A serum factor inducing neurite retraction of morphologically differentiated x glioma NG108-15 cells

A. GHAHARY, P. S. McPHERSON and K. W. CHENG

Department of Physiology, Faculty of Medicine, University of Manitoba, Winnipeg, MB, Canada R3E 0W3

Summary

It is well established that serum plays an important various species studied, including human, cattle, role in cell proliferation and differentiation. In this sheep, rabbit and horse, but not in tissue extracts of study, we have identified a serum factor that in- kidney, heart, lung, skeletal muscle, and brain of duces rapid neurite retraction of morphologically the rat. However, rat spleen and liver homogenates, differentiated NG108-15 cells, cultured in serum- at a protein content of lrngml"1, caused slight free medium containing 1 mM-dibutyryl cyclic neurite retraction. It is noteworthy that NRF is not S AMP. The serum fraction of Mr>30xl0 induces detectable in cerebral spinal fluid. Our data on the neurite retraction in a manner identical to that of properties of serum NRF indicate that it differs the whole serum. The neurite retraction activity in from all of the well-established growth factors, serum appears to be acid- and heat-stable. The namely, NGF, EGF, PDGF, FGF, NSILA, ECGF molecular weight of the serum neurite retraction and TGF. Further studies on purified NRF •will factor (NRF) has been demonstrated to be approxi- delineate the biological role(s) of this serum factor mately 70xl03 by gel permeation on LKB-Ultrogel in the process of maturation and differentiation of AcA-44. The neurite retraction activity is dose- developing neurones. dependent, and the time required for half-maximal activity (fj) is 1*8 min. NRF is present in sera of Key words: neuroblastoma, neurite retraction, serum factor.

Introduction Materials and methods

Neuroblastoma, one of the most malignant solid tumours The neuroblastoma X glioma hybrid NG108-15 cell line was a in children, arises from neuroblasts that fail to differen- gift from Dr B. Schrier, Laboratory of Developmental Neuro- tiate and that undergo malignant changes (Robinson, biology, National Institues of Health, Bethesda, MD, USA. 1980). Neuroblastoma cells have the potential to undergo Tissue culture supplies, including Dulbecco's modified Eagle's morphological changes in vitro upon treatment with medium-high glucose (DMEM), foetal calf serum (FCS), neurite-enhancing agents. Removal of serum from cul- penicillin-streptomycin, L-glutamate and nutrient mixture ture media (Harkins et al. 1972), or addition of bromo- F-12 (HAM) were purchased from GIBCO, Canada. Insulin, transferrin, oleic acid, plasmin, plasminogen activator and deoxyuridine (BrdUrd) (Schubert & Jacob, 1970), di- other reagents were obtained from Sigma Co., St Louis, USA. methylsulphoxide (DMSO) (GYicketal. 1980), dibutyryl Centricon and immersible filters were from Amicon and Milli- cyclic AMP (dBcAMP) (Prasad, 1975), and inhibitors of pore Co., respectively. NGF and PDGF were kindly supplied cyclic AMP phosphodiesterase (Prasad & Sheppard, by Dr L. Murphy, Department of Physiology, School of 1972), among others, has been demonstrated to induce Medicine, University of Manitoba, Winnipeg, Manitoba, morphological differentiation of neuroblastoma, with a Canada. concomitant decrease in proliferation. Although many agents and conditions have been shown to stimulate morphological differentiation of Hybrid NG108-15 cells were maintained in DMEM, containing neuroblastoma cells, little information is available on 6 4 biochemical factors or agents that prevent neuroblastoma 10% FCS, lXlO~ M-hypoxanthine, lXlO~ M-aminopterine, lXlO~SM-thymidine, 100 units penicillin ml"1, 100 ;Ug strepto- differentiation. In this study, we have identified a serum mycin ml~ and 2mM-L-glutamate, at 37°C in a humidified factor that induces neurite retraction of morphologically atmosphere of 10% CC>2-90% air. Morphological differen- differentiated neuroblastoma X glioma hybrid NG108-15 tiation of NG108-15 cells was induced by subculturing the cells cells. This activity appears to be present in serum but not (5 XlO4 cells/well) in the presence of 1 mM-dBcAMP in defined in other tissues, including cerebral spinal fluid. serum-free medium, consisting of 75% DMEM, and 25% Journal of Cell Science 92, 251-256 (1989) Printed in Great Britain © The Company of Biologists Limited 1989 251 nutrient mixture F-12 (HAM), supplemented with insulin majority of the morphologically differentiated cells bore (2S,ugml~ ), transferrin (50[igm\~ ) and oleic acid more than two long neurites upon treatment with 1 1 (10,ug 1 mg" albuminml" ) as described by Wolfe & Sato dBcAMP for 48-72 h (Fig. 1 A). Addition of FCS caused (1982). rapid neurite retraction of morphologically differentiated NG108-15 cells (Fig. 1B-E). After 1 min, retraction of Bioassay of neurite retraction neurites became evident (Fig. IB); after 2min, cells Samples of serum, CSF or extracts of tissues were added to the retracted their neurites to approximately 50 % of their differentiated NG108-15 cells in serum-free medium, contain- original length (Fig. 1C); after 4min., only short rema- ing 1 mM-dBcAMP. Neurite retraction was monitored by nent neurites were observed in some cells (Fig. ID); and measuring neurite length in photographs taken at various time intervals, within 2 min after addition of test sample. One unit of after 8 min, almost all cells had their neurites retracted activity is assigned to be the amount of activity that causes (Fig. IE). In the presence of added 10% serum, the neurites to shorten 50% in 2 min. neurites of all formerly differentiated cells remained retracted after 18 h of culture, even in the presence of Acid and heat stability 1 mM-dBcAMP (Fig. IF). However, it is interesting to For acidification, the pH of FCS was adjusted to 2-5 with 1 M- point out that addition of low concentrations of 1-2% HC1 and, after centrifugation, the supernatant was readjusted FCS to morphologically differentiated cells (Fig. 1A) to pH7-4 with 1 M-NaOH. For heat stability, the activity was stimulated neurite retraction (Fig. 1B-1F); but at this measured after heating the serum sample at 56, 75 and 100°C dose, the cells regained their differentiated morphology for 15 min. Both acidified and heated serum samples were (Fig. 1A) after 18 h of incubation, indicating the revers- centrifuged at 100 000 £ for 60 min to obtain clear supernatants ible nature of this reaction. for assay. In another series of experiments, pools of active fractions of FCS in PBS after gel permeation were similarly Furthermore, cells grown in serum-free medium, in treated for stabilities by acidification at pH35 or boiling the absence of dBcAMP, also became morphologically (100°C) for 15 min, followed by centrifugation and bioassay. differentiated, though to a much lesser degree. Similarly, addition of FCS did stimulate neurite retraction under Fractionation by ultrafiltration these conditions (data not shown), in a manner similar to FCS was separated by Centricon 10X upon centrifugation or that found with cells grown in the presence of dBcAMP Millipore-immersible CX30 under vacuum into fractions of (Fig. 1). However, dBcAMP was able to potentiate molecules of molecular weights less than lOxlO3 or 30X103, significantly the process of differentiation, and was there- 3 respectively. Fractions, containing molecules of Mr< lOxlO , fore used routinely in the neurite retraction assay. 3 3 between 10 and 30X10 , and >30xl0 , were subjected to Fig. 2 indicates the time-dependent nature of neurite bioassay for neurite retraction activity. retraction. The neurite length of differentiated cells decreased approximately 50 % in 2 min upon addition of Gel filtration on Ultrogel AcA-44 10 % FCS and reached its maximum shortening of over For fractionation on gel filtration, samples of FCS, untreated or 90% in 8 min. The time required for half-maximal boiled (100°C) for 15 min, were centrifuged at lOOOOO^for 1 h. activity (ij) is 1-8 min. The supernatant was applied to a column (l-6cm X 110cm) of The effect of serum on neurite. retraction of differen- LKB Ultrogel AcA-44 in PBS (pH 7-4) and eluted at 16 ml h"1. Fractions of 1 ml/tube were collected, and samples of 50—lOOjUl tiated cells was dose-dependent as shown in Fig. 3. FCS of each fraction were used directly for the neurite retraction at 1 % was the lowest dose effective in reducing neurite bioassay. length in 2 min. No neurite shortening was observed by addition of less than 1 % of FCS; whereas, approximately Trypsinization 5-10% of FCS was needed to reduce the neurite length After boiling and centrifugation, the clear supernatant of the of differentiated cells cultured in the presence of 1 mM- serum sample was desalted through Centricon 10X, and trypsin dBcAMP by more than 50% in 2 min (Fig. 3). was added to a final concentration of 2 %. Digestion was carried No neurite retraction was observed upon addition of out by incubating the reaction mixture for 16 h at 37 °C, and homogenates of rat lung, kidney, heart, skeletal muscle terminated by boiling the mixture for 15 min. and brain at a protein concentration of lmgmP1 into morphologically differentiated NG108-15 cells in cul- Extraction of tissues ture. However, spleen and liver homogenates caused Rat lung, brain, kidney, spleen, heart, liver and skeletal muscle slight but significant neurite retraction (Table 1). were washed with PBS and homogenized in 3 vol. of the same Serum samples from rat, sheep, horse and human buffer. The crude homogenate was then centrifuged at 9000u showed various degrees of neurite-retraction activity for 1 h. An appropriate volume of tissue extract, equivalent to 1 mg of protein, was added to the morphologically differen- from 53-5 to 73-8% (Table 1). Horse serum, however, tiated NG108-15 cells in culture to assay neurite retraction appears to possess the highest neurite shortening activity. activity. Samples of human and cat cerebral spinal fluid (CSF) were observed to be void of neurite shortening activity as shown in Table 1. Results Stability of the neurite retraction activity upon treat- ment with acid and heat is summarized in Table 2.. Neuroblastoma X glioma NG108-15 cells cultured in Acidification of FCS to pH 2-5, followed by reneutraliza- serum-free medium, containing 1 mM-dBcAMP, showed tion, reduced the neurite retraction activity from extensive neurite formation, mainly from singlets. A approximately 69% to 41 %. However, heating to 100°C

252 A. Ghahary et al. Fig. 1. Neurite retraction of morphologically differentiated NG108-1S cells. Cells were cultured for 48 h in serum-free medium, supplemented with 1 mM-dBcAMP, to induce differentiation. FCS was added to the culture to a final concentration of 10%, and photographs were taken at various time intervals as indicated. A, Omin; B, 1 min; C, 2min; D, 4min; E, 8min; F, 18 h. X200. for 15 min did not affect the neurite retraction activity. lar weight of approximately 70X103, as shown in Fig. 4. In another set of experiments, the partially purified This active fraction was assayed to be effective at 2-3 fig neurite retraction-active fractions in PBS, obtained from of protein ml"1 in causing a 50 % retraction of neurite untreated whole FCS upon gel permeation (see Fig. 4), length in 2 min. The supernatant of boiled FCS, before were adjusted to pH3-5 for 1 h or boiled at 100°C for gel permeation, was biologically active at 40fig pro- 15 min, followed by centrifugation and bioassay. The tein ml" of culture medium as compared to 120 j.ig neurite retraction activity in these partially purified protein ml" for untreated control serum samples. An fractions was stable to acid and heat treatments in a approximate 60-fold increase in activity on the basis of similar manner to whole serum (Table 2). protein content has been achieved upon boiling, followed The molecular weight of the serum neurite retraction by gel permeation. factor was estimated by gel permeation on a column of Treatment of the clear FCS supernatant with 2% LKB Ultrogel AcA-44, equilibrated in PBS. Neurite trypsin for 16h at 37°C, followed by boiling, abolished retraction activity was eluted in a fraction with a molecu- the neurite retraction activity (Table 2). This observation

A factor inducing neurite retraction 253 100 Table 1. Neurite-retaction activity in serum, organ tissue and cerebral spinal fluid % Retraction Sample of neurites So 60 PBS 0 Serum FCS 69-014-7 3 40 Rat 53-5 ±15-3 U = 1-8 min Human 65-0 ±10-9 Horse 73-8 ±10-0 20 Sheep 60-218-2 Tissue Lung 0 4 6 10 Kidney 0 Time (min) Heart 0 Skeletal muscle 0 Fig. 2. Time course for neurite retraction of differentiated Brain 0 NG108-15 cells. FCS was added to differentiated cells in Spleen 21-9±4-0 culture to a final concentration of 10% and shortening of Liver 14-3 ±3-3 neurites was monitored at various time intervals. Data Cerebral spinal Human 0 represent means ± S.E.M. fluid Cat 0

Samples of serum and cerebral spinal fluid were assayed at a final concentration of 10% (v/v), and tissue homogenates of various 100 organs of the rat were used at a protein concentration of 1 mgml"' Neurite retraction of differentiated NG108-15 cells was measured at 2 min after addition of sample to the culture. Data represent 80 means ± S.E.M. si

ele 60 •c 3 U Table 2. Effects of acidification, boiling and trypsin 40 digestion on the neurite retraction activity

20 % Retraction Sample of neurites PBS 0 6 10 FCS 69-0 ±4-7 % FCS Boiled serum 71-7 ±3-1 Acidified serum 41-716-4 Fig. 3. Dose—response curve of neurite retraction. Neurite NRF fraction (gel permeation) 50-318-5 shortening of morphologically differentiated NG108-15 cells Boiled NRF fraction 50-614-0 was measured at 2 min after the addition of various Acidified NRF fraction 52-3 + 2-4 concentrations of FCS. Data represent means ± S.E.M. Boiled serum + PBS 71-8 + 3-1 Boiled serum + trypsin 19-3 + 3-5 indicates that the neurite retraction factor is protein in PBS + trypsin 17-515-3 nature. No neurite shortening activity was observed for Samples of whole serum or NRF fraction after gel permeation were Nerve Growth Factor (NGF) and Platlet Derived acidified and/or boiled or trypsin-treated as described in Materials Growth Factor (PDGF) under identical assay con- and methods, and assayed at a final concentration of 10% (v/v) for ditions. serum, and lO^gml for NRF fraction. Data represent means! S.E.M.

Discussion (2-8 min) neurite retraction of morphologically differen- It has been established that the removal of serum from tiated NG108-15 cells. culture media results in differentiation and neurite out- Our preliminary studies on the neurite retraction factor growth of neuroblastoma cells (Harkins et al. 1972). (NRF) indicate that it differs from all of the well- Differentiated C1300 neuroblastoma cells gradually characterized growth factors. The molecular weight of (2-4 h) retract processes prior to mitosis upon replace- NRF, as determined by gel permeation on Ultrogel AcA- ment of serum (Seeds et al. 1970). This growth-promot- 44 (Fig. 4), distinguishes it from PDGF, EGF, FGF, ing ability of serum has been shown to be non-dialysable NGF, NSILA and TGF (Deuel & Huang, 1984; Savage and heat-stable (Seeds et al. 1970). To our knowledge, a et al. 1972; Thomas et al. 1984; Gospodarowicz et al. factor(s) promoting neurite retraction of differentiated 1978; Server & Shooter, 1977; Rinderknecht & Humbel, neuronal cells has not been specifically identified and 1976; Roberts et al. 1983; Todaro et al. 1981). Further- characterized. In the present study, we have partially more, our observation that NRF is heat- and acid-stable characterized a serum factor(s) that stimulates rapid (Table 2) also distinguishes it from ECGF, FGF and

254 A. Ghahary et al. differentiation. In contrast, Fauquet et al. (1981) have shown that mesencephalic crest cells, grown in the presence of FCS-supplemented medium, exhibit spiky processes after 7 days in culture. The induction of processes, despite the presence of FCS, may be due to the effects of other factors, such as extracellular matrix components (Sanes, 1983) secreted by these cells, and environmentally derived differentiation factors from other embryonic tissue contaminants in the culture. It has been suggested (Bloch, 1984) that accumulation of proliferation factors (PFs) or differentiation factors (DFs) is responsible for either a cellular proliferation or a differentiation mode, respectively. The PF-mediated activity appears to be regulated by specific exogenous 30 60 90 120 150 180 signals, including serum growth factor(s). DFs are Fraction number (1 ml/tube) reported to be capable of interrupting the proliferation cycle, possibly by deactivating a proliferation factor. Fig. 4. Estimation of molecular weight by gel filtration on Upon longer-term cultures of 2-3 days, NG108-15 cells Ultrogel AcA-44. The centrifuged supernatant of a boiled do differentiate in the presence of PF (5 % FCS) when FCS sample was applied to a column of LKB Ultrogel AcA- 1 the level of DF (1 mM-dBcAMP) is high (data not 44 (l-6cm X 110cm) in PBS, pH7-4, and eluted at 16 ml IT shown). Fractions of 1 ml/tube were collected and samples (100 jttl) of each fraction were assayed directly for neurite retraction. The mechanism by which neurite retraction occurs Maximal activity was at tube 90, corresponding to a upon addition of the serum NRF is unknown. It has been molecular weight of approximately 70X103. Protein markers demonstrated that NG108-15 cells secrete plasminogen 3 {Mr X 10~ ): cytochrome c (12-5); carbonic anhydrase (29); activator (Krystosck & Seeds, 1986), which may convert albumin (66). Blue dextran (V*o). ( ) Absorbance at serum plasminogen to plasmin, leading to the proteolysis 278 nm; ( ) % neurite shortening. of neurite attachment proteins (Gennarini et al. 1984). However, plasminogen is unstable at high temperature NGF (Server & Shooter, 1977; Maehana et al. 1986; (Cleveland, 1957), and addition of purified plasmin to the Maciag et al. 1979). However, a purified preparation of culture has no effect on neurite retraction (unpublished NRF is required for further studies to clarify whether or observation). Replacement of fresh medium prior to not NRF is directly responsible for the growth-promot- addition of serum NRF does not dampen neurite retrac- ing ability of serum or is important in preventing neurite tion activity (unpublished observation), indicating that outgrowth and indirectly potentiating the effects of other NRF does not work through a component(s) in the unidentified serum growth factor(s). conditioned medium. It is noteworthy that NRF has not been detected in Guenther et al. (1985) have purified a glial-derived CSF (Table 2), possibly because the blood-brain barrier factor promoting neurite outgrowth from neuroblastoma restricts the passage of macromolecules from blood cells, and proposed that it stimulates neurite outgrowth circulation into the CSF (Bradbury, 1979). Peripheral by interacting with proteases on the neuroblastoma cell nerves also have a barrier at the endoneurium and the membranes. It has been suggested that the proteases perineurium surrounding the nerve bundles (Rapoport, could affect the turnover of specific membrane proteins 1976). It is conceivable that NRF plays a role in the involved in neurite outgrowth. It is conceivable that development of immature neurones before the functional NRF could stimulate proteases on NG108-15 cells, act as formation of the blood-brain barrier. Maturation of the a protease itself, or in an unknown manner block growth barrier function in relation to parturition is highly cone-substratum adhesion molecules. On the contrary, variable between species. In sheep, the maturation pro- NRF may possibly function by interacting with an ion cess of the barrier occurs during the entire foetal life; channel(s), thereby disrupting the ionic balance neccess- whereas, in the rat, it takes place mostly after birth ary for the maintenance of neurites. However, the (Bradbury, 1979). It seems likely that the accessibility of mechanism of action of this rapidly acting factor remains NRF to neuronal cells during foetal life helps to control speculative. the pathway of proliferation or differentiation of a In summary, we have identified a serum factor that developing neurone. causes rapid neurite retraction of morphologically differ- Our present studies show that serum NRF is active on entiated NG108-15 cells. The factor has a molecular morphologically differentiated neuroblastoma cells under weight of approximately 70X 103 and is relatively stable to serum-deprived conditions. It is well established that heat and acid. Further studies on its isolation and and neuronal cells in culture are able to characterization should clarify the biological role(s) of grow neurites in the presence of serum. However, it has this serum factor in the processes of maturation and been reported (Darmon et al. 1982) that differentiation of differentiation of developing neurones. embryonic carcinoma cells occurs merely as a conse- quence of serum deprivation, since the addition of FCS This research was supported by grant MT-5110 from the to serum-free medium at the time of culturing inhibits Medical Research Council of Canada. The technical assistance

A factor inducing neurite retraction 255 of Mrs E. Ling is gratefully acknowledged. We also thank Dr L. (1986). Purification and characterization of prostate derived growth Murphy for his helpful discussions. factor. Pivc. natn. Acad. Sci. U.S.A. 83, 8162-8166. PRASAD, K. W. (1975). Differentiation of neuroblastoma cells in culture. Biol. Rev. 50, 129-165. References PRASAD, K. W. & SHEPPARD, J. R. (1972). Inhibition of cyclic nucleotide phosphodiesterase induces morphological differenti- ation of mouse neuroblastoma cell culture. Expl Cell Res.73, BLOCH, A. (1984). Induced cell differentiation in therapy. 436-440. Cancer Treat. Rep. 68, 199-205. RAPOPORT, S. (1976). Site and functions of the blood-brain barrier. BRADBURY, M. (1979). The blood-brain barrier during the In Blood-Brain Bonier in Physiology and Medicine (ed. S. development of the individual and the evolution of the phylum. In Rapoport), pp. 43-86. New York: Raven Press. The Concept of the Blood-Brain Barrier (ed. M. Bradbury), pp. RINDERKNECHT, E. & HUMBLE, R. E. (1976). Polypeptides with non- 289-322. New York: J. Wiley & Sons. suppressible insulin-like and cell growth promoting activities in CLEVELAND, 0. (1957). Comparison of certain properties of human human serum: Isolation, chemical characterization and some plasminogen and "proactivator". Proc. Soc. exp. Biol. Med. 95, biological properties of forms. Proc. natn. Acad. Sci. U.S.A. 73, 195-200. 2365-2369. DARMON, M., SATO, G. H., STALLCUP, W. & PITTMAN, Q. J. (1982). ROBERTS, A. B., AMZARIO, M. A., MEYERS, C. A., WIDERMAN, J., Control of differentiation pathways by the extracellular BLOCKER, R., PAN, Y.-C. E., STEIN, S., LEHMAN, S. R., SMITH, J. environment in an embryonal carcinoma cell line. Cold Spring M., LAMB, L. C. & SPORN, M. A. (1983). Purification and Harbor Con}. Cell Prolif. 9, 997-1006. properties of a type-transforming growth factor from bovine brain. DEUEL, T. F. & HUANG, J. S. (1984). Platelet derived growth Biochemistry1 21, 5692-5698. factor: structure, function, and roles in normal and transformed ROBINSON, R. (1980). Tumors of sympathetic . In cells. J. din. Invest. 74, 669-676. Tumors that Secrete Catecholamines (ed. R. Robinson), FAUQUET, M., SMITH, J., ZILLER, C. & LEDOUARIN, N. M. (1981). pp. 11-34. New York: Johan & Sons. Differentiation of autonomic precursors in vitro: SANES, J. R. (1983). Roles of extracellular matrix in neural cholinergic and adrenergic traits in cultured neuronal crest cells. development. A. Rev. Physiol. 45, 581-600. J. Neumsci. 1, 478-492. SAVAGE, C. R., TADASHI, I. & COHEN, S. (1972). The primary GENNARINI, G., ROUGON, G., DEACOSTINI, B. H., HIRN, M. & structure of epidermal growth factor. J. biol. Chem. 247, GORIDIS, C. (1984). Studies on the transmembrane disposition of 7612-7621. the neural cell adhesion molecule N-CAM. Ettr.J. Biochem. 142, SCHUBERT, D. & JACOB, F. (1970). 5-Bromodeoxyuridine-induced 57-64. differentiation of a neuroblastoma. Pmc. natn. Acad. Sci. U.S.A. GLICK, M. C, GIOVANNI, M. Y. & LITTAUER, U. Z. (1980). 67, 247-254. Glycoprotein associated with differentiating neuronal cells in SEEDS, N. W., GILMAN, A. G., AMANO, T. & NIRENBURG, M. W. cultures. In Advances in Neuroblastoma Research (ed. A. E. (1970). Regulation of axon formation by clonal lines of a neural Evans), pp. 171-175. New York: Raven Press. tumor. Pmc natn. Acad. Sci. U.S.A. 66, 160-167. GOSPODAROWICZ, D., BIALECKE, H. & GREENBURG, G. (1978). SERVER, A. C. & SHOOTER, E. W. (1977). Nerve growth factor. Adv. Purification of the fibroblast growth factor activity from bovine Protein Chem. 31, 339-409. brain. J. biol. Chem. 253, 3736-3743. THOMAS, K. A., RIOS-CANDELORE, M. & FITZPATRIK, S. (1984). GUENTHER, J., NICK, H. & MONARD, D. (1985). A -derived Purification and characterization of acidic fibroblast growth factor neurite-promoting factor with protease inhibitory activity. EMBO from bovine brain. Proc. natn. Acad. Sci. U.S.A. 81, 357-361. y. 4, 1963-1966. TODARO, G. J., DELARCO, J. E., FRYLING, C, JOHNSON, P. A. & HARKINS, J., ARSENAULT, M., SCHLESINGER, K. & KATES, J. (1972). SPORN, M. B. (1981). Transforming growth factors (TGF's): Induction of neuronal function: acetylcholine-induced Properties and possible mechanisms of action. J. stipramolec. acetylcholinesterase activity in mouse neuroblastoma cells. Pmc. natii. Acad. Sci. U.S.A. 69, 3161-3164. Struct, cell. Biochem. 15, 287-301. KRYSTOSCK, A. & SEEDS, N. W. (1986). Normal and malignant cells, WOLFE, R. A. & SATO, G. H. (1982). Continuous serum-free culture including , deposit plasminogen activator on the growth of the N18TG-2 neuroblastoma, the C6Bu-l glioma, and the substrata. Expl Cell. Res. 166, 31-46. NG108-15 cell lines. In Croicth of Cells in Honnonally Defined Media (ed. G. H. Sato, B. Pardeca & D. A. Sirbakuj, MACIAG, T. J., ILSLEY, S., KELLEY, R. R. & FORAND, R. (1979). An endothelial cell growth factor from bovine hypothalamus. pp. 1075-1088. New York: Cold Spring Harbor Laboratory. Identification and partial characterization. Pmc. natn. Acad. Sci. U.S.A. 76, 5674-5678. MAEHANA, S., LI, D., NARNI, H., LEYKAM, J. F. & DUEL, T. F. (Received 22 June 1988-Accepted 26 September 1988)

256 A. Chaharv et al.