Up-regulation of glucosylceramide synthesis upon stimulation of axonal growth by basic fibroblast growth factor. Evidence for post-translational modification of glucosylceramide synthase.

We have previously shown that ongoing glucosylceramide (GlcCer) synthesis is required for basic fibroblast growth factor (bFGF) and laminin to stimulate axonal growth in cultured hippocampal neurons (Boldin, S., and Futerman, A. H. (1997) J. Neurochem. 68, 882-885). We now demonstrate that stimulation of axonal growth by bFGF leads to an increase in the rate of GlcCer synthesis. Within minutes of incubation with bFGF, a significant increase in the rate of metabolism of [(14)C]hexanoyl ceramide to [(14)C]hexanoyl GlcCer is detected, but there are no changes in the rate of [(14)C]hexanoyl sphingomyelin synthesis. In vitro analysis of GlcCer synthase activity revealed an approximately 2-fold increase in the rate of [(14)C]hexanoyl GlcCer synthesis upon incubation with either bFGF or laminin; other growth factors, which did not effect the rate of axon growth, had no effect on the rate of [(14)C]hexanoyl GlcCer synthesis. The increased rate of [(14)C]hexanoyl GlcCer synthesis was not affected by preincubation with either cycloheximide or actinomycin, and no elevation of GlcCer synthase mRNA levels was detected, suggesting that GlcCer synthase is up-regulated by a post-translational mechanism. The relevance of these results for understanding the regulation of axonal growth is discussed.

Glucosylceramide (GlcCer), 1 the simplest glycosphingolipid, is synthesized from ceramide (Cer) on the cytosolic leaflet of the Golgi apparatus (1)(2)(3)(4). Glucosylceramide synthase (GCS; UDPglucose:N-acylsphingosine ␤1-1-glucosyltransferase) has recently been cloned (5,6) and the protein isolated (7). The activity of GCS is regulated under certain conditions including substrate depletion (8), keratinocyte differentiation (9), and during the acute phase response in liver (10). In these three cases, regulation of GCS activity occurs by a transcriptional mechanism. Such a mechanism provides a means to regulate GlcCer levels (and as a consequence, levels of down-stream lipids in the glycosphingolipid biosynthetic pathway (8)) over relatively long periods of time, such as which may occur during differentiation and development, but this mechanism could not be involved in regulating GlcCer levels for processes that re-quire a more rapid response.
One example of such a rapid response is the ability of neurons to react to external stimuli by altering the rate of axonal growth (11). When cultured hippocampal neurons are incubated with either basic fibroblast growth factor (bFGF) or laminin, axonal growth rates increase between 2-and 4-fold (12). The stimulatory effects of both factors can be completely abolished by co-incubation with inhibitors of GlcCer synthesis (12). Based on this and other studies (13)(14)(15), we have suggested that GlcCer synthesis may be a rate-limiting step in some facets of axon growth (11). If this is the case, then it might be predicted that stimulation of axon growth by bFGF or laminin would increase the rate of GlcCer synthesis. In the current study we demonstrate, by both in vivo and in vitro analyses, that GCS is indeed activated upon treatment of hippocampal neurons with bFGF and that up-regulation occurs by a posttranslational and not a transcriptional mechanism.
In Vivo Analysis of Sphingolipid Synthesis-Six coverslips containing 2-day-old neurons were removed from dishes containing a glial monolayer and placed with the neurons facing upwards in a 100-mm Petri dish that did not contain a glial monolayer; 4 ml of medium, taken from dishes containing glia, were added to each culture dish. [ 14 C]hexanoyl Cer (synthesized by N-acylation of sphingosine using n-hexanoic acid [1-14 C]N-hydroxysuccinimide ester (18)) was dissolved in ethanol and added to the culture dishes, which were then returned to the incubator for various times. bFGF (1 ng/ml, dissolved in Hank's balanced salt solution (HBSS)) or an equivalent volume (10 l) of HBSS were added to the culture dishes 1 h later. After various times, coverslips were removed from the dishes and washed with HBSS, and neurons were removed by scraping with a rubber policeman into ice-cold water and lyophilized. After resuspension in a small volume of water, lipids were extracted (19) and separated by thin layer chromatography using CHCl 3 /CH 3 OH/9.8 mM CaCl 2 (60:35:8, v/v/v) as the developing solvent. Thin layer chromatography plates were exposed to a 14 Csensitive imaging plate (BAS-TR2040S, Fuji Photo Film Co., Ltd., Japan), lipids were recovered from the plates by scraping, and radioactiv- ity was determined by liquid scintillation counting.
In Vitro Analysis of GlcCer Synthesis-Neurons were placed into new culture dishes as above and treated with or without bFGF or other growth factors for various times. Coverslips were subsequently washed three times with HBSS, neurons were removed by scraping with a rubber policeman, and neurons were homogenized in a handheld Potter-Elvehjem homogenizer in 2-3 ml of TK buffer (25 mM KCl and 50 mM Tris, pH 7.4). Homogenates were used fresh. GlcCer synthesis was assayed as described (2). The standard reaction mixture contained 20 g of protein (determined according to Bradford (20), 5 mM UDP-Glc, 2 M [ 14 C]hexanoyl Cer, and 10 mM MnCl 2 in a total volume of 2-3 ml of TK buffer. The reaction was terminated after 2 h at 37°C by addition of CHCl 3 /CH 3 OH (1:2, v/v). Lipids were extracted (19) and separated by thin layer chromatography as described above.

RESULTS
In Vivo Analysis of GlcCer Synthesis-To determine whether bFGF stimulates sphingolipid synthesis, [ 14 C]hexanoyl Cer was added directly to culture dishes containing 2-day-old neurons. After 1 h, bFGF was added, and [ 14 C]hexanoyl GlcCer and [ 14 C]hexanoyl SM synthesis was analyzed for the next 30 min. Immediately prior to addition of bFGF, the rate of [ 14 C]hexanoyl GlcCer synthesis was 35.8 fmol/min, which increased to 274 fmol/min for a 10-min period after addition of bFGF ( Fig.  1A and Table I). In contrast, there was no increase in the rate of [ 14 C]hexanoyl SM synthesis (Fig. 1A, Table I). 2 A small increase in the rate of [ 14 C]hexanoyl GlcCer synthesis was also observed in control cultures dishes, to which an equivalent volume of buffer (HBSS) was added. The rate of GlcCer synthesis after addition of buffer was 81.6 fmol/min; this is presumably due to removing the culture dish from the incubator, opening the dish for addition of HBSS, and gentle shaking after addition of the buffer.
The rate of uptake of [ 14 C]hexanoyl Cer by neurons was also significantly increased after addition of bFGF ( Fig. 1B and Table I). Two possibilities could explain the increase in the rate of [ 14 C]hexanoyl Cer uptake. (i) bFGF could directly affect the rate of [ 14 C]hexanoyl Cer uptake into neurons, and, as a consequence, the rate of [ 14 C]hexanoyl GlcCer synthesis is elevated simply because of increased availability of intracellular [ 14 C]hexanoyl Cer. (ii) bFGF could directly affect the rate of [ 14 C]hexanoyl GlcCer synthesis, and, as a result, more [ 14 C]hexanoyl Cer is taken up by neurons to provide sufficient substrate for increased [ 14 C]hexanoyl GlcCer synthesis. To distinguish between these two possibilities, which are fundamental to explaining the mechanism of action of bFGF, neurons were incubated with a stereoisomer of Cer, [ 14 C]hexanoyl-Lthreo-Cer, which is not metabolized to GlcCer (12). Addition of bFGF did not affect the rate of uptake of [ 14 C]hexanoyl-L-threo-Cer (Fig. 2B) or the rate of [ 14 C]hexanoyl-L-threo-SM synthesis ( Fig. 2A), demonstrating that bFGF directly affects the rate of [ 14 C]hexanoyl GlcCer synthesis rather than the rate of [ 14 C]hexanoyl Cer uptake.
Although there was an increase in the rate of [ 14 C]hexanoyl GlcCer synthesis immediately after addition of bFGF (Fig. 1A), reaching a maximum after 20 min, the rate of [ 14 C]hexanoyl GlcCer synthesis decreased after this time (Fig. 1A) In Vitro Analysis of GCS-We next examined the specificity of the effect of bFGF by analyzing GCS activity in vitro. In these experiments, neurons were incubated with bFGF or other growth factors for 10 min in culture dishes, subsequently removed from the coverslips by scraping, homogenized, and assayed for GCS activity using [ 14 C]hexanoyl Cer as substrate (2,18). When measured in vitro, GCS activity increased from 2.5 to 4.1 pmol of [ 14 C]hexanoyl GlcCer synthesized per g of protein/h after treatment with bFGF (Table II). A similar increase in the specific activity of GCS was detected when neu- 2 We previously observed little or no metabolism of a short acyl chain fluorescent derivative of Cer (C 6 -NBD-D-erythro-Cer; N-{6-[(7-nitrobenzo-2-oxa-1,3-diazol-4-yl)amino]hexanoyl}-D-erythro-sphingosine) to C 6 -NBD-SM when analyzed in vivo in cultured hippocampal neurons (12). In contrast, [   rons were incubated with laminin, which also stimulates axonal growth in a manner that is dependent on GlcCer synthesis (Ref. 12 and Table II). In contrast, denatured bFGF had no effect on GCS activity and no effect on the rate of axon growth (Table II), and preincubation of bFGF with a neutralizing anti-bFGF antibody also abolished the ability of bFGF to stimulate GCS activity and the rate of axonal growth. Three other growth factors (epidermal growth factor, nerve growth factor, and neurotrophin 3) had no effect on either GCS activity or on the rate of axonal growth. Thus, a direct correlation exists between elevated GCS activity and increased rates of axonal growth; bFGF and laminin stimulate both, whereas inactivated bFGF and other growth factors do not effect either parameter. 3 The activity of GCS at various substrate concentrations was examined in neurons treated with or without bFGF; an approximately 2-fold increase in activity was observed in bFGFtreated compared with control neurons under conditions in which the rate of the reaction was not limited by substrate availability (Fig. 3A). Activity was linear with respect to time of the reaction (Fig. 3B) and protein concentration (not shown).
To examine the time course of the effect of bFGF, neurons were incubated for various times with bFGF prior to assay of GCS activity. GCS activity remained higher in bFGF-treated neurons than in their control counterparts for up to 3 h after addition of bFGF (Table III). However, when bFGF was added to the incubation medium for 10 min and then removed (by washing the coverslips and replacing the medium with medium that did not contain bFGF), GCS activity decreased compared with neurons in which bFGF remained present throughout the 3-h incubation (Fig. 4).

Up-regulation of GCS Activity by a Post-translational Mech-
3 Note that addition of bFGF directly to the reaction mixture had no effect on GCS activity (not shown).  Fig. 1. Rates were calculated by linear regression analysis. After addition of bFGF, there is a significant increase in the rate of both [ 14 C]hexanoyl Cer uptake and of [ 14 C]hexanoyl GlcCer synthesis compared with control neurons; the numbers in parenthesis are the ratios of the rate of uptake or synthesis in the 10-min period after addition of bFGF or HBSS versus the 30-min period immediately prior to their addition. The difference in the rates of uptake and synthesis between 0 -30 min and 30 -60 min is due to the rapid initial rate of uptake of [ 14 C]hexanoyl Cer (18,21) by neurons immediately after its addition to the culture dishes.   The conditions of incubation for each growth factor were as follows: denatured bFGF, bFGF was heated at 95°C for 10 min; bFGF ϩ anti-bFGF, bFGF was preincubated with an anti-bFGF antibody for 2 h at 37°C; EGF, epidermal growth factor; NGF, nerve growth factor; NT3, neurotrophin 3.
b Statistically different from control cells (p Ͻ 0.02).
anism-Because GCS activity was elevated within 5-10 min of addition of bFGF, it appears unlikely that GCS activity is regulated via a transcriptional mechanism as was observed, for instance, upon long-term depletion of substrate levels (8). This was confirmed by semiquantitative RT-PCR analysis, in which no changes in GCS mRNA levels were detected between control neurons and those treated with bFGF for 10 min or 3 h (Fig. 5). Moreover, preincubation with the protein synthesis inhibitor, cycloheximide, or the RNA synthesis inhibitor, actinomycin, had no effect on the bFGF-stimulated increase in GCS activity (Table IV). DISCUSSION The major finding of the current study is that stimulation of axonal growth by bFGF results in an increase in the rate of GlcCer synthesis due to post-translational modification of GCS. In contrast, the rate of SM synthesis is not affected. A role for GlcCer synthesis in regulating the rate of axonal growth in cultured hippocampal neurons has been established (12,14,15), although the precise molecular requirements for which GlcCer synthesis is needed are not known.
It has been reported that the rate of SM synthesis increases compared with GlcCer synthesis during neuronal development (22), perhaps to regulate the supply or the function of glycosylphosphatidylinositol-anchored proteins (23) in growing axons. However, careful comparison of the rates of SM and Glc-Cer synthesis is required before distinct functions can be ascribed to each lipid at different stages of development, particularly as differences in the ratio of SM and GlcCer synthesis can be detected depending on the ceramide analogues used to measure activity and on ceramide concentration. 2 Irrespective of the possible distinct roles for SM and GlcCer synthesis at different stages of neuronal development, the data presented in the current study demonstrate that in young neurons (2 days in culture), GlcCer synthesis is activated by bFGF, whereas the rate of SM synthesis is unaffected. This activation occurs within minutes of bFGF treatment and is presumably a prerequisite for accelerated axonal growth, because incubation with inhibitors of GlcCer synthesis blocks axonal growth (12,14,15).   4. Reversibility of the effect of bFGF on GCS activity. Neurons were incubated with (squares) or without (circles) bFGF for 10 min. One set of coverslips was subsequently washed with HBSS to remove bFGF and placed into new dishes containing medium but no bFGF (open squares). GCS activity (20 g of protein, 2 M [ 14 C]hexanoyl Cer, 2 h reaction) was compared with that of neurons that were incubated with bFGF for the duration of the experiment (closed squares). Note that washing control coverslips had no effect on GCS activity (GCS activity in control neurons after 30 min ϭ 2.14 Ϯ 0.38 pmol of [ 14   In contrast to some previous studies (8 -10) demonstrating transcriptional activation of GCS, our data demonstrate that GCS can also be modulated by a post-translational mechanism. Two earlier studies have suggested a similar mode of activation. In the first (24), a decrease in the activity of GCS was observed upon treatment of Kym-1 rhabdomyasarcoma cells with tumor necrosis factor, presumably to prevent GCS from consuming ceramide generated by the SM cycle (25). In the second study, SM generated by application of endogenous sphingomyelinase to B16 melanoma cells resulted in up-regulation of GCS by both transcriptional and post-translational steps (26). Common to both of these studies is the pathway by which GCS activity is regulated namely as a response to generation of ceramide at the cell surface upon activation of the SM cycle. This appears to be different from the pathway by which GCS is up-regulated in bFGF-treated hippocampal neurons. There is no evidence that the activity of bFGF is mediated by the SM cycle, but rather it appears to be mediated via tyrosine phosphorylation of the cytoplasmic domain of high affinity bFGF receptors (27). In PC12 cells, tyrosine phosphorylation of the bFGF receptor activates mitogen-activated protein kinase (28), and in hippocampal neurons, bFGF causes a rapid increase in tyrosine phosphorylation of a variety of proteins, as yet unidentified (29). Whether activation of one of these pathways is a first step in the rapid up-regulation of GCS minutes after addition of bFGF is not known. However, it should be noted that the active site of GCS is on the cytosolic face of the Golgi apparatus (1-4), rendering the enzyme relatively accessible to intracellular signals and second messengers.
Because GCS has not been crystalized, little secondary and tertiary structural information is available. In a recent study, a histidine residue was detected in the active site of GCS (30). In addition, deletion of the N-terminal domain of GCS (including the transmembrane region) resulted in total loss of activity. As suggested (30), this region may be involved in interaction of GCS with ceramide and may be a potential site for post-translational modification. Unraveling the pathways by which binding of bFGF (and laminin) to their respective cell surface receptors leads to post-translational modification of GCS presents a formidable challenge.