Decreased Ceramide Transport Protein (CERT) Function Alters Sphingomyelin Production following UVB Irradiation*

Increased cellular ceramide accounts in part for UVB irradiation-induced apoptosis in cultured human keratinocytes with concurrent increased glucosylceramide but not sphingomyelin generation in these cells. Given that conversion of ceramide to non-apoptotic metabolites such as sphingomyelin and glucosylceramide protects cells from ceramide-induced apoptosis, we hypothesized that failed up-regulation of sphingomyelin generation contributes to ceramide accumulation following UVB irradiation. Because both sphingomyelin synthase and glucosylceramide synthase activities were significantly decreased in UVB-irradiated keratinocytes, we investigated whether alteration(s) in the function of ceramide transport protein (or CERT) required for sphingomyelin synthesis occur(s) in UVB-irradiated cells. Fluorescently labeled N-(4,4-difluoro-5,7-dimethyl-4-bora-3a,4a-diaza-s-indacene-3-pentanoyl)-d-erythro-sphingosine (C5-DMB-ceramide) relocation to the Golgi was diminished after irradiation, consistent with decreased CERT function, whereas the CERT inhibitor N-(3-hydroxy-1-hydroxymethyl-3-phenylpropyl)dodecanamide (1R,3R isomer) (HPA-12) produced an equivalent effect. UVB irradiation also induced the rapid formation of a stable CERT homotrimer complex in keratinocytes as determined by Western immunoblot and mass spectrometry analyses, a finding replicated in HeLa, HEK293T, and HaCaT cells and in murine epidermis. Ceramide binding activity was decreased in recombinant CERT proteins containing the UVB-induced homotrimer. The middle region domain of the CERT protein was required for the homotrimer formation, whereas neither the pleckstrin homology (Golgi-binding) nor the START (ceramide-binding) domains were involved. Finally like UVB-treated keratinocytes, HPA-12 blockade of CERT function increased keratinocyte apoptosis, decreased sphingomyelin synthesis, and led to accumulation of ceramide. Thus, UVB-induced CERT homotrimer formation accounts, at least in part, for apoptosis and failed up-regulation of sphingomyelin synthesis following UVB irradiation, revealing that inactive CERT can attenuate a key metabolic protective mechanism against ceramide-induced apoptosis in keratinocytes.

decreased sphingomyelin synthesis, and led to accumulation of ceramide. Thus, UVB-induced CERT homotrimer formation accounts, at least in part, for apoptosis and failed up-regulation of sphingomyelin synthesis following UVB irradiation, revealing that inactive CERT can attenuate a key metabolic protective mechanism against ceramide-induced apoptosis in keratinocytes.
UV irradiation represents a major oxidative stressor for mammalian skin. The impact of UV irradiation has been demonstrated in conjunction with the pathogenesis of myriad cutaneous diseases, including photocarcinogenesis, photoaging, and photoallergy (1)(2)(3). Although UV irradiation-induced DNA damage may lead to the development of both melanoma and non-melanoma skin cancers (1,2), UV irradiation also increases apoptosis via activation of death signaling pathways, i.e. NF-B (4), Fas death receptor (5), or mitogen-activated protein kinase (MAPK) activation (6), initiated by alterations of plasma membrane functions (7). As such, UV-induced apoptosis can prevent carcinogenesis by eliminating potentially damaged or mutated cells (8). However, a recent study suggests that increased apoptosis in normal keratinocytes (KC) 3 can promote clonal expansion of mutated cells, resulting in increased tumorigenesis (9). Because both squamous and basal cell carcinoma cells are more resistant to apoptosis (10), the prevention of excess apoptosis in normal KC following oxidative stressors is important for decreasing the risk of tumorigenesis.
Ceramide (Cer) is the constituent backbone structure for all sphingolipids and is ubiquitously distributed in all mammalian cells/tissues. Alterations of cellular Cer levels in response to 3 The abbreviations used are: KC, keratinocytes; Cer, ceramide; CERT, ceramide transport protein; CHK, cultured human keratinocytes; GlcCer, glucosylceramide; SM, sphingomyelin; ER, endoplastic reticulum; START, steroidogenic acute regulatory protein (StAR)-related lipid transfer; PH, pleckstrin homology; MR, middle region; VAP, vesicle-associated membrane protein (VAMP)-associated protein; HPA-12, N-(3-hydroxy-1-hydroxymethyl-3-phenylpropyl)dodecanamide (1R,3R isomer); HPA-12 inactive, inactive 1S,3R isomer of HPA-12; C 5 -DMB-Cer, N- (4,4-difluoro-5,7-dimethyl-4- specific stimuli, e.g. cytokine signaling, radiation, or oxidative stress, lead to cell cycle arrest, cellular differentiation, and apoptosis in a variety of cell types (11)(12)(13) including KC (14 -16). Given that cells, and in particular epidermal KC that reside at the interface with the external environment, are exposed to myriad risks and oxidative stressors, we hypothesized that these key skin cells deploy protective mechanisms against Cer-induced apoptosis. Metabolic pathways regulating the conversion of Cer to either sphingomyelin (SM) (17) or glucosylceramide (GlcCer) (18 -20) and sphingosine to sphingosine 1-phosphate (21) can protect cells from Cer-induced apoptosis. These protective mechanisms exist not only in potentially carcinogenic cells but also in normal mammalian cells. We have shown that increasing the Cer-to-GlcCer conversion by bacterial sphingomyelinase overcomes Cer-induced inhibition of growth of human KC (22). In addition, we recently demonstrated that Cer hydrolysis, accompanied by conversion of sphingosine to sphingosine 1-phosphate, protects KC against UVB-mediated Cer-induced apoptosis. 4 Because Cer is synthesized de novo at/in the endoplastic reticulum (ER) and is further converted to SM and GlcCer at the level of the Golgi, intracellular transport of Cer from ER to Golgi is a primary mechanism for the generation of both GlcCer and SM, involving both ATP-dependent and -independent mechanisms (23,24). Recent studies reveal that the ATP-dependent Cer transport is mediated by the ceramide transport protein, CERT (25). CERT is a member of the family of steroidogenic acute regulatory protein (StAR)-related lipid transfer (START) proteins (26). The carboxyl-terminal region of CERT, consisting of 230 amino acids, contains the START domain and is responsible for stereospecific Cer binding (25,27), whereas the amino-terminal region, consisting of 120 amino acids, contains the pleckstrin homology (PH) domain that binds phosphatidylinositol 4-monophosphate at the Golgi (25). The middle region (MR), between the PH and START domains, has a short peptide (FFAT) motif (25) that interacts with vesicle-associated membrane protein (VAMP)-associated protein (VAP) that is enriched in the ER (28). It has been suggested that sequences within the MR domain may promote the formation of coiled-coil structures (25). Moreover a recent study demonstrates that phosphorylation of serine and threonine residues within the MR domain down-regulates Cer transport (29). Thus, all three domains within CERT are required for Cer transport from the ER to Golgi.
We previously demonstrated that increased de novo Cer production accounts, in part, for the induction of apoptosis in cultured human keratinocytes (CHK) following UVB irradiation (16). However, de novo synthesized Cer also is converted to GlcCer and SM, both of which are major sphingolipid species and non-apoptotic Cer metabolites in keratinocytes. Importantly although GlcCer generation increases in CHK following UVB irradiation, SM generation is not changed (16). In the present study, we investigated 1) whether defects in the conversion of Cer to GlcCer and/or SM occur in CHK following UVB irradiation and thereby promote apoptosis and 2) the mecha-nism(s) responsible for the altered metabolic conversion(s). We demonstrate here that although both SM and GlcCer synthase activities were significantly decreased in CHK following UVB irradiation GlcCer generation increased in these cells. In addition, UVB irradiation decreased Cer transport from the ER to Golgi and induced the formation of a stable CERT homotrimer associated with decreases in Cer binding. Moreover a CERT trimer was generated not only in CHK but also in HeLa, HEK293T, and HaCaT cells as well as in murine epidermis following UVB irradiation. Furthermore the oxidative stressors H 2 O 2 and NO also induced the formation of the stable CERT homotrimer complex in cells. Finally similar to UVB irradiation, blockade of CERT function increased apoptosis in CHK. These results reveal a novel mechanism that down-regulates SM synthesis in response to oxidative stress via alterations of CERT function through the formation of an inactive, stable homotrimer complex.
Cell Culture and Transfection-Normal human keratinocytes were isolated from neonatal foreskins by a modification of the method of Pittelkow and Scott (31) under an Institutional Review Board-approved protocol (University of California, San Francisco). Cells were grown in keratinocyte growth medium supplemented with bovine epidermal growth factor, bovine pituitary extract, insulin, hydrocortisone, and 0.07 mM calcium chloride (Cascade Biologics, Portland, OR). The cultures were maintained at 37°C under 5% CO 2 in air. Immortalized, nontransformed HaCaT cells derived from normal human keratinocytes (a gift from Dr. N. Fusenig, Heidelberg, Germany), HeLa cells, and human embryonic kidney cells (HEK293T) were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. Cells were transfected with plasmids containing cDNA for GFP-fused (carboxyl-terminal) CERT wild type (WT), HA (amino-terminal)-tagged-CERT WT, FLAG (amino-terminal)-tagged CERT WT, FLAGtagged CERT MR domain, or FLAG-tagged CERT ⌬MR (lacking the MR domain), which were cloned into the expression vector pcDNA3.1(ϩ) (25,28) using FuGENE 6 transfection reagent (Roche Diagnostics).
Animals-Hairless mice (Skh:hr-1), 6 -8 weeks old, were purchased from Charles River USA (Hollister, CA). Mice were fed Purina mouse diet (Ralston Purina Co., St. Louis, MO) and water ad libitum. Natural sunlight was excluded, and animals were exposed only to low levels of incandescent light prior to UVB irradiation. All animal experiments were performed under an Institutional Animal Care and Use Committee-approved protocol.
UVB Irradiation-UVB irradiation was performed as described previously (16,32). Briefly cells were seeded (2-4 ϫ 10 4 cells/ml) in 4-well glass chamber slides or 100-mm culture dishes and maintained to 60 -80% confluence. The cells were rinsed with phosphate-buffered saline (PBS) containing 0.07 mM calcium chloride followed by UVB irradiation (emission range, 280 -340 nm; 305 nm maximum; FS 20/T12 bulbs, National Biological Co., Twinsburg, OH) in PBS with 0.07 mM calcium chloride. UVB exposure was measured using a Goldilux Ultraviolet radiometer (Oriel, Stratford, CT). Cells were exposed to a single UVB dose (60 mJ/cm 2 ) in most studies unless indicated otherwise. Immediately after UVB irradiation, PBS was replaced by culture medium. Purified His-tagged recombinant CERT protein (25) or murine epidermis was also irradiated with UVB as above. The dorsal skin of each mouse was irradiated with a single UVB dose equivalent to 150 mJ/cm 2 (32).
GlcCer Synthase and SM Synthase Assay-GlcCer synthase and SM synthase activities were determined as described previously (17). Briefly cell lysates prepared in a buffer containing 20 mM Tris/HCl (pH 7.5), 10 mM EGTA, 2 mM EDTA, and proteinase inhibitor mixture (Complete TM , Roche Applied Science) were used as the enzyme source. GlcCer synthase and SM synthase activities were assayed using fluorogenic Cer (C 6 -NBD-Cer) as the substrate. The reaction products were resolved in a solvent system that contained chloroform, methanol, and 12 mM MgCl 2 in H 2 O (65:25:4, v/v/v). C 6 -NBD-GlcCer and -SM were measured by using a Typhoon 9410 Multimodal Imager (Amersham Biosciences).
Immunofluorescence Microscopy-Cells transfected with CERT-GFP were fixed with 4% paraformaldehyde (w/v) in PBS for 10 min followed by permeabilization with 0.2% Triton X-100. Permeabilized cells were then incubated with mouse monoclonal anti-Golgin 97 at room temperature for 1 h and then with Alexa 680-labeled goat anti-mouse for 45 min. Control staining was performed with untransfected control cells for GFP and secondary antibody alone. Slides were examined with a Carl Zeiss LSM 510 confocal microscope (Carl Zeiss, Oberkochen, Germany). An argon laser at 488 nm and a helium-neon laser at 543 nm were used to excite GFP (emission, 515-540 nm) and Alexa 680 (emission, 680 nm), respectively.
Cer Transport-Cer transport was assessed using C 5 -DMB-Cer as reported previously (25,29,33,34) with modifications. Cells pretreated with or without 1 M HPA-12 for 6 h were labeled with 1 M C 5 -DMB-Cer⅐bovine serum albumin complex (1:1) at 4°C for 10 min and then washed with culture medium and PBS. Cells were irradiated with UVB at 4°C in PBS followed by incubation with or without 1 M HPA-12 at 37°C for 30 min. Cells were fixed and examined by confocal microscopy as described above.
Cer Binding Assay-Binding of CERT to Cer was assessed by the method of Malakhova et al. (35) with modifications. C 5 -DMB-Cer (1 mM) was dissolved in chloroform:methanol (9:1, v/v) in borosilicate glass tubes, dried under nitrogen to generate a Cer film immobilized on the glass, and further dried under vacuum for 1.5 h. Either UVB-treated or sham-treated purified recombinant CERT protein (50 g) (25) in 0.5 ml of Tris/HCl buffer, pH 7.4, was added to the C 5 -DMB-Cer filmcoated tubes and incubated for various times at 37°C. Fluorescence intensity generated in the aqueous protein fraction was measured spectrofluorometrically (excitation at 505 nm and emission at 511 nm).

Protein Identification by In-gel Digestion and Liquid Chromatography/Electrospray Ionization-Tandem Mass Spectrometry (LC/ESI-MS/MS) Analysis-Fractions containing immunoprecipitated CERT from UVB-or sham-irradiated cells were
analyzed by SDS-PAGE as above. The resolved bands, stained with Coomassie Blue, were excised and digested with trypsin overnight at 37°C. The resulting tryptic peptides were analyzed by LC/ESI-MS/MS using an LCQ or an LTQ ion trap mass spectrometer (Thermo Finnigan, San Jose, CA) with a nanoelectrospray ionization source. The LC analysis was conducted using a Micro-LC system (Micro-Tech Scientific, Sunnyvale, CA) or a Thermo Surveyor HPLC system. The resulting MS/MS spectra were searched against the National Center for Biotechnology Information (NCBI) non-redundant protein data base using the Sequest program to identify sequences of peptides.
Lipid Synthesis-To examine de novo sphingolipid synthesis following UVB irradiation, cells were cultured with [ 3 H]serine (1.5 Ci/ml), [9,10-3 H]palmitic acid (1.5 Ci/ml), or [ 3 H]galactose (1.5 Ci/ml) for the final 6 or 8 h. Lipids were extracted by the Bligh and Dyer method (36) with a modification described previously (16), and lipids were fractionated by high performance TLC as described previously (16). Radioisotope incorporated into each lipid fraction was measured by liquid scintillation spectroscopy. For in vivo studies, skin samples were obtained from euthanized animals at various time points before and 24 h after UVB exposure as described previously (37). Subcutaneous fat and fascia were removed by gentle scraping, and samples were floated dermis side down on keratinocyte growth medium containing 500 M sodium acetate (34 Ci of sodium [ 14 C]acetate; Amersham Biosciences) for 2 h. Epidermis was separated from dermis following incubation in PBS (pH 7.4, 60°C) for 30 s. Epidermal lipids were extracted, and the separation of individual lipid species was achieved by high perform-ance TLC (37). Radioactivity was quantitated using a BAS Bioimage Analyzer (Fuji Film, Tokyo, Japan). Data are reported as photostimulated luminescence/cm 2 of epidermis.
Apoptosis Assay-Apoptosis was assessed by determining changes in cellular nuclear chromatin as described previously (38) using 2-(4-amidinophenyl)-1H-indole-6-carboxamidine and by terminal deoxynucleotidyltransferase-mediated digoxigenin-dNTP nick end labeling using the DNA Fragmentation Detection kit following the manufacturer's protocol (EMD Chemicals, Inc., Gibbstown, NJ). Approximately 100 cells chosen at random on each slide were counted per replicate, and the ratio of apoptotic to non-apoptotic cells was determined.
Statistical Analysis-Experiments were repeated at least three times. For each experiment, results from triplicate samples were expressed as the mean Ϯ S.D. Significance between groups was determined with unpaired Student's t test.

SM but Not GlcCer Generation Decreased in Keratinocytes
following UVB Irradiation-Our prior studies demonstrated that increased Cer production accounted for, in part, the induction of apoptosis in CHK following UVB irradiation (16). These studies also indicated that both the content and syntheses of Cer and GlcCer increased in CHK following UVB irradiation; however, SM generation remained unaltered (16). Given that the metabolic conversion of Cer to SM and/or GlcCer opposes Cer accumulation in keratinocytes, we hypothesized that restriction of Cer conversion to SM contributes to the Cer accumulation noted in CHK following UVB irradiation. Because SM synthesis/formation can be regulated by multiple steps, e.g. SM synthase activity or Cer transfer from the ER to the trans-Golgi and SM hydrolysis, we first studied SM synthase and GlcCer synthase activities in CHK 8 h following UVB irradiation when Cer and GlcCer levels both are known to be elevated (16). In vitro SM synthase activities were decreased by 22.4% compared with sham-irradiated control cells. Interestingly in vitro GlcCer synthase activity was also significantly decreased by 63.8% versus control (Table 1). To confirm our prior study showing an increase in GlcCer levels in UVBtreated CHK, GlcCer generation was assessed by [ 3 H]galactose incorporation in CHK following UVB irradiation. Again increased GlcCer generation was evident in these cells (Table  1). Thus, despite significant decreases in in vitro GlcCer synthase activity, increases in GlcCer generation in intact cells fol-lowing UVB irradiation suggest that the available pool of Cer for GlcCer synthesis is greater than that for SM synthesis, possibly involving altered Cer transport within UVB-treated cells.
UVB Irradiation Diminished Cer Transfer from the ER to the Golgi-Because the Cer transport pathways from the ER to the Golgi to synthesize SM and GlcCer are distinctly CERT-dependent and -independent, respectively (25), we next determined whether altered Cer transport from ER to trans-Golgi by a CERT-dependent pathway could account for the decreased SM generation in CHK following UVB treatment. As shown previously (25,33,34,39), C 5 -DMB-Cer was localized to the cytoplasm and the ER in CHK 10 min after incubation at 4°C (Fig. 1A) and was subsequently redistributed to perinuclear regions, i.e. Golgi, 30 min following the increase in ambient temperature from 4 to 37°C (Fig. 1B). In contrast, blockade of CERT function using a specific inhibitor of CERT, HPA-12 (Fig.  1C), showed decreased distribution of fluorescence to Golgi compared with sham-irradiated control cells (Fig. 1B). Moreover the Golgi-localized fluorescence was less intense in CHK irradiated with UVB ( Fig. 1D) compared with that in shamirradiated control cells (Fig. 1B), whereas decreases in fluorescent signals at perinuclear regions were not significant in UVBirradiated cells compared with HPA-12-treated cells. We ascertained that UVB irradiation alone did not alter the fluorescence intensity of C 5 -DMB-Cer because total cellular fluores-  Similar to prior studies demonstrating that C 5 -DMB-Cer concentrates in the Golgi apparatus (25,33,34,39), C 5 -DMB-Cer appears in a punctate, perinuclear pattern in control cells incubated at 37°C for 30 min (B). In addition, as shown in Fig. 2, CERT is also distributed perinuclearly, and this distribution is not altered in cells following UVB irradiation. However, C 5 -DMB-Cer was not concentrated at perinuclear regions in UVB-irradiated CHK (D) as well as in cells treated with CERT inhibitor (C). Additional details are available under "Experimental Procedures." cence did not change (data not shown). These results suggest that UVB irradiation diminishes CERT-dependent Cer transport from ER to Golgi in CHK.
UVB Irradiation Does Not Interfere with CERT Localization at the Golgi-We next investigated whether decreased Cer transfer to the Golgi is due to a modification of cellular CERT localization following UVB irradiation. HEK293T cells transiently transfected with a CERT-GFP expression plasmid were irradiated with UVB. CERT-GFP protein was localized at the Golgi (as determined by anti-Golgin 97 antibody colocalization) in cells at both 15 and 30 min following treatment (Fig. 2). Thus, cellular CERT-GFP localization appears to be normal following UVB irradiation.
UVB Irradiation Induced the Formation of a Stable CERT Homotrimer-To investigate whether UVB irradiation alters CERT protein levels in CHK, Western immunoblot analyses were performed under denaturing conditions. CERT levels were not significantly changed in CHK 7 h following a lower dose irradiation (35 mJ/cm 2 ) but modestly decreased in cells treated with a higher UVB dose (65 mJ/cm 2 ) compared with sham-irradiated control cells (Fig. 3A). However, in addition to a band corresponding to native CERT (Ϸ68 kDa), an unknown band ("X") with an estimated molecular size of Ϸ200 -210 kDa (Fig. 3A) was observed after irradiation at both lower and higher UVB doses. Band X was produced after UVB irradiation not only in CHK but also in other cells such as HA-tagged CERT WT-transfected spontaneously immortalized (HaCaT) human keratinocytes, HeLa cells, and HEK293T cells (Fig. 3B). A time course study revealed that band X was evident as early as 30 min following irradiation and that significant levels remained at least 24 h after irradiation (Fig. 3C).
Because it is likely that band X formation is increased in cells treated with a higher UVB dose (Fig.  3A), we investigated the dose dependence for generation of this unknown band in HeLa cells. Because increases in a single dose of UV irradiation over 100 mJ/cm 2 can cause acute deleterious effects in cells, including cell detachment and/or cell lysis, cells were instead treated mul-  tiple times with a lesser UVB dose of 60 mJ/cm 2 with a 0.5-h time interval between treatments (i.e. one, two, three, or four times). Increases in CERT complex formation were evident with up to three repeat UVB exposures (Fig. 3D).
We next investigated whether band X was comprised of CERT alone, CERT and other protein(s), or nonspecific protein aggregates. Following UVB irradiation, FLAG-CERT was immunoprecipitated from HEK293T cells transfected with FLAG-tagged CERT WT and subsequently resolved by SDS-PAGE. The most intense band with a molecular mass of Ϸ69 kDa present in both UV-treated and sham-irradiated control cells was identified as CERT by in-gel tryptic digestion and LC/ESI-MS/MS analysis (data not shown). Twenty-nine tryptic peptides, spanning between amino acids 49 and 626, were detected that reflected sequence coverage of 60.7%, demonstrating a high degree of data confidence. In addition to the expected Ϸ69-kDa band, a less intense band with a high molecular mass, Ϸ200 -210 kDa, corresponding to band X (see Fig.  3A) was found only in the UVB-treated samples. This band was also identified as CERT. Eight tryptic peptides, spanning between amino acids 49 and 595, were detected with sequence coverage of 25.3%. Peptides from other proteins were not found in this band revealing band X to be a stably cross-linked CERT homocomplex. Estimation of molecular size by SDS-PAGE and size exclusion chromatography (data not shown) suggests that homotrimer formation occurs in CERT following UVB irradiation. These results demonstrate that UVB irradiation induces the formation of a stable cross-linked CERT homotrimer rather than causing a significant decrease in cellular CERT protein levels. Moreover in contrast to CERT protein, only a single band is evident for ␤-actin with no evidence of higher molecular weight complex formation following UVB irradiation (Fig. 4).
Cer Binding Activity Decreases in CERT following UVB Irradiation-To investigate whether the formation of the stable CERT homotrimer accounts for decreased SM generation in CHK following UVB irradiation, we compared the Cer binding ability of UVB-(ϩ) and sham-irradiated (Ϫ) CERT proteins. Purified recombinant CERT protein was treated with UVB in vitro. LC-MS/MS analysis demonstrated that a band at ϳ210 kDa was formed following UVB irradiation consisting solely of the CERT protein (data not shown) as noted above. Approximately Ϸ20% of the CERT protein was converted to the CERT homotrimer form as determined by Western immunoblot analysis. Consistent with the formation of the stable CERT homotrimer at Ϸ20% of total CERT following UVB treatment, Cer binding activity of irradiated CERT (containing both the stable CERT homotrimer and intact CERT in a ratio of ϳ2:8 as above) decreased to Ϸ80% of sham-irradiated CERT control (p Ͻ 0.01) (Fig. 5). Thus, it is likely that the ability of CERT to bind Cer is significantly deceased in the stable homotrimeric protein complex.
MR of CERT Protein Is Required for Homotrimer Formation-CERT protein consists of PH, MR, and START domains (Fig.  6A), which are responsible for binding to/association with the Golgi, ER, and Cer, respectively (25,27). To investigate which domain(s) is(are) required for CERT homotrimer formation, we used both HEK293T and HeLa cells transfected with either FLAG-tagged CERT WT; FLAG-tagged CERT ⌬MR, which lacks the MR domain; or FLAG-tagged CERT MR that expresses only the MR domain. UVB irradiation induced the formation of large CERT multimer complexes in both HEK293T and HeLa cells transfected with either FLAG-tagged CERT WT (Fig. 6B, lanes 4 and 6) or FLAG-tagged CERT MR (molecular size for the resultant peptide is ϳ30 kDa in monomer form) (Fig. 6B, lanes 12 and 14), whereas CERT trimer or multimers were not detected in cells transfected with FLAGtagged CERT ⌬MR (Fig. 6B, lanes 8 and 10). These results reveal that the MR of the CERT protein contains an epitope/ sequences responsible for the UVB-induced stable homotrimer formation, whereas the PH and START domains are not required for this stable association.
UVB Irradiation Does Not Affect CERT Phosphorylation/ Dephosphorylation-Because a recent study demonstrated that the transport activity of Cer from the ER to the trans-Golgi by CERT is regulated by phosphorylation, i.e. down-regulated by  phosphorylation and up-regulated by dephosphorylation (29), we next investigated whether UVB irradiation alters the phosphorylation status of CERT. Consistent with the prior study, two CERT bands, corresponding type I CERT (highly phosphorylated form) and type II CERT (less phosphorylated form) (29), were detected at an approximately equivalent ratio (type I:type II ϭ 0.44:0.56) in sham-irradiated HEK293T cells transfected with FLAG-tagged CERT WT (Fig. 7, lane 1). These CERT proteins are converted to type III CERT (dephosphorylated form) (29) following phage protein phosphatase treatment (Fig. 7,  lane 2). Similar to the results in sham-irradiated control cells, type I and II bands were again present at an approximately equal ratio (type I:type II ϭ 0.50:0.50) in UVB-treated cells (Fig.  7, lane 3), and these proteins were converted to type III by phage protein phosphatase treatment (Fig. 7, lane 4). Moreover the CERT homotrimer band was also shifted to a lower molecular size following phage protein phosphatase treatment (Fig.  7, lanes 1 and 3 versus lanes 2 and 4, respectively). Thus, it is unlikely that alterations of phosphorylation/dephosphorylation affect CERT-mediated Cer cellular trafficking and/or CERT multimer formation following UVB irradiation.
Blockade of CERT Function Induced Apoptosis-We next investigated whether the CERT-dependent conversion of Cer to SM can regulate cellular Cer levels and thereby protect cells against Cer-induced apoptosis in CHK. As anticipated, the inhibition of CERT function using HPA-12 specifically decreased SM synthesis in parallel with increased Cer levels in CHK, whereas GlcCer synthesis was not altered 8 h after treatment (Table 2). In contrast, the inactive stereoisomer of HPA-12 did not affect the synthesis of these lipids (Table 2). Consistent with our prior studies, increased apoptosis was evident in UVB-irradiated cells (Fig. 8 and Ref. 16); in addition, increased apoptosis was also evident in CHK incubated with 10 M HPA-12 (Fig. 8), but the inactive stereoisomer showed no effect, suggesting that the inhibition of CERT function leads to both decreased SM generation and increased Cer levels, which in turn are coincident with increased apoptosis. These results reveal an ongoing role for CERT-mediated transfer of Cer for SM synthesis and a role for CERT in regulating Cer-mediated apoptosis.
Oxidative Stress Induced CERT Trimer Formation-Prior studies demonstrated that other oxidative stressors, i.e. NO (40) and H 2 O 2 (41) as well as UVB irradiation (16), increase Cer-mediated apoptosis. We therefore investigated whether other general oxidative stressors, specifically NO and H 2 O 2 , also induce the formation of the stable CERT homotrimer in cells. Similar to the results of UVB treatment noted above, stable CERT homotrimer was evident in HEK293T cells exposed to either H 2 O 2 (Fig. 9, lane 2) or the NO generator sodium nitroprusside (Fig. 9, lane 3). Thus, oxidative stress induced by UVB irradiation or chemical treatment can promote the formation of the inactive stable CERT homotrimer.
CERT Homotrimer Formation and Decreased SM Generation Occur in Murine Epidermis following UVB Irradiation-Finally we investigated whether the stable CERT homotrimer is formed in intact epidermis following UVB irradiation. Similar to the results in cultured human keratinocytes, a stable CERT homotrimer was evident in epidermis isolated FIGURE 6. Homotrimer formation occurs within the middle region of CERT. A, CERT-FLAG constructs. WT construct expresses all three CERT domains, i.e. the PH domain that binds phosphatidylinositol 4-monophosphate at the trans-Golgi, the MR domain that contains an FFAT motif for binding to VAP enriched in the ER, and the Cer-binding START, whereas MR construct expresses only the MR domain; each construct also contains a FLAG sequence tag. The ⌬MR construct contains the PH and START domains linked to the FLAG tag but lacks the MR domain. B, stable CERT trimer formation was evident in both WT-and MR-but not ⌬MR-expressing cells following UVB irradiation. HEK293T (HE) or HeLa cells (HL) cells were transfected with FLAGtagged CERT WT, FLAG-tagged CERT ⌬MR, or FLAG-tagged CERT MR plasmids followed by irradiation with UVB as above. CERT protein was immunoprecipitated with anti-FLAG antibody and analyzed by Western immunoblot with anti-FLAG antibody. Arrows and asterisks indicate CERT monomer and the stable CERT homotrimer, respectively. Lanes 3-6, FLAG-WT monomer (arrow) and multimer (asterisks) after UVB irradiation; lanes 7-10, FLAG ⌬MR monomer only (arrow) with no multimer formed following UVB irradiation in either the HEK293T or HeLa cell line; lanes 11-14, MR monomer (arrow) and multimer (asterisks) formed when only FLAG-MR domain is expressed. A nonspecific band was observed in all samples (dotted arrows) including non-transfected cells (left lanes). Details are presented under "Experimental Procedures." FIGURE 7. UVB irradiation does not affect CERT phosphorylation. Both type I and II forms of CERT are present at approximately equal levels in HEK293T following sham or UVB irradiation (lanes 1 and 3, respectively), whereas the dephosphorylated form (type III) of CERT is evident in these cells only after phage protein phosphatase (PPase) treatment. HEK293T cells transfected with FLAG-tagged CERT WT were irradiated with UVB as above. Cell lysates were then treated with (lanes 2 and 4) or without (lanes 1 and 3) phage protein phosphatase (800 units) at 30°C for 30 min (29). FLAG-tagged CERT was detected by Western immunoblotting analysis using anti-FLAG antibody. Additional details are presented under "Experimental Procedures." from murine skin irradiated with UVB ex vivo (Fig. 10A). To investigate whether UVB irradiation also leads to similar alterations in CERT function in vivo as noted in CHK (above), we assessed the synthesis of SM in whole epidermis. Because our prior studies showed that epidermal DNA synthesis decreases during the first 0.5-12 h, returns to normal levels at 24 h, and then increases at 36 -48 h following irradiation (32), we determined lipid synthesis at 22-24 h when DNA synthesis is not significantly altered to avoid cellular proliferation-mediated changes in lipid synthesis. Similar to the results observed in UVB-irradiated CHK, SM but not GlcCer synthesis was significantly decreased, whereas Cer generation was increased in epidermis following UVB treatment to intact murine skin (Fig. 10B). In comparison, cholesterol synthesis was not significantly altered by UVB irradiation (Fig. 10B). These results indicate that the formation/ presence of the CERT homotrimer is associated with suppression of SM synthesis and increased Cer production  10. A, Western immunoblot analysis. Dorsal mouse skin was excised from euthanized animals, placed on a culture dish, and irradiated with UVB (0, 100, or 150 mJ/cm 2 ) as described above for CHK. Epidermal homogenates were then prepared from skin 1 h following treatment. SDS-PAGE was performed under denaturing conditions, and a membrane was probed with anti-CERT and alkaline phosphatase-labeled secondary antibody as above. CERT monomer and CERT trimer bands are indicated by lower and upper arrows, respectively; an unknown alkaline phosphatase positive band is also noted (dotted line arrow). B, lipid synthesis. Dorsal mouse skin was irradiated with UVB (150 mJ/cm 2 ) as above. To assess rates of lipid synthesis, skin samples were taken before and 24 h after UVB exposure and then incubated with keratinocyte growth medium containing sodium [ 14 C]acetate for 2 h. Epidermal lipids were extracted, and the separation of individual lipid species was achieved by high performance TLC. Data are reported as photostimulated luminescence (psl)/cm 2 of epidermis (mean Ϯ S.D., n ϭ 5). *, p Ͻ 0.01 versus non-irradiated controls. Additional details are available under "Experimental Procedures." both in epidermis in vivo and in cultured human keratinocytes (shown above).

DISCUSSION
The metabolic conversion of Cer to either SM or GlcCer is not only required for the generation of cellular membranes constituents but also is important for the homeostatic regulation of cellular Cer levels, thereby reducing the risk of Cerinduced apoptosis (17)(18)(19)(20). Our prior studies showed not only increased Cer production in CHK but also significantly elevated GlcCer generation in UV-irradiated cells (16). SM levels, however, were not elevated following UVB irradiation (16), suggesting that the SM and GlcCer synthetic pathways are independently regulated. We report here a mechanism(s) that accounts for the apparent failure to up-regulate SM production following UVB irradiation that appears linked to the subsequent UVBinduced apoptosis. Because the major pathways that transfer Cer from ER to Golgi for either SM or GlcCer synthesis appear to be distinct, i.e. CERT is required for the majority of Cer transfer from the ER to the trans-Golgi for SM synthesis, whereas CERT-independent transport accounts for the bulk of Cer transport for GlcCer synthesis (23,24), we have focused on alterations of CERT function induced by UVB irradiation. We demonstrate that UVB irradiation induced the formation of a stable CERT homotrimer complex associated with decreases in both cellular transfer of Cer from ER to Golgi and Cer binding ability. We also show that blockade of CERT function using a specific CERT inhibitor mimicked UVB-induced apoptosis in CHK. Moreover similar to the results in cultured keratinocytes, in vivo studies also revealed the presence of a stable CERT homotrimer in association with increased Cer formation/content and decreased SM (but not GlcCer) generation in UVBirradiated murine epidermis. Together with our prior studies, these findings strongly suggest that a modification of CERT function by the formation of a stable homotrimer complex contributes to the decrease in SM synthesis and accumulation of Cer, resulting in increased apoptosis following UVB irradiation. In addition, because H 2 O 2 and NO also induced the formation of an equivalent CERT homotrimer, the formation of this (inactive) trimer complex represents a potential risk factor in the apoptosis that occurs subsequent to oxidative stress exposure.
In addition to CERT function, SM synthase or GlcCer synthase activities also represent rate-limiting factors for metabolic conversion of Cer to either SM or GlcCer, respectively. Interestingly both SM synthase and GlcCer synthase activities decreased in UVB-irradiated CHK (Table 1). However, despite the large decrease in GlcCer synthase in vitro enzyme activity, GlcCer generation was significantly increased following UVB irradiation, suggesting that diminished Cer transport in the SM synthesis pathway could provide an increased pool of available Cer for GlcCer production. Our prior studies demonstrated that sphingolipid synthesis (including GlcCer) is up-regulated following acute epidermal permeability barrier perturbation to restore barrier function, whereas serine palmitoyltransferase but not GlcCer synthase activity is increased (42). These results suggest that basal GlcCer activity is sufficient to up-regulate GlcCer production in epidermal keratinocytes without measurable changes in total enzyme activity in vitro. Thus, it is likely that despite a measured decrease in GlcCer synthase activity residual enzyme activity can suffice to generate GlcCer in response to any increases in the available Cer pool in keratinocytes following UV irradiation. It has been suggested that a CERT-independent pathway may also supply Cer for SM synthesis in cells (43). However, inhibition of CERT significantly inhibited SM generation while leaving GlcCer production unaltered in keratinocytes ( Table 2), suggesting that HPA-12-sensitive Cer transport, i.e. CERT-dependent transport, is important for SM synthesis in keratinocytes. Hence although the relative contribution of either decreased SM synthase activity or altered CERT function to the diminished SM generation following UVB irradiation has not been determined, the stable CERT homotrimer formation represents a previously unrecognized negative regulatory mechanism in SM synthesis homeostasis.
Although prior studies have shown that the START domain of CERT binds Cer (25), our present study revealed that the MR domain is required for the stable CERT homotrimer formation that accompanies decreased Cer binding activities. Thus, it appears that conformational changes in the CERT molecule induced by oxidative stress rather than direct effects on the Cer binding domain itself can affect Cer binding. However, the current results do not exclude the possibility that altered interaction between either the PH domain and phosphatidylinositol 4-monophosphate at the trans-Golgi (25) or the FFAT motif in the MR domain and VAP at the ER (28) can affect CERT trafficking between the ER and the trans-Golgi, which could also result in decreased SM generation. Although CERT function also could be modulated by alteration of cellular phosphatidylinositol 4-monophosphate profiles, our results demonstrate that CERT still localizes to the Golgi, whereas abnormal localization of CERT to the nucleus, mitochondria, and/or plasma membrane was not evident following UVB irradiation. Thus, UVB irradiation does not appear to significantly alter the intracellular trafficking of CERT. It appears more likely that decreased overall Cer binding activity, due to the stable homotrimer formation, is the mechanism primarily responsible for decreasing Cer transfer from the ER to the trans-Golgi for SM generation following UVB irradiation and possibly other oxidative stressors.
We have demonstrated here that the CERT homotrimer is stable under both denaturing and reducing conditions as used in electrophoresis. Such results strongly suggest that the CERT complex formation is due to intermolecular covalent bonds, perhaps through dityrosine and/or ditryptophan linkages, rather than hydrogen bonding and/or thioester linkages. Although it is possible that CERT forms stable complexes with other proteins including VAP enriched at the ER, LC-MS/MS analysis revealed that the high molecular weight CERT complex consists solely of CERT proteins. Although the association of the CERT complex with other protein(s) cannot be excluded, the present studies suggest that a molecular interaction between individual CERT molecules occurs more frequently or at higher affinity compared with any interaction(s) between CERT and other protein(s). Because sequence analysis revealed that the region in the MR domain between amino acids 271 and 302 shows a high probability for forming coiled-coil structures