The Journal of Neuroscience, August 1991, 1 I(9): 2443-2452

Interaction of GM1 with the B Subunit of Toxin Modulates Growth and Differentiation of Neuroblastoma N18 Cells

Daniel Masco,l Michele Van de Walle,* and Sarah Spiegel’ ‘Department of Biochemistry and Molecular Biology, Georgetown University Medical Center, Washington, DC 20007, and ‘Biotechnology Unit, LCDB, NIDDK, NIH, Bethesda, Maryland 20892

The present study uses the B subunit of , a of endogenous GM1 on the cell surface by the B subunit is protein that binds specifically to ganglioside GMl, to ex- responsible for these effects on mouse neuroblastoma N 18 amine the role of endogenous GM1 in the process of growth cells. and differentiation of mouse neuroblastoma N 18 cells. Bind- ing of the B subunit to neuroblastoma N18 cells inhibited , -containing , are DNA synthesis with concomitant induction of differentiation. ubiquitous components ofthe plasma membrane of vertebrate The B subunit induced pronounced morphological changes: cells (Sweeleyand Siddiqui, 1977) and are particularly enriched an increase in neurite outgrowth with branched neurites and in neural cells (Ledeen, 1985).The abundanceand the structural spinelike processes. The distinct morphological alterations complexity of gangliosidesin nerve cells, combined with the and neuritogenesis in response to the B subunit were also observation that gangliosidesvary considerably in composition revealed by immunofluorescence with fluorescein-labeled B and concentration during neuronal development and differen- subunit. The mechanism of the B subunit-induced differ- tiation (Hakomori, 1981; Yates, 1986), suggestthat gangliosides entiation is different than that of spontaneous differentiation. have important functional roles in neuronal development in , a serine protease present in normal serum, inhib- vivo. Further support for this contention is that neuronal cells its neurite outgrowth induced by the removal of serum from responduniquely to exogenousgangliosides with profound mor- the medium. In contrast, thrombin did not cause retraction of phological changes characteristic of differentiation (Ledeen, the neurites induced by the B subunit. Thus, thrombin or a 1985). The observations that exogenouslyadministered gangli- thrombin-like protease is not involved in the process of neu- osideshave neuritogenic and neuronotrophic properties capable rite outgrowth mediated through endogenous Gfull. The bi- of influencing neuronal differentiation in vitro and neuronal re- ological effects of the B subunit are due to the binding of pair in vivo have provided possiblelinks to this aspect of gan- the B subunit to ganglioside GM1 and not due to changes glioside function (Ledeen, 1985). in CAMP levels resulting from contaminating A subunit. We Many neuronal cell types respond to exogenousgangliosides used highly purified cloned B subunit that cannot contain with concomitant sprouting and extensionof net&es, including any A subunit because it was isolated from a Vibrio cholerae neuroblastomacell lines Neuro2A (Dimpfel et al., 1981; Roisen mutant that only expresses the B subunit. Neither the cloned et al., 1981; Leon et al., 1982; Byrne et al., 1983; Spoerri, 1983) nor commercial preparations of the B subunit induced in- SB21Bl (Rybak et al., 1983) GOTO, and NB-1 (Tsuji et al., creases of CAMP in these cells. There was a good correlation 1983); neuronal cell lines B103 and 104 (Morgan and Seifert, between the amount of B subunit bound to the cells and the 1979); pheochromocytoma PC12 cells (Ferrari et al., 1983; Ka- biological effect. Finally, treatment with neuraminidase, which toh-Semba et al., 1984); and primary cultures of dorsal root caused a fourfold increase in the level of membrane GM1 ganglia (Roisen et al., 1981) spinal root ganglia (Hauw et al., as determined by iodinated cholera toxin binding, enhanced 1981) and cerebrum (Dreyfus et al., 1984). Exogenousgangli- the biological effect of the B subunit. However, neuramini- osideshave important effects on the regulation of neural differ- dase treatment alone did not have significant effects, either entiation, not only in tissue culture systems, but there is no on DNA synthesis or on morphology of the cells, indicating doubt that they facilitate regenerationof damagedPNS and CNS that elevations in the level of GM1 per se are not sufficient neurons (reviewed in Ledeen, 1985). The monosialoganglioside by themselves to cause significant changes in cell growth GM 1 [Svennerholm’s (1963) system of gangliosidenomencla- or differentiation. It seems most likely that the aggregation ture], in particular, has been shown to causedifferentiation of neuronal cultures, to enhancesprouting of regenerative periph- eral nerves, and to accelerate repair of CNS damage(Facci et Received Oct. 18, 1990; revised Feb. 27, 199 1; accepted Mar. 12, 199 I. al., 1984; Ledeen, 1985). However, though the effects of exog- This work was supported by NIH Research Grants 1 R 29 GM 39718 and 1 enousGM 1 are very well documented, the relationship between R 01 GM 43880. D.M. was supported in part by Alzheimer’s Disease and Related theseobservations and the function of endogenousgangliosides Disorders Association, Inc., Grant PRG 89 087. We thank Dr. J. Shiloach for critically reviewing the manuscript and for advice during the course of this work in the processof cellular differentiation remains unclear, and and Dr. M. Nirenberg for providing neuroblastoma N 18 cells. no connection has yet been establishedbetween the effects of Correspondence should be addressed to Dr. Sarah Spiegel, Department of Bio- chemistry and Molecular Biology, Georgetown University Medical Center, 357 exogenousgangliosides and the role of endogenousgangliosides. Basic Science Building, 3900 Reservoir Road Northwest, Washington, DC 20007. There are conflicting data on the role of endogenousgangli- Copyright 0 1991 Society for Neuroscience 0270-6474/91/l 12443-10$03.00/O oside GM1 in neuritogenesis(Schwartz and Spirman, 1982; 2444 Masco et al. * B Subunit-induced Neuroblastoma Differentiation

Mahadik et al., 1986; Doherty and Walsh, 1987; Spoerri et al., the B subunit of CT (Taylor et al., 1987). Culture supematants were 1988). Doherty and Walsh (1987) found that to GM 1 ultrafiltered and purified on a phosphocellulose column. . The murine neuroblastoma cell line N18 was kindly had no effect on NGF-induced outgrowth in chicken dorsal root provided by Dr. M. Nirenberg, NHLBI, NIH. Stock cultures of cells ganglia. In contrast, others have reported that their anti-GM1 were maintained in DMEM supplemented with 5% (v/v) fetal bovine antibodies blocked NGF-induced sprouting of the same cells serum, respectively, in a humidified atmosphere of 5% CO,, 95% air at (Schwartz and Spirman, 1982; Mahadik et al., 1986) and also 37°C. Upon reaching stationary phase of growth, the cells were sub- inhibited neurite outgrowth from regenerating retinal explants cultured by flushing the cells off their growing substratum with a stream of medium and diluting the cell suspension 1:4 with fresh medium. of the goldfish (Spirman et al., 1982). Recently, it was found Measurement of DNA synthesis. Cells were seeded at a density of 2.5 that monoclonal antibodies to GM 1 inhibit conditioned media- x IO4 cells per well in multicluster plastic tissue culture dishes (24 x mediated neuritogenesis of sensory ganglia (Spoerri et al., 1988). 16 mm wells: Costar. Cambridge. MA) in DMEM containing different Furthermore, it has been recently reported that there is a cor- concentrations of fetal calf serum: In sbme experiments, the cells were grown in chemically defined serum-free medium (medium I), which relation between the level of endogenous GM1 and the ability consists of DMEM supplemented with insulin (5 &ml), transfenin (5 of the cell line to undergo neuritogenesis (Cannella et al., 1988). Fug/ml), and selenium (5 t&ml). At a density of about 6-8 x lo4 cells To gain more insight into the putative role of endogenous per well, the cells were washed with DMEM to remove residual serum ganglioside GM 1, we have used the binding or B subunit of and treated with different concentrations of the B subunit. After 20 hr, cholera toxin (CT). The major advantage of this approach lies the cells were nulsed with 0.5 or 1 uCi of jH-thvmidine for 4 hr. The incorporation bf radioactivity into ‘trichloroaceiic acid-insoluble ma- in the specificity of the B subunit and the very high affinity for terial was measured as described by Spiegel and Fishman (I 987). Values ganglioside GM1 (Fishman, 1982). Previously, we found that are the mean of triplicate determinations, and standard errors were the B subunit stimulated the growth of rat thymocytes (Spiegel routinely less than 10% of the mean. et al., 1985b) and quiescent cultures of murine 3T3 fibroblasts Morphological evaluation of neurite outgrowth. Cells were treated with the B subunit exactly as described above. After 24 hr, the cells were (Spiegel and Panagiotopoulos, 1988). However, in contrast to rinsed with PBS, fixed for 1 hr in PBS containing 4% formaldehyde and its effect on resting cells, the B subunit inhibited the growth of 5% sucrose, then rinsed three times with PBS, and photographs were rus-transformed 3T3 fibroblasts (Ha-, Ki-, and N-rus; Spiegel taken. The numbers of cells bearing neurites were counted in several and Fishman, 1987) and rat glioma C6 cells with elevated levels randomly chosen fields containing approximately 200 cells at a micro- of GM 1 (Spiegel, 1988). Recently, the studies with the C6 cells scopic magnification of 200 x . Neurites were identified by the presence of growth cones. Cytoplasmic extensions were scored as neurites if their were confirmed and further extended by the demonstration that length exceeded twice the diameter (20 wrn) of the neuroblastoma soma neuraminidase (NA) treatment, which increases the amount of (Matta et al., 1986). cell-surface ganglioside GM 1, also induced susceptibility to the Assay of CAMP. Accumulation of CAMP in the cells was determined inhibitory effect of the B subunit (Skaper et al., 1988). Fur- essentially as we described previously (Spiegel et al., I985b; Spiegel, 1988). Briefly, cells were incubated for 5 min at 37°C in 1 ml of medium thermore, the B subunit not only inhibited the growth of as- buffered with 25 mM HEPES and supplemented with 0.5 mM IBMX, troglial cells but also induced marked differentiation to a stellate a phosphodiesterase inhibitor, and 0.0 1% bovine serum albumin. After morphology (Facci et al., 1988; Dekker et al., 1990). incubations with various reagents, CAMP was extracted from the cells The purpose of this study was to extend this approach to the with 0.1 M HCl and measured by radioimmunoassay. The precipitated involvement of membrane ganglioside GM1 in the process of proteins remaining in the dish were dissolved in 0.2 M NaOH and measured (Lowry et al., 195 1). differentiation and growth of neuronal cells. Mouse neuroblas- Measurement ofspec$c binding of iodinated CT. Binding of iodinated toma N 18 cells seem to be especially suitable for this purpose, CT to cells in situ was measured as described (Spiegel et al., I985b; because they have a ganglioside composition similar to that of Spiegel, 1988). Briefly, cells subcultured in 16-mm-diameter wells were primary neuronal cells (Duffard et al., 1977), have large amounts incubated with iodinated CT (0.1-100 nM) in 0.5 ml of serum-free medium buffered with 25 mM HEPES and supplemented with 0.1% of surface ganglioside GM 1 when analyzed by chemical assays bovine serum albumin. After 1 hr at 37”C, the medium was removed, or surface labeling techniques (Miller-Podraza and Fishman, and bound toxin was determined. Nonspecific binding was determined 1982; Miller-Podraza et al., 1982), and can easily be induced by the addition of 0.2 PM unlabeled CT to the assay. All values were to differentiate by various stimuli, such as dibutyryl CAMP, corrected for nonspecific binding and are the means of quadruplicate determinations. tumor promoters, and the removal of serum (Liu et al., 1980). Treatment of cells with thrombin. Cells were washed with medium to The results presented here indicate that binding of the B subunit remove residual serum and incubated with different concentrations of of CT to endogenous plasma membrane ganglioside GM1 can highly purified thrombin at 37°C in serum-free medium for various influence the growth and differentiation of a neuronal cell line. times (Gurwitz and Cunningham, 1988). The cells were then treated with the B subunit as described above. Distribution of the B subunit of CT by fluorescence microscopy. Cells were subcultured on glass coverslips in medium I. At various times after treatment with fluorescein-conjugated B subunit (4 @g/ml), the cells Materials and Methods were rinsed with PBS. fixed for 1 hr in PBS containing 4% formaldehvde and 5% sucrose, and’then rinsed three times with FBS and examined Materials. CT and its B or binding subunit were obtained from Schwarz/ by fluorescence microscopy with a Zeiss microscope modified for epi- Mann Biotech (Cleveland, OH). The B subunit appeared to be holotoxin fluorescence, performed exactly as described previously (Spiegel et al., free bv SDS-PAGE because there was no detectable M. 22.000 band of 1985a). Routine negative controls included treating the cells with excess the A i subunit by silver staining. Fluorescein-conjugatkd g subunit was amounts of unlabeled B subunit or with a large excess of ganglioside from List Biological Labs (Campbell, CA). ‘H-thymidine (55 Ci/mmol) GMl. was purchased from ICN (Irvine, CA). Thrombin, insulin, transferrin, and selenium were from Collaborative Research (Lexington, MA). Dul- becco’s modified Eagle’s medium (DMEM) was from-Advanced Bio- Results technoloeies (Silver Sminn. MD). 1-Methvl-3-isobutvlxanthine (IBMX) The B subunit inhibits DNA synthesis and growth of was from Sigma Chemicz Co. (St. Louis; MO). . neuroblastoma N18 cells Production of highly purijied cloned B subunit. The B subunit was produced as recently described (Van de Walle et al., 1990). Briefly, a Previously, we found that the interaction of the B subunit with defined medium was used to grow a mutant of Vibrio cholerae (0395- endogenous ganglioside GM 1 leads to stimulation or inhibition NI), which carries a deletion in the cbcA gene and thus produces only of proliferation, depending on the particular cell type and even The Journal of Neuroscience, August 1991, 1 I(8) 2445

Figure I. Dose-dependent effectof the B subunit of CT on DNA synthesis cor- relates with iodinated CT binding to Nl8 cells. Cells grown in DMEM me- dium suplemented with 1% FCS were exposed to various concentrations of B subunit obtained commercially (0) or to cloned B subunit (0) and assayed for ‘H-thymidine incorporation as de- scribed in Materials and Methods. The results are expressed as the percent in- hibition of DNA synthesis compared to cells not exposed to the B subunit. The amount of ‘H-thymidine incorporated into control cells was 70 f 5 x lo3 cpm per well. Specific iodinated CT binding 0 was determined on duplicate cultures O.W IiC 1 10 B SUBUNIT (pg/ml) CT (MOW as described in Materials and Methods.

on their growth stage (Spiegel and Fishman, 1987; Spiegel, 1988; cells (data not shown). The B subunit causeda dose-dependent Buckley et al., 1990). The effect of the B subunit on the prolif- inhibition of DNA synthesis (Fig. 1). A significant effect was eration of mouse neuroblastoma cell line N 18 is shown in Table observed at a concentration of 1 &ml, and maximum response 1. Experiments with N18 cells grown in different serum con- to the B subunit was observed at a concentration of 2-4 &ml. centrations revealed that the inhibition of DNA synthesis by While these experiments were carried out using B subunit pu- the B subunit increased as the amount of serum was decreased rified from holotoxin (Schwarz-Mann), an identical dose re- (Table 1). A maximum inhibition of 25%, 24%, 15%, and 17% sponsewas found using a highly purified preparation of the B was found with cells that were cultured in medium containing subunit produced in a mutant of V. cholerae expressingonly O%, l%, 2.5%, and 5% fetal calf serum (FCS), respectively. Al- the B subunit (Fig. 1). The dose dependency for the growth though these experiments were done on cells grown in culture inhibition effect by both preparations ofthe B subunit correlated for 3 d prior to the addition of the B subunit, similar results closely with the dosedependence of iodinated CT binding (Fig. were found when the cells were cultured for 1 d (data not shown). 1). N 18 cells proliferated maximally in chemically defined medium containing insulin, transferrin, and selenium (Table 1). In these Lack of an efecct on the level of intracellular CAMP by the B conditions, the growth of cells was even greater than observed subunit in the presence of serum. A maximum inhibition of DNA syn- Numerous studieshave shown that an elevation in intracellular thesis induced by the B subunit was found with N 18 cells grown CAMP concentration is an important signal for neuritogenesis in this chemically defined medium of 40% and 42% after 1 and in neuronal cells (Richter-Landsberg and Jastorff, 1986). An 3 d of culture, respectively. Thus, the time period in culture increasein the intracellular level ofcAMP can inhibit the growth prior to the addition of the B subunit is not an important factor. of mouse neuroblastoma N18 cells (Prasad and Hsie, 1971). The antiproliferative activity ofthe B subunit was not associated To eliminate the possibility that the inhibition of DNA synthesis with any cytotoxic effects, because the number of viable cells, by the B subunit was due to a small contamination with the A determined by trypan blue exclusion, was always greater than subunit, which is known to activate adenylate cyclaseby ADP- 95%. Also, there was no apparent loss ofcells during this period, ribosylation of the stimulatory G-protein (Gilman, 1984), we and the B subunit had no effect on cell adhesion because identical usedthe cloned B subunit produced in V. cholerae mutants that results were obtained using a cell harvester (data not shown). did not expressthe A subunit. However, it seemsmost likely Uptake studies demonstrated that the effect of the B subunit that the commercial preparation of the B subunit usedin this was not due to a decrease in )H-thymidine transport into the study was also devoid of significant contamination with the A

Table 1. Effects of the B subunit on ‘H-thymidine incorporation in mouse neuroblastoma N18 cells

)H-thymidine incorporation Culture (cpm x 10-j per well) condition Control B subunit % Inhibition P Medium I 86.8 + 5 50.1 + 3.8 42 co.01 0% FCS 15.8 + 0.7 11.8 f 0.5 25 co.01 1% FCS 73.1 f 1.1 56.0 -t 0.2 24 co.01 2.5% FCS 71.4 ? 3.5 60.7 + 3.9 15 co.05 5% FCS 53.0 k 6.3 44.1 k 5.0 17 co.05 Cells were grown for 3 d in the indicated medium and then incubated in the absence (control) or presence of the B subunit (4 &nl) for 20 hr. ‘H-thymidine incorporation was measured as described in Materials and Methods. Each value is the mean k SD of triplicate determinations from a representative experiment. Statistical analyses of the data were done by Student’s I test. Similar results were obtained in at least six additional experiments. 2446 Masco et al. * B Subunit-induced Neuroblastoma Differentiation

400 B subunit on thesecells in which the endogenouslevels of GM 1 are elevated. In agreementwith previous studies (Miller-Pod- raza et al., 1982) when intact cells were treated with NA, there were profound changesin gangliosidecomposition (Table 2). 5 = 300 The amount of GM3 and GDla wasreduced by 73% and 80%, 4Ei=z respectively, and GM1 increased 3.5fold. There was also a corresponding increase in iodinated CT binding to the NA- 5” 2 m 200 treated cells (data not shown). The maximum effect appeared to occur after 2 hr of exposure to NA (0.0 1 U/ml). The increase SE in the level of GM1 correspondedto the increasedinhibition PQ9 of DNA synthesisin responseto the B subunit (Table 3). Exactly I- d 100 the sameresults were obtained usingthe cloned B subunit. Treat- ment with NA alone for 24 hr had no effect on the proliferation in chemically defined medium or in 1%FCS (Table 3). However, when the cellswere grown in higherconcentrations of FCS, there 0 .OOOl .OOl .Ol .l 1 10 100 was a small inhibitory effect of NA treatment. B SUBUNIT OR CT @g/ml) B subunit inducesmorphological d$erentiation in mouse neuroblastomaN18 Figure2. Effectof CT andtwo B subunitpreparations on CAMP pro- duction.Cultures of N18cells were washed with DMEM andincubated Neuroblastoma N18 cells cultured in the presenceof 5% FCS at 37°Cwith the indicatedconcentrations of CT (O), the commercialB have a round and aggregatedmorphology (Fig. 3A). Addition subunit(o), or the B subunitclone (m) in DMEM/HEPES containing of B subunit (4 &ml) to these cells, growing in a compact IBMX (0.5 mM). Cellularcontent of CAMP wasassayed after 2 hr by fashion as bundles of dividing cells, causesmorphological dif- radioimmunoassayasdescribed in Materialsand Methods.The values arethe means(n = 3). The SDSwere less than 10%. ferentiation (Fig. 3B). The aggregategrowth pattern of the N18 cells growing in 5% FCS complicated the evaluation of a more frequently usedparameter for neurite outgrowth, the percentage subunit, basedon the inability of either of the preparations of of cells bearing neurites, becauseof uncertainty about the cel- B subunit to causeany significant accumulation of CAMP in lular origin of the processes.However, the differences in the N18 neuroblastomacells. In agreementwith several previous pattern and morphology of the B subunit-treated cells were studies, (Fishman, 1980, 1982) the intact CT causeda dose- distinct, and the B subunit clearly causedenhancement of neu- dependentincrease in CAMP production in thesecells (Fig. 2) rite outgrowth (cf. Fig. 3A,B). The effect of the B subunit was and CAMP reachedmaximum levels after 2 hr of exposure(data greatly increasedafter pretreatment with NA (Fig. 30). In con- not shown). In contrast, CAMP levels in cells treated with the trast, NA treatment alone had no effect on the morphology or B subunit preparations over a wide concentration range were neuritogenesis(Fig. 3C). These experiments were also carried not significantly different from untreated cells. It shouldbe noted out in cells grown in chemically defined serum-freemedium to that intact CT, even at concentrations of 0.1-10 pg/ml, which avoid complications due to the complex nature of serum and causedup to a 16-fold increasein CAMP production (Fig. 2), its multiplicity of actions. Upon removal of serum,cells rapidly was lesseffective than the B subunit in inhibiting DNA syn- differentiated and acquired morphological characteristics of thesis.In cells grown in 1% FCS for 1 d, a maximum inhibition neurons(cf. Figs. 3A, 4A). Neuritogenesiswas apparent after 12 of DNA synthesisof 34% or 26% was induced by 2 Kg/ml of hr of exposure,but the cells were not scoredfor neurite-bearing either the B subunit or the holotoxin, respectively. cells until 24 hr. After 24 hr treatment with the B subunit, the percent of cells bearing neurites increasedfrom 23.3 f 3% to Inhibition of DNA synthesisby the B subunit in N18 cells after 57 f 4.5%. Cellular differentiation was marked by appearance treatment with NA of more numerous spinelike processesand a more branched BecauseNA treatment is known to increaselevels of ganglioside neurite pattern. Cells with elongatedneurites were also signifi- GM 1 in N18 cells (Miller-Podraza and Fishman, 1982; Miller- cantly increased(Fig. 4B). Again, treatment with NA alone had Podraza et al., 1982) it was of interest to study the effect of the no effect on neuritogenesis.The effect of NA on B subunit-

Table 2. Effect of NA treatment on ganglioside composition in neuroblastoma N18 cells

Sialicacid (nmol/mgprotein) % of total gangliosides Ganglioside Control NA Control NA GDla 1.82 0.38 44 11 GM1 0.56 1.96 13 59 GM2 1.03 0.77 25 23 GM3 0.72 0.20 I7 6 Total 4.13 3.3 N I8 cells were incubated in the presence or absence of NA (0.0 1 U/ml) for 4 hr and analyzed for gangliosides as described previously (Masco and Seifert, 1990). Ths Journal of Neuroschce, August 1991, 17(E) 2447

Figure 3. B subunitinduces morpho- logicaldifferentiation ofneuroblastoma N18 cells.Cells were grown in DMEM containing5% FCS for 1 d and then incubatedin the absence(A) or in the presence(B) of the B subunit(4 &ml). C, NA (0.01 U/ml) alone.D, NA to- getherwith the B subunit.Twenty-four hoursafter the varioustreatments, the cellswere fixed, andphotographs were takenat 100x magnification. induced differentiation in serum-free medium was only mar- ginal. This is likely due to the large degree of differentiation Lack of efect of thrombin on neurite outgrowth induced by the induced by the B subunit alone. In all of theseexperiments, the B subunit results of the effectsof the commercial and the cloned B subunit In view of the studiesindicating that neurite outgrowth activity were indistinguishable. Surprisingly, the effect of the intact CT might be controlled by thrombin (or thrombin-like proteases; on neuritogenesiswas similar to that induced by the B subunit Gurwitz and Cunningham, 1988), we examined the effect of despitethe large increasein CAMP. thrombin on neurite outgrowth producedby the B subunit. When Additional evidence for the distinct morphological alteration cells were cultured for 2 d in the absenceof serum, 15-20% of and neuritogenesisin responseto the B subunit was revealed the cells showedneurites. The addition of 1 &ml of thrombin by immunofluorescencewith fluorescein-labeledB subunit (Fig. causeda fast neurite retraction within 3-4 hr (Fig. 6A,C). The 5). The biological effect of the fluorescein-conjugatedB subunit ability of thrombin to block spontaneousdifferentiation was wascomparable to the unmodified B subunit (data not shown). readily reversible. When thrombin was removed, neurite ex- Similar to the unmodified B subunit, the fluorescein isothio- tension proceededin serum-free medium at the same rate ob- cyanate conjugated(FITC)-B subunit also induced largeincreas- served in cells not treated with thrombin. In sharp contrast to es in the percent of cells bearing net&es. Strong fluorescence its effect on neurite extension induced in responseto serum in the thick neurite trunks and in the numerous and thin fila- withdrawal, thrombin did not retract neurites extended in re- mentsaround the cell body, the growth cone, and especiallyin sponse to the B subunit (Fig. 6C,D). In addition, thrombin spinelike processeswas revealed in cells treated with FITC- treatment did not block the inhibition ofDNA synthesisinduced labeled B subunit for 24 hr (Fig. 5B). Incubations for shorter by the B subunit (Table 4). It is interestingto note that thrombin periods of time did not causedetectable morphological altera- alone has no effect on DNA synthesis,though it can completely tions, which were only apparent after 12 hr. reversespontaneous neurite outgrowth when serumis removed.

Table 3. Effect of NA treatment on the B subunit-mediated inhibition of ‘H-thymidine incorporation in neuroblastoma N18 cells

‘H-thymidineincorporation Culture (cpm x lO-3) condition Control NA B subunit NA plus B subunit MediumI 82.8 f 1 82.9 f 2 51.8 + 3 21.0 * 1 0% FCS 16.1-t 1 19.9-t 1 11.8+ 1 4.7 f 1 1% FCS 71.6 f 5 85.1 -t 2 64.5 + 2 32.2 + 4 2.5% FCS 92.5 + 1 83.6 + 1 19.6 f 2 38.3 + 2 5% FCS 61.1 + 2 51.4 + 2 48.6 + 1 28.1 Z!X1 Cells were grown for 3 d in the indicated medium and then incubated in the absence (control) or presence of NA (0.01 U/ml) or the B subunit (4 &ml) or NA (0.01 U/ml) plus the B subunit (4 &ml) for 20 hr. ‘H-thymidine incorporation was measured as described in Materials and Methods. Each value is the mean + SD of triplicate determinations from a representative experiment. Similar results were obtained in at least five additional experiments. 2442 Masco et al. l B Subunit-hduced Neuroblastoma Dientiaticm

Figure 4. B subunit induces neurite outgrowth in neuroblastoma N 18 cells grown in chemically defined medium. Cells were grown in chemically defined medium for 1 d as described in Mate- rials and Methods and then incubated in the absence(A) or in the presence (B) of the B subunit (4 &ml). C, NA (0.0 1 U/ml) alone. 0, NA together with the B subunit. Twenty-four hours after the various treatments, the cells were fixed, and photographs were taken at 200x magnification. The percentages of neu- rite-bearing cells were untreated con- trol, 23.3 2 396,B subunit, 56.7 If: 9%; NA alone, 30.8 + 8%; NA plus B sub- unit, 57.3 f 9%.

1980; Masco et al., 1989) and that following such insertion it Discussion can act as a functional receptor for CT (Fishman, 1982; Spiegel, Numerous reports have previously shown that both unfmction- 1988) lend support to the contention that in the nervous system ated gangliosides extracted from bovine and highly pu- the effects of exogenous ganglioside GM 1 may reflect the func- rified monosialoganglioside GM 1 stimulate neuronal sprouting tion of its endogenous counterpart. The examination of the role in vitro and in vivo (reviewed in Ledeen, 1989). However, the of endogenous gangliosides in modulating neuritogenesis pre- involvement of endogenous gangliosides in this process remains sents a more difficult task due to the lack of specific probes. One unclear. The observations that exogenous GM 1 is stably inserted notable exception is the B subunit of CT, which displays an into neuronal and glial plasma membranes (To&no et al., absolute affinity for ganglioside GM 1 (Fishman, 1982). The present study uses the B subunit to examine the role of ganglioside GM1 in the process of growth and differentiation of neuroblastoma N 18 cells. Our data demonstrate that binding of the B subunit to N18 cells induces a significant decrease in DNA synthesis and pronounced morphological changes: an in- crease in neurite outgrowth concomitant with more branched neurites and the formation of more spinelike processes. The inhibition of proliferation of the N 18 cells was more pronounced in serum-free medium. However, morphological changes were observed even at high serum concentrations. It should be noted that cells cultured in high serum content (5-1096) grew as ag- gregates and had lower rates of proliferation and decreased abil- ity to extend neurites. Although the underlying molecular mech- anism involved in the process of neuritogenesis is not yet known, the results presented here suggest that the effects are due to the binding of the B subunit to ganglioside GM1 and not due to

Table 4. Lack of effect of thrombin on DNA synthesis induced by the B subunit in neuroblastoma N18 cells

‘H-thymidineincorporation (cpm x lo-) per well) Addition Without B subunit With B subunit None 30.1 * 1 20.6 + 0.81 Figure 5. LlTC-labeled B subunit distribution in neuroblastoma N18 Thrombin 27.0 + 1.3 18.8.f 0.8 cells. N 18 cells grown in chemically defined medium on cover slides as described in Materials and Methods were exposed to FITC-B subunit Neuroblastoma N18 ceils were seeded in chemically defined medium (medium I) at a density of 2 x 10’ cells per well in 96-well clusters. Afier 1 d, the cells were for 24 hr fixed, and then examined by phase-contrast (A) and fluores- exposed to the indicated reagents and ‘H-thymidine incorporation. was measured cence microscopy (B). Treatment of the cells with a large excess of after 20 hr as described in Materials and Methods. Each value is the mean 2 SD unlabeled B subunit or with FITC-labeled B subunit preincubated with oftriplicate determinations from a representative experiment. Similar results were excess ganghoside GM1 completely inhibited the fluorescence. Mag- obtained in at least five additional experiments. The concentrations of the agents nifications, 400 x (top) and 200 x (bottom). were as follows: B subunit, 4 &ml; thrombin, 1 &ml. The Journal of Neuroscience. August 1991, ff(9) 2449

Figure 6. Lack of effect of thrombin on neurite outgrowth induced by the B subunit in neuroblastoma N18 cells. Cells were grown in chemically defined medium as described in Materials and Methods. After 20 hr in the absence IA. C) or presence (B, D) of B subunit-(4 &ml), the cells were treated with thrombin (1 &ml) for 4 hr (B, D) and fixed, and photographs were taken at 200 x magnification. A, Untreated con- trols. B, B subunit. C, Thrombin. D, B subunit plus thrombin. Note the ab- sence of neurite retraction in the B sub- unit-treated cells.

changes in CAMP levels resulting from any contaminating A course of this work, Wu and Ledeen(199 1) reported that treat- subunit. Several lines of evidence support this conclusion. First, ment of mouse neuroblastoma N2A and rat neuroblastomas we used highly purified cloned B subunit that cannot contain BSO and B104 with NA induced enhancedneurite outgrowth, any A subunit because it was isolated from a V. cholerae mutant while N 1A- 103 mouseneuroblastoma was unresponsive to NA. that only expresses the B subunit. Second, neither the com- Furthermore, NA-stimulated neuritogenesiswas blocked by the mercial preparation nor the cloned B subunit induced increases B subunit. These results are not readily reconcilable with the of CAMP in these cells, while under the same conditions, the results presentedhere. However, the possibility that different holotoxin (A and B subunits) caused dramatic increases in CAMP cell types responddifferently to the B subunit cannot be ruled levels. Third, there was a good correlation between the amount out. In this regard, a dual action of the B subunit was recently of B subunit bound to the cells and the biological effect. Finally, observed in different cell types. For example, the growth of treatment with NA, which caused a fourfold increase in the level resting thymocytes and 3T3 fibroblasts was stimulated by the of membrane GM1 as determined by iodinated-CT binding, B subunit, while the growth of transformed cells, such as ras- enhanced the biological effect of the B subunit. transformed 3T3 fibroblastsand C6 cells, wasinhibited (Spiegel However, experiments with NA are subject to some draw- et al., 1985b; Spiegel and Fishman, 1987; Facci et al., 1988; backs. NA not only removes the sialic acid residues of polysi- Skaper et al., 1988; Spiegel, 1988). Furthermore, experiments alogangliosides, converting them to GM 1, it also removes sialic with growing 3T3 fibroblasts revealed that a bimodal response acid from other gangliosides belonging to other series (neolacto, to the B subunit can be observed in the samecell line, depending globo), leading to increases in various other glycosphingolipids, on their growth state(Spiegel and Fishman, 1987).The biphasic and will hydrolyze terminal sialic acids from . The responseto the B subunit by the same 3T3 fibroblasts raised fact that NA treatment alone did not have significant effects, the possibility that endogenousganglioside GM 1 could function either on DNA synthesis or on morphology ofthe cells, indicates as a bimodal regulator of cell growth signals.Based on the ob- that elevations in the level of GM1 per se are not sufficient by servation that the levels of cell-surface GM1 were reduced in themselves to cause significant changes in cell growth or differ- the growing ras-transfectedcells and increasedas the untrans- entiation and that it is either the aggregation of GM1 on the fected NIH 3T3 cells becamecontact inhibited (Matyas et al., cell surface by the B subunit or the displacement of endogenous 1987), it has beenproposed that the ability of gangliosideGM 1 ligands from GM1 that is responsible for the effect. During the to regulate positive and negative signals may depend on its 2450 Masco et al. * B Subunit-induced Neuroblastoma Differentiation

density on the cell surface. However, further studies revealed traction of the neurites induced by the B subunit. Therefore, that this is an oversimplification. Using the unique ability of there is no indication of a possibleinvolvement of thrombin or gangliosides to be functionally inserted into the plasma mem- a thrombin-like protease in the control of neurite outgrowth brane of cells, it was found that, even after the levels of GM1 mediated through GM 1. on the surface of rapidly dividing cells were increased to levels Although the underlying mechanismof the B subunit effect found on confluent cells, the responsiveness to the B subunit remains elusive, cross-linking of GM1 by the B subunit may was not altered. Furthermore, similar results were obtained when modulate ion channels in the plasma membrane. Recently, we the level of endogenous GM 1 was increased by treatment of the found that the B subunit can modulate calcium permeability in cells with NA to convert polysialogangliosidesto GM 1. There- thymocytes as well as in fibroblasts (Dixon et al., 1987; Spiegel fore, these experiments indicate that the bimodal response to and Panagiotopoulos, 1988). The B subunit may have similar the B subunit is not solely a function of the concentration of effects on neuroblastoma cells. Many studies have implicated cell-surface GM 1; rather, it is the growth stage that determines calcium ion in the regulation of neurite outgrowth in both ver- the fate of the signal transduced by the interaction of the B tebrates and invertebrates, and in neurons in culture (Reboul- subunit and ganglioside GM1 (Buckley et al., 1990). In agree- leau, 1986; Kater et al., 1988). Ganglioside GM1 might have ment with this hypothesis, it was recently observed that the B an important role in regulating calcium fluxes and thus regulate subunit has a dual action in quiescent Swiss 3T3 fibroblasts, neurite elongation. This hypothesis is currently under investi- depending on the context ofother growth factors (Spiegel, 1989). gation. While the B subunit potentiated the effect of epidermal growth The results presented here suggest that endogenous ganglio- factor, insulin, bombesin, and platelet-derived growth factor, it side GM 1 may play a role in the differentiation and proliferation diminished the stimulation of DNA synthesis induced by phor- of neural cells. Based on its neural regenerative effects, gangli- bol esters via protein kinase C. Similar results were obtained in oside GM1 has been used successfully in animal models of quiescent rat thyroid FRTL-5 cells, where a biphasic response certain neurological disorders, such as for diabetic neuropathy was also observed (Tetsumoto et al., 1988). (Calcutt et al., 1988) and for acceleration of functional recovery It is clear that the cessation of growth is usually a necessary after CNS damage(Karpiak et al., 1987). The useof ganglioside condition for the onset of differentiation in most cells. The B GM 1 hasrecently been extended to human treatment. Our find- subunit inhibits DNA synthesis of neuroblastoma N 18 cells and ings suggestthat these therapeutic effects of gangliosideGM1 induces enhanced neuritogenesis, an indicator of differentiation. might be a reflection of their normal function in neuronal cells. Similar to the observations presentedhere, it has recently been Further studies addressingthe underlying molecular mecha- shown that in astroglial cells the B subunit inhibited DNA syn- nisms of the role of gangliosides in neural growth and differ- thesisconcomitant with the induction of a classicalstar-shaped entiation are presently under investigation. (stellate) morphology, a processassociated with a more differ- entiated state(Facci et al., 1988). It is possiblethat the B subunit exerts its effect by binding to endogenous ganglioside GMl, References thereby displacing or preventing the binding of a factor that Blackbum CC, Swank-Hill P, Schnaar RL (1986) Gangliosides support normally stimulates growth. Support for this hypothesis comes neural retina cell adhesion. J Biol Chem 26 I :2873-288 1. from a recent study of the effect of exogenousGM 1 on the growth Buckley N, Matyas G, Spiegel S (1990) The bimodal growth response of Swiss 3T3 cells to the B subunit of cholera toxin is independent of S2OY murine neuroblastoma. The portion of the density of its receptor, ganglioside GM I. Exp Cell Res 189: 13- aloneof GM 1 enhancedneuritogenesis of thesecells to the same 21. extent as intact GM 1 (Schengrundand Prouty, 1988). Because Byrne MC, Ledeen RW, Roisen FJ, Yorke G, Sclafani JR (1983) Gan- it is most unlikely that the oligosaccharideportion penetrates glioside-induced neuritogenesis: verification that gangliosides are the active agents, and comparison of molecular species. J Neurochem 4 1: the plasma membrane, it has been suggestedthat S2OY neu- 1214-1222. roblastoma cells have cell-surface gangliosidebinding sitesthat Calcutt N, Tomlinson D, Willars G (1988) Ganghosides treatment of are crucial for cell growth regulation and that recognize the streptozotocin-diabetic rats prevents defective axonal transport of oligosaccharide moiety of GM 1 (Schengrund and Prouty, 1988). 6-phosphofmctokinase activity. J Neurochem 50: 1478-1483. Evidence is also accumulating for the existence of cell-surface Cannella M, Wu G, Vaswani K, Ledeen R (1988) Neuritogenic effects of exogenous gangliosides and synthetic sialoglycolipids: comparison ganglioside binding proteins (Blackbum et al., 1986; Tiemeyer to endogenous ganghoside requirements. In: New trends in ganglioside et al., 1989). research: neurochemical and neurogenerative aspects (Ledeen R, Ho- From another point of view, the B subunit might mimic the gan E, Tettamanti G, Yates A, Yu R, eds), pp 379-390. Padova: effect of an unknown neuronotrophic factor that binds to GM 1. Liviana. Dekker A, Manthorpe M, Varon S (1990) Reversibility of ganglioside In this respect, it was demonstrated that conditioned media effects on astrocyte morphology. J Neurosci Res 26:349-355. derived from several cell types produced neuritogenesisin chick Dimpfel W, Moller W, Mengs U (198 1) Ganglioside induced neurite dorsal root ganglia in culture and that monoclonal antibodies formation in cultured neuroblastoma cells. In: Gangliosides in neu- to GM 1 blocked this effect (Spoeti et al., 1988). These results rological and neuromuscular function, development, and repair (Rap- suggest the presence in the conditioned media of a substance port MM, Gorio A, eds), pp 119-134. New York: Raven. Dixon SJ, Stewart D, Grinstein S, Spiegel S (1987) Transmembrane capableof binding to GM1 and inducing neurite outgrowth. It signaling by the B subunit of cholera toxin: increased cytoplasmic free is tempting to speculate that this endogenousneuritogenesis calcium in rat lymphocytes. J Cell Biol 105: 1153-l 16 1. factor could be similar to the B subunit. The mechanism of B Doherty P, Walsh FS (1987) Ganglioside GM 1 antibodies and B-chol- subunit-induced differentiation is different than that of spon- era toxin bind specifically to embryonic chick dorsal root ganglion neurons but do not modulate neurite regeneration. J Neurochem 4: taneous differentiation. It has been shown that thrombin, a ser- 1237-1244. ine proteasepresent in normal serum,inhibits neurite outgrowth Dreyfus H, Ferret B, Harth S, Gorio A, Freysz L, Massarelli R (1984) induced by the removal of serum from the media (Gurwitz and Effect ofexogenous ganghosides on the morphology and biochemistry Cunningham, 1988). In contrast, thrombin did not cause re- of cultured neurons. In: Ganglioside structure, function, and bio- The Journal of Neuroscience, August 1991, 7 f(8) 2451

medical potential (Ledeen RW, Yu RK, Rapport MM, Suzuki K, Prasad K, Hsie A (197 1) Morphological differentiation of mouse neu- eds), pp 5 13-524. New York: Plenum. roblastoma cells induced in vitro by dibutyryl adenosine 3’:5’cyclic Duffard RO, Fishman PH, Bradley RM, Lauter CJ, Brady RO, Trams monophosphate. Nature (New Biol) 233: 14 1-142. EG (1977) Ganglioside composition and biosynthesis in cultured Reboulleau CP (1986) Extracellular calcium-induced neuroblastoma cells derived from CNS. J Neurochem 28: 1161-l 166. cell differentiation: involvement of phosphatidylinositol turnover. J Facci L, Leon A, Toffano G, Sonnino S, Ghidoni R, Tettamanti G Neurochem 46:920-930. (1984) Promotion of neuritogenesis in mouse neuroblastoma cells Richter-Landsberg C, Jastorff B (1986) The role of CAMP in nerve by exogenous gangliosides. Relationship between the effect and the growth factor-promoted neurite outgrowth in PC 12 cells. J Cell Biol cell association of ganglioside GM 1. J Neurochem 42:299-305. 102:821-829. Facci L, Skaper S, Favaron M, Leon A (1988) A role for gangliosides Roisen FJ, Bartfeld H, Nagele R, Yorke G (198 1) Ganglioside stim- in astroglial cell differentiation in vitro. J Cell Biol 106:82 l-828. ulation of axonal sprouting in vitro. Science 2 14:577-578. Ferrari G; Fabris M, Gorio A (1983) Gangliosides enhance neurite Rybak S, Ginzburg I, Yavin E (1983) Gangliosides stimulate neurite outerowth in PC 12 cells. Dev Brain Res 8:2 15-222. outgrowth and induce tubulin mRNA accumulation in neural cells. FishGan PH (1980) Mechanisms ofthe action ofcholera toxin: studies Biochem Biophys Res Commun 116:974-980. of the lag period. J Membr Biol 54:6 l-72. Schennrund CL. Proutv CP (1988) The oliaosaccharide oortion of Fishman PH (1982) Role of membrane gangliosides in the binding GM> enhances process formation by S20Y-neuroblastoma cells. J and action of bacterial toxins. J Membr Biol 69:85-97. Neurochem 5 1:277-282. Gilman AG (1984) Proteins and dual control of adenylate cyclase. Schwartz M, Spirman N (1982) Sprouting from chicken embryo dorsal Cell 36:577-579. root ganglia induced by nerve growth factor is specifically inhibited Gurwitz D, Cunningham DD (1988) Thrombin modulates and re- by affinity-purified antiganglioside antibodies. Proc Nat1 Acad Sci verses neuroblastoma neurite outgrowth. Proc Nat1 Acad Sci USA USA 79:6080-6083. 85:3440-3444. Skaper S, Facci L, Favaron M, Leon A (1988) Inhibition of DNA Hakomori S (198 1) Glycosphingolipids in cellular interaction, differ- synthesis in C6 glioma cells following cellular incorporation of GM 1 entiation and oncogenesis. Annu Rev Biochem 50:733-764. ganglioside andcholeragenoid exposure. J Neurochem 5 1:688-697. Hauw JJ, Fenelon S, Boutry JM, Nagai Y, Escourolle R (1981) Effects Spiegel S (1988) Insertion of ganglioside GM 1 into rat glioma C6 cells of brain gangliosides on neurite outgrowth in guinea pig spinal ganglia renders them susceptible to growth inhibition by the B subunit of tissue culture and on fibroblast cell culture. In: Gangliosides in neu- cholera toxin. Biochim Biophys Acta 9691249-256. rological and neuromuscular function, development, and repair (Rap- Soieael S (1989) Inhibition of urotein kinase C-deoendent cellular port MM, Gorio A, eds), pp 17 l-l 76. New York: Raven. _priliferacon b; interaction of endogenous ganglioside GM 1 with the Karpiak SE, Li YS, Mahadik SP (1987) Ganglioside treatment: re- s subunit of the cholera toxin. J Biol Chem 264: 165 12-1657. duction of CNS injury and facilitation of functional recovery. Brain Snieeel S. Fishman PH f 1987) Ganaliosides as bimodal regulators of Injury 1:141-170. ‘ceil grdwth. Proc Nat1 kcad’sci USA 84:141-147. - Kater S, Mattson M, Cohan C, Connor J (1988) Calcium regulation Spiegel S, Panagiotopoulos C (1988) Mitogenesis of 3T3 fibroblasts of the neuronal growth cone. Trends Neurosci 11:3 15-32 1. induced by endogenous gangliosides is not mediated by CAMP, pro- Katoh-Semba R, Skaper SD, Varon S (1984) Interaction of GM1 tein kinase C, or phosphoinositides turnover. Exp Cell Res 177:4 14- ganglioside with PC12 pheochromocytoma cells: serum and NGF- 427. dependent effects on neurite growth and proliferation. J Neurosci Res Spiegel S, Yamada KM, Horn BE, Moss J, Fishman PH (1985a) Flu- 12:299-310. orescent gangliosides as probes for the retention and organization of Ledeen RW (1985) Gangliosides of the neuron. Trends Neurosci 8: fibronectin by ganglioside-deficient mouse cells. J Cell Biol 100:721- 169-174. 726. Ledeen RW (1989) Biosynthesis, and biological effects of Spiegel S, Fishman PH, Weber RJ (1985b) Direct evidence that en- gangliosides. In: Neurobiology ofglycoconjugates (Margolis RU, Mar- dogenous GM 1 ganglioside can mediate thymocyte proliferation. Sci- golis RK, eds), pp 43-83. New York: Plenum. ence 230:1285-1287. Leon A, Facci L, Benvegnu D, Toffano G (1982) Morphological and Spirman N, Sela B, Schwartz M (1982) Antiganglioside antibodies biochemical effects of gangliosides in neuroblastoma cells. Dev Neu- inhibit neuritic outgrowth from regenerating goldfish retinal explants. rosci 5:108-l 14. J Neurochem 39:8?4-877. - -- Liu Y, Fiske W, Chen Y (1980) Regulation of cyclic adenosine 3’:5’- Sooerri P. RaDDOrt M. Mahadik S. Roisen F (1988) Inhibition of monophosphate-binding protein in N- 18 mouse neuroblastoma cells. ‘conditidned’&edia-mediated neuritogenesis df sensory ganglia by Cancer Res 40:4100-4108. monoclonal antibodies to GM 1 ganglioside. Brain Res 469:7 l-77. Lowry OH, Rosenbrough NJ, Farr AL, Randall RJ (1951) Protein Spoerri PE (1983) Effects of gangliosides on the in vitro development measurement with the Folin phenol reagent. J Biol Chem 193:265- of neuroblastoma cells: an ultrastructural study. Int J Dev Neurosci 275. 1:383-391. Mahadik SP, Laev H, Rapport MM (1986) Preparation and specificity Svennerholm L (1963) Chromatographic separation of human brain of 11 monoclonal antibodies to GM 1 ganglioside. J Neurochem 47: gangliosides. J Neurochem 10:6 13-623. 1172-l 175. Sweeley CC, Siddiqui B (1977) Chemistry of mammalian . Masco D, Seifert W (1990) Gangliosides in lesion-induced synapto- In: The , Vol 1 (Pigman W, Horowitz MI, eds), pp genesis: studies in the hippocampus of the rat brain. Brain Res 5 14: 459-540. New York: Academic. 84-92. Taylor RK, Miller VL, Furlong DB, Mekanos JJ (1987) Use of PhoA Masco D, Flott B, Seifert W (1989) Astrocytes in cell culture incor- gene fusion to identify a p&s colonization factor coordinately reg- porate GM 1 ganglioside. Glia 2:23 l-240. ulated with cholera toxin. Proc Nat1 Acad Sci USA 84:2833-2837. Matta SG, Yorke G, Roisen F (1986) Neuritogenic and metabolic Tetsumoto T, Takada K, Amino N, Miyai K (1988) Regulation of effects of individual gangliosides and their interaction with nerve proliferation by the cholera toxin B subunit in FRTL-5 cells may growth factor in cultures of neuroblastoma and pheochromocytoma. involve a mechanism independent from the modulation of membrane bev Brain Res 271242-252. receptor function. Biochem Biophys Res Commun 157:605-610. Matvas G. Aaronson S. Bradv R. Fishman PH (1987) Alteration of Tiemeyer M, Yasuda Y, Schnaar R (1989) Ganglioside-specific bind- &colipids in ras-transfecteh NIH 3T3 cells. Pro; NaJ Acad Sci USA ing protein on rat brain membranes. J Biol Chem 264: 167 l-l 68 1. 84:6065-6068. Toffano G, Benvengnu D, Bonetti A, Facci L, Leon A, Orlando P, Miller-Podraza H, Fishman PH (1982) Translocation of newly syn- Ghidoni R, Tettamanti G (1980) Interaction of GM 1 ganglioside thesized gangliosides to the cell surface. Biochemistry 21:3265-3270. with crude rat brain neuronal membranes. J Neurochem 35:861-866. Miller-Podraza H, Bradley R, Fishman PH (1982) Biosynthesis and Tsuji S, Arita M, Nagai Y (1983) GQ 1b, a bioactive ganglioside that localization of gangliosides in cultured cells. Biochemistry 2 1:3260- exhibits novel nerve growth factor (NGFl-like activities in the two 3265. neuroblastoma cell lines. J Biochem‘(Tokyo) 94:303-306. Morgan J, Seifert W (1979) Growth factors and gangliosides: a possible Van de Walle M. Fass R. Shiloach J (1990) Production of cholera new perspective in neuronal growth control. J Supramol Struct 10: toxin subunit B’by a mutant strain of kibrid cholera. Appl Microbial 111-124. Biotechnol 33:389-394. 2452 Masco et al. l B Subunit-induced Neuroblastoma Differentiation

Wu G, Ledeen R (199 1) Stimulation ofneurite outgrowth in neuroblas- Yates AJ (1986) Gangliosides in the nervous system during devel- toma cells by neuraminidase: putative role of GM1 ganglioside in opment and regeneration. Neurochem Path01 5:309-329. calcium-mediated differentiation. J Neurochem 56:95-104.