Sphingomyelinase Treatment Induces ATP-independent Endocytosis

SMase (Bacillus cereus), sodium azide, 2-deoxyglucose, Lucifer yellow, and ATP bioluminescent assay kit were purchased (Sigma Chemical Co., St. Louis, MO). FITC-dextran (70-kD), tetramethylrhodamine isothiocyanate (TRITC)-dextran (70-kD), rhodamine-dextran (70-kD), and 4,4-difluoro-4-bora-3a,4a-diaza-s-indacene-c12-sphingomyelin (BODIPY-C₁₂-SM) were from Molecular Probes (Eugene, OR). SMase (Staphylococcus aureus) was from ICN Biomedicals Inc. (Irvine, CA). Streptolysin O (SL-O) was obtained from Murex Diagnostics (Atlanta, GA). Human transferrin (Tf; Sigma Chemical Co.) was iron-loaded, purified by Sephacryl S-300 (Pharmacia Diagnostics AB, Uppsala, Sweden), and labeled with fluorophore Cy3 (Biological Detection Systems, Pittsburgh, PA) according to the manufacturer's instructions. Rabbit polyclonal antibody to caveolin/ VIP-2 was purchased from Transduction Laboratories (Lexington, KY), and mouse monoclonal antibodies against adaptor protein 2 (AP6) and clathrin (X22) were gifts from F. Brodsky (University of California, San Francisco, CA).

J774 cells (American Type Culture Collection, Rockville, MD) were maintained in spinner culture in DME (Gibco Laboratories, Grand Island, NY) with 10% FCS. TRVb-1 cells (McGraw et al., 1987), a CHO cell line lacking endogenous transferrin receptor and expressing transfected human Tf receptors, were cultured in HF-12 (Gibco Laboratories) with 5% FCS. 2 d before an experiment, cells were plated on polylysine-coated coverslip-bottom dishes (Salzman and Maxfield, 1988). All experiments were performed on day 3 unless otherwise mentioned. ATP measurements were performed on both J774 macrophages and TRVb-1 cells using an ATP bioluminescent assay kit (Sigma Chemical Co.). Cells grown on 12-well culture dishes were pretreated with sodium azide/2-deoxyglucose for 5 min to deplete ATP and then incubated with or without SMase for an additional 10 min at 37°C before ATP assay. To measure ATP in permeabilized cells, cells were preincubated with sodium azide/2-deoxyglucose for 5 min and cooled to 0°C. Cells were then treated with SL-O (0.4 U/ml) for 5 min on ice, rinsed three times with cold PBS, and transferred to 37°C for 10 min. After a 10-min incubation with SMase, cells were washed and assayed for ATP.

In most experiments, cells were preincubated for 5 min in medium alone or containing 2-deoxyglucose and sodium azide. The cells were then incubated in the same medium containing 5 mg/ml FITC-dextran alone, or plus 50 mU/ml SMase for 10 min before being fixed with 4% paraformaldehyde for 5 min in PBS. Cells were immediately viewed using a microscope (model Diavert; Leitz, Wetzlar, Germany) with a 63×, NA 1.4 objective, and recorded with a cooled charge-coupled device (CCD) camera (KAF 1400; Photometrics, Tucson, AZ). To quantify the endocytosed FITC-dextran, fluorescence images of cells were taken using the same system as above but with a 25×, NA 0.8 objective. Total fluorescence intensity from a field of cells (∼70 cells) was measured and, after background correction, divided by the number of cells in the field. Each data point represents an average of four or five fields. For immunofluorescence experiments, cells were fixed with 4% paraformaldehyde in PBS for 10 min and then permeabilized with 0.1 mg/ml saponin before incubating with primary antibodies and a TRITC-conjugated secondary antibody. Confocal fluorescence images were obtained with a laser scanning confocal microscope (model MRC600; Bio-Rad Microscience, Cambridge, MA). FITC-dextran and Cy3-Tf images in Fig. 8 were obtained using an image-intensified video camera (model K51380; Videoscope, Washington, DC) on a fluorescence microscope (model DMIRB; Leica Inc., St. Gallen, Switzerland) using a 63×, NA 1.32 objective.

After incubating with HRP plus SMase for 10 min, ATP-depleted cells were fixed with 3.5% glutaraldehyde in 100 mM Hepes, pH 7.4, for 15 min at 22°C. After three rinses of 5 min each with TBS (50 mM Tris, pH 7.4, 100 mM NaCl), cells were incubated with 10 mg/ml diaminobenzidine in TBS with 0.3% H₂O₂ for 60 min at room temperature with agitation. Cells were rinsed with three washes of 100 mM sodium cacodylate, pH 6.8, and then postfixed with 1% osmium tetroxide and 1.25% potassium ferrocyanide in 100 mM sodium cacodylate for 4 h at room temperature. After two rinses with 100 mM sodium cacodylate, cells were dehydrated in a graded series of ethanol, embedded in Epon, and sectioned in the plane of the monolayer. Thin sections (80 nm) were counterstained with uranyl acetate and lead citrate before observation using a transmission electron microscope (model 1200EX II; JEOL USA, Peabody, MA) operating at 80 kV.

N-([6-{7-nitrobenz-2-oxa-1,3-diazol-4-yl} amino] hexanoyl) sphingosyl phosphocholine (C₆-NBD-gal) was prepared as described previously (Mayor et al., 1993). C₆-NBD-gal/lipid vesicles (100 μM total lipid) were prepared by injecting an ethanolic solution of a 1:1 mixture of C₆-NBD-gal and dioleylphosphatidylcholine (DOPC; 2:5 mM total lipid; Avanti Polar Lipids, Inc., Albaster, AL) into 150 mM NaCl, 20 mM Hepes, pH 7.4. TRVb-1 cells were incubated on ice with 3 μM of this vesicle solution (<0.2% ethanol) in Hepes-buffered, glucose-free Hams HF-12 medium (HF-12; Specialty Media, Inc., Lavallette, NJ) for 30 min. Cells were washed extensively with ice-cold HF-12, incubated with 5 mM 2-deoxyglucose and 5 mM sodium azide for 10 min at 37°C, and then incubated for an additional 10 min at 37°C in the same medium with or without 50 mU/ml SMase. For some samples, the lipid label remaining on the plasma membrane was backexchanged by washing (in six rinses of 5 min each) with ice-cold HF-12 containing 5% BSA. All samples were washed with cold PBS before lysis with 1% Triton X-100, and then C₆-NBD-gal was extracted from the samples by homogenizing the lysates with chloroform. The C₆-NBD-gal content in the chloroform extracts was determined with an SLM 8000C spectrofluorometer (Aminco, Urbana, IL).

Cells were prepared for microscopy as above except that cells were labeled with 0.75 μM C₆-NBD-gal. Cells were fixed with 3.3% paraformaldehyde after the backexchange step, and images were acquired with a cooled CCD camera (model 30427; Princeton Instruments, Princeton, NJ) on a fluorescence microscope (model DMIRB; Leica, Inc.). Because concentration-dependent self-quenching of C₆-NBD-gal could lead to underestimation of total fluorescence, NBD fluorescence emission was integrated over time until photobleaching was complete. This method allows the accurate quantification of total fluorescence (Jovin and Arndt-Jovin, 1989).

43 μg of BODIPY-C₁₂- SM (1 mg/ml in DMSO) was mixed with 20 μl of 5 mM DOPC in ethanol. The mixture was evaporated for 30 min under argon to remove ethanol and then centrifuged for 20 min at 4°C and 100,000 g. The supernatant was then rapidly injected into 1.5 ml PBS to yield a BODIPY-C₁₂-SM stock solution. To label the plasma membrane, TRVb-1 cells were chilled to 0°C for about 20 min and were then incubated with BODIPY-C₁₂-SM (1:5 dilution of stock solution with HF-12 medium) for 30 min at 0°C. Cells were washed with cold PBS three times to remove unbound BODIPY-C₁₂-SM. Some of the cells were further incubated with SL-O (0.4 U/ml) for 5 min at 0°C and then washed with cold PBS three times. Cells were then transferred to 37°C for 3 min, treated with SMase (50 mU/ml)/rhodamine-dextran for 5 min, washed with medium 1 and then immediately observed by fluorescence microscopy.

Fig. 1,a shows the uptake of a fluid endocytosis marker, FITC-dextran, by J774 cells during a 10-min incubation at 37°C. These control cells internalize FITC-dextran and deliver it to endosomes that are seen as bright fluorescent dots. As expected, this uptake is blocked when cellular ATP is depleted by preincubation with sodium azide and 2-deoxyglucose (Fig. 1,b). Under the conditions used for Fig. 1,b, cellular ATP is 2.4 ± 0.5% of control values, as determined by an ATP luminescence assay. When ATP-depleted cells are treated with SMase (50 mU/ml), there is substantial uptake of FITC-dextran into the cells (Fig. 1,c). SMase treatment had no effect on cellular ATP levels. To determine whether FITC-dextran is in sealed vesicles, these cells were either chased in dextran-free medium for long periods of time, or exposed to trypan blue, a membrane-impermeant quencher of FITC fluorescence (Hed et al., 1987). FITC-dextran labeling was not reduced after a 60-min chase, and the fluorescence was not quenched by 2 mg/ml trypan blue (data not shown). It has been shown previously that trypan blue is able to quench fluorescence even in deep surface invaginations (Myers et al., 1993), and the lack of quenching is strong evidence for the sealing of the SMase-induced vesicles. As seen in Fig. 1 c, these vesicles remain in the cell periphery, consistent with the requirement for ATP-dependent motor proteins such as dynein for translocation of vesicles along microtubules into the center of a cell (Holzbaur and Vallee, 1994). When ATP is restored to the cells, these vesicles move rapidly into the center (see below).

To visualize the SMase-induced vesicles by electron microscopy, ATP-depleted J774 cells were incubated for 10 min with HRP and 50 mU/ml SMase. Internalized HRP is visualized after diaminobenzidine treatment as a dark precipitate. As shown in Fig. 2, SMase treatment induces the formation of numerous vesicles that are typically 50–500 nm in diameter and marked by diaminobenzidine reaction product precipitate along the lumenal face of the vesicle membranes. There was not a significant difference in either the size distribution or number of vesicles when the incubation time was reduced from 10 to 2.5 min (data not shown). This suggests that the formation of vesicles is complete in a few min after exposure to SMase. The SMase- induced vesicles exhibit only peripheral staining, and no internal vesicles are seen. We did not observe a discernible coat on the vesicles.

The formation of vesicles in response to SMase is also observed in the TRVb-1 line derived from CHO fibroblasts. An advantage to using CHO cells is that they are less active in fluid uptake compared with macrophages, so the effect of SMase can be examined even in the absence of energy poisons. During a short incubation with FITC-dextran (2.5 min) little uptake is observed in control cells (Fig. 3,a). When SMase is added to these energy-replete cells, there is a burst of vesicle formation (Fig. 3,b), similar to the vesicles induced by SMase treatment of ATP-depleted cells (Fig. 3, c and d). Therefore, SMase-induced vesicle formation does not require ATP depletion, although we have shown that SMase-induced vesicle formation is an ATP-independent event. SMase-induced vesicles in energy-replete TRVb-1 cells (Fig. 3 b) were observed in the cell periphery because of the short incubation (2.5 min); vesicle formation because of SMase is completed within 5 min (earliest time point examined). Electron microscopy using HRP showed that the SMase-induced vesicles in TRVb-1 cells were structurally similar to those in J774 cells (data not shown).

To further verify that SMase-induced vesicle formation is indeed ATP independent, TRVb-1 cells were incubated with sodium azide/2-deoxyglucose and then permeabilized with SL-O before being exposed to SMase. Cellular ATP level in these cells after SMase treatment was down to 0.07 ± 0.01% of the level in intact, untreated cells. In response to SMase treatment, the SL-O–permeabilized cells took up a fluid-phase marker, Lucifer yellow in this experiment, into peripheral vesicles (Fig. 4,a). SL-O permeabilization of ATP-depleted cells without SMase did not induce vesicle formation (data not shown). SMase-treated cells were able to retain Lucifer yellow (mol wt 457) after a 2-h chase in a dye-free medium (Fig. 4 b), indicating these vesicles are completely sealed off from the plasma membrane.

To verify that the observed effects of SMase were because of SMase enzyme activity, we tested another SMase from a different bacterial source. SMase from Staphylococcus aureus had the same effect on TRVb-1 cells and J774 cells as the SMase from Bacillus cereus. Furthermore, when the SMase preparation was size-fractionated by fast protein liquid chromatography, the vesicle-forming activity copurified with the SMase activity (data not shown).

To determine the kinetics of SMase-induced vesicle formation, J774 macrophages were incubated with FITC-dextran with or without SMase for different lengths of time, and total dextran uptake per cell was measured by quantitative fluorescence microscopy. Results are shown in Fig. 5. As expected, in energy-replete cells treated with or without SMase (inset), the uptake of dextran increased approximately linearly with time. ATP depletion completely blocked dextran internalization in control cells (closed squares), but SMase treatment of ATP-depleted cells caused a rapid burst of dextran uptake (closed circles) that is complete within 10 min (earliest time point examined). Cells apparently became insensitive to SMase after the initial burst since longer exposure (>10 min) to SMase did not result in additional internalization. FITC-dextran uptake kinetics in TRVb-1 cells were similar to those of J774 cells (data not shown). The amount of FITC-dextran uptake induced by SMase in TRVb-1 cells during a 5-min incubation was equivalent to the fluid uptake by energy- replete cells over a 20–30-min incubation.

To determine the extent of plasma membrane internalization caused by SMase treatment, a fluorescent membrane-lipid analogue, C₆-NBD-gal, was used to label the plasma membrane of TRVb-1 cells. Cells were incubated on ice with C₆-NBD-gal, depleted of ATP, and then treated with or without SMase. As shown in Fig. 6, in the absence of SMase the plasma membrane was relatively uniformly labeled by C₆-NBD-gal (a), and a backexchange in BSA-containing medium extracted most of this surface label (b). SMase-treated cells (c) show more punctate labeling than untreated cells. After surface C₆-NBD-gal was removed by backexchanging in BSA-containing medium, the peripheral punctate labeling of the SMase-treated cells is retained (d). This retained C₆-NBD-gal label resembles the FITC-dextran uptake pattern in SMase-treated cells (Fig. 3 d). The fact that C₆-NBD-gal in SMase-induced vesicles is not available for backexchange provides further evidence that these vesicles are completely sealed off from the extracellular environment. After quantifying C₆-NBD-gal fluorescence images, we find that 30 ± 5% (average of five fields of cells ± SD) of the plasma membrane C₆-NBD-gal is internalized in response to SMase. In parallel experiments using a spectrofluorometer to detect C₆-NBD-gal extracted from solubilized cells, we find that 14 ± 7% of the C₆-NBD-gal label is resistant to backexchange after treatment with SMase. The difference between the results obtained by fluorometry and digital microscopy may be explained by a small population of cells that take up a large amount of C₆-NBD-gal throughout the cell even on ice (c). C₆-NBD-gal in these cells can be extracted by backexchange. The fluorescence intensity of these cells, therefore, contributes to the total fluorescence intensity in the spectrophotometric analysis but not the microscopic analysis, and this may account for the difference in the values obtained from these two methods. Nevertheless, the results obtained by both methods show that the amount of membrane internalized is much greater than the area covered by clathrin-coated pits and/or caveolae, which typically comprise only 2–4% of the plasma membrane. The total intensities of the membrane C₆-NBD-gal labeling of ATP-depleted cells with or without SMase treatment are not significantly different, indicating that SMase induces endo- but not exovesiculation.

SMase-induced vesicles are formed from part of the plasma membrane, raising the question as to whether they are from some specialized or invaginated plasma membrane regions. To address this, we examined if clathrin or caveolae were preferentially associated with these vesicles. TRVb-1 fibroblasts were treated with SMase in the presence of FITC-dextran, fixed, permeabilized, and stained with antibodies against clathrin or adaptor protein 2 (AP2). We did not find any significant enrichment of either clathrin or AP2 in FITC-dextran–containing vesicles by immunofluorescence microscopy (data not shown). When SMase-treated cells were colabeled with antibody to caveolin/ VIP-21, the major protein in caveolae, there was no significant colocalization between caveolin and FITC-dextran (data not shown). SMase-induced vesicles, therefore, appear to not be enriched in either clathrin-coated pits or caveolae.

We next investigated the fate of internalized receptors and the fluid contents of SMase-induced vesicles. J774 cells were preincubated with TRITC-dextran to label the late endosomes and lysosomes. The cells were then energy-poisoned and incubated with SMase and FITC-dextran. After 10 min, cells were rinsed and chased for an additional 60 min either in complete medium that allows restoration of ATP, or in the continued presence of energy poisons. In the continuous presence of energy poisons, SMase-induced vesicles (FITC-dextran–containing vesicles) remained in the cell periphery (Fig. 7,a), well separated from the TRITC-dextran–labeled late endosomes and lysosomes (Fig. 7,b). When ATP was restored, however, the FITC dextran in the SMase-induced vesicles moved into the cell and merged with TRITC-dextran–containing compartments (Fig. 7, c and d), indicating that these vesicles are able to deliver FITC-dextran to the previously formed late endosomes and lysosomes.

To test if the SMase-induced vesicles are also able to sort receptors for recycling, Cy3-transferrin was bound onto the surface of ATP-depleted J774 cells. The cells were then treated with SMase in the presence of FITC-dextran. Transferrin that remained on the cell surface was stripped off with a mild acid wash (Salzman and Maxfield, 1988). As shown in Fig. 8, Tf and dextran were initially internalized into the same compartments after SMase treatment (Fig. 8, a and b). When the cells were returned to complete medium in the absence of energy poisons, the Tf (Fig. 8,d) was rapidly sorted away from the FITC-dextran compartments (Fig. 8 c) and returned to the cell surface. Thus, the contents of the SMase-induced vesicles can be sorted with recycling receptors returned to the cell surface and the remaining contents being delivered to late endosomes.

To understand the mechanism of SMase-induced vesicle formation, it would be useful to know the lipid composition of these vesicles. As a start on these studies, we have used a fluorescent sphingomyelin to determine if ceramide formed from it enters the induced vesicles. We incorporated BODIPY-C₁₂-SM into the plasma membrane of TRVb-1 cells. When cells were exposed to SMase, a significant fraction of the BODIPY fluorescence translocated to the Golgi region (Fig. 9,a). This is consistent with the behavior of fluorescent acyl chain derivatives of ceramide that have been shown to be transported to the Golgi apparatus by a nonvesicular mechanism (Martin and Pagano, 1994). The sphingomyelin analogue itself does not translocate to the Golgi as demonstrated by the cells in Fig. 9,b, which were incubated at 37°C for the same time as the cells in Fig. 9,a, but without SMase exposure. It is evident that under our experimental conditions, SMase hydrolyzed the majority of BODIPY-C₁₂-SM and turned it into BODIPY-C₁₂-ceramide that could translocate to the Golgi. Since the transport of fluorescent ceramides is thought to occur via flipping to the cytoplasmic leaflet and binding to cytoplasmic carrier proteins, it should be blocked by removal of such proteins. This would allow us to observe the presence of the ceramide in SMase-induced vesicles. We permeabilized BODIPY-C₁₂-SM–labeled cells with SL-O to remove cytosol proteins. As shown in Fig. 9,d, SMase treatment in these cells caused BODIPY fluorescence to leave the plasma membrane and, significantly, to concentrate in SMase-induced vesicles that were labeled by rhodamine-dextran (Fig 9 c). Taken together, these studies indicate that a substantial amount of the BODIPY ceramide formed by SMase enters the induced vesicles, and this BODIPY-C₁₂-ceramide can flip to the cytoplasmic leaflet.

We have shown here that treatment of energy-depleted macrophages and CHO cells with exogenous SMase induces vesicle formation from the plasma membrane. To our knowledge, this is the first instance of an internalization process in living cells that does not require ATP. SMase-induced vesiculation was reported in human erythrocytes, although ATP dependence of the process was not addressed (Allan and Walklin, 1988).

The mechanism by which SMase causes vesicle formation in our system is unclear. One possibility is that the SMase causes an inward deformation of the lipid bilayer. It is known that plasma membrane lipids are asymmetrically distributed with sphingomyelin located mainly in the outer leaflet of the bilayer (Koval and Pagano, 1991). Removal of the head group from sphingomyelin would reduce crowding in the outer leaflet, and the ceramide produced by SMase could flip across the bilayer because of its hydrophobicity (Pagano, 1990). Both of these processes would promote inward curvature (Sheetz and Singer, 1974). Ceramide may form droplets in the hydrophobic core of the plasma membrane in human erythrocytes after SMase treatment (Verkleij et al., 1973), and it can promote lipid fusion in model systems (van Meer, 1993). Contact of ceramide-rich membranes at the neck of an invagination might promote formation of sealed vesicles. Furthermore, it is believed that sphingomyelin preferentially associates with cholesterol, and removal of sphingomyelin from the plasma membrane by SMase may result in the destabilization and subsequent transport of cholesterol to intracellular locations (Slotte and Bierman, 1988). However, it is possible that a substantial component of intracellular cholesterol transport after SMase treatment may also occur by the vesicular carriers described here (Skiba et al., 1996).

Ceramide can also function as a second messenger, triggering a cascade of intracellular events (Obeid and Hannun, 1995). However, it is unlikely that the SMase-induced vesicle formations in our system are related to the signaling functions of ceramide. Ceramide was shown to inhibit both fluid-phase and receptor-mediated endocytosis in a dose-dependent manner (Chen et al., 1995). Interestingly, in the same study SMase was also shown to inhibit endocytosis. This is most likely because of experimental procedures used in this study. The uptake of HRP as an endocytosis marker was measured after a 20-min preincubation with SMase. From our results, we know now that endocytosis should actually be increased within the first 20 min in the presence of SMase and could well be decreased after that because of depletion of sphingomyelin from the plasma membrane.

Endogenous sphingomyelin is believed to be synthesized in the Golgi and transported to the plasma membrane (Mandon et al., 1992). Ceramide produced by exogenous SMase is converted back to sphingomyelin, possibly in the recycling pathway (Kallen et al., 1994). Both processes require ATP. After treating ATP-depleted cells with SMase, sphingomyelin on the plasma membrane would eventually be depleted because of either hydrolysis or internalization. This may explain our observation that uptake of dextran is a time-limited process. Endocytosis reaches its maximum within 10 min after exposure to the enzyme, and further exposure did not generate additional vesiculation (Fig. 3, closed circles). This suggests that the formation of the vesicles after 10 min may be limited by the availability of sphingomyelin on the plasma membrane.

The characterization of endocytosis activated by SMase has several implications. First, our results demonstrate that cellular energy is not directly required for the formation of vesicles that invaginate from the plasma membrane. Since these vesicles apparently have incorporated nonspecific membrane, our results imply that much of the plasma membrane has the information required to recycle receptors and deliver fluid contents to late endosomes. This is consistent with observations that macropinosomes initially contain receptors that are removed from them (Racoosin and Swanson, 1993). The mechanisms for the sorting of the membrane and fluid contents of the SMase-induced vesicles are not known. Secondly, our results indicate that budding and fusion of mammalian cell membranes could occur as a result of altering the lipid composition of the membrane. While budding and fusion in living cells are known to be assisted by proteins such as clathrin and dynamin (Schmid, 1992; Hinshaw and Schmid, 1995), there is increasing evidence that these processes can also be governed by the properties of the lipids. Several diverse biological fusion processes, from sea urchin egg cortical exocytosis to mast cell degranulation, have been shown to be inhibited or enhanced by adding exogenous lipids into the membrane (Chernomordik et al., 1995). It was reported that Semliki Forest virus requires sphingomyelin in the target membrane to fuse (Nieva et al., 1994). It is possible that mammalian cells may use some lipid-modifying enzymes, such as SMase, to change lipid composition locally and, therefore, promote membrane budding and fusion.

The effect of SMase on cholesterol trafficking must now be reexamined. It is well known that SMase treatment can lead to stimulation of cholesterol esterification by transport of free cholesterol from the plasma membrane to intracellular ACAT locations (Slotte and Bierman, 1988). ACAT is thought to be primarily localized in the endoplasmic reticulum, although other intracellular locations have not been excluded. The mechanism by which cholesterol is transported from the plasma membrane to ACAT in response to SMase treatment is not known at present. It is known, however, that energy depletion does not impair this transport (Skiba et al., 1996). How the SMase-induced vesicles are related to the cholesterol translocation is not clear. It is possible that these vesicles are enriched with cholesterol relative to sphingomyelin. This relative enrichment could make cholesterol on these vesicles much more mobile than cholesterol in conventional endosomes, so that it can be easily transferred to cholesterol-poor membranes, such as the endoplasmic reticulum. Thus, the vesicles formed by SMase may act as efficient cholesterol transporters.

Although we used a nonphysiological addition of bacterial enzyme to hydrolyze plasma membrane sphingomyelin, our results may be relevant to several important biological processes. Mammalian cells express acidic and neutral SMases (Das et al., 1984; Merrill and Jones, 1990; Chatterjee, 1993; Hannun, 1994), and activation of mammalian SMases and ceramide production have been implicated in many signal transduction cascades (Hannun, 1994). These regulated SMases were found to be located in either the cytosol or light membrane fraction, possibly the plasma membrane (Liu and Anderson, 1995). Furthermore, a Zn-dependent SMase was recently found to be secreted by several cell lines (Schissel et al., 1996), and it has been found that this secreted Zn-dependent SMase is encoded by the same gene, mRNA, and protein precursor as acid lysosomal SMase (Schissel et al., 1996; Schissel, S., E. Schuchman, K. Williams, and I. Tabas, manuscript submitted for publication). With lipoproteins, treatments such as oxidation or preexposure to phospholipase A₂ cause the sphingomyelin to become a good substrate for secreted SMase at pH 7.4 (Callahan et al., 1983; Schissel et al., 1998). Therefore, even though secreted SMase does not cause extensive ceramide production when added to mammalian cells (our unpublished data), a change in sphingomyelin accessibility to the enzyme may occur under some conditions at the plasma membrane. In summary, there are various forms of SMase in the extracellular medium and in the secretory and endocytic compartments of many cells. Although the mechanisms of activation remain to be worked out, it is likely that one or more of these forms of SMase cause significant hydrolysis of sphingomyelin at the plasma membrane. Our data suggest that among other effects, activation of SMase would be expected to cause endovesiculation.

Recent data point to a specific example where this occurs (Grassme et al., 1997). Infection by Neisseria gonorrhoeae requires entry into mucosal epithelial cells (McGee et al., 1983). It was found that invasion of human epithelial cells and primary fibroblasts was accompanied by hydrolysis of phosphatidyl choline (PC) by a PC-specific phospholipase C, and by hydrolysis of sphingomyelin by SMase (Grassme et al., 1997). Approximately 35% of metabolically labeled sphingomyelin was hydrolyzed within 5 min of exposure to the bacteria, and the sphingomyelin hydrolysis did not occur in fibroblasts from Niemann-Pick Disease type A patients. These fibroblasts lack both the lysosomal and secreted forms of the (acid) SMase. The Niemann-Pick A fibroblasts did not internalize Neisseria gonorrhoeae in contrast to normal human fibroblasts. Transient transfection with the SMase gene restored the SMase activity after contact with bacteria, and it also restored the ability to internalize the bacteria in the transfected cells. Inhibition of PC-specific phospholipase C also reversibly blocked bacterially induced SMase activity and uptake of the bacteria. These data suggest that Neisseria gonorrhoeae can activate a signaling cascade through PC-specific phospholipase C that activates the target cell's SMase, resulting in very extensive and rapid conversion of sphingomyelin to ceramide. This SMase activity is required for internalization of the bacteria (Grassme et al., 1997).

Our results present an obvious mechanism for the SMase-dependent uptake of the Neisseria gonorrhoeae bacteria by the nonphagocytic cells. The massive SMase activity would drive endovesiculation that could engulf the bacteria. Further work will be required to determine if the activation of SMase is localized near the site of bacterial uptake or if there is a widespread activation.

The authors are grateful to T. McGraw (Cornell University Medical College) for critical reading of the manuscript.

This work was supported by National Institutes of Health (NIH) grants to F.R. Maxfield (DK27083) and to I. Tabas (HL39703 and HL21006). X. Zha was supported by a postdoctoral fellowship from the American Heart Association, New York City Affiliate, and P. Skiba was supported in part by an NIH postdoctoral training grant (5T32HL-07343). P.L. Leopold was supported by an Aaron Diamond Foundation postdoctoral fellowship.

ACAT

acetyl-CoA-cholesterol acyl transferase

BODIPY-C₁₂-SM

4,4-difluoro-4-bora-3a,4a-diaza-s-indacene- C12-sphingomyelin

C6-NBD-gal

N-([6-{7-nitrobenz-2-oxa-1,3-diazol-4-yl} amino] hexanoyl) sphingosyl phosphocholine

PC

phosphatidyl choline

SL-O

streptolysin O

SMase

sphingomyelinase

Tf

transferrin

TRITC

tetramethylrhodamine isothiocyanate