Inhibition of endogenous ganglioside synthesis does not block neurite formation by retinoic acid-treated neuroblastoma cells.

Gangliosides are believed to play a critical role in cellular differentiation. To test this concept, we determined the effect of inhibition of endogenous ganglioside synthesis upon neurite formation induced by retinoic acid in LAN-5 human neuroblastoma cells. Ganglioside synthesis and content of LAN-5 cells exposed for 6 days to 10 μM D-threo-1-phenyl-2-decanoylamino-3-morpholino-1-propanol (D-PDMP) (an inhibitor of glucosylceramide synthase) were reduced by >90%. However, these ganglioside-depleted cells were not blocked from forming neurites when exposed to 10 μM retinoic acid. Even more extensive treatment of LAN-5 cells with 20 μM D-PDMP (6 day pretreatment followed by 6 days together with 10 μM retinoic acid) still did not block the retinoic acid-induced neurite formation. An element of neuroblastoma tumor cell differentiation, neurite formation, is therefore dependent neither on an intact cellular ganglioside complement nor on new ganglioside synthesis.

The high concentrations of gangliosides in the central nervous system have led to the widely held view that these molecules may play an important role in the differentiation of neuronal cells (1). There is evidence supporting this view, in that striking changes in ganglioside metabolism have been observed during both spontaneous and induced cellular differentiation (2)(3)(4). The treatment of LAN-5 human neuroblastoma cells with retinoic acid, which causes differentiation of the cells (5,6), results in a marked increase in cellular ganglioside content (7). An increase in concentration of specific gangliosides (8), a switch in ganglioside biosynthetic pathways and the appearance of new gangliosides (9,10), and changes in cellular ganglioside patterns (11) are generally considered to be important in the process of cellular differentiation, particularly in neuronal or neuroblastoma cells. Exogenous gangliosides also influence differentiation (12)(13)(14). However, the question of whether endogenous gangliosides are essential for cellular differentiation under physiological conditions remains to be answered.
To determine whether endogenous gangliosides are essential for the process of cellular differentiation, we developed the following strategy. In a model system, the LAN-5 human neuroblastoma cell line, which can be induced by retinoic acid to differentiate and form neurites (5,6), we studied the effect of cellular ganglioside depletion upon neurite formation. D-threo-1-Phenyl-2-decanoylamino-3-morpholino-1-propanol (D-PDMP), 1 a potent inhibitor of glucosylceramide synthase (15,16), was used to maximally inhibit ganglioside synthesis and deplete cellular ganglioside content. In this situation, despite downregulation of cellular ganglioside synthesis and content of LAN-5 cells, the formation of neurites induced by exposure of these cells to retinoic acid was not blocked.

Preparation of Retinoic Acid, D-PDMP, and Nerve Growth Factor
Stock Solutions-Retinoic acid (all-trans form; Sigma) was dissolved in ethanol to a concentration of 10 Ϫ2 M and kept as stock solution. For each experiment, retinoic acid was diluted from the stock solution directly into the growth medium. The final concentration of ethanol in the culture medium was Յ0.2% (v/v). D-PDMP was dissolved in distilled water at a concentration of 4 mM (15) and stored at 4°C. This stock solution was directly added to culture medium (Յ0.5%, v/v) in each experiment. Nerve growth factor (human, recombinant NGF-␤, Sigma) was dissolved in phosphate-buffered saline with 0.1% bovine serum albumin. This stock solution was filtered and stored at Ϫ70°C.
Induction of LAN-5 Cellular Differentiation-LAN-5 cells were treated with 10 M all-trans-retinoic acid to induce cellular differentiation, which was judged by assessing neurite formation, cell proliferation (5,6), and specific acetylcholinesterase activity (17). Neurite formation was observed and photographed under phase-contrast microscopy. Cells cultured in medium containing 0.2% ethanol were used as the control. Nerve growth factor (250 -1000 ng/ml) was also used to induce neurite formation in LAN-5 cells.
Study of Ganglioside Biosynthesis by Metabolic Radiolabeling-LAN-5 cells were cultured in the presence or absence of D-PDMP for 72 h. D-[6-3 H]Galactose (specific activity, 5 Ci/mmol) and D-[6-3 H]glucosamine hydrocholoride (specific activity, 30.9 Ci/mmol, DuPont NEN, Boston, MA) were added to the culture medium (1 Ci ml Ϫ1 ) during the last 24 h. These radiolabeled cells were washed 3 times and harvested by trypsinization. The resulting cell suspension was centrifuged at 300 ϫ g for 10 min, and the cell pellet was washed once with phosphatebuffered saline prior to processing for ganglioside purification (7,18).
Ganglioside Purification and Analysis-The total lipids of the cells were isolated by chloroform/methanol extraction. Gangliosides were purified by diisopropyl ether/1-butanol partition (19) followed by Sephadex G-50 gel filtration. The purified gangliosides were quantified as nanomole of lipid-bound sialic acid (LBSA) by the modified resorcinol method (20), and ganglioside-associated radioactivity was quantified by ␤-scintillation counting (Betafluor, National Diagnostics, Manville, NJ).

Regulation of Cellular Gangliosides by Retinoic Acid and D-PDMP-LAN-5 human neuroblastoma cells contain G D2
(56%), G M2 (15%), and G T1b (11%) as the main ganglioside components. The minor ganglioside species also seen are G M1 , G D3 , and G D1b (18). To provide the basis for studying the influence of endogenous gangliosides on cellular differentiation in this model system, we first established the effects of each agent on ganglioside synthesis and content. Retinoic acid, which induces differentiation, caused an increase in cellular ganglioside content. When LAN-5 cells were treated with 20 M retinoic acid for 6 days, the cellular ganglioside content increased from 3.95 Ϯ 0.39 (control) to 6.1 Ϯ 0.13 nmol of LBSA/mg of protein ( Fig. 1). In contrast, D-PDMP (which blocks ganglioside synthesis) caused depletion of cellular gangliosides. Exposure of LAN-5 cells to 20 M D-PDMP for 6 days reduced cellular ganglioside content by 90%, to 0.41 Ϯ 0.29 nmol of LBSA/mg of protein ( Fig. 1). Moreover, as shown in Fig. 2, the inhibition of endogenous ganglioside content caused by D-PDMP was rapid and time-dependent; when LAN-5 cells were cultured in medium containing 10 M D-PDMP, the cellular ganglioside content decreased rapidly, with a 42% reduction in cellular ganglioside content observed after 1 day. By day 6, the total cellular ganglioside content was reduced by 92%, to 0.34 from 4.1 nmol of LBSA/mg of protein. This marked cellular ganglioside depletion was confirmed by a metabolic radiolabeling study, in which we measured the inhibitory effect of D-PDMP on ganglioside synthesis. Following the treatment of LAN-5 cells with 10 M D-PDMP for 72 h, ganglioside synthesis was already reduced by 90%, to 378 from 3495 dpm/mg of protein/h by the control cells.
Having found that retinoic acid treatment causes an increase in ganglioside content and that D-PDMP blocks ganglioside synthesis and causes a depletion of cellular gangliosides, it was important to establish the combined effect of these two agents on cellular gangliosides, since the subsequent experiments depended on blocked ganglioside synthesis and cellular ganglioside depletion in order to be able to test the role of endogenous ganglioside synthesis and content in differentiation. Therefore,  Table I, LAN-5 cells treated with both retinoic acid and D-PDMP during the second 6-day period of culture had essentially the same cellular ganglioside content as did the cells treated with D-PDMP alone (0.54 versus 0.40 nmol of LBSA/mg of protein). Thus, the inhibitory effect of D-PDMP on ganglioside synthesis was dominant over the stimulatory effect of retinoic acid on ganglioside synthesis by LAN-5 human neuroblastoma cells. Under this condition, the two agents were not toxic to the cells, as evidenced by cell viability of Ͼ95%. Therefore, any effects on differentiation observed in the subsequent experiments would reflect a state of ganglioside depletion.
Differentiation of Retinoic Acid-treated LAN-5 Cells-We chose the human neuroblastoma cell line, LAN-5, which can be induced to differentiate by retinoic acid, as a model system to study whether depletion of cellular gangliosides prevents cellular differentiation. As previously shown (5,21), LAN-5 human neuroblastoma cells propagated in the usual cell culture medium had large cell bodies and few, short processes (Fig. 3). The effect of retinoic acid treatment (10 M for 6 days) was the induction of striking morphological changes, including rounding up of the cell body, extension of long processes with the appearance of neurites, and cell aggregation into tight clusters (6,21). The cells still retained their long processes for at least 6 days after retinoic acid was removed (not shown). Biochemically, this neurite formation was accompanied by an increase in cellular ganglioside content, as discussed above.
D-PDMP, on the other hand, had no differentiating effect on LAN-5 cells, a finding consistent with a previous report that D-PDMP does not affect differentiation of HL-60 cells (22). Maximal exposure of LAN-5 human neuroblastoma cells to D-PDMP (20 M for 6 days), which caused 92% reduction in cellular ganglioside content and had a growth inhibitory effect resulting in a lower cell density, did not cause any cell morphological changes in the cells compared to the control culture (Fig. 3). Specifically, the cells bodies remained spread out, the processes that were present were short, and despite the lower cell density, which resulted from the growth inhibitory effects of D-PDMP, the cells remained evenly distributed on the surface of the flask (i.e. not aggregated).
Study of the combined effects of D-PDMP and retinoic acid tested the necessity of endogenous gangliosides (inhibited by D-PDMP) for cellular differentiation (induced by retinoic acid). Surprisingly, LAN-5 cells cultured in medium containing both 10 M retinoic acid and 20 M D-PDMP for 6 days underwent exactly the same morphological changes (neurite formation, rounding up of the cell body, and cell aggregation into tight clusters) as did the cells treated with retinoic acid alone (Fig.  3). It should be emphasized that this neurite formation (Fig. 3) occurred despite the fact that D-PDMP had caused a state of almost completely blocked ganglioside synthesis and depleted cellular gangliosides (Figs. 1 and 2).
The observation of lack of inhibition by D-PDMP of induced cell differentiation was confirmed with a second experimental approach, even more prolonged exposure of LAN-5 cells to D-PDMP. In this experiment, LAN-5 cells were exposed to 20 M D-PDMP for an initial 6-day period, and then reseeded and cultured for an additional 6-day period under the four different conditions used in the experiment shown in Table I (0.2%  ethanol, 10 M retinoic acid, 20 M D-PDMP, or 10 M retinoic acid and 20 M D-PDMP). By the end of the initial 6-day culture period in D-PDMP alone, i.e. before the addition of retinoic acid, the cells were essentially depleted of gangliosides (Figs. 1 and 2). Exposed for 6 more days to D-PDMP as well as to retinoic acid, these cells once again demonstrated the neurite formation, rounding up, and aggregation associated with retinoic acid-induced cellular differentiation (Fig. 4). Therefore, the neurite formation of LAN-5 cells usually induced by retinoic acid was not blocked by D-PDMP, despite the depletion of cellular gangliosides caused by D-PDMP. L-threo-PDMP, an isomer of D-threo-PDMP, was also studied. It stimulated ganglioside synthesis, increasing cellular ganglioside content from 3.95 Ϯ 0.39 to 5.06 Ϯ 0.64 (n ϭ 3) nmol of LBSA/mg of protein when the cells were exposed to 20 M L-PDMP for 6 days, which is consistent with a previous report (23). However, it had no effect on neurite formation, nor did it inhibit retinoic acidinduced neurite formation (Fig. 5).
To further assess the influence of D-PDMP on the incidence and length of neurites, LAN-5 cells were seeded at a lower cell density (2 ϫ 10 4 cells/cm 2 ), and neurite response was quantified as described previously (5). The results (Table II) clearly showed the lack of influence of D-PDMP on the number and length of neurites induced by retinoic acid. In addition, LAN-5 cell differentiation was also assessed by measurement of acetylcholinesterase activity (17). Whereas retinoic acid caused a 2-fold increase in the acetylcholinesterase activity, consistent with previous studies (24) (Table II). Together, these findings lead to the conclusion that induced differentiation of neuroblastoma cells does not require either intact endogenous ganglioside synthesis or content.

Stimulation of Neurite Formation in Nerve Growth
Factortreated LAN-5 Cells-We used NGF to induce neurite formation to provide an additional confirmation of these findings, since NGF-dependent trk A tyrosine phosphorylation followed by signal transduction leading to MAP kinase activation (25,26) has been proposed to be an important mechanism for neuronal cell differentiation. NGF has already been shown to stimulate tyrosine phosphorylation of NGF receptor (p140 prototrk ) in LAN-5 cells (25). It seemed likely, therefore, that NGF should stimulate neurite formation in these LAN-5 cells. This was in fact the case; exposure of LAN-5 cells to 250-1000 ng/ml NGF for 8 days resulted in neurite formation. We next determined the effect of inhibition of endogenous ganglioside synthesis by D-PDMP on the NGF-mediated process. Just as in the case of retinoic acid-induced neurite formation, D-PDMP (20 M, which completely inhibits ganglioside synthesis) did not block neurite formation stimulated by NGF (Fig. 6). These results support the conclusion that induced differentiation of neuroblastoma cells, whether by retinoic acid or by NGF, does not require either intact endogenous ganglioside synthesis or content. DISCUSSION There is great interest in the elucidation of biological functions of glycosphingolipids in general, and of gangliosides in particular. The involvement of gangliosides in neuronal cell differentiation of the nervous system is suggested by both qualitative and quantitative changes in ganglioside metabolism during brain development (27). Significant changes in human brain gangliosides have been observed in comparing the ganglioside complement of early fetal brain to that of adult brain (3). The ganglioside accretion is largest during the period of dendritic arborization and synaptogenesis (27). Developmen-  Table I) in this experiment, these cells still underwent cellular differentiation. Magnification: ϫ 400.   (5): cells were scored as morphologically differentiated if they possessed one or more processes at least twice as long as the soma diameter. Results were recorded as % differentiated cells in each culture: 4ϩ Ͼ 76%; 3ϩ ϭ 51-75; 2ϩ ϭ 26 -50%; 1 Ͻ 25%. Acetylcholinesterase (AChE) activity was measured spectrophotometrically as described previously (17). The data are the mean Ϯ S.D. of triplicate samples from a typical experiment. tal changes in ganglioside patterns were also observed in chick retina and brain (28) as well as in primary culture of rat neurons (29 -31). Thus, ganglioside synthesis and expression have been considered to be an index of cellular differentiation in brain (32). Striking changes in ganglioside synthesis have also been observed during cellular differentiation induced by certain pharmacological agents. Exposure of neuroblastoma cells to retinoic acid causes differentiation of the cells (5, 6) as well as a marked increase in cellular ganglioside content (7). More recent studies show that both induced and spontaneous neuritogenesis are associated with enhanced expression of ganglioside G M1 in the nuclear membrane (8), where this ganglioside has been proposed to modulate nuclear calcium homeostasis during neurite outgrowth (12). An increase in concentration of specific gangliosides (8), the appearance of new gangliosides, and a switch in the ganglioside biosynthetic pathway have also been observed in other cell systems (9,10). These changes in glycolipids were accompanied by the regulation of corresponding glycosyltransferases (33)(34)(35). Taken together, these studies clearly demonstrate that the process of cell differentiation, whether normal or pharmacologically induced, influences the biosynthesis of gangliosides.
The converse question is whether gangliosides influence the process of cell differentiation. Numerous studies have demonstrated that exogenous gangliosides are active. They stimulate axonal sprouting in vitro (36) and induce cellular differentiation with neurite formation in neuroblastoma cells (37)(38)(39)(40), although the mechanism is unknown. Another example is that very low concentrations (2-5 nM) of ganglioside G Q1b can induce neuroblastoma cells to differentiate (12). Human leukemia HL-60 cells can also be induced to differentiate, by incubation of the cells with exogenous ganglioside G M3 or sialylparagloboside (13,14). A final example is that of ganglioside G M1 , which enhanced neurite outgrowth elicited by nerve growth factor in two cell lines (41,42).
Modulation of endogenous gangliosides has also been shown to influence the process of cellular differentiation. For example, treatment of neuroblastoma Neuro-2a cells with cholera toxin B-subunit and anti-G M1 antibody inhibited neurite outgrowth (43), and transfection of these same cells with a mammalian expression vector containing the cDNA encoding G D3 synthase caused the expression of G D3 synthase and changes in the cell morphology, consistent with cholinergic differentiation of neuroblastoma cells (44). Other investigators transfected F-11 neuroblastoma cells with the O-acetylesterase gene from influenza C virus (45). This altered the expression of G D3 and O-acetylated G D3 , and caused morphological changes (45). Finally, treatment of HL-60 cells with G M2 and G D3 ganglioside synthase antisense oligomers resulted in alteration of ganglioside synthesis and composition and differentiation of the cells (11). On the other hand, inhibition of glycosphingolipid synthesis by D-PDMP did not affect fish embyo development (46).
The extensive studies reviewed above notwithstanding, the question of whether endogenous gangliosides are essential for cellular differentiation under physiological conditions remained to be resolved. This was the subject of the present investigation. We studied the influence of depletion of cellular gangliosides on cellular differentiation using D-PDMP, a competitive inhibitor of glucosylceramide synthase (15), which is the first enzyme for glycosphingolipid (and ganglioside) synthesis (16). If an intact ganglioside complement and/or ganglioside synthesis are critical for cellular differentiation induced by retinoic acid, the depletion of gangliosides should have prevented the neurite formation. To our surprise, blockade of ganglioside synthesis and depletion of cellular gangliosides had no discernable inhibitory effect on the formation of neurites induced by retinoic acid in human neuroblastoma LAN-5 cells. We confirmed this conclusion by showing that NGF-induced neurite formation (47) in LAN-5 cells likewise was not inhibited by blockade of ganglioside synthesis by D-PDMP.
The process of cellular differentiation includes (i) phenotypic changes of tumor cells associated with arrest of their proliferation, and (ii) phenotypic changes of normal cells in vitro, or most significantly in vivo, as part of maturation. Our studies have addressed the necessity of gangliosides in the former case, and have excluded the requirement of intact endogenous ganglioside metabolism and content for tumor cell differentiation. It will be very interesting now to address this same question in the context of normal cell differentiation, and particularly in vivo under physiological conditions, such as in a ganglioside knockout mouse.