INTRODUCTION

The differentiation of epidermal epithelial cells (keratinocytes) is characterized by a programmed series of profound biochemical and morphological transformations that ultimately produce the protective barrier necessary for terrestrial life. The most thoroughly studied events are the expression of keratins and other proteins integral to the structure of the outermost nonviable layer of the epidermis, the stratum corneum (reviewed in Refs. 1 and 2). The regulation of the complex changes in lipid synthesis and organization that accompany these structural protein alterations and ultimately result in the water impermeability of skin is less understood. As keratinocytes progress from the basal proliferative layer outward toward the stratum corneum, they become enriched in specific lipids and form lamellar granules, specialized organelles that contain stacks of membranous disk-like structures (3). Lamellar granules are particularly enriched in glucosylceramides (GlcCer),¹ and also contain acid hydrolases capable of processing their lipid contents (4, 5, 6). Upon extrusion of the lamellar granule contents into the intercellular spaces at the junction of the viable and nonviable cell layers, GlcCer are converted to ceramides. The extracellular lipids, which are devoid of phospholipids and contain ceramides as a major fraction, organize into lamellar sheets (2). These sheets surround the nonviable corneocytes in the stratum corneum and are responsible for the barrier function of the epidermis (7).

Despite the critical roles that sphingolipid metabolism and trafficking play in epidermal physiology, studies of these processes in keratinocytes have been limited. The increase in ceramide content during differentiation and the de novo synthesis of ceramides and GlcCer have been demonstrated in epidermis (8) and in organotypic keratinocyte cultures (9, 10, 11). Cultured keratinocytes were reported to have a high level of serine palmitoyltransferase activity, which catalyzes the synthesis of the long-chain base precursor of sphingolipids (12). A crucial role for the acid glycosidase, glucosylceramidase (-glucocerebrosidase), in converting GlcCer to extracellular ceramides after extrusion of lamellar granule contents has recently been delineated (13).

The enzyme responsible for the formation of GlcCer has not been studied previously with regard to its role in epidermal sphingolipid metabolism. CerGlc transferase (UDPglucose:N-acylsphingosine D-glucosyltransferase, EC), also termed glucosylceramide synthase, catalyzes the reaction of UDPGlc and ceramide to form UDP and GlcCer, which, in most cell types other than keratinocytes, serves mainly as a precursor for complex glycolipids or as a plasma membrane component. The enzyme has been partially purified from Golgi fractions from porcine submaxillary gland and rat brain, and its dependence on phospholipids, detergents, metals, and other effectors has been studied (14, 15, 16). Just recently, a procedure for the purification of the enzyme to a high specific activity from rat liver Golgi membranes was published (17), but molecular properties of the transferase protein have not yet been described. CerGlc transferase resides on the cytoplasmic face of membranes in the cis-Golgi, and possibly, in a pre-Golgi compartment (18, 19, 20). Studies with DL-threo-1-phenyl-2-(decanoylamino)-3-morpholino-1-propanol and structural analogs (reviewed in Ref. 21), demonstrated that these specific, reversible inhibitors of CerGlc transferase activity reduced glycosphingolipid concentrations in a variety of cultured cells. Effects of these inhibitors on specific cellular processes underscored the importance of the transferase in ganglioside metabolism, cell surface recognition and adhesion properties, and regulation of cell growth (21). A mouse melanoma cell line with deficient GlcCer transferase activity and no detectable glycolipids was reported to show altered morphology and a reduced growth rate as compared to the parental line (22).

Little is known about in vivo regulation of CerGlc transferase. The demand for GlcCer as a precursor of structural quantities of ceramides in the epidermis suggests that differentiating epidermal cells may have the capacity to express high levels of transferase activity. In order to determine how the expression of CerGlc transferase activity relates to keratinocyte differentiation and to develop a better understanding of the role of the enzyme in GlcCer metabolism in epidermis, we studied transferase activity in two culture systems which have been widely used in the characterization of keratinocyte differentiation. At various times after switching to differentiation-inducing medium, cultures were assayed for CerGlc transferase and -glucocerebrosidase activity, examined by electron microscopy, and analyzed for lipid content. The role of CerGlc transferase in keratinocyte GlcCer synthesis was studied further by treating some cultures with the inhibitor, DL-threo-1-phenyl-2-(palmitoylamino)-3-morpholino-1-propanol (PPMP). Finally, the activity of CerGlc transferase in keratinocytes was compared to that in a variety of other types of cultured cells, to help assess their relative potential for GlcCer production. A portion of these results was reported earlier in a preliminary form (23).

EXPERIMENTAL PROCEDURES

[Glc-U-¹⁴C]UDPGlc, 11.8 GBq/mmol, was purchased from DuPont NEN, Boston, MA. 6-((N-(7-Nitrobenz-2-oxa-1,3-diazol-4-yl)amino)-caproyl)sphingosine (NBD-ceramide) was from Molecular Probes. 12-(N-Methyl-N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl))aminododecanoyl sphingosyl--D-glucoside (NBD-GlcCer) was from Sigma. PPMP was from Matreya, Inc., Pleasant Gap, PA. Other lipids, detergents, and biochemicals were purchased from Sigma or were of the highest available grade from other suppliers.

Keratinocytes were obtained from neonatal foreskins by overnight trypsinization in 0.25% trypsin, 0.1% sucrose in phosphate-buffered saline at 4 °C. Cells were plated in 60-mm plastic tissue culture dishes using 0.07 mM calcium keratinocyte growth medium (KGM; Clonetics, San Diego, CA) containing 2% fetal bovine serum (Sigma) and grown at 37 °C in a humidified incubator under a 5% CO₂, 95% air atmosphere. The medium was changed 24 h later to 0.07 mM calcium KGM without serum, and the cultures were then fed every other day. Cultures were passed into 100-mm dishes at 60-70% confluence and used for experiments after the 3rd passage. At near confluence, cultures were switched to either 1.4 mM calcium KGM (nKGM) or to Dulbecco's modified Eagle's medium/Ham's F-12 (3:1) supplemented with 10% fetal bovine serum, 10 µg/ml insulin, 0.4 µg/ml hydrocortisone, 5 units/ml penicillin, and 5 µg/ml streptomycin (cDMEM). Fibroblast cultures were established from human foreskin tissue and maintained in Eagle's minimal essential medium, supplemented with 10% fetal bovine serum, antibiotics, amino acids, and vitamins, as was described previously (24). Cultures of the monocyte-macrophage cell lines, THP-1 and U937, were obtained from the American Type Tissue Culture Collection, Rockville, MD, and were grown in RPMI 1640 medium supplemented with 10% fetal bovine serum. Some of the monocyte-macrophage cultures were differentiated in medium containing 10 nM phorbol 12-myristate 13-acetate (Sigma) for 1 week; medium was changed every other day and cultures were protected from light. Cultured cells from the Caco-2 intestinal epithelial line were grown as described (25) and kindly provided by Dr. Jeffrey Field and Mary Lou Booth.

Fluorescent NBD-ceramide was dissolved in absolute ethanol to give a stock solution of 1 mM. Cultures were labeled with 10 nmol/ml NBD-ceramide in 6 ml of Hank's balanced salt solution containing 0.68 mg/ml fatty acid-free bovine serum albumin for 1 h at 4 °C in the dark. After the labeling period, the medium was removed and cultures were rinsed twice before either harvesting or incubating for various additional times in DMEM/F-12 (3:1) without serum at 37 °C.

After rinsing, cultures were scraped with a Teflon policeman into glass scintillation vials, frozen, and lyophilized. Lipids were extracted using chloroform/methanol mixtures as described previously (26) and separated on 20 × 20 Silica Gel G thin layer chromatography plates (Analtech, Newark, DE) scored into 6-mm lanes. For cDMEM and nKGM culture analyses, the solvent systems were as described previously (9). After development, the chromatograms were dried, sprayed with 50% H₂SO₄, and then heated slowly to 220 °C on a hotplate to char the lipids. The chromatograms were cooled and then scanned at 600 nm using a Shimadzu CS-9000 scanning densitometer with automatic peak quantitation (Shimadzu Scientific Instruments, Columbia, MD). Polar lipids and ceramides were identified by comparison on thin layer chromatography with previously characterized mouse (9) and human (27) epidermal lipids or with commercial standards. Nonpolar lipids were identified by comparison with standards obtained from Sigma. Standard curves were generated with multiple amounts of lipid standards. For sample quantitation, values obtained from the standard curves for 3 to 5 dilutions of samples run in multiple lanes were averaged. For the NBD-lipid separation, the solvent system was chloroform/methanol/water (40:10:1) to 15 cm followed by hexane/ether/acetic acid (70:30:1) to 20 cm. After drying the plate, the lanes were scanned at 460 nm. The densitometer response was linear over a broad range of lipid amounts.

Cultures were fixed in cacodylate-buffered 2.5% glutaraldehyde for 1 h at room temperature or overnight at 4 °C and postfixed in osmium tetroxide for 1 h. After rinsing and dehydration through an ascending ethanol series, the cultures were embedded in Eponate 12. Thin sections were stained with uranyl acetate and lead citrate and examined in a Hitachi H600 electron microscope.

Cultures were rinsed 5 times with 0.15 M NaCl, harvested in saline with a plastic scraper, and centrifuged for 5 min at 1000 × g. Pellets were stored at 20 °C; CerGlc transferase and glucocerebrosidase activity were stable for at least 1 month under these conditions. Cell lysates were prepared in ice-cold buffer, pH 6.9, which contained the following: 0.25 M sucrose, 5 mM MOPS, 1 mM EDTA, 1 mM EGTA, 0.5 mM phenylmethylsulfonyl fluoride, and 1 mM 2,3-dimercaptopropanol. Cell pellets in 1.5-4 ml of buffer were processed for 30 s at the lowest setting with a Polytron PT10/35 homogenizer equipped with a PTA 10 generator (Brinkmann Instruments, Westbury, NY), then sonicated in an ice bath for 3 min (80% power; Model W-370 sonicator equipped with a cup horn; Heat Systems-Ultrasonics, Inc., Plainview, NY). Lysates of more highly differentiated keratinocytes (>1 week in culture) required more processing to permit homogeneous sampling and were passed 5-10 times through a 20-gauge needle.

The assay for CerGlc transferase activity was modified slightly from that published by Matsuo et al. (16). Solid-phase ceramide substrate stock was prepared by dissolving ceramides (bovine brain, Sigma), in chloroform (~2.5 mg/ml) and mixing the lipid solution with Silica Gel 60 (5-µm diameter) at a ratio of 20 µg of ceramides/mg of silica gel. Solvent was evaporated under N₂, and the dried complex was stored at 20 °C. UDPGlc substrate was prepared by quickly evaporating ethanol solvent, under N₂, from the required amount of [¹⁴C]UDPGlc, then adding unlabeled UDPGlc in H₂O to give a 0.22 mM, 0.74 MBq/ml stock solution. The standard reaction mixture contained, at a final concentration, in a volume of 110 µl: 1 mg of ceramide-silica gel, 50 µM UDPGlc (7.4 kBq), 50 mM MOPS, pH 6.5, 5 mM MnCl₂, 2.5 mM MgCl₂, 1 mM NADH, 5 mM dimercaptopropanol, 1% CHAPS, and enzyme source or sucrose homogenization buffer. All reagents, except for sugar nucleotide, were mixed and added to the enzyme source in a 2-ml, v-bottom, polypropylene microcentrifuge tube (Midwest Scientific, St. Louis, MO). The reaction was initiated by addition of 25 µl of the radiolabeled UDPGlc stock, and the mixture was incubated for 30 min at 37 °C, with occasional mixing to suspend the ceramide-silica gel complex. The reaction was terminated by adding 0.5 ml of a cold buffer solution that contained 0.15 M NaCl in 10 mM sodium phosphate, pH 6.8, and centrifuged at 6 °C, 2000 × g, for 5 min. Supernatant liquid was removed, and the pellet was washed 4 times in 0.2 ml of buffered saline with vigorous mixing and centrifugation. The pellet was transferred to a scintillation vial with two 0.15-ml additions of buffered saline, a 5-ml aliquot of scintillation fluid (Budgetsolve; Research Products International) was added, and radioactivity was monitored in a liquid scintillation counter (Packard Tri-Carb Model 1500). The amount of Glc transferred to form radiolabeled product was calculated from the specific radioactivity of the sugar nucleotide substrate.

The dependence of CerGlc transferase activity on substrate concentration was examined using crude extracts prepared from keratinocyte cultures harvested either prior to or 7 days after being switched to cDMEM differentiation medium. Assays were conducted as was described above, except that the concentration of UDPGlc or ceramide in the reaction mixture was varied. UDPGlc at a constant specific activity was tested over a range of 5 to 100 µM. The ceramide concentration was varied from 0.9 to 900 µM by adding 1 mg/reaction of silica gel to which the desired amount of ceramide had been adsorbed. In an alternate procedure, different amounts of the standard ceramide-silica gel complex (20 µg of ceramide/mg of silica) were included in the reaction mixture. Each study consisted of 8 different concentrations of a given substrate tested in duplicate. Apparent values for K_m and V_max were determined from a direct nonlinear least squares fit of the data to the Michaelis-Menten equation.

-Glucocerebrosidase Assay

The procedure was based on a previously described method in which hydrolysis of a fluorescent GlcCer substrate is monitored (28, 29). The reaction mixture contained, in a volume of 100 µl, the following reagents at the indicated final concentrations: 0.015 mM NBD-GlcCer, 0.285 mM GlcCer (human Gaucher spleen, Sigma), 50 mM citrate-phosphate buffer, pH 5.25, 0.25% Triton X-100, and 0.6% sodium taurocholate. Lipids for the required number of assays were dissolved in chloroform/methanol (2:1), and the solvent was removed under N₂. A volume corresponding to 50 µl/assay of citrate-phosphate buffer, pH 5.25, prepared by titrating 0.1 M citric acid with 0.2 M Na₂HPO₄, was added to the lipids with vortex mixing, followed by sonication for 1 min in a cup horn filled with ice water, as described above for cell lysate preparation. The reaction was initiated by addition of 50 µl of the substrate solution to 50 µl of a dilution of cell lysate in H₂O in a 2-ml polypropylene microcentrifuge tube. After incubating the mixture for 30 min at 37 °C, the reaction was stopped by adding 0.23 ml of isopropyl alcohol, 0.75 ml of heptane, and 0.125 ml of H₂O, following each addition with vigorous mixing. Phases were separated by centrifugation for 10 min at 2000 × g, and the upper layer was re-extracted with 0.17 ml of H₂O, and analyzed for fluorescence using a Perkin-Elmer Model 203 spectrofluorimeter set at a wavelength of 460 and 520 nm for excitation and emission, respectively. The concentration of ceramide product was calculated from the fluorescence of a standardized solution of NBD-ceramide prepared from the same solvent extraction system. Reaction rates were linear with respect to time and concentration of samples; the latter was routinely verified by assaying incremental amounts of individual samples.

Protein content of cell lysates was estimated by the method of Lowry et al. (30) with the use of delipidated crystalline bovine serum albumin as a standard. DNA was estimated with a fluorescent dye binding assay (31), using salmon sperm DNA (Type III, Sigma) as a standard. Aliquots of cell lysates were assayed for DNA immediately after cells were disrupted. Both procedures were conducted with amounts of sample that were within the range of standards and under conditions in which sample buffer components were confirmed not to interfere with assay results.

RESULTS

A comparison of the DNA and protein content of nKGM and cDMEM cultures over time is shown in Fig. 1. The DNA content of nKGM cultures increased 3 times before reaching a maximum at ~5 days, whereas that of cDMEM cultures plateaued at ~7 days at a level that was 2-3 times higher than in nKGM cultures. Cultures maintained in low calcium KGM had the same DNA content as those switched to nKGM for the first week and declined to half those values during the second week (Fig. 1 legend). The protein content of nKGM cultures increased ~10-fold to a plateau after 1 week, whereas that of cDMEM cultures continued to increase to a level ~3-4 times that of nKGM cultures after 2 weeks. Results of most of our biochemical analyses are normalized with respect to DNA, rather than protein content, because of the huge increase in structural proteins that characterizes keratinocyte differentiation.

A comparison of nKGM with cDMEM cultures by electron microscopy after 7-10 days (Fig. 2) showed that cDMEM cultures had more and thicker cell layers (Fig. 2, a and b) and also demonstrated keratohyalin granules, a marker of terminal epidermal differentiation. There were numerous lamellar granules in cDMEM cultures (Fig. 2c), but only rare lamellar granules in nKGM cultures. By day 10, a prominent cornified envelope and extrusion of lamellar granule contents into the intercellular spaces was noted only in cDMEM cultures (Fig. 2d). By morphologic criteria, cDMEM cultures clearly demonstrate a much higher level of differentiation than do nKGM cultures even though a well-developed stratum corneum does not form.

Results from experiments in which the lipid composition of cDMEM cultures was analyzed over a 10-day period are presented in Table I. A nearly 3-fold increase in total lipids over the period from day 3-10 was accompanied by a marked increase in the neutral sphingolipid fractions, primarily at the expense of diacylglycerophospholipid fractions, with little change in sphingomyelin. The ceramide fraction increased 6- to 9-fold over the time course to a level that was nearly 7% of total lipids by day 10 in one experiment. GlcCer, which are precursors of the extracellular ceramides in the stratum corneum, were increased 3-8 times and represented 2.8-4.4% of total lipids by day 10. Of particular interest was the time-dependent appearance of acyl-GlcCer and its metabolite, acylceramide, lipids unique to epidermis and markers of terminal epidermal differentiation (32, 33).

Table I. Lipid composition of human keratinocyte cultures over time after switching to cDMEM Keratinocyte cultures grown to near-confluence in low calcium KGM were switched to cDMEM and harvested for lipid analysis after 3, 7, and 10 days. Results shown are the average values of data from two separate experiments; four to seven 100-mm cultures were combined for each time point. Numbers in parentheses indicate the deviation of the extremes from the mean. Lipid speciesa Lipid content 3 days 7 days 10 days % total SM 13.7 (0.2) 13.6 (0.5) 14.0 (0.5) PC 41.9 (2.2) 37.5 (3.5) 34.8 (0.4) PE 27.5 (2.5) 22.8 (2.8) 18.8 (3.2) GlcCer 0.97 (0.62) 3.7 (0.6) 3.6 (0.8) AGlcCer NDb 0.36 (0.12) 0.56 (0.03) Other Cer 0.73 (0.06) 2.8 (0.4) 5.3 (0.6) ACer ND 0.12 (0.03) 0.52 (0.27) CHOL 12.6 (1.6) 15.1 (1.0) 17.9 (0.5) FFA 1.00 (0.31) 0.57 (0.09) 0.85 (0.15) TG 1.35 (0.05) 2.7 (0.9) 2.4 (0.3) CME 0.33 (0.02) 0.88 (0.13) 1.5 (0.2) Total lipid, µg/100-mm culture 802 (33) 1728 (11) 2169 (140) a  SM, sphingomyelin; PC, phosphatidylcholine; PE, phosphatidylethanolamine; GlcCer, glucosylceramide; AGlcCer, acylglucosylceramide; Cer, ceramide; ACer, acylceramide; CHOL, cholesterol; FFA, free fatty acids; TG, triglycerides; CME, cholesterol monoesters. b  ND, not detected.

Time-dependent increases in the neutral sphingolipid fractions were much less pronounced to absent in nKGM cultures, as is illustrated for the GlcCer content in Fig. 3. By day 7 after the medium switch, the total lipid content, normalized to cell DNA, of nKGM and cDMEM cultures, was increased approximately 1.5 times and 2 times, respectively, over cultures that were maintained in low calcium KGM. In contrast to the relatively small change in total lipid/DNA, the GlcCer/DNA content of cDMEM cultures increased 6-fold, whereas that of nKGM cultures showed only a slight increase over the low calcium KGM cultures. The increased GlcCer content in the more highly differentiated, cDMEM cultures suggested that the expression of CerGlc transferase might differ in the two culture systems.

The preparation of homogeneous samples from the more differentiated keratinocyte cultures posed a difficulty in view of the instability of CerGlc transferase activity with prolonged homogenization. A homogenization protocol was maximized for retention of CerGlc transferase activity and was strictly followed. In anticipation of possible effects on the reaction rate due to changes in sample composition during culture differentiation, the linearity of reaction rates with the concentration of sample was routinely verified. Catalytic properties of the CerGlc transferase in crude keratinocyte extracts were similar to those of a partially purified detergent-solubilized Golgi extract from brain tissue as studied by Matsuo et al. (16) using the same assay system. Only trace radioactivity was detected when reaction mixtures were devoid of an enzyme source or contained boiled keratinocyte extract. Maximal activity was found at pH values from ~6.3-7, with a shoulder at ~60% maximum extending to pH values up to 7.7.

Values for CerGlc transferase activity over time in culture are shown in Fig. 4. Data from several experiments with overlapping sampling times were combined. Over a 2-week period following the medium switch, the mean transferase activity, normalized to DNA content, in nKGM cultures was not significantly elevated over the day 0 value, although in some individual experiments up to 2 times increases in activity were observed during the second week. By contrast, after a lag of ~3 days, transferase activity in cDMEM cultures began to increase and reached a maximum level at 8 days that was >6 times the activity at day 0 in the combined data and >10 times in some individual experiments. Variability in the data are evident during periods when activity is changing the most and may be due, in part, to the variability of the primary cultures which were expanded for use in individual experiments.

1

1

Kinetic properties of the transferase activity from undifferentiated versus differentiated cultures were determined as described under ``Experimental Procedures.'' Values for the apparent K_m were 8.0 and 7.2 µM, with respect to UDPGlc, and 5.9 and 5.2 µM, with ceramide, for activity from cultures harvested at 0 and 7 days after the switch to cDMEM, respectively. The apparent V_max values were 24 and 85, with respect to UDPGlc, and 14 and 71, with ceramide, for the transferase in the undifferentiated versus differentiated cultures, respectively. The near identity of K_m values with each substrate for enzyme in the two culture systems and the similar change in V_max values with each substrate is compatible with the notion that the amount of transferase is increased during differentiation. A less likely possibility is that the change in V_max values reflects a differentiation-induced modification of the transferase structure that equally affects the catalytic efficiency with respect to both substrates, without affecting the K_m values. To further explore the mechanism of the increased transferase activity during differentiation, cultures were treated with cycloheximide 4 days after the switch to cDMEM. Following two 4-h applications of cycloheximide over the range of 3-100 µg/ml, CerGlc transferase activity was reduced 60-80% from the control activity. A fit of data points obtained at 4, 8, 12, and 24 h after addition of 30 µg/ml cycloheximide to the culture medium, to the equation for first order decay, revealed a half-life for the transferase activity of approximately 9 h. These results suggest that during culture differentiation, the transferase activity undergoes a relatively rapid turnover, and that continued protein synthesis is required to provide for the increased level of transferase activity.

Taken together, the kinetic data and the cycloheximide results suggest that the increased expression of transferase activity during differentiation may be explained by synthesis of new transferase protein. However, other possibilities, such as activation of pre-existing enzyme by a cycloheximide-sensitive process or the above-mentioned modification of catalytic efficiency cannot be ruled out by the current results.

Comparison of Glucosyltransferase and -Glucocerebrosidase Activity during Differentiation

Behavior of glucocerebrosidase, which converts extruded GlcCer to ceramides at the junction between the granular layer and the stratum corneum, was compared to that of CerGlc transferase under the same conditions. In contrast to the profile seen for the transferase, glucocerebrosidase activity increased ~5 times by day 2 in both cDMEM and nKGM cultures (Fig. 5).

-Glucocerebrosidase activity in lysates of keratinocyte cultures with time after switching to nKGM or cDMEM.

1

1

PPMP has been shown to be a relatively potent, specific, and reversible inhibitor in in vitro assays of CerGlc transferase activity. Its decanoyl analog has been widely applied to analysis of GlcCer metabolism of intact cells in several different culture systems and in whole animals (21). When added to the assay for keratinocyte CerGlc transferase activity, PPMP inhibited the reaction rate 40% and 75% at concentrations of 10 and 30 µM, respectively (data not shown). To study the effect of the inhibitor in intact keratinocyte cultures, 10 µM PPMP was added to cDMEM cultures from days 3-7, when activity of CerGlc transferase showed the greatest rate of increase. Analysis of lipids from extracts of the cultures that were harvested on day 7 (Table II) showed that the GlcCer content was reduced to <15% that of the untreated control. In addition, production of acyl-GlcCer was obliterated, and ceramides and cholesteryl monoesters were also reduced. Values for total lipid and other lipid fractions were unchanged, with the exception of a small increase in phosphatidylethanolamine. Further evidence for an effect of PPMP on GlcCer synthesis was obtained from another approach in which the conversion of exogenous NBD-ceramide to NBD-GlcCer in keratinocyte cultures was examined (Fig. 6). On day 7 after the switch to cDMEM, cultures were pretreated for 2 h with PPMP, pulsed with NBD-ceramide, and then chased for 30 and 90 min in the presence of PPMP. Cultures treated with the inhibitor had 5 and 9%, respectively, of the label in GlcCer, compared to 46 and 66% in the untreated controls. PPMP stimulated an increase in the fractional labeling of sphingomyelin and intracellular ceramide in this short-term pulse-chase experiment.

Table II. Effect of PPMP treatment of keratinocyte cultures on lipid composition Human keratinocyte cultures were switched to cDMEM after reaching near-confluence in low calcium KGM. Cultures were treated with a final concentration of 10 µM PPMP in ethanol or 0.1% solvent control on days 4, 5, and 6, and harvested on day 7. Results shown are from a single experiment in which four 100-mm cultures from each group were combined for lipid analysis. Abbreviations are as in Table I. Lipid species Lipid content Control PPMP % total SM 13.7 13.2 PC 32.3 33.8 PE 28.7 32.5 GlcCer 5.1 0.78 AGlcCer 0.58 NDa Other Cer 1.21 0.85 ACer ND ND CHOL 14.0 13.4 FFA 0.45 0.54 TG 3.2 4.8 CME 0.73 0.15 Total lipid, µg/100 mm culture 2479 2335 a  ND, not detected.

These results are consistent with the premise that addition of PPMP to keratinocyte cultures inhibited transferase catalysis, thus blocking the formation of GlcCer.

Analysis of CerGlc transferase activity in extracts of selected human cultures other than keratinocytes revealed low levels of activity, in the general range of those found for undifferentiated keratinocyte cultures (Table III). The samples selected included foreskin fibroblasts, monocyte-macrophage cell lines before and after differentiation with phorbol ester, and the Caco-2 intestinal epithelial cell line. By contrast, the cultures had varying levels of glucocerebrosidase activity, with fibroblasts expressing >20 times that of undifferentiated keratinocyte cultures and ~5 times that of 10-day nKGM and cDMEM cultures. Although differentiation of monocyte-macrophage cultures by treatment with phorbol ester did not significantly affect transferase activity, glucocerebrosidase activity was stimulated, as has been shown for other lysosomal enzymes in U937 cultures (34). In contrast to the results in the more highly differentiated keratinocyte cultures, relatively low levels of CerGlc transferase activity were found in homogenates of intact foreskin tissue, dermis, and trypsin-isolated epidermal cells.

Table III. Glucosyltransferase and -glucocerebrosidase activity of selected human cell cultures and foreskin tissue Procedures for cell culture and enzyme analysis are described under ``Experimental Procedures.'' Confluent fibroblast and Caco-2 cultures were switched to cDMEM for 1 week prior to harvesting cells for enzyme analysis. Results are given as the mean value ± S.E. for triplicate determinations of samples from at least 2 individual cultures. A unit of activity is defined as the amount of product, in picomoles, for glucosyltransferase, and in nanomoles, for glucocerebrosidase, that was generated per min under standard conditions. Cell culture or tissue Enzyme activity Glucosyltransferase Glucocerebrosidase units/mg DNA Keratinocyte cultures KGM, low calcium 28  ± 16 25  ± 6 KGM, normal calcium, 10 days 23  ± 4 106  ± 31 cDMEM, 10 days 154  ± 48 161  ± 22 Cell lines Foreskin fibroblasts 5.1  ± 1.6 566  ± 32 Caco-2 intestinal epithelial 6.8  ± 0.1 25  ± 2 THP-1 monocyte-macrophage 4.1  ± 0.7 24  ± 2 THP-1 + phorbol ester 1.7  ± 0.4 73  ± 6 U937 monocyte-macrophage 1.5  ± 0.4 9  ± 3 U937 + phorbol ester 1.6  ± 0.9 107  ± 32 Foreskin Intact 7.3  ± 2.3 20  ± 2 Epidermis 4.0  ± 4.7 21  ± 4 Dermis 5.3  ± 2.4 31  ± 4

DISCUSSION

Numerous recent studies of the cell biology and biochemistry of sphingolipids have demonstrated the crucial involvement of ceramides and other sphingosine derivatives in the regulation of diverse cellular processes (reviewed in Refs. 35 and 36). In most cell types, the glucosylation of ceramide serves primarily to provide a precursor for complex glycolipid synthesis (37). In epidermis, collective evidence from lipid analyses, electron microscopy, and enzymology has indicated a quantitatively important role for GlcCer as precursors of ceramides, the major component of the lipids responsible for the cutaneous permeability barrier (2, 7). The mechanisms involved in epidermal lamellar granule formation and enrichment with respect to GlcCer, and the subsequent extrusion of their contents into the extracellular space between the viable layers of the epidermis and the nonviable stratum corneum are not understood. The crucial role for GlcCer in the process suggests that its synthesis may be under unique regulatory control in differentiating keratinocytes.

Results from the present study directly demonstrate CerGlc transferase activity for the first time in keratinocytes and show that the activity is induced during keratinocyte differentiation. Analyses of steady-state kinetic parameters and intracellular turnover of the transferase activity suggest that the mechanism for enhanced expression of the activity during differentiation involves synthesis of new transferase protein. To our knowledge, regulation of GlcCer transferase activity has not been examined previously with respect to cell differentiation in any epithelial culture model, although induction of the enzyme has been demonstrated in a nerve regeneration model (38). Our results indicate that CerGlc transferase induction correlates with keratinocyte differentiation as shown by the different response of the enzyme in two culture systems. KGM cultures have been studied extensively and previously shown to support specific differentiation-related events (39, 40). DMEM-based media have been used mainly in conjunction with more highly differentiated, organotypic cultures (32, 41, 42). We have shown that cDMEM cultures grown on plastic are more highly differentiated than nKGM cultures and exhibit much higher transferase activity.

Changes in transferase activity paralleled those in GlcCer content. Thus, the more highly differentiated cultures were more enriched with respect to GlcCer, and the time course of the GlcCer increase in cDMEM cultures, beginning at ~3 days and plateauing after ~7 days (Table I), was roughly comparable to the increase in transferase activity (Fig. 4). Ceramides also accumulated in culture during this period and continued to increase after 7 days. This is consistent with the known precursor-product relationship of GlcCer and ceramides, in which precursor GlcCer are packaged into lamellar granules to form a ``steady-state pool'' from which ceramide product is formed as the lamellar granule contents are extruded into a cell-associated extracellular compartment. The electron microscopic images shown in Fig. 2 illustrate this unique pathway, with lamellar granules becoming most prominent around 7 days and extruding their contents in the 7-10-day time period.

Correlation between culture GlcCer content and transferase activity was also apparent in experiments with PPMP. PPMP inhibited keratinocyte CerGlc transferase activity in keratinocyte lysates with roughly the potency reported in other cell types. When applied to keratinocyte cultures, PPMP blocked the synthesis of NBD-GlcCer from exogenously added NBD-ceramide in short-term experiments and prevented an increase in GlcCer mass over a period of several days. These effects of the inhibitor are consistent with many of the findings for PPMP and analogs in other cell systems (21).

In the short-term study, blocking the conversion of NBD-ceramide pulse label to NBD-glucosylceramide would be expected to result in its diversion to sphingomyelin synthesis (the major alternative pathway of ceramide metabolism) as well as a delayed decay of NBD-ceramide during the 90-min span of the chase period, results which were observed (Fig. 6). After long-term PPMP inhibition, in addition to a profound decrease in GlcCer content, there was also a decrease in culture ceramide content (Table II). Remembering that, in epidermis, the majority of measured ceramide is extracellular and is a product of GlcCer hydrolysis, long-term PPMP inhibition of GlcCer synthesis would be expected to ultimately result in a decrease in its extracellular ceramide product. The lack of change in sphingomyelin mass during long-term inhibition may reflect the fact that cellular sphingomyelin levels are tightly controlled (as are levels of free ceramide); homeostatic mechanisms likely operate to restore normal levels during long-term experiments. Although PPMP experiments need to be interpreted carefully in view of certain nonspecific effects that have been demonstrated for this class of inhibitors (43), taken together, our results indicate that regulation of CerGlc transferase activity is an important determinant of the GlcCer content of keratinocytes and, thus, would be expected to affect production of ceramides, as well as the barrier function of the epidermis.

Glucocerebrosidase activity, which is required for conversion of GlcCer to ceramide in the stratum corneum, has been studied previously with relation to epidermal differentiation. Enzyme activity was increased in the outermost layers of skin (44, 45) and has also been demonstrated in lamellar granule preparations (4). In our studies, induction of GlcCer transferase activity was shown to be more strongly correlated with the morphological and lipid properties that characterize culture differentiation than was glucocerebrosidase activity. Glucocerebrosidase activity increased earlier than transferase activity and reached nearly the same maximum level in cDMEM and nKGM cultures, indicating that the requirements for stimulation of the two enzymes are different.

The expression of other markers of epidermal differentiation have been studied, including the structural proteins, keratin (46), loricrin (47), involucrin (48), and filaggrin (49); transglutaminase (50), believed to mediate cross-linking of cornified envelope proteins; and cholesterol sulfotransferase (50, 51), required for the formation of cholesterol sulfate, a barrier lipid whose hydrolysis is necessary for normal desquamation (2). A distinctive time course for induction of the individual markers (1, 46, 47, and references therein) indicates a complex programmed sequence of events in the differentiation process which remains poorly understood. Insight may be provided in the near future from the results of numerous investigations centering on the profound effects of retinoids on epidermal differentiation (52, 53).

Low levels of transferase activity were found in other cultured cell types as compared with that in differentiated keratinocytes, suggesting that epidermal cells have the capacity for high activity when there is a demand for GlcCer. Apical plasma membranes of intestinal brush border cells have been demonstrated to be enriched with unusually high levels of glycosphingolipids, which are putatively required for membrane stabilization and protection of the lumenal surface against digestion by phospholipase A₂ (reviewed in Ref. 54). This observation suggested that intestinal cells, as well as Caco-2 intestinal epithelial cultures, which also exhibit a polarized membrane glycolipid distribution (55), might express a high level of CerGlc transferase activity, similar to that of differentiated keratinocytes. This was not supported by our results for Caco-2 cultures (Table III), although sufficient data on lipid quantitation in this system are not available to enable correlation of transferase activity with glycosphingolipid content. Interestingly, the glucocerebrosidase activity of Caco-2 cells is much lower than that of fibroblasts and the other differentiated cells that were examined, suggesting that the cultured intestinal cells may regulate their glycolipid content, in part, by minimizing GlcCer breakdown.

The low level of transferase activity in homogenates of intact foreskin, dermis, and trypsin-isolated epidermal cells, as compared to that in differentiated keratinocyte cultures, may reflect some instability of the enzyme with respect to the harsh conditions necessary for disruption of the whole tissue. On the other hand, cDMEM cultures may accentuate a period in the differentiation process when keratinocytes are actively growing and producing the lipids needed to establish barrier function. This most likely represents the situation in regenerating epidermis, such as following barrier disruption or during wound healing, rather than the steady state condition of intact epidermis. Disruption of the epidermal permeability barrier with organic solvent in an in vivo experimental murine model has been shown to transiently stimulate the synthesis of DNA and of several lipids important in barrier formation (56); values returned to baseline once the barrier was re-established. A recent study of murine fetal differentiation demonstrated a transient increase in cholesterol sulfotransferase followed by a return to baseline shortly after birth (50).

A number of diseases characterized by abnormalities of epidermal lipids are recognized (57, 58). The only disorder in this group with a demonstrated defect in the ceramide pathway is the severe, neuronopathic type of Gaucher disease, resulting from deficient glucocerebrosidase activity (13, 59). Biochemical abnormalities in CerGlc transferase activity or in the other steps involved in the production and processing of stratum corneum ceramides, whether primary or secondary, would be expected to result in aberrant epidermal structure and function. Such abnormalities could possibly have a role in, for example, harlequin ichthyosis, in which lamellar granules are known to be defective (60), or multisystem triglyceride storage disease (61), inherited disorders affecting skin that have not yet been explained at the biochemical level. Epidermal ceramide abnormalities have been reported in such diverse conditions as psoriasis (62), atopic dermatitis (63, 64), and aging (64), but the biochemical basis for these abnormalities remains unknown.

The current work has established conditions under which CerGlc transferase activity is expressed at a high level by cultured human keratinocytes, and our results indicate an important role for this enzyme during epidermal differentiation. The use of this system will facilitate studies of the regulation of CerGlc transferase activity and its impact on keratinocyte function. Study of the unique aspects of the ceramide pathway in keratinocytes should further understanding of the biochemistry and cell biology of sphingolipids in general.