H292 cells and increased TGF-alpha in cell culture supernatant. From these results, we conclude that HNE-induced MUC5AC mucin production occurs via its proteolytic activation of an EGFR signaling cascade involving TGF-alpha.     

Rapid stimulation of pinocytosis in human carcinoma cells A-431 by epidermal growth factor

Horseradish peroxidase (HRP) uptake was used to measure fluid-phase pinocytosis in monolayers of human epithelioid carcinoma cells (A-431). Histochemistry confirmed that cell-associated HRP was restricted to intracellular vesicles. Biochemical methods showed that HRP uptake in control cultures was directly proportional to the duration of exposure. The addition of low concentrations of epidermal growth factor (EGF) to the incubation media produced a 10-fold increase in the initial rate of pinocytosis. The EGF effect was rapid (within 30 s) but transient; the rate of pinocytosis returned to control levels within 15 min. Metabolic inhibitors reduced the EGF-stimulated rate of pinocytosis by greater than 90%. A conjugate of EGF and ferritin (F:EGF) was used to simultaneously compare the intracellular locations of EGF and HRP. Much of F:EGF was internalized in approximately 100-nm vesicles, while most of the HRP was located in much larger vesicles (range 0.1--1.2 micrometer) which also contained F:EGF. The tumor-promoter 12-0-tetradecanoyl-phorbol-13-acetate, which shares several biological activities with EGF, was also effective in stimulating an increase in the rate of pinocytosis.     

Aspects of the metabolism of the epidermal growth factor receptor in A431 human epidermoid carcinoma cells.

The biosynthesis and posttranslational metabolism of the epidermal growth factor (EGF) receptor were examined in the A431 human epidermoid carcinoma cell line. Polyclonal antibody against the receptor specifically immunoprecipitated two [35S]methionine-labeled proteins of Mr = 160,000 and 170,000. Pulse chase experiments showed the Mr = 160,000 protein to be a precursor of the Mr = 170,000 protein. Preincubation with tunicamycin resulted in immunoprecipitation of a single band of Mr = 130,000, whereas monensin inhibited maturation to the Mr = 170,000 form. Digestion of the Mr = 160,000 and 170,000 proteins with endoglycosidase H resulted in the appearance of Mr = 130,000 and 165,000 proteins, respectively. Prolonged pulse-chase experiments indicated that the half-life of the receptor is ca. 20 h in the absence of EGF and 5 h in the presence of EGF. Approximately three- to five-fold more phosphate is incorporated into the mature receptor upon addition of EGF, due primarily to increases in levels of phosphotyrosine and phosphoserine. Phosphate was also present on the Mr = 160,000 protein and the Mr = 130,000 protein found in the presence of tunicamycin.     

Epidermal growth factor and potent phorbol tumor promoters induce epidermal growth factor receptor phosphorylation in a similar but distinctively different manner in human epidermoid carcinoma A431 cells

When human epidermoid carcinoma A431 cells labeled with 32Pi to steady state specific activity were treated either with epidermal growth factor (EGF) or with active phorbol ester tumor promoters such as 12-O-tetradecanoylphorbol 13-acetate (TPA) or phorbol 12,13-dibutyrate, labeling of 160 kDa EGF receptors isolated by immunoprecipitation with monoclonal anti-EGF receptor IgG was increased 2- to 3-fold. These treatments produced no significant increase in 32Pi labeling of acid-precipitable material present in detergent extracts of the cells. Phosphoamino acid analysis of radiolabeled EGF receptors isolated from these cells revealed several differences: the relative abundance of phosphotyrosine in EGF receptors was increased in cells treated with EGF, but decreased in cells treated with TPA; the overall relative abundance of phosphothreonine in EGF receptors was decreased in cells treated with EGF, but remained constant within the limits of experimental detection in cells treated with TPA. Two-dimensional mapping of the radiolabeled phosphopeptides produced from EGF receptors isolated by immunoprecipitation and treated with trypsin revealed 9 independent labeled regions, 2 of which contained phosphothreonine and were present only in EGF- or TPA-treated cells. These two phosphopeptide regions were more highly labeled in cells treated with TPA than with EGF.     

Biosynthesis of the epidermal growth factor receptor in human epidermoid carcinoma-derived A431 cells

Using human-specific antibody reagents, we have examined the biosynthesis of the epidermal growth factor receptor in human epidermoid carcinoma-derived A431 cells. Four Mr species (Mr = 70,000, 95,000, 135,000, and 145,000) are detected when cells are subjected to a brief pulse of L-[35S]methionine; an Mr = 165,000 species is detected after 45-60 min of exposure of cells to radiolabel. In pulse-chase experiments, the four lower Mr species appear to bear a precursor relation to the Mr = 165,000 protein. The molecule acquires N-linked oligosaccharide cotranslationally, and two of the species (Mr = 95,000 and 145,000) are susceptible to digestion with endo-beta-N-acetylglucosaminidase H. The Mr = 145,000 and Mr = 165,000 proteins, which become labeled with 125I-epidermal growth factor after treatment of intact cells with a bifunctional cross-linking reagent, are phosphorylated at serine and threonine on identical tryptic peptides.     

The hydrolysis of cell surface glycosphingolipids by endoglycoceramidase reduces epidermal growth factor receptor phosphorylation in A431 cells

This paper presents a new method to evaluate the biologicalsignificance of glycosphingolipids (GSLs) using a GSL-spedficenzyme, endoglycoceramidase (EGCase), by which GSL-sugar chainsare removed from the cell surface of living cells. In this report,the effects of EGCase on epidermal growth factor (EGF)-dependenttyrosine-specific EGF receptor (EGFR) phosphorylation of A431cells are described. After treatment of A431 cells with EGCaseII (20 mU/ml) in the presence of the activator for 12 h, allacidic GSLs tested were reduced to     

Ceramide stimulates epidermal growth factor receptor phosphorylation in A431 human epidermoid carcinoma cells. Evidence that ceramide may mediate sphingosine action

Recent studies suggest the existence of a signal transduction pathway involving sphingomyelin and derivatives (Kolesnick, R. N. (1989) J. Biol. Chem. 264, 7617-7623). The present studies compare effects of ceramide, sphingosine, and N,N-dimethylsphingosine on epidermal growth factor (EGF) receptor phosphorylation in A431 human epidermoid carcinoma cells. To increase ceramide solubility, a ceramide containing octanoic acid at the second position (C8-cer) was synthesized. C8-cer induced time- and concentration-dependent EGF receptor phosphorylation. This event was detectable by 2 min and maximal by 10 min. As little as 0.1 microM C8-cer was effective, and 3 microM C8-cer induced maximal phosphorylation to 1.9-fold of control. EGF (20 nM) increased phosphorylation to 2.1-fold of control. Sphingosine stimulated receptor phosphorylation over the same concentration range (0.03-3 microM) and to the same extent (1.8-fold of control) as ceramide. The effects of C8-cer and sphingosine were similar by three separate criteria, phosphoamino acid analysis, anti-phosphotyrosine antibody immunoblotting, and phosphopeptide mapping by high performance liquid chromatography. Phosphorylation occurred specifically on threonine residues. N,N-Dimethylsphingosine, a potential derivative of sphingosine, was less effective. Since sphingosine and ceramide are interconvertible, the level of each compound was measured under conditions sufficient for EGF receptor phosphorylation. C8-cer (0.1-1 microM) induced dose-responsive elevation of cellular ceramide from 132 to 232 pmol.10(6) cells-1. In contrast, cellular sphingosine levels did not rise. This suggests that C8-cer acts without conversion to sphingosine. Exogenous sphingosine (0.1-1 microM) also increased cellular ceramide levels to 227 pmol.10(6) cells-1, but did not increase its own cellular level of 12 pmol.10(6) cells-1. Higher sphingosine concentrations that induced no further increase in EGF receptor phosphorylation produced very large elevations in cellular sphingosine. Hence, at effective concentrations, both compounds elevated cellular ceramide but not sphingosine levels. Additional studies performed with [3H]sphingosine demonstrated that cells contain substantially less N,N-dimethylsphingosine than free sphingosine and, during short term incubation, convert less than 5% of added sphingosine to N,N-dimethylsphingosine. These studies provide evidence that ceramide may have bioeffector properties and suggest sphingosine may act in part by conversion to ceramide.     

Differential tyrosine phosphorylation of phospholipase D isozymes by hydrogen peroxide and the epidermal growth factor in A431 epidermoid carcinoma cells.

The regulatory mechanism through which the phospholipase D (PLD) isoforms PLD1 and PLD2 are activated is poorly understood. We investigated the possibility that the PLD isozymes are differentially regulated in response to pharmacologic stimulants in cells. In this report, we demonstrate for the first time that H2O2 and EGF differentially induce tyrosine phosphorylation of the PLD isozymes in A431 cells, which express both PLD1 and PLD2. H2O2 induced tyrosine phosphorylation of PLD1 and PLD2, whereas EGF only caused the tyrosine phosphorylation of PLD2. Both agents also induced phosphorylation of the EGF receptor. Interestingly, the PLD isozymes were associated with the EGF receptor and PKC-alpha in a ligand independent manner. Activation of PLD by H2O2 and EGF nearly correlated with tyrosine phosphorylation of the protein in PLD1 immune complexes. Activation of PLD by both agents was inhibited by the PKC inhibitor, Ro 31-8220, and by the down-regulation of PKC. Pretreatment of the cells with the tyrosine kinase inhibitor tyrphostin AG1478 resulted in inhibition of the H2O2 and EGF-induced tyrosine phosphorylation and PLD activation. These results indicate that H2O2 and EGF induce differential tyrosine phosphorylation of PLD isozymes. Also, the activation of PLD by these agonists involves tyrosine phosphorylation and PKC activation.     

Presence of mannose phosphate on the epidermal growth factor receptor in A-431 cells

In this report, we demonstrate a novel post-translational modification of the epidermal growth factor (EGF) receptor. This modification involves the presence of phosphate, previously thought to exist only on amino acid residues in the EGF receptor, on oligosaccharides of the receptor. We have utilized several independent approaches to determine that mannose phosphate is present on the EGF receptor in A-431 cells. Following metabolic labeling with 32P, immunoisolation of the EGF receptor, and digestion with Pronase radioactivity was determined to be present on high mannose type oligosaccharides by concanavalin A chromatography. Also, after acid hydrolysis of in vivo 32P-labeled EGF receptor, radioactivity was detected that co-migrated with mannose 6-phosphate on two-dimensional thin layer electrophoresis. This radiolabeled material co-eluted with a mannose 6-phosphate standard from a high pressure liquid chromatography anion exchange column. Last, an acid hydrolysate of [3H]mannose-labeled EGF receptor contained two radiolabeled fractions, as analyzed by thin layer electrophoresis, and the radioactivity in one of these fractions was substantially reduced by alkaline phosphatase treatment prior to electrophoresis. These experiments indicate that the mature EGF receptor in A-431 cells contains mannose phosphate. This is a novel modification for membrane receptors and has only been reported previously for lysosomal enzymes and a few secreted proteins.     

Similarities in glycosylation and transport between the secreted and plasma membrane forms of the epidermal growth factor receptor in A-431 cells

We have studied the synthesis and oligosaccharide processing of the 110,000 dalton form of the epidermal growth factor (EGF) receptor that is secreted into the medium of A-431 cells. Its 90,000 dalton precursor is soluble within the lumen of intracellular membrane vesicles shortly after synthesis, indicating that it lacks a membrane anchor. Analysis of labeled glycopeptides reveals that the glycosylation of the 110,000 dalton, secreted receptor is very similar to that of the 170,000 dalton, plasma membrane receptor. Based on Concanavalin A-Sepharose elution profiles of its glycopeptides, the secreted receptor has both complex and high-mannose N-linked oligosaccharides. Also, like the plasma membrane receptor, the secreted receptor contains N-acetylgalactosamine residues in its complex chains. Not only are major features of oligosaccharide processing of the soluble and membrane-bound forms of the receptor similar, but the kinetics of transport to the cell exterior is the same for each. These data indicate that the glycosylation pattern and kinetics of cellular transport of the EGF receptor are determined by factors other than the sequence of its cytoplasmic and transmembrane domains.     

Characterization of the metabolic turnover of epidermal growth factor receptor protein in A-431 cells

The metabolism of the receptor for epidermal growth factor (EGF) in A-431 cells has been measured by labeling the receptor in vivo with radioactive amino acid precursors and then determining, by immunoprecipitation with specific anti-EGF receptor antisera, the rate of degradation of the receptor when the cells are placed in a nonradioactive medium. The rate of EGF receptor degradation (t1/2 = 20 hr) was faster than the rate of degradation of total cell protein (t1/2 = 52 hr). When EGF was added at the beginning of the chase, the half-life of prelabeled receptor decreased to 8.9 hr. This decrease was specific, as the level of total cellular protein and another plasma membrane protein, the transferrin receptor, were relatively unaffected by EGF. The carbohydrate portion of the receptor is degraded, in the presence or absence of EGF, at approximately the same rate as the protein moiety. The amount of EGF receptor protein in A-431 cells has been quantitated by radiolabeling total cellular protein and quantitating the immunoprecipitable receptor. The EGF receptor constitutes approximately 0.15% of the total cell protein in A-431 cells. These cells, therefore, have approximately 30 times more EGF receptor protein than fibroblasts. The EGF receptor constitutes an even higher proportion of 3H-glucosamine- or 3H-mannose-labeled macromolecules in A-431 cells, 1.5% or 5.2%, respectively. The EGF receptor from A-431 cells can easily be identified by submitting carbohydrate-labeled, solubilized cells to electrophoresis as described by Laemmli (1970).     

Activation of the calcium channels of A-431 cells by epidermal growth factor

The cell-attached version of patch-clamp technique was used to search for calcium permeable channels in human carcinoma A-431 cells. With 100 mM CaCl2 in pipette, the inward currents were recorded having the mean unitary conductance of 2.8 pS and the reversal potential (obtained by linear extrapolation) equal to +25.5 mV. Application of the epidermal growth factor (EGF) into the bath extracellular solution produced a transient increase in probability for these channels to be open. The effect developed with a delay of about 20 seconds to last thereafter for 36 seconds (mean values). We propose that these channels mediate EGF-induced increase in concentration of cytosolic free calcium.     

Regulation of casein kinase II activity by epidermal growth factor in human A-431 carcinoma cells

In order to characterize more fully the mechanism by which casein kinase II is regulated in mammalian cells, the effect of epidermal growth factor (EGF) on the activity of the kinase in human A-431 carcinoma cells was examined. Treatment of cells with EGF prior to lysis consistently resulted in a transient 4-fold increase in the activity of cytosolic casein kinase II. Activity rose sharply between 20 and 30 min, peaked at approximately 50 min, and returned to basal levels by approximately 120 min. Similar results were obtained using the casein kinase II specific peptide substrate, Arg-Arg-Arg-Glu-Glu-Glu-Thr-Glu-Glu-Glu, or DNA topoisomerase II (which is specifically modified by the kinase in vivo and serves as a high affinity substrate in vitro) as the phosphate acceptor in assays. Identification of casein kinase II as the stimulated activity was confirmed by partial proteolytic mapping and phosphoamino acid analysis of modified topoisomerase II, by inhibition at nanomolar levels of heparin or micromolar levels of nonradioactive GTP, and by the ability to employ radioactive GTP as a direct phosphate donor. The EGF stimulation of casein kinase II was dependent on the availability of intracellular (but not extracellular) calcium. In addition, hormonal action was modulated by calcium/phospholipid-dependent protein kinase (protein kinase C). Casein kinase II stimulation did not require an increase in the concentration of the kinase, protein synthesis, the continual presence of a small effector molecule, or a direct interaction with the EGF receptor/tyrosine kinase. In contrast, hormonal activation of the kinase was dependent on the phosphorylation of casein kinase II or a terminal stimulatory factor.