Characterization of serine palmitoyltransferase activity in Chinese hamster ovary cells

284

Blochimrcu et Bloph~vsiru Acru, 754 (1983) 2846291 lsevl~r

BBA 51534

CHARACTERIZATION OF SERINE PALMITOYLTRANSFERASE HAMSTER OVARY CELLS ALFRED Department (Received

H. MERRILL, of Biochemistry,

ACTIVITY IN CHINESE

Jr. Emory University, School of Medicine, Atluntu,

GA 30322 (U.S.A.)

July 4th, 1983)

Key words: Serine palmitoyltransferase;

Sphingolipid synthesis; Long-chain

base; (Chinese hamster ouqv)

Serine palmitoyltransferase (palmitoyl-CoA: L-serine C-palmitoyltransferase (decarboxylating) EC 2.3.1.50) catalyzes the first unique and regulatory reaction of sphingolipid biosynthesis. Its activity was demonstrated in Chinese hamster ovary cells (CHO-Kl) by measuring the incorporation of radiolabel from L-l”HJserine into 3-ketosphinganine, which was found to be the predominant chloroform-soluble product under optimal assay conditions. Most of the total activity (14.8 k 4.2 pmol/min per lo6 cells, measured with sonicated cells) was recovered in particulate fractions, with the highest percentage (54%) and specific activity (102 pmol/min per mg) in the high-speed (airfuge) pellet. The greatest activity was obtained with palmitoyl-CoA; however, other fatty acyl-CoA thioesters were also utilized. Serine palmitoyltransferase required pyridoxal S-phosphate for activity, but was apparently fully saturated with this coenzyme when assayed with sonicated cells. Regardless of whether the CHO cells were grown in culture medium containing whole serum with or without sphinganine addition, lipoprotein-depleted serum, lipid-extracted serum, low-density lipoproteins, or no serum, the activities of this enzyme were identical. This finding was confirmed using human fibroblasts. Hence, these results establish that CHO cells, and probably others, are engaged in long-chain base synthesis de novo and that serine palmitoyltransferase activity is not regulated by the availability of such compounds in the culture media.

Introduction Membrane lipids can be grouped into three major classes: glycerolipids, sphingolipids and cholesterol. The latter two are primarily in the plasma membrane and functionally related structures, such as endocytotic and secretory vesicles, lysosomes and the Golgi; whereas, the glycerolipids are distributed in all membranes [l]. Since lipid compositions are highly characteristic of different membranes but can depend upon the physiological state of the cell, it is important to underAbbreviations. Hepes, 4-(Zhydroxyethyl)-l-piperazineethanesulfonic acid; LDL, low-density lipoproteins; CHO, Chinese hamster ovary. 0005-2760/83/$03.00

0 1983 Elsevier Science Publishers

B.V.

stand the regulation of these phenomena. The biochemical information requisite for this understanding is most lacking for sphingolipids, and in particular, for the backbone of this class of compounds - the long-chain bases. Although the reactions of long-chain base synthesis and degradation were elucidated over a decade ago [2,3], few subsequent investigations have concerned the biochemistry and regulation of this pathway in mammalian systems [4-61. We have recently begun studies of the first unique and committed reaction of long-chain base synthesis, the condensation of L-serine with a fatty which is catalyzed by serine acyl-CoA, palmitoyltransferase [7]. It is our hypothesis that this enzyme will regulate both the amount and, by

285

its selectivity for fatty acyl-CoA thioesters, the type of long-chain bases that are made by cells. Evidence for the former is that activity has been found to be highest in regenerating liver and hepatoma 7777, compared to normal liver [8], and in rat tissues that contain the highest proportions of sphingolipids (Merrill et al., unpublished observation). It is not yet known if its activity can be attenuated by the uptake of long-chain bases by cells, although evidence predicting this possibility has been recently reported by Verdery and Theolis [9]. Secondly, the activities of serine palmitoyltransferase with different fatty acyl-CoA thioesters exhibit a clear optimum for palmitoylCoA [7,8,10], as is expected from the predominance of 18-carbon bases in most mammalian sphingolipids [ll]. This manuscript extends these characterizations to include a model more amenable to studies of the regulation of serine palmitoyltransferase, Chinese hamster ovary (CHO) cells, and reports that its activity was not found to be affected by long-chain bases or sphingolipids in the culture media. Materials and Methods Chinese hamster ovary cells (CHO-Kl, ATCC No. CCL 61) were obtained from the American Type Culture Collection; human fibroblasts (GM3348 A) were from the N.I.G.M.S. Human Mutant Cell Repository. Both were grown routinely in Ham’s F12 medium (Gibco) supplemented with 10% fetal calf serum (Gibco), penicillin-G (100 units/ml), streptomycin sulfate (100 pg/ml) and NaHCO, (1.76 g/l) (Gibco). Cultures were maintained at 37°C in a 5% CO, atmosphere at 100% humidity. The cells (at approx. 80% confluence) were released from the plates for reculturing by washing the cells with Puck’s versene, incubating the cells in the same plus 0.04% trypsin until the cells began to detach from the plate, and then diluting the cell suspension into culture media to yield approx. 0.1 . 10’ cells/plate. The fibroblasts were recultured at a ratio of four dishes per original culture, with six total passages in the course of these experiments. The suspended CHO cells were frozen in 20% glycerol at - 80°C and separate aliquots were used for each different experiment described herein. The cell numbers were de-

termined using a Model ZF Coulter counter, which was calibrated by comparison with a certified counter in the clinical labs of Emory University Hospital. For the growth of CHO cells in medium without serum, the suspended cells were added to a 50-fold greater volume of a “modified” medium, which contained all of the components above except serum and was supplemented with 10 mM sodium selenite, 0.6 mM CaCl,, 3 mM L-glutamine, and bovine pancreatic insulin (20 pg/ml) (Sigma). With these modifications the medium approximated that reported by Hamilton and Ham [12] for the serum-free growth of CHO cells. After 24 h, the modified medium was used to wash and grow the cells (the cells doubled approx. every 18 h). Human serum was obtained from healthy adult volunteers who had fasted for 18 h before donating blood. Lipoprotein-depleted serum was prepared by adding NaBr until the density of the serum was 1.215 g/ml, centrifuging it for 18 h at 125 000 x g, and collecting the lower half of the sample. Lipid-depleted serum was prepared by organic extraction as described by Cham and Knowles [13], except that optimal removal of the sphingomyelin was found when the two phases were left in the separatory funnels overnight to clarify the aqueous phase. It was estimated that at least 95% of the sphingomyelin and other phospholipids were removed by these treatments because none was visible by thin-layer chromatography of organic solvent extracts [14] of the depleted serum. The low density lipoproteins were harvested from plasma in the density range 1.020 to 1.063 g/ml and established to be pure by polyacrylamide gel electrophoresis with visualization of the LDL with Sundan Red. The serum and lipoprotein samples were dialyzed repeatedly against 0.15 M NaCl/0.02 M potassium phosphate (pH 7.4) and filter sterilized (0.22 pm) before use. For the assay of serine palmitoyltransferase, the medium was removed from the culture plates and the cells were washed with 0.05 M potassium phosphate (pH 7.4)/0.15 M NaCl/lO mM EDTA. To each plate was added 0.3 ml 0.05 M Hepes (pH 8.0 at 4”C)/lO mM EDTA/S mM dithiothreitol; the cells were detached using a rubber policeman and transferred to a test tube on ice. Residual cells

286

were recovered in an additional 0.2 ml of buffer and were added to the first cell suspension. The cells were disrupted by three 5 s bursts in a Bransonic 12 that contained ice/water. The cells were placed on ice for the 1-2 min between bursts. Enzymatic activities were determined immediately thereafter by adding 50 ~1 of the sonicated cells to 50 ~1 0.2 M Hepes (pH 8.0 at 37”C)/2 mM L-[ 3H]serine (purchased from ICN Radiochemicals, 10 Ci/mmol, purified by chromatography on Dowex SOW-X8, and diluted to 40000 cpm/nmol with unlabelled L-serine)/S mM EDTA (pH adjusted to 7.4 with NaOH)/lO mM dithiothreitol/ 100 PM pyridoxal 5’-phosphate/ 0.4 mM palmitoyl-CoA. After a lo-min incubation at 37°C. 0.2 ml 0.5 N NH,OH was added and the lipid products were extracted by a minor modification (inclusion of 25 pg of sphinganine as carrier) of the method of Bligh and Dyer [15]. An aliquot of the CHCl, extract was transferred to a scintillation vial and dried, and the radioactivity was measured by scintillation counting using Econofluor (New England Nuclear). The counting efficiency was approx. 40%. Control tubes contained all of the assay components except palmitoyl-CoA, and the resultant cpm were subtracted to correct for the background radioactivity, which was primarily due to a degradation product of the [3H]serine. The assays were conducted in triplicate for at least four plates per experimental group. To determine the cell number and viability (by Trypan blue exclusion) an equal number of plates were treated with trypsin as described above

TABLE

and counted. The significance of differences between groups was determined by Student’s z-test. To determine the amount of [ ‘H]serine incorporated into the long-chain bases of sphingomyelin, cells were grown in culture flasks containing medium with 5% fetal calf serum, then washed twice over 24 h with medium containing dialyzed serum (5%) or an equivalent amount of organicsolvent extracted serum. New media that also contained 0.25 mCi (10 Ci/mmol) [“Hlserine were added and the flasks were sealed and incubated at 37°C for 24 h. The cells were then recovered using trypsin, the cell numbers were determined, and the lipids were extracted and the amounts of radiolabelled long-chain bases were determined as described by Verdery and Theolis [9]. Protein was determined by the method of Bensadoun and Weinstein [16] with bovine serum albumin as the standard. Unless otherwise noted, all the biochemicals were purchased from Sigma Chemical Co. Results Serine palmitoyltransferase activity was present in pellets obtained both from low-speed (approx. 8000 x g) and from high-speed (approx. 100 000 x g) centrifugation of sonicated CHO cells (Table I). The greatest recovery (54%) and the highest specific activity were found in the airfuge pellet. Little or no activity was detected in the soluble fraction; hence, it appears that serine palmitoyltransferase is predominantly membrane-associated.

I

SERINE

PALMITOYLTRANSFERASE

ACTIVITY

IN CHOFKl

CELLS

Cells grown in medium containing approx. 5X fetal calf serum were disrupted by sonication and centrifuged for 10 min in a microfuge (approx. 8000~ g), then the supernatant was centrifuged for 1 h in an airfuge ( = 100000x g). Assays were conducted using aliquots from each fraction. Protein (mg/106

Fraction

Activity cells)

pmoI/min per lo6 cells

pmol/min per mg 12.3

Sonicated

cells

0.91

11.2

Microfuge

pellet

0.12

3.0

25 102

Airfuge

pellet

0.06

6.1

Airfuge

supernatant

0.63

cc 0.5

< 0.8

287

Since the sum of the activities of the subcellular fractions agreed well with the activity obtained with sonicated cells, the feasibility of assaying this enzyme in unfractionated samples was determined. The properties of the radiolabelled compounds recovered in the organic-solvent extracts of assays using sonicated cells are shown in Fig. 1. The radiolabel was found almost entirely in the region of the chromatograms corresponding to the 3-ketosphinganine standard, with a small fraction (5%) at the solvent front. The latter was a chromatographic artifact, as was first established by Braun and Snell 1171,since sodium borohydride reduction of the products yielded only sphingamine (Fig. 1, dashed line), As is shown in Fig. 2, product formation was essentially proportional to time and the number of cells under the assay conditions. Activities with varying concentrations of paImitoyl-CoA were measured to determine optimal assay conditions because this amphiphatic compound is known to inhibit many enzymes when added at high con-

Fig. 1. Thin-layer chromatography of the radiolabelled products obtained by assays of serine PaI~toyltransfera~ before (0) and after (0) reduction with sodium borohydride NaBH,. A 1.0 ml assay mixture containing 0.5 ml of sonicated CH0 cells was incubated at 37°C for 10 mm and the CHCl,-soluble products were extracted. Most of the solvent was removed under a stream of N,, and to half of the sample was added an equal volume of ethanol containing 2 mg NaBH,. After 5 min. both were applied to a silica gel chromatoplate, which was developed with CHCI, /methanol/2 N NH,OH (40 : 10 : 0.6). The standards were visualized with I, vapor; then plates were cut into portions and the lipids were extracted and counted. The abbreviations for the standards are: PS, phosphatidylse~ne (N- and O-palmitoylserine and ethanolamine migrated with PS); Sa, sphinganine; S, sphingosine; KS, 3-ketosp~nganine; and C. ceramide.

7.2

L; tz :

20 0.8

z* T. z

10 0.4

3 :

0

0 5

10

15

20

0

0.3

Fig. 2. The dependence of the product time (A) and number of cells (B).

0.6

formation

0.9

and rate on

centrations. The activity increased until approx. 0.2 mM palmitoyl-CoA, then decreased precipitously (Fig. 3). The data presented in this report were obtained using 0.2 mM palmitoyl-CoA; however, many of the experiments were confirmed using this substrate at 0.1 mM. Palmitoyl-CoA yielded the greatest activities of the fatty acyl-CoA thioesters examined (Table 11). Activities were only slightly lower with the 15 and 17-carbon homologs, but decreased considerably when the alkyl chain length differed from paImitoyl-CoA by two carbons. Myristoleoyl-, palmitoleoyl- and oleoyl-CoA’s yielded substantially lower activities than their fully saturated counterparts. Despite the low values with some of the fatty acyl-CoA’s, they were significantly higher than that obtained with a control which did not

Fig. 3. The dependence of activity on the concentration of palmitoyi-CoA. The pahnitoyl-CoA concentration was varied as shown, and the values for the minus palmitoyl-CoA control were not subtracted in calculating the final activities.

288

TABLE

difference was found. However, the activity was completely abolished by incubation of the cells with 1 mM r_-cysteine for 15 min prior to the assays that omitted pyridoxal 5’-phosphate (the final cysteine concentration was 0.1 mM). Cysteine forms a stable thiazoldine derivative of pyridoxal 5’-phosphate that is inactive with enzymes that require this coenzyme [18]. To determine if the serine palmitoyltransferase activities of CHO cells were affected by the presence of sphingolipids in the media, cells were cultured under the conditions described in Table III. When cells were grown to approx. 0.1 . 10’ cells/plate in medium containing 5% fetal calf serum (FCS), then washed and replenished with new media with or without fetal calf serum and LDL, no effect on the activity was found, whereas the cell numbers did vary somewhat (Table III, A). Similar results were obtained when delipidated (DL-FCS) or lipoprotein-depleted (LD-FCS) fetal calf serum was added. If a correlation exists for these data, it would appear to be that the serine palmitoyltransferase activities increase with the cell number (i.e., growth rate). Since the possibility existed that the residual serum lipids and/or lipoproteins could have masked the effects of their readdition, the CHO

II

SERINE PALMITOYLTRANSFERASE ACTIVITIES DIFFERENT FATTY ACYL-CoA THIOESTERS Fatty acyl-CoA’s with varying alkyl chain lengths of saturation were substituted for palmitoyl-CoA.

WITH

and degrees

Name

Alkyl chain length Number of carbon atoms (unsaturation)

Activity a (pmol/min

Myristoyl-CoA Pentadecanoyl-CoA Palmitoyl-CoA Heptadecanoyl-CoA Stearoyl-CoA

14 15 16 17 18

0.9 * 0.1 8.7 + 0.6 12.1 kO.3 9.OAO.3 2.4kO.l

Myristoleoyl-CoA Palmitoleoyl-CoA Oleoyl-CoA

14(A9) 16(A9) 18(49)

per lo6 cells)

0.5 + 0.1 2.8 k 0.1 1.3kO.2

a Means f S.D. for three determinations.

receive fatty acyl-CoA’s (0.12 pmol/min per lo6 cells). Hence, they were not artifacts of the utilization of endogenous substrates and/or other side reactions involving [ 3H]serine. Assays of sonicated cells were conducted with and without the inclusion of supplemental pyridoxal 5’-phosphate to determine whether or not the enzyme was saturated with its cofactor. No TABLE SERINE

III PALMITOYLTRANSFERASE

The cells were grown in F12 medium were replenished every 24 h.

ACTIVITIES

OF CHO-Kl

with the initial supplements

Supplements

A

B

Initial

New

5% FCS

(1) 5% FCS (2) 5% FCS+ 100 pg/ml (3) F12 only

5% DL-FCS

(I) 5X DL-FCS (2) 5% DL-FCS+ (3) F12 only

C

5% LD-FCS

(1) 5% LD-FCS

D

SeC

(1) none (2) 100 pg/mI (3) 5% FCS

LDL

100 pg/ml

LDL

LDL

a AI1 groups had over 99% cell viability. ’ Means* S.D. for 4-8 separate determinations. ’ Se refers to the selenium-containing (serum-free)

supplement.

CELLS

GROWN

IN DIFFERENT

CULTURE

for 5 days then for 3 days with the new supplements.

Cell number a,h (X 10Fh/plate)

Activity h (pmol/min

0.811 kO.027 0.973 k 0.072 0.463 k 0.055

14.8*4.2 13.9+ 1.1 11.6k2.7

0.568 & 0.051 0.730 f 0.076 0.221 f 0.176

13.2k3.9 19.2 f 2.9 9.9 f 4.3

1.48 kO.22

15.8 * 5.4

0.220 * 0.020 0.363 f 0.035 0.410 f 0.027

12.7+2.1 14.6 + 3.1 16.3 + 2.2

MEDIA The new media

per lo6 cells)

289

cells were also grown in media without serum. A modified formulation based upon Ham and Hamilton [12] supported cell growth at a somewhat lower rate than did media containing serum. Nonetheless, supplementation of the medium with 100 pg/ml of LDL or whole fetal calf serum did not alter the activities significantly (Table III, D). The effect of adding long-chain bases directly to the culture medium was examined. As is shown in Fig. 4, sphinganine was highly toxic to CHO cells, despite the presence of albumin in the media. At a concentration of 1 PM sphinganine, which was selected since it did not exhibit such toxicity, serine palmitoyltransferase activities were not affected by the presence of the exogenous supply of long-chain base. The CHO cells were also grown in media containing dialyzed fetal calf serum or organic-solvent extracted serum and the incorporation of [ 3H]serine into cellular sphingomyelin was quantitated. After 24 h the mean cpm/106 cells (k S.D.) for these two groups were 3382 k 1445 and 4129 f 1394, respectively. This indicated that somewhat larger amounts of the long-chain bases were formed de novo by cells that were cultured in medium without these molecules; however, the differences between the groups were not statistically significant (P < 0.25). Only slightly greatly

1 a- 0.4 '0 v-

x 0.3 Q)

l

G 0.2 P \ i 0.1

0

: -LA 0

2 VM

4

6

100

Sphinganine

Fig. 4. Effects of adding erythro-sphinganine to cultures of CHO cells. Sphinganine in 25 ~1 of ethanol was added to 10 ml of culture medium (containing 10% FCS) to yield the final concentrations shown. The CHO cells were grown to approx. 0.1.106 cells/plate in culture medium, changed to medium containing sphinganine, and after 48 h the cells were recovered by mild trypsin treatment and counted.

differences were reported by Verdery and Theolis [9] using human fibroblasts. For comparison, human fibroblasts for which lipoprotein uptake has been well characterized [19] were examined and the results are given in Table IV. These cells were grown to near confluence then washed and changed to media with or without whole or lipoprotein-depleted serum. The serine palmitoyltransferase activities were similar to those for CHO cells. Furthermore, the absence of serum or serum lipoproteins did not result in activities higher than the control, which contained whole serum. Similar results were obtained when the cells were cultured under these conditions for 5 days (data not shown). Discussion Sphingomyelin constitutes approx. 9.9% of the total phospholipid of CHO cells [20] which is relatively high compared to other types of tissue [ 11. Thus, plus the accumulating information about lipid metabolism by these cells and derived mutant strains, would recommend these cells as a useful model for studies of long-chain base biosynthesis. This report describes the first analyses of the serine palmitoyltransferase in cultured mammalian cells, and the first experimental test of the hypothesis that exogenously supplied sphingolipids shut off long-chain base synthesis at the level of this enzyme. The specific activity in the particulate fraction (i.e., airfuge pellet) from sonicated CHO cells was greater than for rat liver microsomes (= 60 pmol/ min per mg) [7] despite the likelihood that the former contains a less homogeneous subcellular fractionation. This could be related to either the higher sphingomyelin content of these cells, or their rapid growth rate and, hence, requirement for membrane lipid synthesis. That such high activities were obtained despite the addition of 5-108 fetal calf serum to the culture media suggested that serum sphingolipids do not strongly suppress the activity of this enzyme, if at all. More definitive evidence against such suppression was the finding of essentially identical activities in cells grown without serum, with lipid and/or lipoprotein-depleted serum, with LDL, with sphinganine and with whole serum. Since whole cell extracts

290

were assayed, it is not likely that regulatory factors were lost, although their effects could have been diminished upon dilution. Since this cell line is known to be fully competent in the uptake of lipoproteins and the inhibition of de novo cholesterol biosynthesis [21], we conclude that long-chain base synthesis is not regulated by the uptake of exogenous sphingolipids, as has been proposed by Verdery and Theolis [9] based upon the decreased incorporation of radiolabelled serine or palmitate into long-chain bases of fibroblasts when the cells utilize LDL. This contradiction was not caused by differences in the cell types examined because the incorporation of [ 3H]serine into sphingomyelin long-chain bases was decreased slightly by culturing CHO cells in whole serum, and because the serine palmitoyltransferase activities of human fibroblasts were also unaffected by serum lipoproteins (Table IV). It appears likely that the effects of plasma lipoproteins [9] reflect dilution of the pool of radiolabelled long-chain bases (or of the precursors serine and palmitate) by those supplied exogenously. The long-chain bases would be released from the sphingolipids by lysosomal enzymes [22] and appear in the cytosol and endoplasmic reticulum, where they could be degraded by phosphorylation and cleavage [2,3] or reconverted into ceramides. The extent to which the long chain bases of complex sphingolipids are reused versus degraded is not known; however,

TABLE

IV

SERINE PALMITOYLTRANSFERASE HUMAN FIBROBLASTS CULTURED DIFFERENT SUPPLEMENTS

ACTIVITIES OF IN MEDIA WITH

The cells were grown in media containing 10% fetal calf serum for 7 days, then for 3 days in media containing the supplements shown. Supplement

Cell number +.b ( X lo6 cells/plate)

Activity b (pmol/min per lo6 cells)

None 5% Lipoprotein - depleted 5% Whole serum

0.383 * 0.053 0.569 f 0.039 0.528 f 0.046

18.1 f4.6 13.6+ 5.3 27.6 f 6.2

a Al1 groups had over 96% cell viability. b Means f SD. for four plates/group.

Stoffel et al. [23] have found that only a small fraction of administered sphingosine was recovered in sphingolipids. Based upon the reported cholesterol content of these cells when provided with LDL (115 nmol/ mg protein, calculated from Ref. 24 and including cholesterol and cholesterol esters), and the known lipid ratios of human LDL (15.4 : 2.5 : 1 for cholesterol and cholesterol esters : phosphatidylcholine : sphingomyelin) [25], the maximal uptake of sphingomyelin, assuming that all of the cellular cholesterol is obtained by LDL uptake, would be 7.5 nmol. This is approx. one-third of the sphingomyelin content of these cells (22 nmol/ mg, calculated from Ref. 26). Hence, even if all of the long-chain bases were recycled, the major portion must be provided by de novo biosynthesis. It should be noted that CHO cells cultured in lipoprotein-depleted media with LDL supplementation have a 3-fold greater content of cholesterol and cholesterol esters [24]. Since it has been proposed that the sphingomyelin and cholesterol levels of cells may be related [27], our finding of greater activities with cells provided LDL could reflect a compensatory increase in sphingomyelin biosynthesis de novo. This possibility is currently under investigation. These experiments have also shown that serine palmitoyltransferase exhibited greatest activities with fully saturated fatty acyl-CoA thioesters having a linear alkyl chain length of 16 k 1 carbon atoms. Identical behavior has been observed with rat liver [7] and brain [4,10] microsomes and microorganisms [2,28,29]. That palmitoyl-CoA yielded the highest activity is in excellent agreement with the prevalence of l&carbon long-chain bases in mammalian sphingolipids. Lesser amounts of 14- through 20-carbon homologs have been observed [ll], which can be explained by the lower activities with the other fatty acyl-CoA’s. It is somewhat surprising, however, that the long-chain bases that would be formed from oleoyl-CoA, which is a major constituent of cells, have not been reported. This could reflect an oversight of previous analyses, that oleoyl-CoA is not used in vivo, or that the resultant long-chain bases are modified or handled selectively by subsequent enzymes. Tests of these hypotheses are now feasible using CHO cells and constitute ongoing studies of the

291

enzymology and palmitoyltransferase.

regulation

of

serine

Acknowledgements The author thanks Dr. Jack Kinkade and Mrs. Kathy Barnes for advice and support with the cell culture, and Ms. L. Shulin for testing the CHO cells for contamination by mycoplasma. Thanks are also due to Drs. Jeff Esko and C.R.H. Raetz for warning the author about the toxicity of sphingosine to cells, and Dr. Robert M. Bell for suggesting comparison of the lipids of lipoproteins and CHO cells. This work was supported by a Biomedical Research Grant (RR 5364) from the National Institutes of Health and funds from the Georgia Affiliate of the American Heart Association References Rouser, G., Nelson, G.J. and Fleischer, S. (1968) in Biological Membranes (Chapman, D., ed.) pp. 5-69, Academic Press, New York Snell, E.E., DiMari, S.J. and Brady, R.N. (1970) Chem. Phys. Lipids 5, 116-138 Stoffel, W. (1970) Chem. Phys. Lipids 5, 139-158 Braun, P.E., Morel& P. and Radin, N.S. (1970) J. Biol. Chem. 245, 335-341 Kanfer, J.N. and Bates, S. (1970) Lipids 5, 718-720 Krisnangkura, K. and Sweeley, SC. (1976) J. Biol. Chem. 251, 1597-1602 Williams, R.D., Wang, E. and Merrill, A.H., Jr. (1983) Arch. Biochem. Biophys. (in the press) Williams, R.D., Nixon, D.W. and Merrill, A.H., Jr. (1983) J. Nutr. Growth Cancer (in the press)

9 Verdery, R.B. and Theolis, R. (1982) J. Biol. Chem. 257, 1412-1417 10 Merrill, A.H., Jr. and Williams, R.D. (1984) J. Lipid Res., in the press 11 Karlsson, K.-A. (1970) Chem. Phys. Lipids 5, 6-43 12 Hamilton, W.G. and Ham, R.G. (1977) In Vitro 13,537-547 13 Cham, B.E. and Knowles, B.R. (1976) J. Lipid Res. 17, 176-181 14 Spanner, S. (1973) in Form and Function of Phospholipids (Ansell, G.B., Hawthorne, J.N. and Dawson. R.M.C., eds.), pp. 43-65, Elsevier, Amsterdam 15 Bligh, E.A. and Dyer, W.J. (1959) Can. J. Biochem. 37, 911-917 16 Bensadoun, A. and Weinstein, D. (1976) Anal. Biochem. 70, 241-250 17 Braun, P.E. and Snell, E.E. (1968) J. Biol. Chem. 243, 3775-3783 18 Federiuk, C.S. and Shafer, J.A. (1981) J. Biol. Chem. 256, 7416-7423 19 Barak, L.S. and Webb, W.W. (1981) J. Cell Biol. 90.595-604 20 Polokoff, M.A., Wing, D.C. and Raetz, C.R.H. (1981) J. Biol. Chem. 256, 7687-7690 21 Sinensky, M. (1979) FEBS Lett. 106, 129-131 22 Chen, W.W., Moser, A.B. and Moser, H.W. (1981) Arch. Biochem. Biophys. 208, 444-455 B. and Heimann, G. (1973) 23 Stoffel, W., Hellenbrioch, Hoppe-Seyler’s Z. Physiol. Chem. 354, 1311-1316 24 Chin, D.J., Luskey, K.L., Anderson, R.G.W., Faust, J.R., Goldstein, J.L. and Brown, M.S. (1982) Proc. Natl. Acad. Sci. U.S.A. 79, 1185-1189 25 Eisenberg, S. (1976) in Low Density Lipoproteins (Day, C.E. and Levy, R.S. eds.), pp. 76-92, Plenum Press, New York 26 Wattenberg, B.W.,+nd Wilbert, D.F. (1983) J. Biol. Chem. 258, 2284-2289 28 DiMari, S.J., Brady, R.N. and Snell, E.E. (1971) Arch. Biochem. Biophys. 143, 553-565 29 Lev, M. and Milford, A.F. (1981) Arch. Biochem. Biophys. 212, 424-431