Induction of glucosylceramide synthase by synthase inhibitors and ceramide

ELSEVIER

Biochimica et Biophysica Acta 1299 (1996) 333- 341

Biochi~mic~a et BiophysicaA~ta

Induction of glucosylceramide synthase by synthase inhibitors and ceramide Akira Abe a, Norman S. Radin

a,

James A. Shayman a,b,*

a Nephrology Division, Department of Internal Medicine, Universi~ of Michigan, 1150 W. Medical Center Dr., Ann Arbor, MI 48109-0676, USA b V.A. Medical Center, Ann Arbor, M148109, USA

Received 19 May 1995; accepted 12 September 1995

Abstract

Glucosylceramide (GlcCer) synthase acts on the sphingolipid, ceramide, to transfer a glucose moiety from UDP-glc, thus forming the first member of a large family of glucosphingolipids. Two inhibitors of the enzyme, D-threo-1-phenyl-2-decanoylamino-3-morpholino- 1propanol (D-threo-PDMP) and N-butyldeoxynojirimycin (NBDN), have been found to induce an elevated level of the synthase in MDCK cells. In cells treated with 20/zM PDMP, then assayed for synthase activity under conditions in which the absorbed PDMP was partially diluted out, the assay showed that the enzyme's specific activity had risen considerably in only l h and reached a maximum of about three times the control activity within 6 h. Both cycloheximide and actinomycin D, inhibitors of translational and transcriptional protein synthesis, caused much of the synthase activity to disappear in 6 h, presumably because of normal catabolic destruction. However, simultaneous inclusion of PDMP or NBDN in the cell medium slowed the rate of synthase disappearance. L-Cycloserine, which blocked the synthesis of ceramide, nevertheless allowed PDMP to elevate the synthase activity. Thus the inductive effect appears to be due, in part at least, to resistance of the enzyme-inhibitor complex to the normal process of enzyme degradation. Two other inhibitors of GlcCer synthase, more active than PDMP, did not produce detectable induction because they could not be dissociated from the enzyme during the cell washing and diluting steps. Agents that produced a large increase in endogenous cell ceramide level (DL-erythro-PDMP, N-acetylsphingosine, and bacterial sphingomyelinase) also induced an elevated level of GlcCer synthase. The latter two agents did not protect the synthase from catabolism in the presence of cycloheximide. These findings suggest the existence of a second mechanism of enzyme induction, enhanced synthesis of the enzyme due to the increased availability of the enzyme's lipoidal substrate. The possibility is raised that events involving ceramide in cell signalling may be mediated in part by changes in glucosphingolipid levels. Keywords: Glucosylceramide synthase induction; Acetylsphingosine effect; Glucosylceramide synthase inhibitor; Sphingomyelinase effect; l-Phenyl-2-decanoylamino-3-morpholino- 1-propanol; Ceramide elevation; 1-Phenyl-2-palmitoylamino-3-pyrrolidino-1-propanol; D-threo-1-Pyrrolidino-l-deoxyceramide

1. Introduction

Glucosylceramide (GlcCer) is the precursor of hundreds of different glucosphingolipids (GSLs). The galactosyl analog, prevalent in brain, is the precursor of only a few galactosphingolipids. GlcCer is synthesized from ceramide (N-acyl sphingol) and UDP-glc by a glucosyltransferase that is strongly inhibited by a group of compounds resembling ceramide in structure [1-3]. The most thoroughly

Abbreviations: C2Cer, N-acetyl sphingosine; CHX, cycloheximide; GSL, glucosphingolipid; NBDN, N-butyl deoxynojirimycin; P4, (R,R)1-phenyl-2-palmitoylamino-3-pyrrolidino-1-propanol; PDMP, (R, R)- lphenyl-2-decanoylamino-3-morpholino-l-propanol; TLC, thin-layer chromatography. Corresponding author (at address a). Fax: + 1 (313) 7630982. 0005-2760/96/$15.00 © 1996 Elsevier Science B.V. All rights reserved SSDI 0 0 0 5 - 2 7 6 0 ( 9 5 ) 0 0 2 1 7 - 0

studied inhibitor is D-threo-PDMP ((R, R)- 1-phenyl-2-decanoylamino-3-morpholino-l-propanol), which resembles ceramide in having adjacent fatty acylamide and alcohol groups. A phenyl group replaces the long alkenyl chain of sphingosine and a cyclic tertiary amine replaces the primary alcohol group. Subsequent work showed that substitution of the more natural palmitoyl group for the decanoyl group increased the inhibitory power [4,5]. Replacing the morpholine ring with a pyrrolidine ring improved the inhibitor still further [6]. In addition, replacing the aromatic propyl backbone with the o-threo isomeric form of sphingosine yielded an inhibitor of similar efficacy [7]. A ketonic variant of PDMP, in which the secondary alcohol group was replaced by a carbonyl group, was less active but it completely inactivated the synthase, possibly by a covalent interaction [1].

334

A. Abe et al. / Biochimica et Biophysica Acta 1299 (1996) 333-341

When mice were injected with the ketone, the synthase activity in liver was found to be depressed ~ 48% within 2 - 6 h, but by 24 h, the enzyme's specific activity had rebounded to ~ 20% above the control value [8]. This suggested the possibility that the synthase had been induced. A marked increase in the level of ceramide (46%) was visible even 2 h after injection, raising the possibility that ceramide was instrumental in stimulating the production of glucosyltransferase molecules. In the case of mice injected with PDMP, the enzyme activity was normal 24 h later [9]. This difference between the two inhibitors might be due to the shorter-lived inhibitory action of the hydroxylic compound [10]. Kidney levels of ceramide were only slightly (but significantly) elevated, when measured at 7 h. Ceramide has recently been implicated as an important intracellular messenger, mediating a diverse array of cellular functions, including differentiation, apoptosis, activation of nuclear transcription factors, and oxidant release in neutrophils (e.g., [11,12]). It thus seemed important to consider whether ceramide also acts to rapidly stimulate the synthesis of GSLs, which might be involved in some of the phenomena attributed to ceramide. When MDCK cells in serum-free culture medium were incubated with PDMP, large accumulations of ceramide were observed [13]. Thus, the possibility arose that these cells had responded like the mice, inducing an elevated level of GlcCer synthase activity. We report here that ceramide accumulation, produced by a variety of methods, appears to result in elevated levels of the synthase. A second mechanism of induction, possibly resulting from slowed catabolism of the enzyme when complexed with inhibitors, also appears to be active.

supplemented medium. After 24 h the medium was replaced with fresh medium (for control cells) or flesh medium containing the material to be tested as an inducer. The GlcCer synthase inhibitors and related compounds were added as equimolar BSA complexes [15] made from fatty acid poor BSA (Sigma A6003). The complexes were made by diluting a 50 mM solution of inhibitor in EtOH with 25 vol. of 2 mM sterile-filtered BSA, then stored at -20°C. Portions of the solution, when needed, were diluted with growth medium [4]. BSA and alcohol were included in the control medium too. After the incubation period ended, the cells were washed with 2 × 8 ml of cold PBS, scraped from the dishes, and transferred with a small amount of cold PBS to a 15-ml conical plastic tube. NBDN was so soluble in medium that BSA was not required. Usually, the treated cells from 10 dishes were pooled, centrifuged at 600 × g, suspended in 0.3-0.5 ml of cold water, and sonicated in an ice bath with a tip probe for 5 × 10 s. Protein concentration was determined by the copper/Folin method [16]. In some experiments, the cell lipids were extracted and examined by thin-layer chromatography on 10 X 20 cm Merck HPTLC plates, followed by charring and computer-video densitometry [6]. Partial hydrolysis of sphingomyelin in the cell surface to form ceramide was accomplished by incubating the cells with sphingomyelinase. The enzyme was dissolved in growth medium and added to each incubation dish to produce a concentration of 0.4 mU/ml. In other experiments, C2Cer, acetylsphinganine, and sphingosine were added to cells as ethanolic solutions (final concentration of EtOH 1 /zl/ml).

2. Materials and methods

2.3. Enzyme assays

2.1. Chemicals

Reagents and their suppliers were: N-acetylsphingosine (C2Cer) from Matreya (Pleasant Gap, PA); actinomycin D from Gibco; dioleoylphosphatidylcholine from Avanti Polar L i p i d s (Pelham, AL); DL-erythro-sphinganine, Derythro-sphingosine, sphingomyelinase from S. aureus, UDP-glc, stearoyl-CoA, and cycloheximide, from Sigma; UDP-[3 H]glc and [3- 3H]sphingosine from New England Nuclear; P4 compounds from Biomol Research Laboratories (Plymouth Meeting, PA); the aliphatic inhibitor, Dthreo-l-pyrrolidino-l-deoxyceramide, from the laboratory of Dr. Bruce Ganem [7]. N-Acetylsphinganine was made from DL-erythro-sphinganine [14]. N-Butyldeoxynojirimycin (NBDN) was a gift from Searle (Skokie, IL). Other substances were made here or purchased. 2.2. Cell growth

MDCK cells were grown as described [13], starting with 105 cells in 8 ml of serum-free DMEM (Mediatech)

GlcCer synthase in the cell sonicate was assayed by incubation in a thermostatted ultrasonic bath [17] with octanoyl sphingosine as acceptor lipid and UDP-[ 3H]glc as sugar donor [18]. The lipoidal substrate (85 /zg) was added in liposomes made from 570 /zg dioleoylphosphatidylcholine and 100 /zg of Na sulfatide; ~ 0.2 mg of protein was used in each assay tube. The [3H]GlcCer formed was isolated by partitioning between water, isopropanol, and t-butyl methyl ether, then assayed by liquid scintillation. The enzyme activity data, usually based on cells incubated for 6 h under various conditions, are presented as the mean and S.D. of three separate assays. The S.D. values in Figs. 2 and 3 are shown as short vertical lines. Comparisons on a percentage basis were made against control cultures that had been incubated the same length of time but without the added reagent. Tests with BSA and ethanol as the sole additives showed that they had no effect on the enzyme. Ceramide synthase was assayed with freshly purified [3H]sphingosine and stearoyl-CoA [19].

335

A. Abe et al. / Biochimica et Biophysica Acta 1299 (1996) 333-341 350

3. Results

3.1. PDMP effects 300

Cells treated with 20 /~M D-threo-PDMP for 24 h were harvested, homogenized, and assayed for glucosyltransferase activity. The control cells had a specific activity of 2.15 _+ 0.05 n m o l / h per mg protein and the specific activity in the PDMP-treated cells was > 3 times greater, 6.94 _+ 0.10. Another induction experiment, using a 6-h exposure to 20 tzM D-threo-PDMP, yielded a 166% increase in transferase activity, as well as a 54% loss of GlcCer and a 56% increase in ceramide concentration. The control values, per mg of protein content, were 3.12 _+ 0.012 nmol/h, 2.03 /zg, and 0.75 ktg, respectively. A time study of the phenomenon showed that marked induction was visible within as little as 1 h (Fig. 1). The increase in enzyme activity approached its maximum in only 6 h. Determination of the synthase's K m for UDP-glc, using 6 to 200 /~M UDP-glc, showed that the K m w a s not markedly different as the result of exposing the cells to 20 ~ M PDMP for 6 h (38 /zM in controls, 47 /zM after PDMP). The apparent V,,~x, however, was ~ 6 × as high in the treated cells. If the increased level of synthase activity was due to a new isoform, the two forms apparently have similar K,, values. To characterize the PDMP effect further, we grew cells for 6 h in medium containing 10 /~g/ml CHX (Fig. 2). The cells maintained in the absence of PDMP lost most of their GlcCer synthase during this period. The loss of activity is presumably due to the normal catabolic breakdown or inactivation of the enzyme. Inclusion of PDMP with the CHX partially prevented the enzyme's breakdown (39% residual activity in cells containing PDMP vs. 14% residual activity in cells lacking PDMP). It may be noted

i

I

i

t

I

i

,

,

i

I

i

,

,

i

[

i

i

i

i

I

i

i

£

i

20 IZM D - t h r e o - P D M P

6

4

.

v

Controls

2 1

e2 '

'

'

,

i

S

. . . .

i "

10

'

'

,

i

'

'

'

15

'

i

20

. . . .

25

Incubation time (h)

Fig. 1. Activity of glucosyltransferase in MDCK cells as a function of duration of exposure to (R,R)-l-phenyl-2-decanoylamino-3-morpholinol-propanol (D-threo-PDMP).

25O

.o

g

2OO

~

>-

150

~

100

50

0 ÷+ °

Fig. 2. GlcCer synthase activity in cells grown 6 h in medium containing cycloheximide (10 /zg/ml), with and without 200 /zM NBDN, 20 /~M D-threo-PDMP, or 40 /xM DL-ervthro-PDMP. The error bars signify the S.D. from the values obtained from five dishes.

that de novo synthesis of new enzyme molecules could not take place during exposure to the protein synthase inhibitor, regardless of whether PDMP was or was not present. A similar, but less complete loss of activity ( - 6 6 % ) was seen in a shorter, 4-h, incubation with the mRNA synthesis inhibitor, actinomycin D (Fig. 3). Here too PDMP furnished significant protection against enzyme loss, with 73% of the control activity still remaining. During a 24 h incubation with MDCK cells, 40 /~M DL-erythro-PDMP also induced a marked increase in glucosyltransferase specific activity: 14.6_+ 0.3 n m o l / h per mg protein in treated cells vs. 2.35 + 0.03 n m o l / h per mg in control cells. This 520% increase took place despite considerable slowing of cell growth (the protein content/dish was lowered 78%). Uemura et al. had previously found that this isomer inhibited the growth of rabbit skin fibroblasts [20]. erythro-PDMP does not inhibit GlcCer synthase [2], so its inductive and growth-inhibiting effects could not be attributed to a decrease in cellular GSL content. In fact, TLC showed that the treated cells

336

A. Abe et al. / Biochimica et Biophysica Acta 1299 (1996) 333-341

,oo

mogenate were then assayed for synthase activity in the usual way. The specific activity of the enzyme dropped from the zero-time value of 1.8 n m o l / h per mg protein by 53% in 15 min and by 96% in 40 min. The erythro-PDMP exerted no protective effect.

lii!iiiii!i!!iliiiiiiii!iii!iil

. . . .

!!iif:!~izi~G:i~{iiii~i{i{i~iiifi

o

3.2. Effects of PDMP derivatives and other transferase inhibitors

iiii!!iiiii!iii!iii!iiiiiiiiliiiii~iiiiiiiiii:~

C 0

i

0

"ililiiiii~ii!~i~i~ili~ili~!i~iiiiiiiiiiii:'

ii~i~+i~....................... i~ ililil

v

It)

,< ,,f.

+i_ :::::: i i i ili!.i.i.i.i.i.i:.+i. .i!.~i. i i +i i~!~i~i~'

l,-

":+!i!iii!iiiiii

-:::::::::::::::::::::::::::::::::::::::

>.z

°+ 5O

0

,~

~,, ~;

,,,,, ,<

I: 0

Pt O.

+

+

~

~-

o

0-

o

,,,, ,,It

o

Fig. 3. GlcCer synthase activity in cells grown 6 h in medium containing 1 g g / m l actinomycin D, with and without 20 /xM D-threo-PDMP. The control activity of GlcCer synthase was 2.26+0.09 n m o l / h per mg protein.

had accumulated a significant amount of ceramide and GlcCer. Exposure of cells to the erythro compound for a shorter period, 6 h, produced a 154% increase in synthase activity (21.1-+-0.2t vs. 8 . 3 0 + 0 . 1 9 n m o l / h per mg protein), raised the ceramide level by 98% (1.48 vs. 0.75 /xg/mg protein), and exerted only minor modification of GlcCer. It is likely that the reduced effect on GlcCer level in these cells is due to the shorter period of exposure. DL-erythro-PDMP also protected the cells against loss of synthase in cells exposed to CHX (Fig. 2, right set of columns). In this experiment, in which the cells were incubated for 6 h, the inductive effect (in the absence of CHX) was again visible and the losses of enzyme activity in the presence of CHX were similar to those seen in the experiment with D-threo-PDMP (center columns of Fig. 2). An experiment was performed to test the possibility that the above compounds protected the synthase from normal catabolic breakdown due to thermal inactivation. An MDCK cell sonicate was prepared in water and 400/xl (2.65 mg protein) were incubated at 42°C in the ultrasonic bath for 0, 15, and 40 min, with and without 40 /xM DL-erythro-PDMP. Three 120-#1 portions of the ho-

We have developed variants on the PDMP structure that are more effective as inhibitors of GlcCer synthase and cell growth [6]. 1-Phenyl-2-palmitoylamino-3-pyrrolidino1-propanoi (DL-threo-P4) strongly slows the growth of MDCK cells at concentrations > 3 / z M and, below 1 /zM, drastically reduces the content of GSLs. At the higher concentration, both the threo- and erythro-stereoisomers cause ceramide accumulation but erythro-P4 has little or no effect on GlcCer synthase activity when assayed with cell homogenates [6]. When MDCK cells were incubated with 4 /xM DLthreo-P4 for 6 h and the GlcCer synthase was assayed, the observed activity was only 10% of the control activity. This apparent failure to induce the enzyme (in contrast with D-threo-PDMP) was found to be the result of persistent binding of the absorbed inhibitor. TLC examination of the cell homogenate showed that there was a substantial amount of P4 present (more than the amount of cellular phosphatidylethanolamine or lecithin). When PDMP was used, much less inhibitor could be seen, evidently because it was largely removed by the washing procedure. The uptake a n d / o r binding affinity of P4 may be much greater; it is certainly much less soluble in water. Being a more effective inhibitor in vitro, the bound threo-P4 made any induction effect undetectable. It is possible that the small amount of PDMP left in the washed cells was sufficient to produce some inhibition; thus the extent of enzyme activity increase may be even greater than we observed. When cells were incubated with 4 /xM erythro-P4 (which does not inhibit GlcCer synthase) under the same conditions, the synthase activity was 420% of the control activities, an inductive effect similar to that seen with DL-erythro-PDMP. TLC showed that both erythro compounds caused accumulation of ceramide and that they were clearly present in the homogenate (somewhat less of the smaller molecule) [6]. Examination of the effect of erythro-P4 on the loss of enzyme activity in CHX-treated cells in a separate experiment (Fig. 4, right side of figure) indicated that it did not protect the enzyme from degradation. This is in contrast to the protective action of

erythro-PDMP. The recently described inhibitor of GlcCer synthase, NBDN [21], was found to be a relatively weak inhibitor under our assay conditions. It produced 13% inhibition at 3 /zM and 51% inhibition at 30 /zM. However, use of the compound at 200 /xM for 6 h did result in transferase induction (Fig. 2, left side) and it also showed protective

A. Abe et al. / Biochimica et Biophysica Acta 1299 (1996) 333-341 250

200

.o eQ iso

x100 O3 o 50

0 --

~

o 0

~a

*

*

o

¢~

--

~r

X

o 0

~

.

..4

o

X

,

Fig. 4. GlcCer synthase activity in cells grown 6 h in medium containing cycloheximide D (10 /zg/ml), with and without 10 /xM Nacetylsphingosine or 4 /zM DL-er),thro-P4 (DL-l-phenyl-2-palmitoylamido-3-pyrrolidino- 1-propanol).

action against catabolic breakdown in the presence of CHX. This inhibitor, which is relatively nonlipoidal and probably binds to the enzyme more weakly, was evidently washed out during the enzyme assay and thus could reveal the induction effect. There was no accumulation of ceramide, which would seem to show that induction can occur without ceramide accumulation. The aliphatic analog of D-threo-P4, (R,R)-l-pyrrolidino-l-deoxyceramide, which is about as active as an inhibitor, did not inhibit the growth of MDCK cells or cause the accumulation of ceramide [6]. Assay for GlcCer synthase after 6 h of incubation with 4 /zM inhibitor yielded a low activity. TLC analysis showed that this was the result of retention of the (relatively insoluble) inhibitor by the membranes. The 2S,3S enantiomer of pyrrolidinodeoxyceramide, which showed only minor inhibition of the synthase in vitro [6], did not affect the enzyme or ceramide level. The same lack of effect was seen with 10 /xM sphingosine.

3.3. Uptake of inhibitors by cell membranes An attempt was made to remove the strongly-bound inhibitors more effectively by more intensive washing and thus, perhaps, demonstrate that they too induced the syn-

337

thase. We incubated cell homogenate (1.47 mg of cell protein) with a medium containing 150 /zM UDP-glc, 40 mM Tris-C1- (pH 7.4), 15 mM MgC12, 1.5 mM DDT, 0.75 mM EGTA and enough liposomes to give a final concentration of 5 /xM inhibitor. The liposomes were made from 150 nmol of inhibitor, 4.76 mg of DOPC, and 1 mg of sodium sulfatide in 1 ml of water. The UDP-glc and other ingredients were included to stabilize the glucosyltransferase. After incubation at 37°C for 30 rain in a thermostatted ultrasonic bath [16], the suspension was cooled in ice and centrifuged at 14000 × g for 5 min. The resultant pellet was resuspended in 1 ml of cold water and centrifuged again, to wash out additional inhibitor. The pellet from this step was then sonicated in 0.5 ml water with a probe for 3 × 10 s and used for enzyme assay. The observed enzyme activities were normal in the case of PDMP treatment, showing that it had been washed out adequately, but threo-P4 treatment again exhibited considerably reduced enzyme activity ( - 7 2 % ) . The aliphatic, sphingol-based inhibitor was even more inhibitory ( - 9 1 % ) . TLC analysis again showed the presence of ample amounts of the longer chain inhibitors but only a trace of PDMP. These results show that the more lipoidal inhibitors bind rapidly and fimlly to GlcCer synthase and possibly to other cell components. A study with a fluorescent analog of PDMP indicated that it binds to Golgi-like membranes (which contain GlcCer synthase and ceramide), as well as to lysosome-like particles [22]. We have earlier shown that the myristoyl homolog of PDMP is more rapidly absorbed by cells than the decanoyl homolog [4]; thus the chain length of the inhibitor is important.

3.4. Ceramide effects We have previously reported that a short chain ceramide, octanoyl sphingosine, evoked marked increases in MDCK cells of natural (long chain) ceramide, sphingosine, and GlcCer [23]. We report here that the shorter chain ceramide, C2Cer, also produced an elevated level of ceramide. This was observed by examination of the chromatographed lipids from MDCK cells after incubation for 6 h with 10 ~ M C2Cer. The truncated homolog also produced a marked increase in the activity of GlcCer synthase: 240% of control activity (Fig. 3, left side). As with PDMP, the increase was blocked by CHX, as well as by 1 m g / m l of actinomycin D. An important difference between PDMP and C2Cer is that there was no protective effect of the latter; that is, the degradation of the enzyme seems to have been unaffected by the short chain ceramide. N-Acetyl-DL-sphinganine, the saturated version of C2Cer, had no visible effect on the cell lipids nor did it induce the synthase. TLC showed that the cells had taken up the compound, so its inactivity cannot be attributed to lack of uptake.

A. Abe et al. / Biochimica et Biophysica Acta 1299 (1996) 333-341

338

Table 1 Correlation between GlcCer synthase induction and levels of ceramide Treatment

Percent of control cells Ceramide

GlcCer

GlcCer synthase

Expt. 1: 5 k~M PDMP 20/zM PDMP 20/xM PDMP + cycloserine Expt. 2:

99 118 63

56 47 46

220 + 3 256_+4 264 + 4

e~thro-PDMP erythro-PDMP+ cycloserine

205 85

88 71

224 5:2 242_+ 2

MDCK cells were seeded into plastic dishes and incubated 24 h. The medium was replaced with fresh medium + inhibitors and the cells were incubated 6 h. The control contents in the two experiments, per mg protein, were 1.22 and 0.93 p~g for ceramide and 1.99 and 2.75 /xg for GlcCer; protein per dish was 0.76 and 0.47 mg. The control GlcCer specific activities were 5.03+0.03 and 6.605:0.13 n m o l / h per mg protein, respectively. The values shown are the means and S.D. values from five dishes.

The role of ceramide accumulation in the induction by

D-threo-PDMP was studied by blocking ceramide accumulation with L-cycloserine. This inhibits the synthesis of ketosphinganine, the parent sphingolipid that leads to ceramide formation. Thus its use prevents the synthesis of all the more complex sphingolipids. In the first test, MDCK cells were incubated 6 h with 0, 5, and 2 0 / x M PDMP. The levels of GlcCer synthase, ceramide, and GlcCer were then determined (Table 1). GlcCer levels decreased slightly and appreciable induction of GlcCer synthase was seen even at the low concentration of PDMP. In cells incubated with 20 /~M PDMP plus 1 mM t-cycloserine, ceramide decreased yet induction of the synthase still took place. The lower PDMP concentration (5 /zM) did not affect ceramide, while the higher concentration (20 /zM) produced an elevation in ceramide. (Here the increase was somewhat less than those seen in the other experiments). The major observation is that PDMP produced enzyme induction

even in cells in which the ceramide concentration was unchanged or partially depleted ( - 3 7 % ) . In a second experiment (Table 1) the noninhibitory isomer, Dt-erythro-PDMP, was also tested with and without cycloserine. Similar findings were seen: ceramide accumulated in the presence of the isomer and cycloserine depressed the level of ceramide. In both cases, ceramide accumulation was not necessary for induction of the synthase.

3.5. Effect of exogenous sphingomyelinase Since some of our data supported the idea that one method of transferase induction involved endogenous accumulation of ceramide, we tried treating cells with microbial sphingomyelinase, which acts at the plasma membrane's surface. After the cells were grown for 24 h in standard medium, they were treated for 6 h with enzyme, then harvested and assayed. The ceramide level was 280% higher than in the controls, the GlcCer level was 170% higher, and the synthase specific activity was 143% higher than the control. (The control concentrations, per mg protein, were 0.45 /zg, 1.41 /zg, and 2.74___ 0.03 nmol/h, respectively.) The specificity of our batch of sphingomyelinase was confirmed by TLC of the cell lipids, which showed a small loss of sphingomyelin and distinct increases in ceramide and GlcCer, while the lecithin level was unaffected. It is possible that the various compounds producing an increase in cellular ceramide caused induction of ceramide synthase. This enzyme was assayed in cells treated with 20 izM D-threo-PDMP for 6 h. The level of the enzyme was enhanced only to a minor extent: the activities in control and in treated cells were 14 _ 1 and 17 + 1 n m o l / h per mg protein, respectively. We had previously tested Othreo-PDMP, Dt-erythro-P4, DL-threo-P4, and the aliphatic inhibitor in contact with MDCK homogenates and had

Table 2 Experimental findings in support of ceramide as an inducer Treatment

A GlcCer transferase specific activity

A ceramide concentration

A GlcCer concentration

Conclusion

PDMP

+ 139% + 154% -- 90%

+ 56% + 98% high

- 54% -- 14% = - 90%

+ + -

high

Induction might be due to Cer increase Induction might be due to Cer increase Cer may have induced the enzyme, but induction not detectable because enzyme-inhibitor complex is undissociable Induction due to Cer increase Fast catabolism of synthase is evident No protection against enzyme degradation Induction due to Cer increase No protection against enzyme degradation

DL-eo'thro-PDMP DL-threo-P4

DL-erythro-P4 CHX

DL-erythro-P4 + CHX N-acety|sphingosine N-acetylsphingosine + CHX Sphingomyelinase

320% also + 54% 81% 79% 140% 84%

+ 143%

high

high

+ 380%

+ 170%

Induction due to Cer increase

Cer, natural (long chain acyl) ceramides; CHX, cycloheximide; P4, phenylpalmitoylaminopyrrolidinopropanol; PDMP, ~threo-phenyldecanoylaminomorpholinopropanol used at 20 /zM. Data from 6-h incubations.

A. Abe et aL / Biochimica et Biophysica Acta 1299 (1996) 333-341

found no direct effect on ceramide synthase with 5 /xM inhibitor [6].

4. Discussion The induction of GlcCer synthase appears to involve two mechanisms, ceramide accumulation and a protective effect that slows the catabolic degradation of the synthase. The first mechanism (summarized in Table 2) was seen in the experiments that produced ceramide accumulation and an elevated level of synthase. D-threo-PDMP, erythro-PDMP, erythro-P4, bacterial sphingomyelinase, and CeCer, each of which elevated the cellular concentration of ceramide, also produced a high specific activity of synthase within 6 h or less. Ceramide is the lipoidal substrate of GlcCer synthase and the ability of an enzyme's substrate to induce production of the enzyme is well known. Historically, the first examples of enzyme induction were made under such circumstances. This supports the idea that ceramide actually stimulates the synthesis of the enzyme. The mechanisms by which the elevations in ceramide levels were produced are still obscure except in the case of sphingomyelinase treatment. With the GlcCer synthase inhibitor, D-threo-PDMP, one would expect ceramide to accumulate because it could not be glucosylated. In the case of the compounds inactive against the synthase, such as erythro-P4, one might postulate that accumulation of ceramide was due to inhibitory activity against ceramide hydrolase or sphingomyelin synthase. However, direct tests of 5 /xM erythro-P4 in MDCK cell homogenates did not yield significant effects [6]. At 50 /xM there was distinct inhioition of sphingomyelin synthase ( - 54%) and a stimulation of neutral sphingomyelinase (+42%); both effects would act to cause ceramide accumulation. Perhaps the concentration of erythro-P4 in intact cells can reach an

339

effective level even at low concentrations in the medium. If this is so, erythro-P4 might prove useful in studies of the role of neutral sphingomyelinase in cell signalling. It may be noted that there apt:,ears to be no feedback inhibition of ceramide synthase by the metabolic product, GlcCer. Gaucher disease patients, who accumulate much GlcCer because of a glucosidase deficiency, are not known to exhibit a low level of ceramide. Thus, it is unlikely that the accumulation of ceramide by treatment with D-threoPDMP is due to release of feedback inhibition. The large increase in normal (long chain fatty acyl) ceramide resulting from treatment of the cells with CzCer cannot yet be explained. A test with [3H]acetate-labeled C2Cer in Jurkat T cells [24] was interpreted to mean that this ceramide is not metabolized, although it should be noted that the authors accounted for only ~ 38% of the added lipid. In our system, it was evident from the TLC plates that the amount of natural ceramide formed, as the result of treatment with C2Cer, was less than the amount of C2Cer taken up by the cells. Our previous study with N-octanoylsphingosine [23] and a new report based on CHO cells [25] showed that free sphingosine as well as long chain ceramide was formed by cells exposed to short-chain ceramides. In addition, N-acetylsphinganine (labeled by tritiation of the double bond in C2Cer) was converted to long chain ceramide and sphingomyelin in CHO cells [26]. These observations suggest that some N-acetylsphinganine is converted by hydrolysis and reacylation, or by acyl exchange, to natural, long-chain ceramides. The second mechanism (summary in Table 3) was seen in the experiment with NBDN, which produced depletion of GlcCer and an elevation in synthase, yet did not cause accumulation of ceramide. The same relationship was also seen in the experiment (Table 1) in which cycloserine lowered ceramide levels below normal, yet did not interfere with the induction of synthase by threo-PDMP or its

Table 3 Experimental findings supporting induction by blocking degradation of GlcCer synthase Treatment

A GlcCer synthase specific activity

CHX CHX + PDMP Actinomycin D Actinomycin D + PDMP DL-erythro-PDMP + CHX DL-erythro-P4 + CHX NBDN 0.2 mM NBDN + CHX PDMP PDMP + 1 mM cycloserine e~thro-PDMP erythro-PDMP + cycloserine

- 84% -61% - 66% - 27% - 62% - 79% + 108% - 62% + 156% + 164% + 124% + 142%

A ceramide concentration

A GlcCer concentration

no effect

- 48%

+ 18% - 27% + 105%

-53% - 54% - 12%

-

-29%

15%

NBDN, N-butyldeoxynojirimycin. Other abbreviations as in Table 2.

Conclusion

Rapid degradation of enzyme Slower degradation of enzyme Inhibition of protein synthesis less complete than CHX Slower degradation of enzyme Slower degradation of enzyme No protection against enzyme degradation Induction due to Cer increase Protection against degradation Protection against degradation; Cer increase has little or no effect Protection against degradation, Cer loss does not block induction Induction due to Cer increase OR to protection against degradation Protection against degradation, Cer loss does not block induction

340

A. Abe et al. / Biochimica et Biophysiea Acta 1299 (1996) 333-341

DL-stereoisomer. In this study and in a previous one [6], it was found that low concentrations of GlcCer synthase inhibitors could deplete cells of their GlcCer without causing ceramide accumulation. It is clear that synthase induction does not necessarily require ceramide accumulation. This second method of induction may be due to protection of the enzyme against the normal process of inactivation or degradation rather than by specific gene activation. Many examples are known of inhibitors that stabilize enzymes in vitro; the effect has proved very useful in the isolation of enzymes. This phenomenon explains the results with o - t h r e o - P D M P and NBDN in cells exposed to protein synthesis inhibitors (Fig. 2). At first glance, the columns in this figure would seem to show simply that PDMP induced an elevated level of synthase even in the presence of CHX. However, when the cells were treated only with CHX, protein synthesis stopped, so any hypothetical inductive action of PDMP and NBDN involving the production of additional new enzyme molecules was ineffectual. It has been known for > 30 years that even lower concentrations of CHX block protein synthesis in mammalian cells almost completely, taking effect very soon after addition to cell media [27]. In this situation, disappearance of the synthase simply reflects the normal mechanism of catabolism, although there must also be some simultaneous degradation of the catabolizing enzymes. Thus the decreased rate of catabolism seen in the presence of the two inhibitors seems to indicate that they exerted a protective action against the degradative process. Some alternative explanations of these observations can be proposed: (a) degradative enzymes (such as proteases) are directly inhibited by PDMP or NBDN, or (b) the inhibitors stimulate conversion of a proenzyme to form active GlcCer synthase. It is possible that the addition of protease inhibitors to the growth medium would block synthase degradation. Other inhibitors might do the same thing if synthase degradation requires removal of carbohydrate or phosphate moieties before proteolysis can proceed. A more definitive explanation must await the availability of purified synthase or antibody to the enzyme. NBDN inhibits the a-glucosidases involved in the anaholic processing of N-linked glycoproteins [28], so it is conceivable that NBDN caused accumulation of GlcCer synthase by blocking a posttranslational step, such as deglucosylation of an intermediate form of the enzyme. In order to explain the apparent increase in GlcCer synthase one would have to postulate that the intermediate form is more active than normally-processed synthase, o-threoPDMP, on the other hand, does not block glycoprotein synthesis [5] although it slows the transport of newly processed glycoproteins to the cell periphery [22]. o L - e r y t h r o - P D M P , which is not an inhibitor of GlcCer synthase, nevertheless protected the enzyme against inactivation (Fig. 2). This can be ascribed to protective, albeit noninhibitory, binding to a modulatory site on the enzyme. Evidence for the existence of such a site was obtained by

kinetic analysis with D-threo-PDMP [2]. The failure of erythro-P4 to produce similar protection (Fig. 3) was unexpected; perhaps it binds poorly to the modulatory site. Ceramide accumulation alone, which might be expected to also protect the synthase against catabolism, apparently did not exert this effect (Fig. 3). Our test of protection against thermal inactivation, at 42°C, did not eliminate the possibility that the normal catabolism of the synthase involves initial phosphorylation or simple proteolysis or secretion of enzyme into the medium. Shah and Peterson [29] showed that CHX injected into rats rapidly produced a low level of GlcCer synthase activity in brain. Mixing control and low-activity microsomes from the brains did not disclose the presence of a (dissociable) inhibitor in the treated rats. This is further evidence for a genuine, rapid loss of synthase even in intact animals. One possible explanation for the induction effect is that PDMP stimulates formation of a glucosyltransferase activator. A study of GlcCer synthase in the cystic kidneys of genetically impaired mice yielded evidence for the presence of such an activator [30]. Supplementing a control cell homogenate in the present study with the high-speed cytosol from cells stimulated 6 h with PDMP did not result in enhancement of the synthase activity. One might expect an increased amount of activator to enhance binding of the substrate to the enzyme, but we did not find any change in the enzyme's K m. The level of the proteins (saposins) in mouse liver that activate GlcCer glucosidase is rapidly raised by an inhibitor of the hydrolase, conduritol B epoxide, and by direct injection of GlcCer [31]. Thus the possible involvement of an activator needs more study. There is some possibility that a decrease in cellular GSLs is by itself responsible for a third mechanism of induction. One would expect a feedback process by which a cell can respond to a loss of a metabolite as central as GlcCer. However, we also observed induction in the face of elevated GlcCer. Injection of mice with testosterone produced rapid rises in GlcCer synthase and decreases in GlcCer hydrolase [32]. Thus, tissues rapidly and coordinately control the net level of GlcCer by inducing and repressing appropriate enzymes. Similar rapidity of changes in the GlcCer level in intact mice [8] and neuroblastoma cells [33] indicates that GlcCer also has a high turnover rate. Our finding that GlcCer synthase can be induced rapidly by the enzyme's inhibitors may have important implications if the inhibitors find some clinical use. Assorted potential uses have been proposed, such as therapy of cancer, Gaucher's disease, Fabry's disease, bacterial and viral infections, kidney hypertrophy in diabetics, and psoriasis [34-36]. As with other drugs that produce increased enzyme levels, tachyphylaxis and rebound might occur on discontinuation of treatment. The rapid induction of GlcCer synthase seen in this study suggests that the levels of the enzyme and its

A. Abe et al. /Biochimica et Biophysica Acta 1299 (1996) 333-341

product m a y be physiologically important. The studies reporting a role for ceramide as a second messenger, such as those using interferon, v i t a m i n D-3, and tumor necrosis factor, m a y point to control of the rate of glucosphingolipid synthesis as a critical p h e n o m e n o n . O n l y a few studies of these factors have considered interactions with GSLs. For example, i n t e r f e r o n - y greatly increased the surface expression of gangliosides, while p r o d u c i n g corres p o n d i n g l y large decreases in the expression of neutral GSLs [37]. T u m o r necrosis factor-c~ e n h a n c e d the expression of ganglioside G D 3 in cultured melanocytes [38]. The factor seems to induce G S L galactosyltransferase(s) [39]. Perhaps our findings will p r o m p t others to investigate the functional significance of G l c C e r synthase induction.

Acknowledgements This work was supported by National Institutes o f Health grants D K 3 9 2 5 5 and D K 4 1 4 8 7 , the Glycolipid Research Fund, and a Merit A w a r d from the D e p a r t m e n t of Veteran Affairs.

References [1] Vunnam, R.R. and Radin, N.S. (1980) Chem. Phys. Lipids 26, 265-278. [2] Inokuchi, J. and Radin, N.S. (1987) J. Lipid Res. 28, 565-571. [3] Radin, N.S. and Shayman, J.A. (1993) in NeuroProtocols: A Companion to Methods in Neurosciences (Fisher, S.K. and Bleasdale, J.E., eds.), Vol. 3, pp. 145-155, Academic Press, Orlando. [4] Abe, A., Inokuchi, J., Jimbo, M., Shimeno, H., Nagamatsu, A., Shayman, J.A., Shukla, G.S. and Radin, N.S. (1992) J. Biochem. 111, 191-196. [5] Paul, P., Bordmann, A., Rosenfelder, G. and Towbin, H. (1992) Anal. Biochem. 204, 265-272. [6] Abe, A., Radin, N.S., Shayman, J.A., Wotring, L.L., Zipkin, R.E., Sivakumar, R., Ruggieri, J.M., Carson, K.G. and Ganem, B. (1995) J. Lipid Res. 36, 611-621. [7] Carson, K.G., Ganem, B., Radin, N.S., Abe, A. and Shayman, J. A. (1994) Tetrahedron Lett. 35, 2659-2662. [8] Hospattankar, A.V., Vunnam, R.R. and Radin, N.S. (1982) Lipids 17, 538-543. [9] Shukla, G., Shukla, A., Inokuchi, J. and Radin, N.S. (1991) Biochim. Biophys. Acta 1083, 101-108. [10] Shukla, A. and Radin, N.S. (1991) J. Lipid Res. 32, 713-722.

341

[11] Hannun, Y.A. (1994)J. Biol. Chem. 269, 3125-3128. [12] Kolesnick, R. and Golde, D.W. (1994) Cell 77, 325-328. [13] Shayman, J.A., Deshmukh, G., Mahdiyoun, S., Thomas, T.P., Wu, D., Barcelon, F.S. and Radin, N.S. (1991) J. Biol. Chem. 266, 22968-22974. [14] Gaver, R.C. and Sweeley, C.C. (1966) J. Am. Chem. Soc. 88, 3643-3647. [15] Pagano, R.E. and Martin, O.C. (1988) Biochemistry 27, 4439-4445. [16] Lowry, O.H., Rosebrough, N.J., Farr, A.L. and Randall, R.J. (1951) J. Biol. Chem. 193, 265-275. [17] Radin N.S. and Shukla, G.S. (1991) Enzyme 45, 67-70. [18] Shukla, G.S. and Radin, N.S. (1990) Arch. Biochem. Biophys. 283, 372-378. [19] Morell, P. and Radin, N.S. (1970) J. Biol. Chem. 245, 342-350. [20] Uemura, K., Sugiyama, E., Tamai, C., Hara, A., Taketomi, T. and Radin, N.S. (1990) J. Biochem. (Tokyo) 108, 525-530. [21] Platt, F.M., Neises, G.R., Dwek, R.A. and Butters, T.D. (1994) J. Biol. Chem. 269, 8362-8365. [22] Rosenwald, A.G., Machamer, C.E. and Pagano, R.E. (1992) Biochemistry 31, 3581-3590. [23] Abe, A., Wu, D., Shayman, J.A. and Radin, N.S. (1992) Eur. J. Biochem. 210, 765-773. [24] Dbaibo, G., Obeid, L.M. and Hannun, Y.A. (1993) J. Biol. Chem. 268, 17762-17766. [25] Ridgeway, N.D. and Merriam, D.L. (1995) Biochim. Biophys. Acta 1256, 57-70. [26] Ladenson, R.C., Monsey, J.D., Allin, J. and Silbert, D.F. (1993) J. Biol. Chem. 268, 7650-7659. [27] Bennett, Jr., L.L., Smithers, D. and Ward, C.T. (1964) Biochim. Biophys. Acta 87, 60-69. [28] Elbein, A.D. (1987) Annu. Rev. Biochem. 56, 497-534. [29] Shah, S.N. and Peterson, N.A. (1971) Biochim. Biophys. Acta 239, 126-131. [30] Deshmukh, G., Radin, N.S., Gattone, V.H. and Shayman, J.A. (1994) J. Lipid Res. 35, 1611-1618. [31] Datta, S.C. and Radin, N.S. (1986) Lipids, 21, 702-709. [32] Shukla, A., Shukla, G.S. and Radin, N.S. (1992) Amer. J. Physiol. 262, F24-F29. [33] Radin, N.S, Shayman, J.A. and Inokuchi, J. (1993) in Advances in Lipid Research; Sphingolipids in Signaling, Part B (Bell, R.M., Hannun, Y.A. and Merrill, A.H., eds.), Vol. 28, pp. 183-213, Academic Press, Orlando. [34] Radin, N.S. and Inokuchi, J. (1988) Biochem. Pharmacol. 37, 28792886. [35] Radin, N.S. (1994) Molec. Chem. Neuropathol. 21, 111-127. [36] Gillard, B.K., Jones, M.A., Turner, A.A., Lewis, D.E. and Marcus, D.M. (1990) Arch. Biochem. Biophys. 279, 122-129. [37] Uemura, K., Sugiyama, E. and Taketomi, T. (1991) J. Biochem. (Tokyo) I10, 96. [38] Furukawa, K., Arita, Y., Satomi, N., Eisinger, M. and Lloyd, K.O. (1990) Arch. Biochem. Biophys. 281, 70-75. [39] van de Kar, N.C., Kooistra, T., Vermeer, M., Lesslauer, W. and Monnens, L.A. (1995) Blood 85, 734-743.