Vitamin C Stimulates Sphingolipid Production and Markers ofBarrier Formation in Submerged Human KeratinocyteCultures

Yoshikazu Uchida,* Martin Behne,* Daniele Quiec,* Peter M. Elias,* and Walter M. Holleran*²*Dermatology Service and Research Unit, Department of Veterans Affairs Medical Center, and Departments of *Dermatology, School of Medicine and

Pharmaceutical Chemistry, School of Pharmacy, University of California, San Francisco, CA, U.S.A.

Human keratinocytes differentiate in vitro in responseto a variety of stimuli, but neither the levels nor thespectrum of ceramides approach those seen in vivo.Ceramide production increases when human kera-tinocytes are grown at an air±liquid interface, andalterations in ceramide content occur when vitaminC is added to air-exposed, organotypic culture sys-tems (Ponec et al. J Invest Dermatol 109:348, 1997).Here, we assessed whether vitamin C stimulatessphingolipid production in human keratinocytesindependent of differentiation and air exposure.When submerged, human keratinocytes were grownin 1.2 mM calcium and serum-containing mediumwith vitamin C (50 mg per ml) for 9 d, total lipidcontent remained unchanged, but both glucosylcera-mide and ceramide content increased. Moreover,selected ceramide and glucosylceramide species:i.e., nonhydroxy ceramide 2 and both a- andw-hydroxylated sphingolipids, increased prefer-entially [ceramide 4 (6-hydroxy-acylceramide), cera-mide 5 (a-hydroxyceramide), ceramide 6(4-hydroxy-a-hydroxyceramide), and ceramide 7

(6-hydroxy-a-hydroxyceramide); and acylglucosyl-ceramide, glucosylceramide-B, and glucosylcera-mide-D], whereas ceramide 1, ceramide 3,glucosylceramide-C, and sphingomyelin remainedunchanged. Synthesis of the corresponding ceramideand glucosylceramide fractions was enhanced byvitamin C, attributable, in part, to increased cera-mide synthase activity (over 2-fold, p= 0.01); bothserine palmitoyltransferase and glucosylceramidesynthase activities remained unaltered. Finally,increased vitamin C-stimulated sphingolipid produc-tion correlated with the presence of lamellar bodieswith mature internal contents, an increase in cova-lently bound w-hydroxyceramide, and the appear-ance of prominent, corneocyte-bound lipidenvelopes, whereas corni®ed envelope formation wasunchanged. Thus, in submerged human keratino-cytes, vitamin C induces both increased sphingolipidproduction and enhancement of permeability barrierstructural markers. Key words: epidermis/ceramide/glu-cosylceramide/keratinocyte/vitamin C. J Invest Dermatol117:1307±1313, 2001

The three major lipids in mammalian stratum corneum(SC) are ceramides, cholesterol, and free fatty acids(FFA), which are organized into extracellular bilayersthat subserve the epidermal permeability barrierfunction (Elias and Menon, 1991). Prior studies have

provided indirect and direct evidence for the importance ofepidermal ceramides for the barrier: First, mammalian epidermissynthesizes unusually large amounts of ceramides (Schoephoerster etal, 1985; Uchida et al, 1988), which ultimately comprise 50% of SClipids by weight. Second, epidermal ceramides comprise aheterogeneous mixture of molecules, including some species that

are unique to the epidermis; e.g., w-hydroxyceramide (w-OH-Cer) and linoleate-containing w-(o-linoleoyl)-Cer (Wertz andDowning, 1983b; Bowser et al, 1985). w-OH-Cer are the primarylipids that are covalently attached to the corni®ed envelope (CE)(Wertz and Downing, 1986, 1987, 1989a), forming the so-calledcorneocyte lipid envelope (CLE) (Wertz et al, 1989b; Downing,1992; Elias et al, 2000), a structure that is required for normalbarrier function (Behne et al 2000). Third, epidermal ceramidesynthesis is stimulated in parallel with barrier recovery followingeither acute or chronic barrier perturbations (Holleran et al, 1991a,b), and epidermal ceramide synthesis is required for normal barrierrecovery after such perturbations (Holleran et al, 1991b). Finally,alterations in the content and distribution of epidermal ceramideoccur in skin diseases that display barrier abnormalities, such asatopic dermatitis (Imokawa et al, 1991; Yamamoto et al, 1991) andpsoriasis (Motta et al, 1993). Therefore, it is likely that both totalceramide content and the molecular heterogeneity of ceramides areimportant for normal permeability barrier homeostasis. Hence,identi®cation of factors that in¯uence the production of ceramidescould contribute both to a further understanding of normal barrierhomeostasis, and to potential therapeutic interventions that couldmodulate barrier formation.

Manuscript received November 5, 2000; revised July 3, 2001; acceptedfor publication July 17, 2001.

Reprint requests to: Dr. Walter M. Holleran, Dermatology Service &Research Unit (190), UCSF/Veterans Affairs Medical Center, 4150Clement Street, San Francisco, CA 94121, U.S.A. Email: walth@itsa.ucsf.edu

Abbreviations: Vit C, vitamin C; OH-GlcCer, hydroxyglucosylcera-mides; OH-Cer, hydroxyceramides; FFA, free fatty acids; SC, stratumcorneum; LB, lamellar bodies; CE, corni®ed envelope; CLE, corneocytelipid envelope.

0022-202X/01/$15.00 ´ Copyright # 2001 by The Society for Investigative Dermatology, Inc.

1307

Cultured human keratinocytes (CHK) and cultured murinekeratinocytes have been employed extensively to investigateepidermal growth and differentiation. It is well established that anincrease in exogenous calcium stimulates keratinocyte differenti-ation in vitro (Hennings et al, 1980). Moreover, when submergedcultures are grown at an air±medium interface, they generate a stillmore-differentiated, ``organotypic'' phenotype, including moreabundant ceramide content and a partially competent barrier(Madison et al, 1989; Fartasch and Ponec, 1994; Bouwstra et al,1995; Vicanova et al, 1998). Such lifted cultures, when they aresupplemented with vitamin C (Vit C), demonstrate an increase inceramides as a percentage of total lipids, as well as changes in thespectrum of ceramide and glucosylceramide fractions to moreclosely resemble ceramide content and distribution in vivo; i.e., anincrease in more polar ceramide species (Ponec et al, 1997).Whether Vit C affects sphingolipid synthesis, independent of airexposure and/or the extent of differentiation, has not beeninvestigated. In this study, we show that the addition of asupraphysiologic dose of Vit C to postcon¯uent CHK, grown inhigh calcium and serum stimulates bulk sphingolipid production,with enhancement of selected ceramide and glucosylceramidespecies, particularly of a- and w-hydroxylated sphingolipids. Thenet result is a spectrum of sphingolipids that closely resembles thatobserved in vivo. This stimulation of sphingolipid production isaccompanied, in turn, by increased formation of covalently boundlipids, as well as enhanced lamellar body (LB) production, secretion,and postsecretory membrane structural reorganization, markers ofpermeability barrier formation. Finally, we show that thesebiochemical and morphologic changes may be attributable to VitC-induced stimulation of the enzyme, ceramide synthase, the ®nalstep in ceramide synthesis.

MATERIALS AND METHODS

Cell culture Human epidermis was isolated from newborn foreskins byincubation in Dispase, and a suspension of keratinocytes was obtained byincubation in 10 mM ethylenediamine tetraacetic acid and subsequenttrypsinization as previously described (Pittelkow and Scott, 1986).Second passage cells were plated on to 100-mm plastic dishes in serum-free keratinocyte growth medium (Cascade Biologics, Portland, OR),containing 0.07 mM calcium, and grown to 90±100% con¯uence. Themedium then was switched to Dulbecco's modi®ed eagle's medium andHam F-12 (2:1, vol/vol), modi®ed from 3:1, vol/vol (as per Sando et al,1996), containing 1.2 mM calcium, supplemented with 10% fetal bovineserum, 10 mg insulin per ml, and 0.4 mg hydrocortisone per ml.Immediately after switching the medium, either Vit C (®nalconcentration of 50 mg per ml in phosphate-buffered saline) (Ponec et al,1997) or vehicle (phosphate-buffered saline) was added to parallelcultures, and cells were cultured for 9 d with medium changes everyother day. Preliminary studies also demonstrated that Vit Cconcentrations below 50 mg per ml induced lesser changes in ceramideand glucosylceramide production. Higher concentrations of Vit C (i.e.,100±500 mg per ml) did not further increase sphingolipid production.

Lipid analysis Total lipids were extracted from CHK and fromsunburn scale of normal human subjects (n = 3), as described previously(Bligh and Dyer, 1959), and separation of individual lipid species wasachieved by high-performance thin-layer chromatography (HPTLC),followed by quantitation by scanning densitometry as describedpreviously (Holleran et al, 1991b, 1997), as modi®ed (Ponec andWeerheim, 1990). Assignment of ceramides and glucosylceramides andtheir nomenclature are based on their comparative mobilities vs knownstandards (Wertz and Downing, 1983c). a-hydroxy- and nonhydroxy-ceramide and nonhydroxy-glucosylceramide standards were from Sigma(St Louis, MO). Phospholipids were separated using chloroform/methanol/acetic acid/water (50:30:8:4, vol/vol/vol). Triglycerides andcholesterol were fractionated by development in n-hexane/diethyl ether/acetic acid (80:20:1, vol/vol/vol). Covalently bound lipids were isolatedusing the method of Wertz et al (1989a), and resultant ceramide fractionswere quantitated by HPTLC-scanning densitometry, as above.

To assess rates of lipid synthesis, cells were incubated with 100 mMsodium acetate (5 mCi, [14C]sodium acetate; Amersham, ArlingtonHeights, IL) for 8 h. To measure hydroxyglucosylceramide (OH-GlcCer)and OH-Cer synthesis, lipids extracts ®rst were fractionated on an

aminopropyl column, following elution ®rst of phospholipids with n-hexane, by elution with chloroform±isopropanol (2:1, vol/vol) to obtainceramides. Further fractionation of ceramides were performed byHPTLC as above. For analysis of newly synthesized amide-linked fattyacids of glucosylceramides and ceramides, the combined glucosylcera-mide/ceramide fraction was scraped from the plate, hydrolyzed withaqueous methanolic HCl, treated with 2,2¢-dimethoxypropane, contain-ing methanolic HCl, and the fatty acid methyl esters were separated byHPTLC [in n-hexane/diethylether/acetic acid (70:30:1, vol/vol/vol), aspreviously reported (Wertz and Downing, 1983a)], and measured with aBeckman LS-1800 scintillation counter.

Serine palmitoyltransferase, glucosylceramide synthase, andceramide synthase activities For the enzyme activity determinationsbelow, CHK were ®rst washed with phosphate-buffered saline (3 3),harvested with a cell scraper, homogenized (15 s 3 3) using a PolytronPCU2 Tissue Homogenizer (Kinematica, GmbH, Lucerne, Switzerland),and then sonicated (35%, 10 s 3 2) using a Fisher Sonic dismembranator(Artec Corp, Farmingdale, NY). Protein content of the resultanthomogenates were determined using BCA protein assay method (Pierce,Rockford, IL), using bovine serum albumin as standard. Our proceduresfor the isolation and quantitation of serine palmitoyltransferase fromcultured keratinocytes have been described in detail elsewhere (Holleranet al, 1990). The reaction product, [3H]-3-ketodihydrosphinganine wasisolated and counted, as above. The assay for glucosylceramide synthaseactivity was described recently (Chujor et al, 1998). Incorporation of[14C]uridine diphosphate glucose (American Radiolabeled Chemical, StLouis, MO; Spec. Act. 20,000 dpm per nmol) into glucosylceramideswas determined by counting, as above. The assay for ceramide synthasewas described recently (Wang and Merrill, 2000). Brie¯y, CHKhomogenates (1.5 mg of protein) were incubated with 50 mM[14C]lignoceric acid (C24:0) (American Radiolabeled Chemical, Spec.Act. 9±10,000 dpm per nmol), 1 mM coenzyme A, 5 mM adenosinetriphosphate, 1 mM NADPH, 1 mM NADH and 50 mM HEPESbuffer, pH 7.4 at 37°C (60 min). Enzyme reactions were terminated byadding chloroform±methanol (1:2, vol/vol). Incorporation of[14C]lignoceric acid into the amide-linked fatty acid of ceramides wasanalyzed as described above.

CE formation The formation of CE as a marker for the terminaldifferentiation of keratinocytes was determined by the method of Kinget al (1986), as previously modi®ed (Pillai et al, 1988). CHK wereincubated in Dulbecco's modi®ed eagle's medium-Ham F-12 asdescribed above containing [35S]-L-methionine (1.5 mCi medium per ml,ICN Radiochemicals Inc, Irvine, CA) for 46 h. The cells were rinsedwith phosphate-buffered saline and cultured in Dulbecco's modi®edeagle's medium-Ham F-12 containing 10 mM ionomycin for 2 h. Thecross-linked, 2% sodium dodecyl sulfate and 25 mM dithiothreitol-insoluble pellets were collected on ®lters and quantitated by liquidscintillation spectroscopy.

Electron microscopy CHK samples were ®rst ®xed in situ by theaddition of half-strength Karnovsky's ®xative to the Petri dish for 1 h(room temperature), transferred to a glass tube, further ®xed overnight(4°C), and post®xed in both ruthenium tetroxide and 2% aqueousosmium tetroxide, both containing 1.5% potassium ferrocyanide, aspreviously described (Hou et al, 1991). After ®xation, all samples weredehydrated in graded ethanol solutions, and embedded in an Epon-epoxymixture. Ultrathin sections were examined, with or without furthercontrasting with lead citrate, in an electron microscope (Zeiss 10A, CarlZeiss, Thornwood, NY) operated at 60 kV. Quantitative morphometry(i.e., counting of LB numbers per cell) was performed on ®ve separate®elds from ®ve control and nine Vit C-treated sample sections; thenumbers of LB were counted in a total of 47 control cells and 86 Vit C-treated cells.

Statistical analysis Statistical analyses were performed using a Mann±Whitney test or an unpaired Student's t test, as appropriate.

RESULTS

Glucosylceramide and ceramide content increase with Vit Csupplementation We ®rst determined the effects of 9 d of Vit Csupplementation (50±500 mg per ml) on total lipid, protein, andDNA content in submerged, postcon¯uent CHK. Nine days of VitC treatment did not alter total cellular lipid, protein, or DNAcontent as compared with vehicle controls (not shown). In contrast,the total content of both glucosylceramide and ceramide contentincreased signi®cantly in 9 d Vit C-supplemented vs control CHK.

1308 UCHIDA ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

The total glucosylceramide content of Vit C-treated CHKincreased 1.8-fold vs control levels after 9 d (49.96 6 SD 0.80 vs28.95 6 0.20 mg per 100 mm cultured dish, respectively; p< 0.0002). Glucosylceramides in Vit C supplemented vs controlCHK could be separated into four major fractions by HPTLC[acylGlcCer (GlcCer-A; the most hydrophobic component),GlcCer-B, GlcCer-C, and GlcCer-D (the most hydrophiliccomponent)] (Wertz and Downing, 1983c; Wertz et al, 1984).GlcCer-A, GlcCer-B, and GlcCer-D, but not GlcCer-C, increasedsigni®cantly in Vit C-treated vs vehicle-treated CHK (Fig 1A; i.e.,2.79-fold, 1.46-fold, and 2.09-fold; p < 0.001 vs vehicle,respectively).

The total ceramide levels in Vit C-treated cells also increased1.7-fold over control levels after 9 d (i.e., 66.54 6 6.50 vs40.50 6 4.54 mg per dish; p < 0.02). Moreover, speci®c ceramidespecies increased, including Cer 2 [N-nonhydroxyacyl sphingosine(Cer-NS); 1.23-fold, p < 0.02], Cer 4 [N-(O-linoleoyl) w-hydroxyacyl dihydroxysphingosine (Cer-EOH); 5.6-fold, p< 0.001], and Cer 5 [N-a-hydroxyacyl sphingosine (Cer-AS);2.87-fold, p < 0.01], whereas Cer 1 [N-(O-linoleoyl) w-hydro-xyacyl sphingosine (Cer-EOS) and Cer 3 [N-nonhydroxyacylphytosphingosine (Cer-NP)] levels did not change (Fig 1B).[Ceramide nomenclature as per Motta et al (1993) and Robson etal (1994).] Neither Cer 6 nor 7 were detected in controlpreparations, but both became prominent in Vit C-treated cells

(Fig 1B). Finally, sphingomyelin content did not change following9 d of Vit C supplementation (not shown).

Vit C increases the synthesis of bulk and speci®c sphingo-lipids We next studied the effects of Vit C supplementation onglucosylceramide and ceramide synthesis. First, incorporation of[14C]acetate into both total glucosylceramides and total ceramidesincreased signi®cantly (i.e., 3.55-fold and 1.76-fold, respectively; p< 0.001 and 0.01), changes that mirrored the Vit C-inducedalterations in lipid content (cf. Fig 1). The corresponding,individual glucosylceramide species also increased signi®cantly(Fig 2A): GlcCer-A, 5.94-fold (p < 0.001); GlcCer-B, 3.03-fold(p < 0.0001); and GlcCer-D, 3.72-fold (p < 0.0001). But as withthe lipid content (above), the increase in GlcCer-C production didnot achieve statistical signi®cance (Fig 2A). Vit C also increasedthe synthesis of the four most polar ceramide fractions (Fig 2B);i.e., Cer 4±7 (e.g., Cer 4, 5.08-fold, p < 0.01; Cer 5, 2.59-fold, p< 0.01). In fact, Cer 6 and 7 syntheses were only evident in Vit C-treated CHK (Fig 2B). Finally, consistent with the content studies,sphingomyelin synthesis also was not altered by Vit Csupplementation (Fig 2C cf. Fig 1B). These results indicate thatthe Vit C-induced changes in sphingolipid content re¯ect increasedsynthesis of bulk and speci®c ceramide species.

A characteristic of epidermal sphingolipids is that the more polarglucosylceramide and ceramide species are enriched in a- and

Figure 2. Vit C stimulates glucosylceramide,ceramide, and cholesterol syntheses innormal human keratinocytes cultured in asubmerged system. Cells were cultured inserum-containing medium with or without 50 mgVit C per ml for 9 d. To assess lipid synthesis,cells were incubated with [14C]sodium acetate for8 h. Panel A:glucosylceramide; panel B:ceramide;panel C:phospholipids (LPC, lysophosphatidyl-choline; SM, sphingomyelin; PC, phosphatidyl-choline; PS, phosphatidylserine; PI, phos-phatidylinositol; PE, phosphatidylethanolamine);panel D:cholesterol and triacylglycerol. Valuesrepresent mean [14C]acetate incorporation intolipids per 100 mm cultured dish (n = 3; 6 SD).*p < 0.01; **p < 0.001; ***p < 0.0001 each vsvehicle-treated control.

Figure 1. Vit C increases glucosylceramideand ceramide content in normal humankeratinocytes cultured in a submergedsystem. Cells were cultured in serum-containingmedium with or without 50 mg Vit C per ml for9 d. Panel A:glucosylceramides; panel B:ceramides.Values represent mean mg of sphingolipids per100 mm cultured dish (n = 3; 6 SD). *p < 0.02;**p < 0.01; ***p < 0.001, each vs vehicle-treatedcontrol, respectively.

VOL. 117, NO. 5 NOVEMBER 2001 VITAMIN C STIMULATES KERATINOCYTE SPHINGOLIPID PRODUCTION 1309

w-hydroxylated, amide-linked fatty acids. We next assessedwhether Vit C supplementation enhances [14C]acetate incorpor-ation into hydroxylated vs nonhydroxylated, amide-linked fattyacids of these sphingolipid species. Whereas Vit C stimulated theincorporation of radiolabel into all nonhydroxylated, amide-linkedfatty acids in sphingolipids by approximately 1.4-fold over vehiclecontrol values (i.e., 14.8 6 0.5 vs 10.2 6 2.0 CPMx10,000 per100 mm dish; p < 0.05), the synthesis of hydroxylated fatty acid-containing ceramide increased still further (i.e., a-OH fatty acid:7.5-fold, 1.75 6 0.17 vs 0.5 6 0.15, p < 0.0001; and w-OH fattyacid: 3.5-fold, 3.5 6 0.2 vs 0.47 6 0.08; p < 0.001). These resultsshow that Vit C stimulates not only sphingolipid production, butalso the preferential formation of hydroxylated glucosylceramideand ceramide species.

Finally, we next assessed whether the synthesis of nonsphingo-lipid lipids also changes with Vit C supplementation. Vit Cincreased cholesterol synthesis 1.5-fold (Fig 2D; p < 0.01), anddecreased triacylglycerol (Fig 2D; p < 0.01) and phosphatidylcho-line production (» 30% below control; Fig 2C). Together, theseresults demonstrate that Vit C supplementation modulates thesynthesis of other epidermal lipids to achieve a pro®le that furtherre¯ects SC lipid pro®les (Lampe et al, 1983a, b).

Vit C stimulates ceramide synthase To assess the metabolicstep(s) at which Vit C might exert its effects on sphingolipidsynthesis, we next measured the activities of three key enzymes ofepidermal sphingolipid metabolism (i.e., serine palmitoyltransferase,ceramide synthase, and glucosylceramide synthase) following Vit Csupplementation. Whereas neither serine palmitoyltransferase norglucosylceramide synthase activities increased during the 9 d of VitC treatment (Table I), ceramide synthase activity increasedsigni®cantly in Vit C-treated cells vs vehicle-treated cells (2.3-fold; p < 0.01). (Activity includes increased formation of both a-and w-OH-containing, as well as nonhydroxylated ceramidespecies.) These results show that the N-acylation of

dihydrosphingosine by ceramide synthase, leading to thegeneration of ceramides, is a step in ceramide production that isupregulated by Vit C.

Vit C increases the amount of w-OH-Cer bound to theCE To evaluate the potential consequences of increasedsphingolipid production by Vit C treatment for epidermalstructure and function, we ®rst determined the levels of w-OH-Cer that form a modi®ed membrane bilayer, external to the CE.Although the levels of total protein-bound (i.e., base-hydrolyzable)lipids were elevated at 9 d (1.8-fold over controls; p < 0.02), afurther increase was evident in 12 d Vit C-treated cells (4.70-fold vslevels in control cells; p < 0.001) (Table II). In addition, althoughonly N-w-OH acylsphingosine was observed in control cells at12 d, N-w-OH acyldihydroxysphingosine and w-OH FFA alsowere contained in Vit C-treated cells. As the backbone structures ofN-w-OH acylsphingosine and N-w-OH acyldihydroxysphingosineare the same as Cer 1 and Cer 4, respectively, an increase in N-w-OH acyldihydroxysphingosine could be due to Vit C-inducedproduction of Cer 4 (cf. Figure 1B); however, the ratio of Cer Ato Cer B in Vit C-treated CHK is lower in human SC (i.e., CHK,4.1 vs SC 1.7±2.2) (Wertz et al, 1989a; Elias et al, 2000). Inaddition, w-OH-GlcCer, which are precursors of N-w-OHacylsphingosine and N-w-OH acyldihydroxysphingosine, alsoincreased in Vit C-treated cells. Finally, the boundnonhydroxylated fatty acids RM also increased in Vit C-treatedcells (1.48-fold vs levels in control cells: 16.97 6 0.65 vs11.37 6 0.36 mg per 100 mm culture dish, respectively; p< 0.001). Together, these results demonstrate that Vit C enhancesthe amount of ceramide bound to the CE, even under submergedconditions.

Vit C treatment increases LB production and secretion, andCLE formation We next assessed whether the increasedsphingolipid production in Vit C supplemented cultures is linkedto permeability barrier formation. Nine day postcon¯uent CHKtreated with vehicle or serum alone exhibited few LB, and little orno secreted lamellar contents (not shown). In contrast, CHKtreated for 9 d with Vit C demonstrated numerous suprabasal cellswith large numbers of LB, but few of these organelles exhibitedreplete, internal lamellar contents (Fig 3, panel B, arrows).Morphometric analysis of 9 d cultures revealed signi®cantlyincreased LB numbers (p < 0.003) in Vit C-treated vs controlcells; i.e., 4.3 6 0.5 vs 2.3 6 0.6 LB per cell for Vit C and controlsamples, respectively. Although foci of newly secreted lamellae alsooccurred in the 9 d Vit C cultures (Fig 3, panel A, arrows), nomature, lamellar structures were evident. Prolongation ofpostcon¯uent growth from 9 to 12 d resulted both in a furtherampli®cation in the number of cytosolic LB, and the appearance ofa large proportion of organelles with replete internal contents (notshown); however, mature extracellular lamellar membranestructures still were not evident at 12 d (not shown). In contrast,whereas 14 d cultures displayed no further increase in LB numbersor contents, these cultures frequently displayed extracellular

Table I. In¯uence of Vit C on sphingolipid syntheticenzymesa

Enzyme activity(pmol per mg protein per min)

Control +Vit C

Serine palmitoyltransferase 148.5 6 12.20b 117.5 6 24.70Ceramide synthase 0.50 6 0.05 1.15 6 0.06c

Glucosylceramide synthase 30.80 6 1.88 27.52 6 0.44

aCells were cultured serum-containing medium without (Control) or with VitC for 9 d.

bValues represent mean 6 SD (n = 3±4).cp < 0.01 vs control.

Table II. Vit C increases the amount of CLEa

Covalently bound lipidb

N-w-OH-Cer-Ac w-OH-Cer-B w-OH-GlcCer w-OH FFA Total

Control 1.24 6 0.31 NDd trace trace 1.17 6 0.24Vit C 4.35 6 0.45 1.05 6 0.12 0.47 6 0.08 1.00 6 0.07 5.48 6 0.52e

aCells were cultured serum-containing medium with or without Vit C for 12 d.bCovalently bound lipids recovered following base-hydrolysis (see Materials and Methods); values represent the meanmg per 100 mm cultured dish 6 SD (n = 3).cDe®nitions: w-OH-Cer-A: N-w-OH acyl sphingosine, w-OH-Cer-B: N-w-OH acylhydroxy-sphingosine (Robson et al, 1994), w-OH-GlcCer: w-OH acylglucosyl-

ceramide (Doering et al, 1999a; Doering et al, 1999b; Uchida et al, 1999).dNot detected.ep < 0.001 vs control.

1310 UCHIDA ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

lamellar bilayers (not shown). These results show that Vit Csupplementation leads not only to increased sphingolipidproduction, but also to apparent downstream accumulation ofsphingolipids in barrier-related structures.

Although some CLE became evident around corneocytes inboth control and Vit C-supplemented cultures by day nine, thesestructures became much more prominent in day 12 and 14 cultures(not shown). Together, these results show that the Vit C stimulatedincrease in covalently bound lipids is re¯ected in enhanced CLEformation. Finally, as ceramide itself can induce differentiation(Wakita et al, 1994), we next assessed whether Vit C treatmentsimultaneously produced changes in CHK differentiation; how-ever, Vit C treatment did not signi®cantly increase CE formation(Vit C: 205 6 8 vs vehicle: 173 6 28 cpm per mg protein; notsigni®cant). These results show that the Vit C-induced increase insphingolipid synthesis is not linked either to prior or concurrentchanges in CHK differentiation, re¯ected by CE formation.

DISCUSSION

The extracellular lamellar membranes in the SC of mammalianepidermis, which are enriched in cholesterol, ceramides, and FFA,are an essential component of the epidermal permeability barrier(Elias and Menon, 1991). Among these lipids, ceramides constitutenearly 50% of the lipid mass in the SC (Elias and Menon, 1991).Although the molecular structures of epidermal glucosylceramidesand ceramides have been studied extensively (Wertz and Downing,1983b, >c; Abraham et al, 1985; Bowser et al, 1985; Ponec et al,1988; Uchida et al, 1988, 2000; Hamanaka et al, 1989; Motta et al,1993; Robson et al, 1994), the mechanisms that regulate ceramidesynthesis still are not known. The heterogeneity of ceramidestructures also plays a crucial part in the formation of lamellarmembrane structures (Bouwstra et al, 1998). The formation of theselamellar structures occurs in four sequential steps: (i) the synthesis of

these lipids as their polar precursors (e.g., glucosylceramides andsphingomyelin); (ii) the packaging of these lipids into nascent LB;(iii) the extrusion of LB contents into the extracellular spacesbetween the stratum granulosum and SC; and ®nally (iv) theprocessing of these secreted lipids into their products, leading to theformation of bilayer structures. Although these steps parallelkeratinocyte differentiation, sphingolipid production and LBformation/secretion remain very low in immersed keratinocytecultures, even when such cultures display ample evidence ofdifferentiation (Pillai et al, 1988; Ponec et al, 1988; Fartasch andPonec, 1994).

Over the past decade, organotypic culture systems in whichsubmerged keratinocytes are lifted to the air±medium interface,have been developed to replicate epidermis in vitro further. Indeed,such reconstituted organotypic models display variable sphingolipidproduction, LB formation, and appearance of extracellular lamellarbilayers (Ponec et al, 1988; Williams et al, 1988; Madison et al,1990; Boyce and Williams, 1993; Fartasch and Ponec, 1994). Afurther increase in sphingolipid and LB production occurs whenthe external humidity surrounding organotypic cultures is loweredfurther (Mak et al, 1991). Likewise, organ cultures of fetal skinaccelerate barrier formation, attributable to increased sphingolipidproduction and LB generation, when they are grown at an air±medium interface (Hanley et al, 1997). Although these studiessuggest that increased sphingolipid/LB production is linkedspeci®cally to a reduction in external humidity, LB also occur infollicular epithelia (Schmitz and MuÈller, 1991), in epidermalinclusion cysts (Madison et al, 1987), in mucosal epithelia(Schmitz and MuÈller, 1991), and in marine mammalian epidermis(Menon et al, 1986), where stimulation from a lowered externalhumidity is not operative. Furthermore, in these air-exposedculture systems, neither the total sphingolipid content nor thedistribution of ceramides/glucosylceramides achieves that found

Figure 3. Vit C increases the number of LBand extrusion of their contents. Electronmicrographs of normal human keratinocytescultured in serum-containing medium with 50 mgVit C per ml for 9 d. Panel A: foci of newlysecreted lamellae are evident, whereas matureextracellular lamellar structures were not present.Panel B: abundant number of LB were present inthe cytosol of cells. Panel A, ruthenium tetroxide;panel B, osmium tetroxide; Mag bars = 0.25 mm.

VOL. 117, NO. 5 NOVEMBER 2001 VITAMIN C STIMULATES KERATINOCYTE SPHINGOLIPID PRODUCTION 1311

in vivo. Hence, other factors, such as speci®c enzyme cofactor(s) oractivators of key transcription factors, could regulate or coregulatesphingolipid/LB production. Recently, supplementation of anorganotypic culture system with Vit C was shown both to increasetotal ceramide content and to alter ceramide distribution to moreclosely mirror that observed in vivo (Ponec et al, 1997); however,these studies left unanswered whether Vit C increases sphingolipidproduction in relation to differentiation, air exposure, and/or byunrelated mechanism(s). We show here that increased productionof epidermal-type ceramides and glucosylceramides in Vit C-supplemented CHK occurs independently of air exposure; i.e., insubmerged cultures. Moreover, although keratinocyte differenti-ation per se reportedly stimulates ceramide/glucosylceramide syn-thesis (Madison et al, 1989; Sando et al, 1996), we show here thatCE formation, a marker of keratinocyte terminal differentiation(Sun and Green, 1976; Pillai et al, 1988), does not changesigni®cantly with Vit C treatment. Thus, the observed, Vit C-induced elevation in ceramide and glucosylceramide syntheses isnot likely to be linked to a prior or concurrent increase ofkeratinocyte differentiation. Instead, Vit C treatment appears tostimulate sphingolipid production directly, increasing not only bulksphingolipids through an increase in ceramide synthase, butparticularly those sphingolipid subclasses that contain a- and w-hydroxylated species. Some nonhydroxylated sphingolipids alsoincreased with Vit C treatment, but to a lesser extent. The netresult is a spectrum that closely resembles that in Vit C-treatedorganotypic cultures (Ponec et al, 1997).

The basis for the overall increase in ceramide production appearsto be stimulation in the activity of a key enzyme in ceramidegeneration; i.e., ceramide synthase. Whether regulation occurs at atranscriptional and/or post-translational level could not be deter-mined, because the responsible gene for ceramide synthase has yetto be cloned. Whereas it is possible that the reducing environment,resulting from Vit C treatment, might favor post-transcriptionalregulation of ceramide synthesis, it is important to note thataddition of other reducing agents, such as vitamin E, did notstimulate sphingolipid production in organotypic CHK (Ponec etal, 1997). Thus, it is more likely that Vit C is a speci®c cofactor forceramide synthase, as it is for key enzymes of collagen synthesis(Barnes and Kodicek, 1972). In contrast, Vit C treatment did notappear to alter either serine palmitoyltransferase or glucosylcer-amide synthase activities, two other key enzymes of ceramide andglucosylceramide synthesis, respectively. It is intriguing that theactivity of serine palmitoyltransferase, reported as the rate-limitingstep in epidermal sphingolipid production (Holleran et al, 1991a,1991b), was unaffected by Vit C supplementation. This resultsuggests either that constitutive levels of these enzymes may suf®ceto accommodate increased ceramide and glucosylceramide pro-duction (i.e., via substrate regulation), or that keratinocyte cultureconditions may alter the relative importance of contributingenzymes in the sphingolipid synthetic pathway. Both of thesepossibilities merit further investigation.

Vit C also could in¯uence other relevant enzymes nonspeci®c-ally as a reducing agent. Such activity could account for thedisproportionate increase in a- and w-OH-Cer levels observed inVit C-treated cultures. Previous structural studies have shown thatthe amide-linked fatty acids of Cer 4 through 7 are hydroxylated atspeci®c positions (Abraham et al, 1985; Bowser et al, 1985; Robsonet al, 1994; Stewart and Downing, 1999). As Vit C increased notonly the production of hydroxylated sphingolipids, but also thelevels of certain nonhydroxylated sphingolipids its effects on fattyacid hydroxylation appear to be distinct from those on ceramidesynthase.

The proportion of three major lipid species, i.e., ceramides,cholesterol, and FFA, is important for the formation of lamellarmembrane structures in mammalian SC, and the formation of thebarrier to water loss (Bouwstra et al, 1996; Man et al, 1996).Addition of serum and epidermal growth factor to organotypickeratinocyte cultures both elevates triacylglycerol, and reducesceramide and FFA levels (Gibbs et al, 1997). Conversely, Vit C

supplementation, while increasing ceramides, also reduced triacyl-glycerol in an organotypic model (Ponec et al, 1997). As Vit Cinduced similar normalization of triacylglycerol levels undersubmerged conditions (this study), this effect of Vit C appearsindependent of air exposure. Although phospholipids are known tobe key precursors in the generation of SC FFA (Mao-Qiang et al,1995), the role of triglycerides in this process remains unresolved.Thus, the present results suggest that Vit C may alter epidermal orSC lipase activity leading to the further normalization of barrierlipid composition.

This work was supported by National Institutes of Health Grant AR 39448. Dr

Quiec was supported as visiting postdoctoral research fellows by L'Oreal Research

(France). The authors thank Ms. Debbie Crumrine for expert assistance with

electron microscopy, and Ms. Vanessa Collins (Undergraduate student, University

College Cork, Ireland) for valuabe technical assistance.

REFERENCES

Abraham W, Wertz PW, Downing DT: Linoleate-rich acylglucosylceramides of pigepidermis: structure determination by proton magnetic resonance. J Lipid Res26:761±766, 1985

Barnes MJ, Kodicek E: Biological hydroxylations and ascorbic acid with specialregard to collagen metabolism. Vitam Horm 30:1±43, 1972

Behne M, Uchida Y, Seki T, de Montellano PO, Elias PM, Holleran WM: Omega-hydroxyceramides are required for corneocyte lipid envelope (CLE) formationand normal epidermal permeability barrier function. J Invest Dermatol 114:185±192, 2000

Bligh EG, Dyer WJ: A rapid method of total lipid extraction and puri®cation. Can JBiochem Physiol 37:911±917, 1959

Bouwstra JA, Gooris GS, Weerheim A, Kempenaar J, Ponec M: Characterization ofstratum corneum structure in reconstructed epidermis by X-ray diffraction. JLipid Res 36:496±504, 1995

Bouwstra JA, Gooris GS, Cheng K, Weerheim A, Bras W, Ponec M: Phase behaviorof isolated skin lipids. J Lipid Res 37:999±1011, 1996

Bouwstra JA, Gooris GS, Dubbelaar FE, Weerheim AM, Ijzerman AP, Ponec M:Role of ceramide 1 in the molecular organization of the stratum corneumlipids. J Lipid Res 39:186±196, 1998

Bowser PA, Nugteren DH, White RJ, Houtsmuller UM, Prottey C: Identi®cationisolation and characterization of epidermal lipids containing linoleic acid.Biochim Biophys Acta 834:419±428, 1985

Boyce ST, Williams ML: Lipid supplemented medium induces lamellar bodies andprecursors of barrier lipids in cultured analogues of human skin. J InvestDermatol 101:180±184, 1993

Chujor CS, Feingold KR, Elias PM, Holleran WM: Glucosylceramide synthaseactivity in murine epidermis: quantitation, localization, regulation, andrequirement for barrier homeostasis. J Lipid Res 39:277±285, 1998

Doering T, Holleran WM, Potratz A, Vielhaber G, Elias PM, Suzuki K, Sandhoff K:Sphingolipid activator proteins are required for epidermal permeability barrierformation. J Biol Chem 274:11038±11045, 1999a

Doering T, Proia RL, Sandhoff K: Accumulation of protein-bound epidermalglucosylceramides in beta-glucocerebrosidase de®cient type 2 Gaucher mice.FEBS Lett 447:167±170, 1999b

Downing DT: Lipid and protein structures in the permeability barrier of mammalianepidermis. J Lipid Res 33:301±313, 1992

Elias PM, Menon GK: Structural and lipid biochemical correlates of the epidermalpermeability barrier. Adv Lipid Res 24:1±26, 1991

Elias PM, Fartasch M, Crumrine D, Behne M, Uchida Y, Holleran WM: Origin ofthe corneocyte lipid envelope (CLE): observations in harlequin ichthyosis andcultured human keratinocytes. J Invest Dermatol 115:765±769, 2000

Fartasch M, Ponec M: Improved barrier structure formation in air-exposed humankeratinocyte culture systems. J Invest Dermatol 102:366±374, 1994

Gibbs S, Vicanova J, Bouwstra J, Valstar D, Kempenaar J, Ponec M: Culture ofreconstructed epidermis in a de®ned medium at 33 degrees C shows a delayedepidermal maturation, prolonged lifespan and improved stratum corneum. ArchDermatol Res 289:585±595, 1997

Hamanaka S, Asagami C, Suzuki M, Inagaki F, Suzuki A: Structure determination ofglucosyl beta 1-N-(omega-O-linoleoyl)-acylsphingosines of human epidermis.J Biochem (Tokyo) 105:684±690, 1989

Hanley K, Jiang Y, Elias PM, Feingold KR, Williams ML: Acceleration of barrierontogenesis in vitro through air exposure. Pediatr Res 41:293±299, 1997

Hennings H, Michael D, Cheng C, Steinert P, Holbrook K, Yuspa SH: Calciumregulation of growth and differentiation of mouse epidermal cells in culture.Cell 19:245±254, 1980

Holleran WM, Williams ML, Gao WN, Elias PM: Serine-palmitoyl transferaseactivity in cultured human keratinocytes. J Lipid Res 31:1655±1661, 1990

Holleran WM, Feingold KR, Man MQ, Gao WN, Lee JM, Elias PM: Regulation ofepidermal sphingolipid synthesis by permeability barrier function. J Lipid Res32:1151±1158, 1991a

Holleran WM, Man MQ, Gao WN, Menon GK, Elias PM, Feingold KR:

1312 UCHIDA ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Sphingolipids are required for mammalian epidermal barrier function.Inhibition of sphingolipid synthesis delays barrier recovery after acuteperturbation. J Clin Invest 88:1338±1345, 1991b

Holleran WM, Uchida Y, Halkier-Sorensen L, Haratake A, Hara M, Epstein JH,Elias PM: Structural and biochemical basis for the UVB-induced alterations inepidermal barrier function. Photodermatol Photoimmunol Photomed 13:117±128,1997

Hou SY, Mitra AK, White SH, Menon GK, Ghadially R, Elias PM: Membranestructures in normal and essential fatty acid-de®cient stratum corneum:characterization by ruthenium tetroxide staining and x-ray diffraction. JInvest Dermatol 96:215±223, 1991

Imokawa G, Abe A, Jin K, Higaki Y, Kawashima M, Hidano A: Decreased level ofceramides in stratum corneum of atopic dermatitis: an etiologic factor in atopicdry skin? J Invest Dermatol 96:523±526, 1991

King I, Mella SL, Sartorelli AC: A sensitive method to quantify the terminaldifferentiation of cultured epidermal cells. Exp Cell Res 167:252±256, 1986

Lampe MA, Burlingame AL, Whitney J, Williams ML, Brown BE, Roitman E, EliasPM: Human stratum corneum lipids: characterization and regional variations. JLipid Res 24:120±130, 1983a

Lampe MA, Williams ML, Elias PM: Human epidermal lipids: characterization andmodulations during differentiation. J Lipid Res 24:131±140, 1983b

Madison KC, Swartzendruber DC, Wertz PW, Downing DT: Presence of intactintercellular lipid lamellae in the upper layers of the stratum corneum. J InvestDermatol 88:714±718, 1987

Madison KC, Swartzendruber DC, Wertz PW, Downing DT: Murine keratinocytecultures grown at the air/medium interface synthesize stratum corneum lipidsand ``recycle'' linoleate during differentiation. J Invest Dermatol 93:10±17, 1989

Madison KC, Swartzendruber DC, Wertz PW, Downing DT: Sphingolipidmetabolism in organotypic mouse keratinocyte cultures. J Invest Dermatol95:657±664, 1990

Mak VH, Cumpstone MB, Kennedy AH, Harmon CS, Guy RH, Potts RO: Barrierfunction of human keratinocyte cultures grown at the air±liquid interface. JInvest Dermatol 96:323±327, 1991

Man MM, Feingold KR, Thornfeldt CR, Elias PM: Optimization of physiologicallipid mixtures for barrier repair. J Invest Dermatol 106:1096±1101, 1996

Mao-Qiang M, Feingold KR, Jain M, Elias PM: Extracellular processing ofphospholipids is required for permeability barrier homeostasis. J Lipid Res36:1925±1935, 1995

Menon GK, Grayson S, Brown BE, Elias PM: Lipokeratinocytes of the epidermis of acetacean (Phocena phocena). Histochemistry, ultrastructure, and lipidcomposition. Cell Tissue Res 244:385±394, 1986

Motta S, Monti M, Sesana S, Caputo R, Carelli S, Ghidoni R: Ceramidecomposition of the psoriatic scale. Biochim Biophys Acta 1182:147±151, 1993

Pillai S, Bikle DD, Hincenbergs M, Elias PM: Biochemical and morphologicalcharacterization of growth and differentiation of normal human neonatalkeratinocytes in a serum-free medium. J Cell Physiol 134:229±237, 1988

Pittelkow MR, Scott RE: New techniques for the in vitro culture of human skinkeratinocytes and perspectives on their use for grafting of patients withextensive burns. Mayo Clinic Proc 61:771±777, 1986

Ponec M, Weerheim A: Retinoids and lipid changes in keratinocytes. MethodsEnzymol 190:30±41, 1990

Ponec M, Weerheim A, Kempenaar J, Mommaas AM, Nugteren DH: Lipidcomposition of cultured human keratinocytes in relation to theirdifferentiation. J Lipid Res 29:949±961, 1988

Ponec M, Weerheim A, Kempenaar J, Mulder A, Gooris GS, Bouwstra J, MommaasAM: The formation of competent barrier lipids in reconstructed human

epidermis requires the presence of vitamin C. J Invest Dermatol 109:348±355,1997

Robson KJ, Stewart ME, Michelsen S, Lazo ND, Downing DT: 6-Hydroxy-4-sphingenine in human epidermal ceramides. J Lipid Res 35:2060±2068, 1994

Sando GN, Howard EJ, Madison KC: Induction of ceramide glucosyltransferaseactivity in cultured human keratinocytes. Correlation with culturedifferentiation. J Biol Chem 271:22044±22051, 1996

Schmitz G, MuÈller G: Structure and function of lamellar bodies, lipid-proteincomplexes involved in storage and secretion of cellular lipids. J Lipid Res32:1539±1570, 1991

Schoephoerster RT, Wertz PW, Madison KC, Downing DT: A survey of polar andnon-polar lipids of mouse organs. Comp Biochem Physiol [B] 82:229±232, 1985

Stewart ME, Downing DT: A new 6-hydroxy-4-sphingenine-containing ceramidein human skin. J Lipid Res 40:1434±1439, 1999

Sun TT, Green H: Differentiation of the epidermal keratinocyte in cell culture:formation of the corni®ed envelope. Cell 9:511±521, 1976

Uchida Y, Iwamori M, Nagai Y: Distinct differences in lipid composition betweenepidermis and dermis from footpad and dorsal skin of guinea pigs. Jpn J ExpMed 58:153±161, 1988

Uchida Y, Sidransky E, Ginns EI, Elias PM, Holleran WM: Formation of the lipid-bound envelope (LBE): Insights from glucocerebrosidase-de®cient Gauchermouse epidermis. J Invest Dermatol 112:543a, 1999

Uchida Y, Hara M, Nishio H, et al: Epidermal sphingomyelins are precursors forspeci®c ceramides in the stratum corneum. J Lipid Res 41, 2071±2082, 2000

Vicanova J, Boelsma E, Mommaas AM, et al: Normalization of epidermal calciumdistribution pro®le in reconstructed human epidermis is related toimprovement of terminal differentiation and stratum corneum barrierformation. J Invest Dermatol 111:97±106, 1998

Wakita H, Tokura Y, Yagi H, Nishimura K, Furukawa F, Takigawa M: Keratinocytedifferentiation is induced by cell-permeant ceramides and its proliferation ispromoted by sphingosine. Arch Dermatol Res 286:350±354, 1994

Wang E, Merrill AH Jr: Ceramide synthase. Methods Enzymol 311:15±21, 2000Wertz PW, Downing DT: Acylglucosylceramides of pig epidermis: structure

determination. J Lipid Res 24:753±758, 1983aWertz PW, Downing DT: Ceramides of pig epidermis: structure determination. J

Lipid Res 24:759±765, 1983bWertz PW, Downing DT: Glucosylceramides of pig epidermis: structure

determination. J Lipid Res 24:1135±1139, 1983cWertz PW, Downing DT, Freinkel RK, Traczyk TN: Sphingolipids of the stratum

corneum and lamellar granules of fetal rat epidermis. J Invest Dermatol 83:193±195, 1984

Wertz PW, Downing DT: Covalent attachment of omega-hydroxyacid derivatives toepidermal macromolecules: a preliminary characterization. Biochem Biophys ResCommun 137:992±997, 1986

Wertz PW, Downing DT: Covalently bound omega-hydroxyacylsphingosine in thestratum corneum. Biochim Biophys Acta 917:108±111, 1987

Wertz PW, Madison KC, Downing DT: Covalently bound lipids of human stratumcorneum. J Invest Dermatol 92:109±111, 1989a

Wertz PW, Swartzendruber DC, Kitko DJ, Madison KC, Downing DT: The role ofthe corneocyte lipid envelopes in cohesion of the stratum corneum. J InvestDermatol 93:169±172, 1989b

Williams ML, Brown BE, Monger DJ, Grayson S, Elias PM: Lipid content andmetabolism of human keratinocyte cultures grown at the air±medium interface.J Cell Physiol 136:103±110, 1988

Yamamoto A, Serizawa S, Ito M, Sato Y: Stratum corneum lipid abnormalities inatopic dermatitis. Arch Dermatol Res 283:219±223, 1991

VOL. 117, NO. 5 NOVEMBER 2001 VITAMIN C STIMULATES KERATINOCYTE SPHINGOLIPID PRODUCTION 1313