HepG2 cells are highly differentiated hepatoma cells that have retained an apical, bile canalicular (BC) plasma membrane polarity. We investigated the dynamics of two BC-associated sphingolipids, glucosylceramide (GlcCer) and sphingomyelin (SM). For this, the cells were labeled with fluorescent acyl chainlabeled 6-[N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl)- amino]hexanoic acid (C6-NBD) derivatives of either GlcCer (C6-NBD-GlcCer) or SM (C6-NBD-SM). The pool of the fluorescent lipid analogues present in the basolateral plasma membrane domain was subsequently depleted and the apically located C6-NBD-lipid was chased at 37°C. By using fluorescence microscopical analysis and a new assay that allows an accurate estimation of the fluorescent lipid pool in the apical membrane, qualitative and quantitative insight was obtained concerning kinetics, extent and (intra)cellular sites of the redistribution of apically located C6-NBD-GlcCer and C6-NBD-SM. It is demonstrated that both lipids display a preferential localization, C6-NBD-GlcCer in the apical and C6-NBD-SM in the basolateral area. Such a preference is expressed during transcytosis of both sphingolipids from the apical to the basolateral plasma membrane domain, a novel lipid trafficking route in HepG2 cells. Whereas the vast majority of the apically derived C6-NBD-SM was rapidly transcytosed to the basolateral surface, most of the apically internalized C6-NBD-GlcCer was efficiently redirected to the BC. The redirection of C6-NBD-GlcCer did not involve trafficking via the Golgi apparatus. Evidence is provided which suggests the involvement of vesicular compartments, located subjacent to the apical plasma membrane. Interestingly, the observed difference in preferential localization of C6-NBD-GlcCer and C6NBD-SM was perturbed by treatment of the cells with dibutyryl cAMP, a stable cAMP analogue. While the preferential apical localization of C6-NBD-GlcCer was amplified, dibutyryl cAMP-treatment caused apically retrieved C6-NBD-SM to be processed via a similar pathway as that of C6-NBD-GlcCer.

The data unambiguously demonstrate that segregation of GlcCer and SM occurs in the reverse transcytotic route, i.e., during apical to basolateral transport, which results in the preferential localization of GlcCer and SM in the apical and basolateral region of the cells, respectively. A role for non-Golgi–related, sub-apical vesicular compartments in the sorting of GlcCer and SM is proposed.

HepG2 cells are well-polarized hepatoma cells which are regarded as a model system for hepatocytes (Chiu et al., 1990; Sormunen et al., 1993; Zaal et al., 1994). In polarized cells, the plasma membrane (PM)1 is composed of two distinct domains, the basolateral and the apical membrane domain, of which the basolateral membranes form a continuous planar surface. HepG2 cells differ from other polarized cells because the apical membranes are arranged into closed compartments located between two adjacent cells, thus representing the bile canalicular space (BC). Such apical BC membrane domains are believed to originate from intracellular microvilli-lined vesicles that mature under specific conditions of growth. It has been proposed that these vesicles are vectorally transported to the cell surface where they fuse with a similar, cell surface located microvilli-lined vesicle of an adjacent cell (Chiu et al., 1990).

Each membrane domain has its specific protein and lipid composition. To maintain such a unique composition, sorting mechanisms that regulate trafficking of molecules to the appropriate membrane domain must be operational. From studies on intracellular membrane traffic of proteins, a number of factors involved in the trafficking, sorting, and targeting of such molecules now emerges (Pimplikar and Simons, 1993; Mostov and Cardone, 1995). By contrast, still little is known about lipid trafficking pathways and lipid sorting mechanisms.

Routes of lipid transport can be conveniently monitored by using fluorescently tagged lipids (Pagano and Sleight, 1985; Hoekstra and Kok, 1992). By this means we have previously shown that in HepG2 cells BC membranes are a target for lipids endocytosed from the basolateral surface (Zaal et al., 1993). Transcytosis of lipids has also been demonstrated in polarized Madin-Darby canine kidney (MDCK) cells (van Genderen and van Meer, 1995). However, whereas in these cells lipid transcytosis was shown to be bidirectional, in hepatic cells only basolateral to apical transcytosis has been demonstrated thus far. In addition to transcytosis, Golgi-derived de novo synthesized (glyco)- sphingolipids are transported to both the basolateral and apical domain in HepG2 (Zaal et al., 1993), MDCK (van Meer et al., 1987), and intestinal epithelial (Caco-2) cells (van 't Hof and van Meer, 1990). In the latter two cell lines it was demonstrated that newly synthesized glucosylceramide (GlcCer) and sphingomyelin (SM) were sorted for either the apical or basolateral PM domain. A sorting model was postulated in which the clustering of GlcCer in the lumenal leaflet of the trans-Golgi network forms so-called microdomains. Subsequently, these microdomains, relatively poor in SM, bud off, thus giving rise to vesicles that are destined for the apical domain (van Meer and Simons, 1988). Current evidence suggests that also glycosylphosphatidylinositol (GPI)-anchored proteins can be part of these glycosphingolipid-enriched microdomains, and, moreover, that the preferential interaction of the GPI-anchor with these microdomains would account for their apical enrichment (Lisanti and Rodriguez-Boulan, 1990). Although direct, i.e., nontranscytotic, targeting of some Golgi-derived GPI-anchored proteins (Ali and Evans, 1990) as well as ceramide metabolites (Zaal et al., 1993) to BC has also been demonstrated in hepatocytes, specific sorting of GlcCer, and its subsequent targeting to BC has not been demonstrated in these cells.

In the present study, evidence is presented for the existence of a transcytotic route via which fluorescent acyl chain- labeled 6-[N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl)amino] hexanoic acid (C6-NBD) derivatives of GlcCer (C6-NBD-GlcCer) and SM (C6-NBD-SM) are transported from the apical BC to the basolateral PM. Apical to basolateral trans-cytosis has, to our knowledge, never been demonstrated before in hepatic cells. It is shown that in this reverse transcytotic route C6-NBD-GlcCer and C6-NBD-SM are sorted and subsequently flow to distinct PM domains, each lipid thus displaying a specific preference for localization. The experimental data indicate the involvement of vesicular compartments located subjacent to the BC, regulating trafficking and sorting in the apical to basolateral transport pathway.

Materials And Methods

D-sphingosine, sphingosylphosphorylsphingosine, 1-β-d-glucosylsphingosine, bovine serum albumin (BSA, fraction V) and monensin were obtained from Sigma Chemical Co. (St. Louis, MO). 6-[N-(7-nitrobenz- 2-oxa-1,3-diazol-4-yl)amino]hexanoic acid (C6-NBD) was from Molecular Probes (Eugene, OR). Na2S2O4 (sodiumdithionite) was obtained from Merck (Darmstadt, Germany). Dibutyryl cyclic AMP (db-cAMP) was purchased from Boehringer Mannheim (Mannheim, Germany). DMEM was from GIBCO BRL (Life Technologies, Paisley, Scotland) and FCS was bought from BioWhittaker, Verviers, Belgium. All other chemicals were of analytical grade.

Cell Culture

HepG2 cells were grown in DMEM with 4.5 g of glucose per liter, supplemented with 10% FCS, and antibiotics. Media were changed every other day. For experiments, cells were plated in plastic culture flasks or, for microscopy, onto glass coverslips at low density (±20% of surface occupied). In all experiments cells were used 3 d after plating. At this time the cells had reached an optimal ratio of polarity vs density, i.e., a maximal ratio of BC/cell.

Synthesis of C6-NBD-labeled Sphingolipids

C6-NBD-glucosylceramide, C6-NBD-sphingomyelin, and C6-NBD-ceramide were synthesized from C6-NBD and 1-β-d-glucosylsphingosine, sphingosylphosphorylcholine and D-sphingosine, respectively, as described elsewhere (Kishimoto, 1975; Babia et al., 1994). The lipids were purified by thin layer chromatography (TLC, see next section) until a single spot was obtained, cochromatographing with appropriate reference lipids (Babia et al., 1994). They were stored at −20°C and routinely rechecked for purity. The C6-NBD-lipids were quantified spectrophotometrically in an SLM fluorometer at excitation and emission wavelengths of 465 nm and 530 nm, respectively.

Incubation of Fluorescent Lipids with Cells

C6-NBD-Cer.

Cells, grown on glass coverslips, were washed three times with a PBS solution. C6-NBD-Cer was dried under N2, redissolved in absolute ethanol and injected into Hank's balanced salt solution (HBSS, final concentration ethanol, 0.5% vol/vol) under vigorous vortexing. The cells were incubated with 4 μM C6-NBD-Cer under a CO2-containing, humidified atmosphere at 37°C for 30 min. Then, cells were washed three times with PBS to remove non-internalized probe, and further incubated at 37°C for 1 h. To allow microscopical examination of fluorescently labeled intracellular structures, the basolateral PM pool of fluorescent lipids was removed by incubating the cells in 5% (wt/vol) BSA-containing HBSS at 4°C for 2 × 30 min (back exchange). Subsequently, cells were washed three times with ice-cold HBSS and examined microscopically using an Olympus Provis AX70 microscope.

C6-NBD-GlcCer and C6-NBD-SM.

Cells, grown on glass coverslips, were washed three times with PBS and incubated with 4 μM C6-NBD-GlcCer or C6-NBD-SM at 37°C (lipid dispersion was prepared as described above). After 30 min, the cells were washed with ice-cold PBS and the basolateral PM pool of C6-NBD-lipid was removed by back exchange at 4°C for 2 × 30 min, using 5% BSA-containing HBSS. Subsequently, cells were washed with ice-cold HBSS and examined microscopically. Occasionally, the cells were, subsequent to the back exchange and a PBS wash, rewarmed to 37°C and further incubated (chased) in HBSS at 37°C for various time intervals. In some experiments, the cells were chased at 37°C in HBSS, supplemented with 5% BSA. Note that during a 90-min incubation at 37°C in HBSS in the presence of BSA, only 0.7% of the total BSA fraction was taken up by the cells, as determined by measuring cellular uptake of rhodamine B isothiocyanate (RITC)-labeled BSA (which was added as a tracer to the BSA-medium; 0.5% of the nonlabeled BSA). Hence, these data exclude a potential interference of the cellular lipid pool with a lipidanalogue fraction that might have re-entered the cells after back exchange. Finally, artifacts in lipid flow and sorting resulting from treatment of the cells with BSA (fraction V) per se, were similarly excluded. Thus, identical results with respect to the fate of the lipid analogues, as described in this study, were obtained when the back exchange was carried out with either 15% FCS or nonlabeled dioleoylphosphatidylcholine vesicles (SUV, 500 nmol/ml) in HBBS.

After all final incubations, cells were washed with ice-cold HBSS and examined microscopically. For confocal laser scanning microscopy a TCS Leica (Heidelberg, Germany) apparatus equipped with an argon/crypton laser coupled to a Leitz DM IRB-inverted microscope was used. Images were converted to tagged-information-file format and printed on a Fujix P3000 printer.

To determine whether metabolic conversion of C6-NBD-GlcCer or C6NBD-SM occurred during the incubations, cells, grown in culture flasks, were washed three times in PBS and incubated with 4 μM of either C6NBD-lipid at 37°C for 30 min. After a subsequent back exchange procedure, cells were rewarmed and further incubated at 37°C for 90 min. After gently scraping the cells with a rubber “policeman”, cells and incubation media were extracted according to the method of Bligh and Dyer (1959). The lipids were analyzed by thin layer chromatography, using CHCI3/ methanol/NH4OH/H2O (35:15:2:0.5, vol/vol/vol/vol) as running solvent. Lipid amounts were quantified by scraping the spots from the TLC plates, followed by vigorous shaking in 1% (vol/vol) TX-100 in H2O for 60 min. After removal of silica particles by centrifugation, NBD-fluorescence was measured as described above.

Quantification of Fluorescent BC Labeling with C6-NBD-SM or C6-NBD-GlcCer

Fluorescent BC labeling was quantified by the use of two different approaches: (1) Determination of the percentage NBD-positive BC and (2) calculation of BC-associated NBD-fluorescence. To determine the percentage of BC labeled with C6-NBD-lipid, BC were first identified by phase contrast illumination, and then classified as being NBD-positive or NBD-negative under epifluorescence illumination using a filter set for blue excitation (BP 470-490/LP515). BC were only classified as being NBD-positive when labeled microvilli could be observed (cf. Figs. 1 and 2 B). Quantitative determination of BC-associated C6-NBD-lipid was performed by using the capacity of sodiumdithionite to rapidly reduce NBDfluorescence in the lumen and/or lumenal leaflet of the BC (see Results). For this, cells were grown in plastic 25 cm2 culture flasks and labeled with the fluorescent lipid analogues as described above. To determine the amount of BC-associated lipid, the cells were first subjected to a BSA/ HBSS back exchange procedure, as described above, to remove the basolateral PM pool of C6-NBD-lipid. Note that BSA has no access to the BC domain (see Results). By contrast, we observed that the irreversible NBDquencher sodiumdithionite (McIntyre and Sleight, 1991) does acquire access to this membrane domain. Hence, after a BSA-back exchange, the cells were subsequently incubated with 30 mM sodiumdithionite in HBSS (diluted from a stock solution of 1 M dithionite in 1 M Tris-buffer, pH 10) at 4°C for 7 min. As a control, cells were incubated in HBSS in the absence of sodiumdithionite at 4°C for 7 min. After extensively washing (15 times with HBSS), cells were scraped, lipid was extracted, and fluorescence was measured as described above. The amount of C6-NBD-lipid (expressed as pmol/mg protein) that was associated with BC was calculated from the following equation: NBD-fluorescenceBC = NBD-fluorescencecontrol − NBDfluorescencedithionite-treated.

Miscellaneous Procedures

Monensin Treatment.

In the C6-NBD-Cer–labeling experiments, cells were pretreated with 10 μM monensin at 37°C for 30 min. Monensin was present during the entire experiment. In the experiments in which transport of apically derived C6-NBD-GlcCer or C6-NBD-SM was investigated, monensin was added to the back exchange medium and was kept present in all following steps.

Db-cAMP Treatment.

Cells were incubated with 4 μM C6-NBD-SM or C6-NBD-GlcCer in HBSS at 37°C for 30 min. After washing with ice-cold PBS, C6-NBD-lipid residing in the basolateral PM was back exchanged. During the back exchange procedure 100 μM db-cAMP was included to preincubate the cells. After the second back exchange, cells were washed, rewarmed, and further incubated at 37°C in the presence of 100 μM dbcAMP.

Results

Basolateral to Apical Transcytosis and Quantification of Apically Located C6-NBD-GlcCer and C6-NBD-SM

To determine the intracellular fate of BC-associated C6NBD-GlcCer and C6-NBD-SM, HepG2 cells were first incubated with 4 μM C6-NBD-GlcCer or C6-NBD-SM at 37°C for 30 min. After this time interval, both lipids were found intracellularly in vesicular structures (punctate fluorescence) and prominently labeled the BC and the basolateral PM (Fig. 1, a–d; arrowheads). The confocal micrographs, shown in Fig. 2, are entirely consistent with these observations and provide further support for the dominant localization of the lipid analogues in BC and basolateral membranes. Note the typical “microvilli-like” appearance of fluorescence located in a BC formed between two adjacent cells (Fig. 2 b) when scanning through the plane of focus.

After the initial incubation at 37°C, C6-NBD-GlcCer or C6-NBD-SM residing in the outer leaflet of the basolateral PM was back exchanged by treatment of the cells with 5% BSA-containing HBSS at 4°C for 2 × 30 min. This procedure (see Materials and Methods section “C6-NBD-GlcCer and C6-NBD-SM”) removed all NBD-fluorescence (>96%) from the basolateral PM, but not from the apical membrane (cf. Zaal et al., 1993). Using either lipid analogue, 70–75% of the BC, initially identified by phase contrast (see Materials and Methods), was scored as NBD-positive.

Because of the inaccessibility of C6-NBD-lipids residing in the lumen and/or lumenal leaflet of BC to the lipid scavenger BSA or acceptor vesicles, the quantification of BCassociated NBD-labeled lipid was considered as virtually impossible in this system. Recently, however, sodiumdithionite, which very slowly permeates across membranes, has been introduced as a useful agent to specifically and irreversibly quench fluorescent NBD-lipid analogues (McIntyre and Sleight, 1991; Pomorski et al., 1995). Here, we tested the potency of sodiumdithionite to quench C6-NBD-GlcCer and C6-NBD-SM that was located in the lumen and/or lumenal leaflet of the BC. For this, cells were incubated with C6-NBD-GlcCer or C6-NBD-SM to label the BC, as described above. After a back exchange, the cells were incubated in HBSS supplemented with 30 mM sodiumdithionite at 4°C for 7 min. Subsequently, the cells were extensively (15 times) washed with ice-cold HBSS to remove the sodiumdithionite. Fluorescence microscopical analysis revealed a complete elimination of BC-associated fluorescence (Fig. 1, e–h; arrowheads). Presumably due to its slow membrane-permeating properties, sodiumdithionite did not significantly interfere with intracellularly located fluorescence, as reflected by an unchanged appearance of intracellular punctate fluorescence. However, these results do indicate that sodiumdithionite is apparently able to pass the tight junctions in HepG2 cells and, consequently, reaches the BC in a paracellular manner. Hence, by comparing NBD-fluorescence in cells before and after treatment with sodiumdithionite, the amount of BC-associated C6-NBDGlcCer and C6-NBD-SM could be calculated (see Materials and Methods). We thus determined that in cells, labeled with C6-NBD-GlcCer or C6-NBD-SM as described above, 40.2 ± 5.0% of the total pool of cell-associated fluorescence was associated with BC (see below). This implies that sorting of either lipid during the early (30 min) transcytotic events does not take place, i.e., both lipids seem to follow the bulk flow in the transcytotic pathway.

After reaching the BC membrane, the question arises whether the observed lipid distribution reflects a steady state distribution or that a redistribution may occur, governed by a preferential residence of either lipid in the apical area. The next experiments were undertaken to address this issue.

Apically Derived C6-NBD-GlcCer and C6-NBD-SM Are Preferentially Located at Different Plasma Membrane Domains

After labeling the cells (30 min, 37°C) with either C6NBD-GlcCer or C6-NBD-SM and a subsequent depletion of the C6-NBD-lipid from the basolateral PM, the cells were rewarmed to 37°C and incubated in HBSS for 30, 60, or 90 min. Over a 90-min incubation period, the percentage of BC labeled with C6-NBD-GlcCer, as determined by fluorescence microscopy, remained nearly constant (Fig. 3, hatched bars). In cells labeled with C6-NBD-SM (Fig. 3, cross-hatched bars), the percentage of fluorescently labeled BC progressively decreased from 71.5 ± 8.8% at time zero to 44.0 ± 2.4% after 60 min, after which the number of fluorescently labeled BC remained constant. However, although the percentage of C6-NBD-SM–labeled BC, as detected by microscopical examination, did not further decline, the BC-associated fluorescence intensity did. As shown in Fig. 4, after a 60-min incubation period, the amount of BC-associated C6-NBD-SM was hardly detectable anymore, as revealed by using the sodiumdithionite quenching assay. By contrast, and consistent with the microscopic observations, the amount of BC-associated C6NBD-GlcCer slightly decreased from 40 to ∼30% of the total pool of cell-associated fluorescence (Fig. 4, filled symbols). Indeed, fluorescence microscopical analysis revealed that C6-NBD-GlcCer remained mainly associated with BC and was also found in clusters of vesicles surrounding the BC (Fig. 5,a). Only a relatively faint labeling of the basolateral PM could be observed. After 60 min, C6NBD-SM was found to label mostly the basolateral PM while labeling of BC was hardly distinguishable (Fig. 5,b, compare to Fig. 1 d). Apparently, apically located C6NBD-GlcCer and C6-NBD-SM were segregated for either the apical or basolateral domains in HepG2 cells, respectively. Moreover, this segregation step appears to occur subsequent to basolateral to apical transport of the lipid analogues, i.e., after insertion of the lipid analogues in the apical membrane.

Segregation of C6-NBD-GlcCer and C6-NBD-SM Occurs during Apical to Basolateral Transcytosis

To clarify the extent to and the mechanisms by which C6NBD-GlcCer and C6-NBD-SM were transported from the BC to the basolateral surface, the following experiment was carried out. After BC labeling with C6-NBD-GlcCer or C6-NBD-SM as described above, the C6-NBD-lipid was chased at 37°C in HBSS, supplemented with 5% BSA to prevent re-internalization of the C6-NBD-lipid once the analogue reached the basolateral membrane (see Materials and Methods). In a parallel experiment, the lipids were chased at 4°C. At 37°C, but not at 4°C (not shown), the percentage of BC labeled with C6-NBD-GlcCer gradually decreased from 69.6 ± 1.0% at time zero to 68.3 ± 2.0%, 45.0 ± 4.9%, and 32.6 ± 3.0% after 30, 60, and 90 min, respectively (Fig. 6). At these conditions, i.e., in the presence of BSA, the absolute amount of BC-associated C6-NBDGlcCer was stable for 30 min after which it could no longer be detected as probed by the sodiumdithionite quenching assay (not shown). However, in cells labeled with C6NBD-SM, the percentage of NBD-positive BC decreased much faster, i.e., from 67.5 ± 6.8% at time zero to 42.0 ± 3% and 31.5 ± 4.1% after 30 and 60 min, respectively (Fig. 6). Prolonged incubation for up to 90 min did not result in a further decrease of the percentage of C6-NBD-SM–positive BC, although BC-associated fluorescence intensities decreased in time. In fact, for both lipid analogues a nearly complete loss of the overall intracellular fluorescence was observed after 90 min (not shown). The loss of intracellular fluorescence could be entirely accounted for by the fluorescence recovered in the BSA-containing incubation medium, which is consistent with the notion that re-entry of BSA-complexed lipid is negligible (see Materials and Methods). Since these results were not observed at 4°C (data not shown), paracellular leakage of C6-NBD-SM or C6-NBD-GlcCer across the tight junctions could be excluded, and the data thus support a vesicular nature of lipid transport in the reverse transcytotic pathway.

During the time span of our experiments, only 10 and 7% of the total cell-associated pools of C6-NBD-SM and C6-NBD-GlcCer, respectively, were degraded after 2 h, with C6-NBD-Cer as the main (i.e., >90%) metabolic product, as determined by lipid extraction and HPTLC analysis (see Materials and Methods). This implies that the observed redistribution of BC-associated C6-NBD-SM or C6-NBD-GlcCer was not due to conversion of the lipids. Based on these results we conclude that both C6-NBD-GlcCer and C6NBD-SM can be retrieved from the apical domain after which they can be transcytosed to the basolateral surface. Note that the entire C6-NBD-lipid content of at least 60– 70% of the BC is involved in the reverse transcytotic pathway, as these fractions for both lipids can be efficiently removed in the presence of BSA (Fig. 3 vs Fig. 6).

Intracellular Sites Involved in the Segregation of C6-NBD-GlcCer and C6-NBD-SM in the Apical to Basolateral Transcytotic Route

Studies on apical endocytosis in various polarized cells have demonstrated that (a) apically internalized membrane proteins can recycle from a tubulovesicular compartment, located subjacent to the apical PM (Hunziker et al., 1990; Apodaca et al., 1994; Barroso and Sztul, 1994) and (b) transport between sub-apical compartments and the apical PM can be affected by modulators of distinct signal transduction pathways (Barroso and Sztul, 1994; Cardone et al., 1994; Hansen and Casanova, 1994). cAMP and stimulators of protein kinase A (PKA) which are both members of the adenylate cyclase-cAMP-PKA signal transducing pathway, have been shown to positively modulate apically directed transport (Brignoni et al., 1995; Pimplikar and Simons, 1994). To investigate a possible role for cAMP in the segregation of C6-NBD-GlcCer and C6-NBD-SM in the reverse transcytotic pathway, cells were labeled with either lipid after which the basolateral PM pool of C6-NBD-lipid was depleted by a back exchange as described above. During the back exchange procedure, which is carried out at 4°C, cells were preincubated with 100 μM db-cAMP. After the back exchange, the cells were rewarmed and further incubated in HBSS in the presence of 100 μM db-cAMP. As shown in Fig. 7, the percentage of NBD-positive BC in cells labeled with C6-NBD-GlcCer remained constant for 90 min, as was the case in control cells (cf. Fig. 3). However, whereas the percentage of NBD-positive BC in cells labeled with C6-NBD-SM rapidly decreased in control cells (cf. Fig. 3), this percentage of labeling remained virtually the same, for up to 90 min, in db-cAMP–treated cells (Fig. 7). Although (fractions of) both lipids could still be transcytosed to the basolateral surface, as reflected by their back exchangeability after chasing BC-associated C6-NBD-lipid in the presence of BSA, basolateral trafficking of both lipids was severely slowed down with regard to rate and extent (Fig. 8) in comparison to the basolateral flow in nontreated cells (cf. Fig. 6). Microscopic examination revealed that in cells, not treated with dbcAMP, fluorescently labeled vesicles surrounding the BC were most prominently seen in cells labeled with C6-NBD-GlcCer (Fig. 5). Interestingly, although essentially no BC-surrounding C6-NBDSM–labeled vesicles could be observed in nontreated cells (Fig. 5,b), such sub-apical NBD-fluorescent vesicles were readily distinguishable in db-cAMP–treated cells (Fig. 9). Hence, db-cAMP favored and/or promoted apically directed trafficking of both sphingolipids.

Sorting of C6-NBD-GlcCer in a Sub-apical Compartment Does Not Involve the Golgi Complex

In hepatocytes, the Golgi apparatus is located near the biliary pole. The organelle is well known as the main location for the assembly of biosynthetic products and secretory compounds (Farquhar, 1985). It has also been shown that endocytosed lipids can be transported to the Golgi apparatus (Schwarzmann and Sandhoff, 1990; Kok et al., 1991; Trinchera et al., 1991). Furthermore, after synthesis from their common precursor, C6-NBD-Cer, C6-NBD-SM, and C6-NBD-GlcCer can be transported from the Golgi, where sorting is thought to occur, to both PM domains via direct, nontranscytotic routes in various polarized cell lines (van Meer et al., 1987; van 't Hof and van Meer, 1990; Zaal et al., 1993). To elucidate whether the sub-apical vesicular structures observed in our experiments might be Golgi-related and, thus, could be involved in governing the preferential localization of C6-NBD-GlcCer in the apical area, we performed the following experiments. HepG2 cells were labeled with 4 μM C6-NBD-Cer at 4°C for 30 min, washed, and incubated at 37°C for 60 min in the presence of BSA. This resulted in an intense fluorescent labeling of the Golgi apparatus and the metabolic conversion of C6-NBDCer into C6-NBD-GlcCer and C6-NBD-SM, as described previously (Zaal et al., 1993). C6-NBD-GlcCer and C6NBD-SM, produced by the de novo synthesis, labeled approximately 66 ± 4% of the BC. However, when the cells had been pretreated with monensin (10 μM; Table I), only 17 ± 3% of the BC was faintly labeled. Monensin is known to block vesicular transport from the Golgi apparatus to the PM and was shown to bring about a decrease in biliary lipid secretion (Casu and Camogliano, 1990). Since monensin treatment did not affect C6-NBD-Cer metabolism (data not shown), our data show that monensin treatment of HepG2 cells, in agreement with results reported previously (Lipsky and Pagano, 1985; Kok and Hoekstra, 1994), inhibits (vesicular) lipid transport exiting from the Golgi. In the next experiment, cells were incubated with 4 μM C6-NBD-GlcCer at 37°C for 30 min to label BC as described above. During the back exchange procedure 10 μM monensin was added to prevent transport of C6-NBDGlcCer from the Golgi to the BC in the subsequent steps. After this treatment, the cells were rewarmed to 37°C and incubated in the presence of 10 μM monensin and 5% BSA for 30 and 60 min. Compared to control cells, no difference in the percentage of labeled BC or BC-associated C6-NBD-GlcCer was observed after the various incubation times (cf. Fig. 3). Similar results were obtained with cells that had been labeled with C6-NBD-SM (data not shown). Hence, these results demonstrate that neither sorting of apically derived C6-NBD-GlcCer nor apical to basolateral transport of C6-NBD-SM involves the Golgi complex.

Discussion

In this study, we have presented evidence for the existence of a novel, vesicular lipid trafficking route by which (glyco)- sphingolipids are transcytosed from the apical, bile canalicular domain to the basolateral surface in HepG2 cells. Most importantly, in this apical to basolateral transcytotic route GlcCer and SM were segregated and preferentially targeted to the apical and basolateral PM domain, respectively. By contrast, sorting and a preferential transcytotic basolateral to apical transport of either GlcCer or SM was not observed after initial basolateral insertion. At those conditions, the lipids appear to follow bulk flow, similarly as observed for the flow of these sphingolipids in polarized MDCK cells (van Genderen and van Meer, 1995). The experimental data indicated that multiple sites were involved in governing the preferential apical localization of C6NBD-GlcCer, including the basolateral PM from which a rapid reflux of the lipid analogue could occur, which became apparent when monitoring its fate, following delivery to the apical membrane. The percentage of labeled BC, as well as the amount of BC-associated C6-NBD-GlcCer, was more or less stable during a chase in HBSS (Fig. 3). Note that in the presence of BSA (Fig. 6) the percentage of C6-NBD-GlcCer–labeled BC was, similarly, fairly constant after an initial chase of ∼30 min. Subsequently, a rapid decline in the apical pool of C6-NBD-GlcCer became apparent in the presence of the scavenger, but not in its absence. By contrast, irrespective of the presence of BSA, the patterns obtained for the percentages of C6-NBD-SM–labeled BC were similar (Fig. 3 vs Fig. 6). Taken together, these data indicate that C6-NBD-SM is rapidly transcytosed from the apical to the basolateral PM, whereas reverse transcytosis of C6-NBD-GlcCer is more gradual. Thus, the latter lipid displays a tendency to reside in and near the apical domain, whereas the fate of C6-NBD-SM reflects a basolateral preference. The data also indicate that after a prolonged chase time (>30 min), a substantial fraction of the initial pool of apical C6-NBD-GlcCer, may reach the basolateral membrane (Fig. 6). Evidently, as revealed by monitoring the chase in the absence of a scavenger, this pool apparently displays a rapid reflux to the apical membrane (Fig. 3).

In case of C6-NBD-SM, the equilibrium that was established after a 60-min chase in HBSS did show a higher percentage of NBD-positive BC compared to the percentage seen after a chase in BSA-containing HBSS. This result indicates that also a reflux of basolateral arrived C6-NBDSM to BC occurred. The reflux of C6-NBD-GlcCer, however, was much more extensive than that of C6-NBD-SM. This suggests that the basolateral PM contributes to the distinct preferential localization of both lipids, consistent with the basolateral PM being a major site for sorting of apical membrane proteins in hepatocytes (Bartles et al., 1987; Schell et al., 1992). Yet, the delay in basolateral delivery of apical C6-NBD-GlcCer in comparison to that of C6-NBD-SM, as noted above, indicates that segregation of apically derived C6-NBD-GlcCer and C6-NBD-SM had occurred before these lipids reached the basolateral PM. Hence, the data suggest that multiple sites are involved in the distinct targeting of apically derived C6-NBD-GlcCer and C6-NBD-SM to their preferential domains.

Apically derived C6-NBD-GlcCer appeared to be efficiently redirected to BC via non-Golgi–related sub-apically located vesicular compartments. Indeed, the inertness of sphingolipid trafficking of apically derived C6-NBD-GlcCer and C6-NBD-SM towards monensin treatment strongly supports the view that the Golgi apparatus was not involved in the intracellular processing of the apical pool of lipids. Furthermore, a “typical” Golgi fluorescence pattern, as reported previously for HT29 cells (Kok et al., 1991), was never observed. Although further work is needed, current evidence strongly suggests that the nature of the sub-apical compartment involved very likely relates to the compartments, described as part of the basolateral to apical transcytotic pathway (Hunziker et al., 1990; Apodaca et al., 1994; Barroso and Sztul, 1994; Hansen and Casanova, 1994). In fact, similar sub-apical tubulovesicular compartments, involved in transcytosis of IgA, have also been described in hepatocytes (Geuze et al., 1984; Hoppe et al., 1985). Moreover, also in hepatocytes, elevated concentrations of cAMP have been found to stimulate translocation of vesicles to the canalicular membrane (Boyer and Soroka, 1995). Indeed, dbcAMP promoted apically directed trafficking of both sphingolipids.

Given an apical enrichment, most C6-NBD-GlcCer presumably underwent multiple rounds of recycling between the apical membranes and these sub-apical compartments. By contrast, C6-NBD-SM was rapidly transcytosed from the BC to the basolateral surface and labeling of sub-apical compartments was never observed when C6-NBD-SM was chased from BC (Fig. 5). Thus, transcytosis of C6-NBD-SM might involve a pathway, entirely different from that of C6-NBD-GlcCer, i.e., the analogue may bypass sub-apical compartments. However, treatment of the cells with dbcAMP did reveal a sub-apical accumulation of C6-NBDSM, very reminiscent of that seen for C6-NBD-GlcCer, in which C6-NBD-SM accumulates before a redirection of trafficking to the apical membrane (Fig. 8). Hence, these observations would favor a scenario in which sub-apical compartments participate in the reverse transcytotic pathway of both lipid analogues. In such a compartment, C6NBD-GlcCer may be sorted in GlcCer-enriched microdomains which eventually bud off to be rerouted to the BC, whereas C6-NBD-SM is rapidly transcytosed to the basolateral surface. Conditions where intracellular levels of cAMP are elevated (e.g., during biogenesis of apical [bile canalicular] membranes) may then provide a signal for the targeting to and subsequent (apical) recycling of C6NBD-SM–labeled vesicles via sub-apical compartments. However, when the transcytotic flow of C6-NBD-SM would entirely bypass that of C6-NBD-GlcCer, the conclusion is still justified that sorting between both sphingolipids must occur, in this case at the level of the apical membrane. Hence either scenario, as depicted in this paragraph, clearly reveals the occurrence of sphingolipid sorting in the transcytotic pathway, a phenomenon which has not been shown before. The novel transport route as well as conditions that affect the flow through and sorting in this transcytotic pathway, as described in the present work, should thus pave the way to obtain clues as to the mechanisms of sphingolipid trafficking and sorting in polarized cells. Such work is currently in progress.

References

References
Ali
N
,
Evans
WH
Priority targeting of glycosyl-phosphatidylinositol-anchored proteins to the bile-canalicular (apical) plasma membrane of hepatocytes. Involvement of ‘late' endosomes
Biochem J
1990
271
193
199
[PubMed]
Apodaca
G
,
Katz
LA
,
Mostov
KE
Receptor-mediated transcytosis of IgA in MDCK cells is via apical recycling endosomes
J Cell Biol
1994
125
67
86
[PubMed]
Babia, T., J.W. Kok, and D. Hoekstra. 1994. The use of fluorescent lipid analogues to study endocytosis of glycosphingolipids. In Receptor Research Methods. B. Greenstein, editor. Harwood Academic Publishing/London, UK. pp. 155–174.
Barroso
M
,
Sztul
ES
Basolateral to apical transcytosis in polarized cells is indirect and involves BFA and trimeric G protein sensitive passage through the apical endosome
J Cell Biol
1994
124
83
100
[PubMed]
Bartles
JR
,
Feracci
HM
,
Stieger
B
,
Hubbard
AL
Biogenesis of the rat hepatocyte plasma membrane in vivo: comparison of the pathways taken by apical and basolateral proteins using subcellular fractionation
J Cell Biol
1987
105
1241
1251
[PubMed]
Bligh
EG
,
Dyer
WJ
A rapid method of total lipid extraction and purification
Can J Biochem Biophys
1959
37
911
1017
Boyer
JL
,
Soroka
CJ
Vesicle targeting to the apical domain regulates bile excretory function in isolated rat hepatocyte couplets
Gastroenterology
1995
109
1600
1611
[PubMed]
Brignoni
M
,
Pignataro
OP
,
Rodriguez
ML
,
Vega-Salas
DE
,
RodriguezBoulan
E
,
Salas
PJI
Cyclic AMP modulates the rate of ‘constitutive' exocytosis of apical membrane proteins in Madin-Darby canine kidney cells
J Cell Sci
1995
108
1931
1943
[PubMed]
Cardone
MH
,
Smith
BL
,
Song
W
,
Mochly-Rosen
D
,
Mostov
KE
Phorbol myristate acetate-mediated stimulation of transcytosis and apical recycling in MDCK cells
J Cell Biol
1994
124
717
727
[PubMed]
Casu
A
,
Camogliano
L
Glycerophospholipids and cholesterol composition of bile in bile-fistula rats treated with monensin
Biochim Biophys Acta
1990
1043
113
115
[PubMed]
Chiu
J
,
Hu
C
,
Lui
W
,
Lo
S
,
Chang
C
The formation of bile canaliculi in hepatoma cell lines
Hepatology
1990
11
834
842
[PubMed]
Farquhar
MG
Progress in unraveling pathways of Golgi traffic
Annu Rev Cell Biol
1985
1
447
488
[PubMed]
Geuze
HJ
,
Slot
JW
,
Strous
GJ
,
Peppard
J
,
von Figura
K
,
Haslik
A
,
Schwartz
AL
Intracellular receptor sorting during endocytosis: comparative immunoelectron microscopy of multiple receptors in rat liver
Cell
1984
37
195
204
[PubMed]
Hansen
SH
,
Casanova
JE
Gsα stimulates transcytosis and apical secretion in MDCK cells through cAMP and protein kinase A
J Cell Biol
1994
126
677
687
[PubMed]
Hoekstra
D
,
Kok
JW
Trafficking of glycosphingolipids in eukaryotic cells; sorting and recycling of lipids
Biochim Biophys Acta
1992
1113
277
294
[PubMed]
Hoppe
CA
,
Connolly
TP
,
Hubbard
AL
Transcellular transport of polymeric IgA in the rat hepatocyte: biochemical and morphological characterization of the transport pathway
J Cell Biol
1985
101
2113
2123
[PubMed]
Hunziker
W
,
Male
P
,
Mellman
I
Differential microtubule requirements for transcytosis in MDCK cells
EMBO (Eur Mol Biol Organ) J
1990
9
3515
3525
[PubMed]
Kishimoto
Y
A facile synthesis of ceramides
Chem Phys Lip
1975
15
33
36
[PubMed]
Kok
JW
,
Hoekstra
D
Glycosphingolipid trafficking in the endocytic pathway
Curr Top Membr
1994
40
503
537
Kok
JW
,
Babia
T
,
Hoekstra
D
Sorting of sphingolipids in the endocytic pathway of HT29 cells
J Cell Biol
1991
114
231
239
[PubMed]
Lipsky
NG
,
Pagano
RE
Intracellular translocation of fluorescent sphingolipids in cultured fibroblasts: endogenously synthesized sphingomyelin and glucocerebroside analogues pass through the Golgi apparatus en route to the plasma membrane
J Cell Biol
1985
100
27
34
[PubMed]
Lisanti
MP
,
Rodriguez-Boulan
E
Glycosphingolipid membrane anchoring provides clues to the mechanism of protein sorting in polarized epithelial cells
Trends Biochem Sci
1990
15
113
118
[PubMed]
McIntyre
JC
,
Sleight
RG
Fluorescence assay for phospholipid membrane assymetry
Biochemistry
1991
30
11819
11827
[PubMed]
Mostov
KE
,
Cardone
MH
Regulation of protein traffic in polarized epithelial cells
Bioessays
1995
17
129
138
[PubMed]
Pagano
RE
,
Sleight
RG
Defining lipid transport pathways in animal cells
Science (Wash DC)
1985
229
1051
1057
[PubMed]
Pimplikar
SW
,
Simons
K
Role of heterotrimeric G proteins in polarized membrane transport
J Cell Sci Suppl
1993
17P
27
32
[PubMed]
Pimplikar
SW
,
Simons
K
Activators of protein kinase A simulate apical but not basolateral transport in epithelial Madin-Darby canine kidney cells
J Biol Chem
1994
269
19054
19059
[PubMed]
Pomorski
T
,
Herrmann
A
,
Zimmermann
B
,
Zachowski
A
,
Mueller
P
An improved assay for measuring the transverse redistribution of fluorescent phospholipids in plasma membranes
Chem Phys Lip
1995
77
139
146
[PubMed]
Schell
MJ
,
Maurice
M
,
Stieger
B
,
Hubbard
AL
5′nucleotidase is sorted to the apical domain of hepatocytes via an indirect route
J Cell Biol
1992
119
1173
1182
[PubMed]
Schwarzmann
G
,
Sandhoff
K
Metabolism and intracellular transport of glycosphingolipids
Biochemistry
1990
29
10865
10871
[PubMed]
Sormunen
R
,
Eskelinen
S
,
Lehto
V
Bile canaliculus formation in cultured HepG2 cells
Lab Invest
1993
68
652
662
[PubMed]
Trinchera
M
,
Carrettoni
D
,
Ghidoni
R
A part of glucosylceramide formed from exogenous lactosylceramide is not degraded to ceramide but recycled and glycosylated in the Golgi apparatus
J Biol Chem
1991
266
9093
9099
[PubMed]
van Genderen
I
,
van Meer
G
Differential targeting of glucosylceramide and galactosylceramide analogues after synthesis but not during transcytosis in Madin-Darby canine kidney cells
J Cell Biol
1995
131
645
654
[PubMed]
Van Meer
G
,
Simons
K
Lipid polarity and sorting in epithelial cells
J Cell Biochem
1988
36
51
58
[PubMed]
Van Meer
G
,
Stelzer
EHK
,
Wijnaendts-van-Resandt
RW
,
Simons
K
Sorting of sphingolipids in epithelial Madin-Darby canine kidney cells
J Cell Biol
1987
105
1623
1635
[PubMed]
Van 't Hof
W
,
Van Meer
G
Generation of lipid polarity in intestinal epithelial (Caco-2) cells: sphingolipid synthesis in the Golgi complex and sorting before vesicular traffic to the plasma membrane
J Cell Biol
1990
111
977
986
[PubMed]
Zaal
KJM
,
Kok
JW
,
Sormunen
R
,
Eskelinen
S
,
Hoekstra
D
Intracellular sites involved in the biogenesis of bile canaliculi in hepatic cells
Eur J Cell Biol
1993
63
10
19
Zaal
KJM
,
Kok
JW
,
Kuipers
F
,
Hoekstra
D
Lipid trafficking in hepatocytes: relevance to biliary lipid secretion
Adv Mol Cell Biol
1994
8
133
150

1.  Abbreviations used in this paper: BC, bile canalicular; GlcCer, glucosylceramide; GPI, glycosylphosphatidylinositol; PM, plasma membrane; SM, sphingomyelin.

Please address all correspondence to D. Hoekstra, Department of Physiological Chemistry, University of Groningen, A. Deusinglaan 1, 9713 AV, Groningen, The Netherlands. Tel.: 31 50 3632741. Fax: 31 50 3632728. E-Mail: D.Hoekstra@med.rug.nl