Brefeldin A-induced increase of sphingomyelin synthesis. Assay for the action of the antibiotic in mammalian cells.

Brefeldin A leads to an increase of sphingomyelin in Chinese hamster ovary cells. The antibiotic is known to cause a dramatic morphological change of the endomembrane system in various mammalian cells resulting in a redistribution of Golgi resident proteins to the endoplasmic reticulum (Lippincott-Schwartz, J., Donaldson, J. G., Schweizer, A., Berger, E. G., Hauri, H. P., Yuan, L. C., and Klausner, R. D. (1990) Cell 60, 821-836). A strict correlation was found between the brefeldin A-induced increase of sphingomyelin and the biochemical criteria that apply for this morphological change. From our data we conclude that the increase in sphingomyelin caused by brefeldin A reflects translocation of the enzyme sphingomyelin synthase from the Golgi apparatus to the endoplasmic reticulum. Using a radioactively labeled truncated ceramide this increase in sphingomyelin synthesis is easily detectable, and thus this method can serve as a convenient biochemical assay for the action of brefeldin A in mammalian cells.

The antibiotic brefeldin A (BFA)' has attracted the interest of biochemists and cell biologists because it acts on the equilibrium state of intracellular organelles (1-3) and thus interferes with vesicular protein transport (4). These effects on the distribution of intracellular proteins as well as on transport have been demonstrated by fluorescence and electron microscopy using labeled antibodies against marker proteins (1)(2)(3) or by measuring the transport of proteins which have been pulse-labeled in vivo (2,3,5,6). Such experiments, though very effective, are either not quantitative at all or technically complicated and time consuming.
The molecular mechanism of action of the antibiotic is still unclear. An immediate effect is to cause dissociation of coat proteins (7) from the Golgi (4,8). Subsequently, the organelle appears to disintegrate by vesiculation, and after a few min-* This work was supported by Grant SFB 352 from the Deutsche Forschungsgemeinschaft. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Utes Golgi enzymes are translocated to the endoplasmic reticulum (ER) (2, 3). These effects are fully reversible provided that ATP is not depleted in the cells (1-3). The dissociation of coat proteins from the Golgi caused by BFA can be inhibited by preincubation with GTPyS or AlF;, indicating that trimeric G-proteins are somehow involved in this process (4,9). Nocodazole, a microtubuli-disassembling drug, inhibits translocation of Golgi enzymes to the ER but does not inhibit BFA-induced dissociation of coat proteins from the Golgi (10). These findings have supported a model of recycling of Golgi constituents to the ER via an intermediate (10,11) or salvage (12) compartment.
We have developed a ceramide analogue that is truncated in both hydrophobic chains (13,14). This water-soluble and amphiphilic compound CsCs-Cer readily permeates biological membranes and is converted in vivo to yield the sphingolipid analogues "truncated sphingomyelin" (CsC8-SM) and "truncated glucosylceramide" (CsCs-GlcCer).
Biosynthesis of sphingomyelin in CHO cells occurs by transfer of a phosphorylcholine moiety from phosphatidylcholine to ceramide.' The enzyme sphingomyelin synthase is located in the cis (15) or cis and medial (16) Golgi, whereas both substrates, phosphatidylcholine and ceramide, are synthesized in the ER (17,18), a membrane system of about 10 times the size of the Golgi apparatus, with correspondingly more phosphatidylcholine and ceramide present than in the Golgi. If the availability of one of these substrates is rate-limiting in physiological sphingomyelin synthesis in the Golgi, then translocation of sphingomyelin synthase from the Golgi to the ER caused by BFA should lead to an increase of sphingomyelin. This has led us to probe sphingomyelin synthesis in the presence of BFA. We show here that sphingomyelin synthesis from endogenous ceramide is indeed increased after administration of BFA to CHO cells. An even more drastic increase is observed with the exogenously added analogue CsCs-Cer. Stimulation of CaC8-SM biosynthesis observed after administration of BFA was found to be in accordance with all biochemical criteria so far defined for the action of BFA on the endomembrane system. Using the model substrate C8C8-Cer, sphingomyelin synthase activity can quickly and quantitatively be followed. This assay may be useful for an easy and effective screening of cells for their sensitivity to BFA as well as for drugs antagonistic to BFA.

EXPERIMENTAL PROCEDURES
Materials-BFA and its derivatives were kindly provided by R. E. Kneusel and Dr. U. Matern (Freiburg). Forskolin and nocodazole were purchased from Sigma, Munchen. All cell culture reagents were from Biochrom KG, Berlin. Truncated ceramide (CsCs-Cer) and [3H] C8Cs-Cer were prepared as described (13).
CeU Culture-CHO cells (ATCC CCL 61) were grown in suspension cultures in a-MEM containing 7.5% fetal calf serum and 100 units of penicillin plus 100 units of streptomycin per ml. Cells were harvested at a density of 6-8 X lo5 cells/ml and resuspended at 5 X 10' cells/ ml in a-MEM without fetal calf serum and without antibiotics but supplemented with 20 mM Hepes buffer (pH 7.2).

Labeling of CHO Cells with PHJClwline-1 X IO5 CHO cells each
were plated out in plastic Petri dishes (3.5 cm) in the medium described under "Cell Culture" and grown to near confluency (>go%). After preincubation with BFA (5 pg/ml) at 37 "C for 15 min, 5 WCi of [3H]choline (75 Ci/mmol, Amersham Corp.) was added, and the cells were kept at 37 "C for 1, 3,6, and 16 h, respectively, in a humidified atmosphere of 5% CO,. Thereafter, the cells were harvested using trypsin as described above for PtK cells. Membrane lipid extraction was carried out according to Ref. 19 with the following modifications. The cell pellets were resuspended in 3 volumes of water, then methanol and chloroform were added to obtain ratios of 0.8/2/1 for H,O/ CH30H/CHC13. After sonication for 5 min (Sonorex, Bandelin, Berlin), more CHCl3 was added to give ratios of 0.8/2/2 for H20/CH30H/ CHCl3. After another 5 min of sonication, water was added to a final ratio of 1.8/2/2 for H20/CH30H/CHCla, and the two phases were separated. The lipids under study (phosphatidylcholine and sphingomyelin) were found exclusively in the CHC1, phase and subsequently separated by TLC (HPTLC Silica Gel 60 plates, Merck) using CHCl3/CH30H/H20 (65/25/4) as eluant. The amounts of phosphatidylcholine and sphingomyelin were determined by two-dimensional TLC radioscanning (Berthold, Wildbad). In order to calibrate the radioscanner, in early experiments single spots were scraped off using an autoradiograph as a guide, and the radioactivity was determined by liquid scintillation counting. A spot of 1000 counts/h on the two-dimensional TLC scanner corresponds to 1800 cpm in the scintillation counter.
Standard Assay-2.5 X lo6 CHO cells in 50 pl of a-MEM were incubated in the presence or absence of 5 pg of BFA/ml for 15 min at 37 "C. Then 5 pl of a premix of [3H]C8Cs-Cer/CeCs-Cer in a-MEM was added to give a final radioactivity of 2.8 pCi of [3H]C8Cs-Cer and a final concentration of 150 p~ unlabeled C8Ca-Cer. After 30 min at 37 'C, the cells were chilled on ice, centrifuged (30 s, 1,200 X g), and the cell pellet was extracted with 50 pl of methanol/water (l/l). After centrifugation at 13,000 X g for 5 min, 5-pl aliquots of the cell extracts and of the corresponding media were subjected to TLC on Silica Gel 60 (Merck) with butanone/water/acetone/formic acid (6/1/0.6/0.02) as the solvent system. The radioactivities of CsCs-SM and CaCs-GlcCer were determined by two-dimensional radioscanning as described above.

Sphingomyelin Synthesis Is Increased in the Presence of
BFA-Two different pathways of sphingomyelin synthesis are discussed in the literature: (i) transfer of phosphorylcholine from phosphatidylcholine to ceramide (both reaction partners are generated in the ER (17,18) or (ii) direct transfer of phosphorylcholine from CDP-choline to ceramide (20). We have ruled out the latter pathway to occur in CHO cells.' Thus, for physiological sphingomyelin synthesis to take place in the cis or in the cis and medial Golgi, both substrates are likely to be provided by the ER.
As BFA leads to a redistribution of Golgi-resident proteins into the ER, we have measured sphingomyelin biosynthesis in the absence and presence of BFA by labeling CHO cells with [3H]choline and subsequent determination of the labeled products. The result is shown in Table I. Addition of BFA leads to an approximate 2-fold increase of sphingomyelin. The reaction measured in this assay is rather complex because it depends both on the newly formed [3H]phosphatidylcholine and on endogenous ceramide; each substrate could be ratelimiting in the Golgi. In order to assess the effect of BFA on sphingomyelin synthesis independent of endogenous ceramide, and, as a fringe benefit, to exploit the complete (unlabeled) pool of phosphatidylcholine, we have employed [3H] CsCs-Cer as a precursor. This compound can be added in excess to cells exogenously and is converted into [3H]CaCs-SM and [3H]CsCs-GlcCer in vivo. The products are soluble in water and will eventually be secreted into the medium (13). Using this assay system, sphingomyelin synthesis was found  to be drastically increased by BFA (Fig. 1). A maximal cellular content of [3H]CsCs-SM representing almost 5-fold the amount found in control cells is already observed 30 min after the addition of BFA. In contrast, the level of [3H]CaCs-GlcCer is essentially uninfluenced by the drug. Does this increase reflect the described BFA-induced fusion of the Golgi with the ER or rather a direct influence of BFA on the activity of the enzyme sphingomyelin synthase? In order to exclude the latter possibility, we investigated [3H]CaC8-SM synthesis in vitro using isolated intact CHO Golgi membranes as described (15). Membranes were incubated at 37 "C for 15 min with various concentrations of BFA (0-200 pglml), and subsequently [3H]CsCa-SM synthesis was determined. No influence of the antibiotic on sphingomyelin synthase was detected. This finding suggests that the increase of sphingomyelin observed after administration of BFA in vivo indeed reflects fusion of ER with Golgi membranes, i.e. translocation of sphingomyelin synthase from the Golgi to the mixed ER-Golgi organelle, in which a large supply of phosphatidylcholine is available for increased sphingomyelin synthesis. This concept was corroborated by the experiments described in the following sections.
BFA-induced Increase in Sphingomyelin Is Specific to BFA-In order to assess the specificity of the observed effect, structural analogues of BFA were tested. BFA proved to be highly specific because at 5 pg/ml neither 7-oxo-BFA, 4 , l -Brefeldin A-induced Increase of Sphingomyelin Synthesis dioxo-BFA, nor BFA-acid caused an increase of C8C8-SM synthesis (not shown). These substances are all naturally occurring derivatives of BFA. The inability of the synthetic derivatives 4-epi-BFA and @methylated BFA species to block non-clathrin-coated vesicle formation has been shown in a cell-free system (4), suggesting a highly specific binding of BFA to its target. BFA-induced Increase in Sphingomyelin Is Prevented by Nocodazole and by Fluoride-Dissociation of the Golgi can be prevented by preincubating the cells with the microtubulidestabilizing reagent nocodazole before addition of BFA (10). A similar inhibition of the antibiotic's effect by GTPrS or AlF; has been described (9). Therefore, nocodazole and fluoride were tested for their ability to inhibit BFA-induced sphingomyelin increase. CHO cells were incubated with nocodazole or fluoride, followed by BFA, and then sphingomyelin synthase activity was probed by the addition of [3H]CaC8-Cer. As shown in Fig. 2, both drugs cause a strong reduction of the BFA effect when compared with the control. In addition, fluoride completely blocks the formation of C8Cs-GlcCer (not shown). This indicates that the high concentration of fluoride used has led to energy depletion, because C8C8-GlcCer synthesis depends on UDP-Glc, the formation of which is dependent on the energy charge of a cell. Thus, the inhibition of BFA-induced C8C8-SM synthesis by fluoride may be explained by two effects: interference with trimeric G-proteins (21) and/or energy depletion. Both conditions have been described to inhibit the antibiotic's effect on the endomembrane system (2,9).
BFA-induced Increase in Sphingomyelin Is Antagonized by Forskolin and Is Reversible-Forskolin, a drug known for its activation of adenylate cyclase has been shown to antagonize the BFA-induced Golgi dissociation/ER fusion process in a CAMP-independent way (22). This drug inhibits translocation of Golgi markers into the ER when applied before adding BFA and leads to a relocation to Golgi structures of Golgi markers when administered after the addition of BFA. Likewise, the effects of BFA can be reversed by simply washing out the drug (1-3). In order to probe the described BFAinduced increase in sphingomyelin synthase under the above mentioned conditions, cells were either pretreated with forskolin and then challenged with BFA, or forskolin was added after the challenge. The results are depicted in Fig. 3; forskolin CsC8-SM content was determined. F200, forskolin was present at a concentration of 200 pg/ml for 30 min at 37 "C.

Radioactivity lcts/lOOOl
at 200 or 20 pg/ml clearly reduced the BFA effect when added before BFA. In addition, the BFA-induced increase in sphingomyelin was partially reversed by adding forskolin after BFA. We have never observed a total block of the C8C8-SM increase in accordance with the idea that forskolin acts as a, competitive antagonist to BFA (22). Likewise, a single washing of the cells with a BFA-free medium caused a strong reduction of [3H]CsC8-SM even after a recovery time as short as 15 min.

BFA-resistant PtK Cells Do Not Respond to BFA with an
Increase in Sphingomyelin-Recently, PtK-1 cells (rat kangaroo kidney cells from the marsupial Potorous tridactylis) have been described to be resistant against BFA. Even at high concentrations, the antibiotic (up to 50 pg/ml) did not induce translocation of Golgi markers into the ER (23). If the sphingomyelin increase described above is a consequence of the BFA-induced ER/Golgi fusion, this increase must not occur in PtK cells. Therefore PtK-2 cells were incubated with various concentrations of BFA, and their [3H]C8C,-SM synthesis was determined. As clearly shown in Fig. 4, no increase in sphingomyelin synthesis is detectable even at a BFA concentration of 200 pg/ml, indicating BFA resistance of both established PtK cell lines. It may be of interest to test if this resistance is restricted to PtK cells or if other marsupials are resistant against BFA as well.
Taken together, these data provide strong evidence that the increase in sphingomyelin synthesis induced by BFA reflects dissociation and fusion of the Golgi with the ER. The most straightforward explanation for the increased synthesis of sphingomyelin is the translocation of sphingomyelin synthase into a mixed ER-Golgi organelle, where a larger pool of substrate is available than in the Golgi alone. However, the amount of endogenous free ceramide in this mixed compartment might still be limiting because addition of an excess of exogenous membrane-permeable ceramide analogue further increased the levels of the sphingomyelin analogue. Under physiological conditions, the truncated sphingolipids are efficiently secreted from the cells. In the experiments with BFA presented here, the ratio of the cellular content of sphingomyelin to that of the corresponding media was increased by a factor of 2 (usually measured 30 min after the addition of BFA) (cf. Fig. 1). This cellular accumulation indicates that C8Cs-SM is retained. This is in line with previous observations that BFA inhibits the transport of proteins from the ER via the Golgi to the plasma membrane. On the other hand, transport of some membrane lipids has been reported to be uninfluenced by BFA (24,25).
The increase of [3H]CsC8-SM induced by BFA can easily and quickly be determined by TLC. This system provides a practical assay to screen cell lines or organisms for their sensitivity to BFA or to screen drugs for an antagonistic action. This will lead to a better understanding of the still unknown mechanism of the dramatic action of this antibiotic. In addition, the assay should help to characterize the cellular target of BFA, in semi-intact cells or (as a classical biochemical complementation assay) duringprotein purification. Presently we attempt to isolate and characterize the mixed ER-Golgi organelle using this assay.