Sphingomyelin Synthesis in Rat Liver Occurs Predominantly at the cis and medial Cisternae of the Golgi Apparatus*

The intracellular site of sphingomyelin (SM) synthe- sis was examined in subcellular fractions from rat liver using a radioactive ceramide analog N-([1-14C]hexa-noyl)-D-erythro-sphingosine. This lipid readily trans- ferred from a complex with bovine serum albumin to liver fractions without disrupting the membranes, and was metabolized to radioactive SM. To prevent deg- radation of the newly synthesized SM to ceramide, all experiments were performed in the presence of EDTA to minimize neutral sphingomyelinase activity and at neutral pH to minimize acid sphingomyelinase activity. An intact Golgi apparatus fraction gave an 85-98- fold enrichment of SM synthesis and a 58-83-fold enrichment of galactosyltransferase activity. Con- trolled trypsin digestion demonstrated that SM synthe- sis was localized to the lumen of intact Golgi apparatus vesicles. Although small amounts of SM synthesis were detected in plasma membrane and rough microsome

The intracellular site of sphingomyelin (SM) synthesis was examined in subcellular fractions from rat liver using a radioactive ceramide analog N-([ Subfractions of the Golgi apparatus were obtained and characterized by immunoblotting and biochemical assays using cisjmedial (mannosidase II) and trans (sialyltransferase and galactosyltransferase) Golgi apparatus markers.
The specific activity of SM synthesis was highest in enriched cis and medial fractions but far lower in a trans fraction. We conclude that SM synthesis in rat liver occurs predominantly in the cis and medial cisternae of the Golgi apparatus and not at the plasma membrane or endoplasmic reticulum as has been previously suggested.

Sphingomyelin
(SM)' is ubiquitous in animal tissues and has been found in almost every cell and membrane examined. However, the intracellular site of SM synthesis is not well established. In a number of cell types and using a variety of substrates the plasma membrane (PM) (l-5), endoplasmic reticulum (ER) (6,7), and Golgi apparatus (5,(8)(9)(10)(11)(12)) have been suggested as major sites of SM synthesis. The current study was undertaken to resolve this issue and is particularly significant given current ideas about the regulation of protein kinase C by diacylglycerol (13) and sphingosine (14, 15), products of SM synthesis and degradation, respectively. In addition, localization of the site of SM synthesis may have important implications for intracellular lipid traffic (16). This study differs from previous work in several significant ways. First, we examined the synthesis of SM in subcellular fractions from rat liver, which, in contrast to the subcellular fractions from cultured cells (17) used in some of the previous studies (l-5), are highly enriched for particular organelles and are well characterized.
Second, SM synthesis was measured using N-( [ l-l%] hexanoyl)-D-erythro-sphingosine (["Cl hexanoyl Cer), a radioactive analog of the naturally occurring stereoisomer of ceramide (Cer). By virtue of the short-chain (hexanoyl) fatty acid, this analog rapidly and spontaneously transferred into biological membranes and did not destroy the integrity of the membranes. This is in contrast to some previous studies in which lipid substrates were incorporated into membranes by dispersion with detergents (1, 3) or by extensive incubations (2-6 h) with protein complexes (2,4,5, 7). Finally, special consideration was given to inhibiting PMassociated neutral sphingomyelinase activity (l&20), allowing accurate comparison of the amounts of SM synthesis in different subcellular compartments.
Using these approaches, we demonstrate that the major site of SM synthesis in rat liver is the lumenal side of the cis and medial cisternae of the Golgi apparatus.   (Fig. 1, A and B). Therefore, all subsequent assays of ['4C]hexanoyl SM synthesis were performed in the presence of 0.5 mM EDTA, while sphingomyelinase assays were performed in the presence of 5 mM MgC12. An intact Golgi apparatus fraction was obtained by a rapid (5-6 h) one-step isolation procedure (26) which allows easy quantification of recoveries of marker enzymes used to identify subcellular compartments. This fraction is essentially devoid of membranes derived from other organelles, with the exception of small amounts of endosomes (42). The isolation procedure was modified by omitting M$+ from all sucrose solutions to reduce neutral sphingomyelinase activity (see above). There was no difference in marker enzyme distribu- tion in fractions prepared with or without Mg*+ (not shown) and 45-60% of galactosyltransferase activity was recovered in fraction b, the intact Golgi apparatus fraction (see Table II).3 Electron microscopy of samples prepared with or without Mg*+ revealed intact Golgi stacks with similar morphology (not shown). When sphingomyelinase activity was assayed with the addition of Mg*+ to the incubation buffer the majority of the activity was in fraction e (Fig. 2)   ]hexanoyl SM was detected in fraction e, but this activity was probably due to Golgi apparatus membranes as 18% of the recovered galactosyltransferase activity was found in fraction e in addition to the majority of the recovered alkaline phosphodiesterase and glucose-6-phosphatase activities (Table II). f4C]Hexanoyl SM Synthesis in Subcellular Compartments of Rat Liver-To determine the contribution of various subcellular compartments to SM synthesis in rat liver, ['"Cl hexanoyl SM synthesis was compared in enriched Golgi apparatus, PM, and ER fractions. An enriched PM fraction was prepared (29) without Mg*+. Marker enzyme recovery (Table  III) was similar to that reported (29) with alkaline phosphodiesterase enriched by 12-30-fold. Small but variable amounts of galactosyltransferase were detected (Table III) and electron microscopy revealed the presence of some Golgi apparatus cisternae. Relatively large amounts of galactosyltransferase and alkaline phosphodiesterase were observed in smooth microsomes prepared by two methods (30,31). Rough microsome fractions were considerably purer than the smooth microsome fractions, and rough microsomes prepared according to Adelman et al. (31) showed similar marker enzyme distribution (Table III) and appeared similar by electron microscopy to that reported (31), with glucose-6-phosphatase enriched by 2.5-3-fold. Electron microscopy of the rough microsome fraction also revealed the presence of some Golgi apparatus cisternae.

EXPERIMENTALPROCEDURES
[W]Hexanoyl SM synthesis was compared in the enriched fractions and in a homogenate as a function of protein concentration (Fig. 3, A and D), time (Fig. 3, B and E) and substrate concentration (Fig. 3, C and F    Fraction-The topology of the enzyme responsible for SM synthesis was determined by examining the sensitivity of [Wlhexanoyl SM synthesis to trypsin in intact and permeabilized Golgi apparatus vesicles. In intact Golgi apparatus vesicles either harvested directly from the gradient or subsequently dialyzed against 0.25 M sucrose, trypsin had no significant effect on [Wlhexanoyl SM synthesis4 or galactosyltransferase activity (32) ( Table V). A number of methods were attempted to permeabilize the Golgi apparatus including addition of detergents, freeze-thaw, dialysis against either 10 mM NaHC03 (pH 11) followed by dialysis against 5 mM Tris (pH 7.4), or dialysis against 5 mM Tris (pH 7.4) alone. Detergents could not be used as those tested (Triton X-100, sodium taurodeoxycholate, n-octyl-a-D-glucopyranoside) completely inhibited [W]hexanoyl SM synthesis at concentrations which permeabilized the Golgi apparatus. The other methods did not render all of the galactosyltransferase activity sensitive to trypsin. However, dialysis against water was found to render over 95% of the galactosyltransferase activity sensitive to trypsin, demonstrating that the Golgi apparatus vesicles were permeabilized.
N-j6-[(7-Nitrobenzo-Z-oxa-I,3-diazol-4-yl)amino]caproylJ-D-erythro-sphingosine undergoes rapid transbilayer movement in Iiposomes (43) and cells (8,9,43)   and autoradiography. Results are expressed as relative specific activities, defined as the ratio of specific activity in the fraction compared to specific activity in Gn, with the latter normalized to 1. Data represent means + S.D. The cisternae in which most of each enzyme activity is localized within the Golgi apparatus are shown in parentheses. [ Golgi elements (26,28,44,45,47), even though considerable amounts of each were present in Gn and Gi.
To further characterize the Golgi subfractions, we determined the distribution of a &/medial Golgi marker, mannosidase II (21,48), by immunoblotting. In contrast to the trans Golgi markers, the highest specific activity of mannosidase II was in Gn and Gi (Table VI), but was 60% lower in GL. In some experiments, mannosidase II was barely detectable in GL (not shown).
The specific activity of ['4C]hexanoyl SM synthesis paralleled that of mannosidase II and was 2-3 times higher in Gu and Gi than GL (Table VI) and in some experiments there was no detectable synthesis of ['YJJhexanoyl SM in GL. These results strongly suggest that cis and medial elements of the Golgi apparatus are enriched in SM synthesis but that little or no synthesis of ['4C]hexanoyl SM occurs in the trans-Golgi in rat liver.

The Golgi Apparatus
Is the Major Site of SM Synthesis-In this paper, we demonstrate that the Golgi apparatus is the major site of SM synthesis in highly enriched and well characterized subcellular fractions from rat liver. In addition, we show that SM synthesis occurs on the lumenal side of the cis and medial cisternae of the Golgi apparatus. These results are in agreement with previous studies implicating the Golgi apparatus as the site of SM synthesis based on the metabolism, translocation, and intracellular distribution of a fluorescent analog of Cer in cultured fibroblasts (8)(9)(10)(11) and in Madin-Darby canine kidney cells (12). However, they are in disagreement with earlier suggestions that the PM or ER is the major site of SM synthesis. Several earlier studies implicated the PM as the major site of SM synthesis (l-5). However, the conclusions from these studies need reevaluating in light of the following: (i) Limitations of subcellular fractionation in cultured cells. It is difficult to obtain highly purified subcellular fractions from cultured cells (17) and PM fractions may be contaminated with membranes derived from other organelles. (ii) Inadequate characterization of the subcellular fractions. In the study which established phosphatidylcholine as the proximal donor of the phosphorylcholine moiety in the pathway of SM synthesis (1) significant amounts of SM were synthesized by a PM fraction. However, this fraction was characterized only on the basis of 5'-nucleotidase activity, which is known to be present in both Golgi and PM fractions (49,50). In several studies, the purity of the PM fractions and the contamination by Golgi apparatus membranes was not assessed (l-4) but in a study in which PM fractions were characterized for Golgi apparatus contamination, SM synthesis was found in a "Golgi-PM" fraction (5). (iii) Hydrolysis of SM at the PM. The effect of PM-associated neutral sphingomyelinase (18)(19)(20) in degrading newly synthesized SM was not evaluated in any of the earlier studies (l-5) and thus the quantification of SM synthesis could not be accurately determined. In the current study, neutral sphingomyelinase activity was inhibited and the PM fraction used was well characterized and highly enriched for a defined PM marker (Table III). After correcting for contamination by Golgi apparatus membranes (Table IV), a small amount of SM synthesis was associated with the PM fraction. At present we cannot estimate the contribution that this small amount of SM synthesis makes to the total SM pool at the PM since the rate of turnover of SM at the PM of rat liver is not known.
The ER has also been suggested as the intracellular site of SM synthesis in subcellular fractions from rat (6) and mouse (7) liver. However, the conclusions from these studies also need reevaluating due to the following: (i) Incorrect choice of substrates. In a study using rat liver (6), a nonnaturally occurring stereoisomer of Cer was used. Furthermore, CDPcholine was used as the phosphorylcholine donor, but it was subsequently shown (I) that CDP-choline is not involved in the pathway of SM synthesis. (ii) Inadequate characterization of the microsome fraction. Using mouse liver (7), no attempt was made to differentiate between activity due to ER and Golgi apparatus membranes in the microsomal fractions. In our study, all of the SM synthesis in a rough microsome fraction could be accounted for by contaminating Golgi apparatus membranes.
We conclude that the Golgi apparatus is the major site of SM synthesis in rat liver, although the possibility that other intracellular sites (e.g. PM or ER) contribute to the total SM pool in other cell types remains to be established. In addition, the effects of other parameters (e.g. levels of endogenous substrates, see Ref. 5) may need to be more fully evaluated.
The finding that the Golgi apparatus is the major site of SM synthesis is of particular importance in light of recent suggestions that the synthesis and degradation of SM may be coordinately regulated and that the products of SM metabolism may be involved in modulating the activity of protein kinase C (14-16). The suggestion of a coordinated role of the products of SM metabolism was based upon the assumption that most intracellular SM is synthesized at the PM (15), not at the Golgi apparatus. However, a subspecies of protein kinase C has recently been found to be associated primarily with the Golgi apparatus in some cells (13) raising the intriguing possibility that SM synthesis may be involved in the regulation of protein kinase C at this organelle.
Topology and Distribution of SM Synthesis within the Golgi Apparatus-In contrast to some other enzymes involved in lipid synthesis, which are localized to the cytosolic side of the Golgi apparatus (33), the transferase responsible for SM synthesis was localized to the lumenal membranes of this organelle (Table V). This is similar to the localization of a number of glycoprotein sugar transferases (32, 51, 52) and glycolipid sugar transferases (53). As previously suggested (16), the localization of SM synthesis on the lumenal side of the Golgi apparatus is consistent with the fact that the majority of SM is on the external leaflet of the PM (54, 55), which is topologically equivalent to the lumenal leaflet of the membranes of the Golgi apparatus.
While previous studies have used Golgi apparatus subfractions to examine the distribution of some phospholipids and to examine the synthesis of phosphatidylcholine (44, 56) the present study represents the first attempt to examine SM synthesis in Golgi apparatus subfractions. The procedure used for the preparation of these subfractions (28) takes advantage of the density shifts imparted to Golgi elements by enclosed lipoprotein particles. Due to the unavailability of a cis/mediul Golgi apparatus marker until recently (21), the assignment of the Golgi apparatus subfractions as &-enriched (Gn), mediulenriched (GJ, and truns-enriched (GL) was based upon morphological criteria (28), the sequential progression of secreted proteins through the fractions (47), and upon the distribution (27,(44)(45)(46) of the terminal glycosyltransferases, sialyl-, galactosyl-, and GlcNAc-transferase, enzymes that reside in the tram Golgi (57,58). We have now confirmed these assignments by immunoblotting with cislmediul (mannosidase II, Ref. 48) and tram (sialyltransferase, Ref. 58) Golgi apparatus markers. We found that GL is truns-enriched but does contain small amounts of contamination by cis/mediul elements.
Using these subfractions, we found enrichment of SM synthesis in ci.s/mediul elements of the Golgi apparatus (Table  VI). This result is consistent with earlier observations (9) that monensin, an inhibitor of glycoprotein transport between medial and tram regions of the Golgi apparatus (59,60), blocks the appearance of newly synthesized fluorescent SM at the PM without inhibiting its synthesis from a fluorescent Cer precursor. We cannot exclude the possibility that the tram-enriched subfraction of the rat liver Golgi apparatus represents a subset of tram elements as the yield of tram Golgi markers obtained in this fraction is rather low but this seems unlikely due to the high specific activity of the truns Golgi markers in this fraction (Table VI).
We are currently attempting to isolate the enzyme responsible for SM synthesis in rat liver, phosphatidylcholine:Cer cholinephosphotransferase, obtain antibodies, and subsequently examine the distribution of the enzyme in the Golgi apparatus by immunolocalization and electron microscopy. Isolation of this enzyme would provide a means for further studying the role of the Golgi apparatus in sphingolipid synthesis, regulation, and transport.