110215 (251)

Biosci. Biotechnol. Biochem., 75 (8), 110215-1–8, 2011

Orally Administered Glucosylceramide Improves the Skin Barrier Function by Upregulating Associated with the Tight Junction and Cornified Envelope Formation

y Ritsuro IDETA, Tomohiro SAKUTA, Yusuke NAKANO, and Taro UCHIYAMA

Shiseido Functional Food Research and Development Center, 2-12-1 Fukuura, Kanazawa-ku, Yokohama 236-8643, Japan

Received March 18, 2011; Accepted May 9, 2011; Online Publication, August 7, 2011 [doi:10.1271/bbb.110215]

Dietary glucosylceramide improves the skin barrier mammalian skin barrier function through their role as function. We used a microarray system to analyze the intracellular lipids.6) The skin barrier is essential for mRNA expression in SDS-treated dorsal skin of the protecting against physical stimuli, thermal challenge, hairless mouse to elucidate the molecular mechanisms ultraviolet light (UV), chemical substances and micro- involved. The transepidermal water loss of mouse skin organisms, as well as for preventing water loss.7) The was increased by the SDS treatment, this increase being barrier function is mainly localized in the stratum significantly reduced by a prior oral administration of corneum (SC) which is formed in the outermost layer of glucosylceramides. The microarray-evaluated mRNA the epidermis and consists of the cornified envelope expressionAdvance ratio showed a statistically significant View in- (CE) and intercellular multilamellar lipids. CE forms crease in the expression of genes related to the cornified a highly durable and flexible barrier8) comprising a envelope and tight junction formation when compared 15-nm-thick structure composed of such insoluble with all genes in the glucosylceramide-fed/SDS-treated as involucrin, loricrin, and small proline-rich mouse skin. We then examined the contribution of proteins that are covalently cross-linked by transgluta- glucosylceramide metabolites to the tight junction minases (TGases).9) SC intercellular lipids, which con- formation of cultured keratinocytes. The SDS treatment sist mainly of ceramide and such other components as of cultured keratinocytes significantly decreased the cholesterol esters and free fatty acids, are also known to transepidermal electrical resistance, this decrease being influence the skin barrier function. Maturation of CE is significantly ameliorated in the presence of sphingosine essential for a proper barrier function,10) and it is known or phytosphingosine, the major metabolites of glucosyl- that a decrease ofProofs ceramide causes impairment of the ceramide. These results suggest that an oral adminis- barrier function in human skin.11) tration of glucosylceramide improved the skin barrier Tight junctions (TJ)s in the granular layer of the function by up-regulating genes associated with both the epidermis also contribute to the skin barrier func- cornified envelope and tight junction formation. tion12,13) by controlling the paracellular permeability of ions and water, as well as larger molecules. TJ is formed Key words: glucosylceramide; sphingoid base; cornified by a variety of proteins, including structural trans- envelope; tight junction; skin barrier membrane components (claudins (Cldns), , junctional adhesion molecules (JAMs) and tricellulin) Amorphophallus konjac (A. konjac) is a perennial and scaffolding proteins for undercoating TJ and for the plant native to eastern Asia (from Japan and China, assembly of transmembrane proteins (ZO-1, 2, 3, Mupp- south to Indonesia). It forms a large corm which 1, Magi-1 and so on),14–16) as well as some contains around 40% glucomannan gum,1) and is used complexes which regulate the set-up of the polarity, to prepare flour and konjac jelly. Konjac jelly is a aPKC/Par3/Par6 complex and Crb3/Pals1/Patj com- popular health food in Japan, because it has almost no plex.16,17) TJ and TJ proteins form zipper-like structures calories, but is very high in fiber content. in epithelial cells which firmly fasten adjacent cells to A. konjac is also a rich source of glucosylceramides each other, but divide between the apical space and (GCs), which are structurally constituted by sphingoid basal space to control the paracellular passage of soluble bases, long-chain fatty acids and sugar moieties, and factors. TJ proteins in mammalian skin contribute to occur in animals, fungi and plants.2,3) GCs are essential various skin functions, including barrier formation, structural components of mammalian cell membranes polarity, expression, proliferation, differentiation, and are mostly found at the cell surface; they participate and vesicular transport.14) in such biological functions as immunomodulation4) and Interestingly, an oral intake of GC has been reported insulin resistance.5) They also serve to maintain the to improve the skin barrier function; for example, GC

y To whom correspondence should be addressed. Fax: +81-45-788-7284; E-mail: [email protected] Abbreviations: AJ, adherens junction; aPKC, atypical protein kinase C; CE, cornified envelope; Cldn, claudin; Cy, cyanine; dNHEK, differentiated normal human epidermal keratinocytes; GC, glucosylceramide; GSEA, gene set enrichment analysis; JAM, junctional adhesion molecule; Magi, membrane-associated guanylate kinase; Mupp, multiple PDZ domain protein; NHEK, normal human epidermal keratinocytes; PAGE, parametric analysis of gene set enrichment; SC, stratum corneum; SD, standard deviation; Sprr, small proline-rich protein; TER, transepithelial electrical resistance; TEWL, transepidermal water loss; TGase, transglutaminase; UV, ultraviolet light; TJ, tight junction; ZO, zonula occludens 110215-2 R. IDETA et al. improved the recovery of SC flexibility and transepider- (Thermo Fisher Scientific, Waltham, MA, USA) and an Agilent mal water loss (TEWL) in acutely barrier-perturbed Bioanalyzer (Agilent Technologies, Palo Alto, CA, USA). mice.18) A konjac extract, which contains GC, has also 19) cRNA amplification and labeling. Total RNA was amplified and improved TEWL in healthy human subjects. labeled with Cyanine 5 (Cy5) and Cyanine 3 (Cy3) by using a Low We performed a microarray analysis in this study to RNA Input linear amplification kit (Agilent Technologies) according to evaluate mRNA expression in the SDS-treated (barrier- the manufacturer’s instructions. RNAs for each sample were individ- perturbed) skin of GC-fed and control-fed mice in order ually coupled to both Cy3 and Cy5 dyes so that a dye swap comparison to clarify the mechanism for the barrier-improving effect could be made. Briefly, 500 ng of total RNA was reverse-transcribed to of orally administered GC. We evaluated two sets of double-stranded cDNA by using a poly dT-T7 promoter primer and MMLV-RT enzyme. The cDNA products were used as templates for genes associated with CE formation and TJ formation in vitro transcription to generate fluorescent cRNA by using T7 RNA and function, and calculated their average induced polymerase and Cy5-labeled or Cy3-labeled CTP. Labeled cRNAs expression ratio by GC feeding against the average were purified by using Qiagen RNeasy mini spin columns and eluted expression ratio of all genes on the microarray. We also in nuclease-free water. The cRNA quantity and cyanine incorporation confirmed the contribution of GC metabolites to TJ were determined using the Nanodrop ND-1000 and Bioanalyzer formation by examining their effect on the transepithe- instruments. lial electrical resistance (TER) of normal cultured Hybridization of sample cRNAs and data processing. Two labeled human epidermal keratinocytes. cDNA samples, one from a GC-fed mouse and the other from a control-fed mouse, were combined. Each hybridization used 825 ng of Materials and Methods labeled cRNAs that were mixed, fragmented, and hybridized at 65 C for 17 h to an Agilent 4 44 K Whole Mouse Genome microarray Materials. Normal human epidermal keratinocytes (NHEK) were (Agilent 14868). After washing, the microarray was scanned with an obtained from Kurabo Co. (Osaka, Japan). NHEKs were cultured in a Agilent DNA microarray scanner. Feature Extraction software version low-Ca2þ (0.15 mM Ca2þ) medium, HuMedia-KG2 (Kurabo). Cells at 9.1.3.1 (Agilent Technologies) was used to assess the fluorescent passage three were used in this study. hybridization signals and to normalize the signals by using linear AdvanceKonjac extracts containing various concentrations Viewof GC (12%, regression and a Lowess curve-fitting technique. The reproducibility 66% and 100%) were presented by Unitika Limited (Osaka, Japan) and and reliability of each microarray were assessed by using Quality suspended in 1% of tragacanth gum (Wako Pure Chemicals Industries, Control report data in Feature Extraction. Osaka, Japan) at respective final GC concentrations of 30, 165 and 250 mg/mL. Statistical analysis. The scanned data were analyzed by using Sphingoid bases, sphingosine and phytosphingosine, were pur- Genespring microarray data processing software (Agilent Technol- chased from Avanti Polar Lipids (Alabaster, AL, USA). ogies). Normalized signals were processed after first eliminating the flagged signals. Since some genes have multiple probes on the Agilent platform, gene-level expression values were calculated by using the Animals. This study was approved by the ethics committee of ‘‘Gene-level experiment’’ function in Genespring. Expression values Shiseido Research Center in accordance with the guidelines of the below 200 were floored to 200.20) The processed signal values for the National Institute of Health. Hairless male mice (HR-1) were GC-fed mice were divided by the processed signal values for the purchased from Hoshino Laboratory Animals (Ibaraki, Japan), and references, and these values were used as expression ratio data. Log 5-week-old HR-1 were fed with the AIN-93G rodent diet (Oriental Proofs base 2 of the expression ratio (log ratio) was used for further analysis, Yeast, Tokyo, Japan). and Grubb’s method was used to eliminate outliers.21) The mean and standard deviation values for parent populations were calculated, and a SDS treatment and TEWL measurement. Twenty mice were z-test statistical analysis was performed between CE-related genes and assigned to four groups of 5 animals each. A 0.5-mL amount of the overall genes and between TJ-related genes and overall genes by using appropriate GC suspension (0, 30, 165 and 250 mg/mL final) was orally the ZTEST function in Microsoft Excel (Microsoft, Redmond, WA, administered once a day to the mice in each group. The dorsal skin on USA) according to the method described by Kim et al.20) Ingenuity the right side of each mouse was treated with 10% SDS for five Pathway database analyzing software (Ingenuity Systems, Redwood minutes daily after 2 weeks. After four days (four SDS treatments), City, CA, USA) was used to search the knowledge-based relationships TEWL of the dorsal skin was measured on both sides with a among genes and to prepare the figures. The data discussed in this Tewameter (Courage+Khazaka, Germany), and the increase of TEWL study have been deposited in NCBIs Gene Expression Omnibus (GEO, on the SDS-treated side (right side) versus the non-treated side (left http://www.ncbi.nlm.nih.gov/geo/) and are accessible through GEO side) was used as a measure of the perturbation of the skin barrier series accession no. GSE28086. Dunnett’s multiple-comparison test function by SDS. The procedures for the SDS treatment and TEWL was used to evaluate the TEWL and TER data. measurement have previously been described in detail.19) All treat- ments were performed under anesthesia. Transepithelial electrical resistance assay. NHEKs were seeded into 12-well Transwell-Clear polyester membrane inserts (Corning Total RNA preparation for the microarray analysis. Sixteen mice International, Akasaka, Japan) and cultured in a low-Ca2þ medium. were assigned to two groups of 8 animals each which were fed for 14 d The cells were then cultured at 80% confluency in a high-Ca2þ with 0.5 mL of a 250 mg/mL GC-containing tragacanth gum suspen- (1.80 mM) medium for 3 d to induce differentiation. The differentiated sion or no GC-containing suspension. These mice were treated with NHEKs (dNHEKs) were treated with 0.001% SDS for 3 h in order to SDS as already described after 12 d. Dorsal skin samples from the inhibit the TJ function. After this SDS treatment, the cells were SDS-treated side were taken from both groups before and 2 d after the incubated for 72 h in the presence of sphingoid bases (sphingosine and beginning of the SDS treatment (12 and 14 d after the beginning of the phytosphingosine) at a concentration of 106 mM, and their barrier GC administration). The skin samples were frozen in liquid nitrogen function was measured with a Millicell-ERS instrument (Millipore, and then stored at 80 C. Skin samples from the GC-fed mice were Billerica, MA, USA). used as test samples and those from the control diet-fed mice as references. A frozen sample (5 cm2) was crushed in a Cryo-Press Results (Microtech Co., Chiba, Japan), suspended in 1 mL of Isogen reagent (Nippon Gene, Toyama, Japan), and total RNA was isolated as Effect of dietary GC on TEWL of SDS-treated mouse recommended by the manufacturer. Total RNAs were further purified by using an RNeasy fibrous tissue mini kit (Qiagen, Valencia, CA, skin USA) according to the manufacturer’s instructions. The RNA quantity The mice were fed with different concentrations of and quality were evaluated by a Nanodrop ND-1000 spectrophometer konjac GC, the dorsal skin on one side was then treated Oral Glucosylceramide Uptake Improves Skin Barrier Function 110215-3 21) 25 observing a Z-score over 1.96 is under 0.05. In this case, many CE-related genes showed increases, and the 20 Z-score for the gene set was 8.26. The calculated p- /h)

2 value for the mean of CE-related genes with respect to 15 the mean of all genes was extremely low (1:11 1016) * (Table 1). 10 We also selected a set of 30 genes considered to be TEWL (g/m

∆ related to TJ formation and maintenance, including 5 claudin-1, 3, 4, 5, 7, 8, 10, 12 and 23 from the claudin 28) 29) 0 family, occludin, tricellulin, and junctional adhe- 30) 0 (Control) 30 165 250 sion molecules (JAM1, 2 and 3) as transmembrane GC concentration (µg/mL) proteins in TJ. Several genes for cytoplasmic scaffolding proteins which undercoat and organize TJ proteins, i.e., Fig. 1. Effects of Glycosylceramide on TEWL of SDS-Treated ZO proteins (Tjp1, 231) and Tjp332)), multiple PDZ Mouse Dorsal Skin. domain protein (Mupp-1) and membrane-associated HR-1 mice were fed for a week with a no-GC control diet or with 16) a diet containing GC (30, 165 or 250 mg/mL). The TEWL increase guanylate kinase-1 (Magi-1), were also selected. We on the SDS-treated side versus the untreated side was measured in also included other genes known to be constitutively each animal group. All data are expressed as the mean SD associated with TJ proteins and considered to be (n ¼ 5). An asterisk indicates statistical significance according to involved in the assembly and function of TJ, i.e., Dunnett’s multiple-comparison test ( p < 0:05). symplekin,33) members of the aPKC/Par3/Par6 complex (Prkci, Prkcz, Pard3, Pard6a, Pard6b and Pard6g), Crb3/ 4 times with SDS, and TEWL was measured on both Pals1/Patj complex (Crb3, Mpp5 and Inadl)16) and sides. TEWL of the SDS-treated side was increased in .34) theAdvance control mice, while this increase was inhibited View in the Among these genes, claudin-1, 4, 10 and occludin GC-fed mice; the effect reached statistical significance showed relatively high log ratios (0.36, 0.35, 0.54, and in the mice fed with 250 mg/mL of GC (Fig. 1). 0.35, respectively). The respective mean log ratios for the set of 30 genes at 0 d and 2 d were 0.02 and 0.14. The Microarray results and differences of mean expres- respective Z-score and p-value of the average difference sion ratios of CE- and TJ-related genes from the mean of TJ-related genes against the average of all genes were expression ratios of all genes 4.27 and 9:93 106 for the 2 d data (Table 2). Skin from the mice fed with or without 250 mg/mL of GC was used to obtain microarray data. Skin samples Transepithelial electrical resistance assay were obtained just before the SDS treatment (0 d), and Treating dNHEK with 0.001% SDS for 3 h signifi- after two SDS treatments (2 d). One sample for 0 d was cantly decreased TER (272.9 cm2 to 159.1 cm2). As lost during the hybridization procedure, so that, three shown in Fig. 2,Proofs subsequent treatment with sphingoid replicates of the 0 d samples and four replicates of the bases, sphingosine and phytosphingosine, for 72 h 2 d samples were used for further analysis. We obtained significantly enhanced the recovery of TER (201.6 and respective gene level log ratio data for 14,518 and 221.3 cm2, respectively (n ¼ 4)). 14,547 genes from the 0 d and 2 d samples after data processing. The values for the average log ratio for all Discussion genes at 0 d (n ¼ 3) and at 2 d (n ¼ 4) were close to zero (Tables 1 and 2). We have already reported that an oral intake of a We selected 29 genes as a CE-related gene set. These konjac extract containing GC improved the skin barrier included genes of the CE precursors, i.e., loricrin, condition in mice and humans. The increase in TEWL of involucrin, filaggrin, cystatin-A, elafin, desmoplakin, SDS-treated (barrier-perturbed) mouse skin was signifi- envoplakin, periplakin, small proline-rich proteins cantly reduced by the konjac ingestion. A drink (Sprr1a and 1b, Sprr2a1 to 2k and Sprr4), trichohyalin, containing a konjac extract reduced TEWL on the cheek keratin-1, 5 and 10,22) annexin 123) and S-100 proteins of healthy human volunteers in a randomized double- (S100a10 and S100a11).24) We also included genes for blind placebo-controlled trial.19) We examined in the TGase-1 and 325) and sulfhydryl oxidase, because they present work the effects of dietary GC derived from formed cross-links among the CE precursors.26) Cas- konjac extracts on a mouse model of SDS-treated skin. pase-14 also plays a role in CE maturation as it is GC at 250 mg/mL produced a significant improvement involved in the cleavage of profilaggrin into filaggrin in TEWL, so we used this concentration for the units to stabilize keratin intermediate filaments.27) As subsequent experiments to elucidate the molecular shown in Table 1, the respective log ratios for loricrin, mechanism for the improvement of skin barrier function involucrin, filaggrin and TGase-1 at 2 d were 0.52, 0.41, by the oral GC treatment. 0.45 and 0.41. The respective mean log ratios for the We hypothesized that the genes involved in the CE-related genes at 0 d and 2 d were 0.02 and 0.27. We maintenance and formation of SC would be up-regulated calculated the Z-score and p-value by applying the by GC or its metabolites. Indeed, we have observed a ZTEST function of Microsoft Excel to the 2 d data. The significant increase of TGase-1 in the skin of GC-fed Z-score is the distance from a sample mean to the parent mice following barrier perturbation by UV exposure.35) population mean in units of the standard error, and it can TGase-1 is known to catalyze the formation of ester be used to estimate the probability of observing a sample bonds between specific glutaminyl residues of involu- mean by chance; for example, the probability of crin which is localized at the outmost face of the CE and 110215-4 R. IDETA et al. Table 1. List of CE-Related Genes

Log ratio Gene expression Entre Gene Description/ Gene name (log2 (GC/Control)) ratio (GC/Control) gene ID symbol Function 0d 2d 0d 2d 16952 Anxa1 annexin A1 0.017 0.023 1.012 1.016 CE component 209294 Csta cystatin A 0:010 0.436 0.993 1.353 CE component 109620 Dsp desmoplakin 0.117 0.390 1.085 1.310 CE component 14027 Evpl envoplakin 0:074 0.331 0.950 1.258 CE component 14246 Flg filaggrin 0.279 0.452 1.213 1.368 CE component 16447 Ivl involucrin 0:139 0.414 0.908 1.332 CE component 16678 Krt1 keratin 1 0.109 0.724 1.078 1.651 CE component 16661 Krt10 keratin 10 0.363 0.597 1.286 1.513 CE component 110308 Krt5 keratin 5 0.147 0.283 1.107 1.217 CE component 16939 Lor loricrin 0.108 0.520 1.078 1.434 CE component 19041 Ppl periplakin 0.014 0.346 1.010 1.271 CE component 20194 S100a10 S100 calcium binding protein A10 0:153 0:142 0.899 0.906 CE component (calpactin) 20195 S100a11 S100 calcium binding protein A11 0:002 0.035 0.999 1.024 CE component (calgizzarin) 20753 Sprr1a small proline-rich protein 1A/cornifin A 0:196 0.327 0.873 1.254 CE component 20754 Sprr1b small proline-rich protein 1B/cornifin B 0:081 0.068 0.946 1.048 CE component 20755 Sprr2a1 small proline-rich protein 2A1 0.128 0.092 1.093 1.066 CE component 20756 Sprr2b small proline-rich protein 2B 0.091 0.115 1.065 1.083 CE component 20758 Sprr2d small proline-rich protein 2D 0.302 0.215 1.233 1.160 CE component 20759 Sprr2e small proline-rich protein 2E 0.106 0.056 1.077 1.040 CE component 20760 Sprr2f small proline-rich protein 2F 0.090 0.027 1.064 1.019 CE component Advance20761 Sprr2g small proline-rich protein 2G View 0.000 0.000 1.000 1.000 CE component 20762 Sprr2h small proline-rich protein 2H 0.015 0.303 1.010 1.234 CE component 20765 Sprr2k small proline-rich protein 2K 0.109 0.074 1.078 1.052 CE component 229562 Sprr4 small proline-rich protein 4 0:129 0.420 0.915 1.338 CE component 99681 Tchh trichohyalin 0:494 0.339 0.710 1.265 CE component 21816 Tgm1 transglutaminase 1, K polypeptide 0:018 0.405 0.988 1.324 CE maturation 21818 Tgm3 transglutaminase 3, E polypeptide 0:002 0.188 0.999 1.139 CE maturation 104009 Qsox1 quiescin Q6 sulfhydryl oxidase 1 0:022 0.209 0.985 1.156 CE maturation 12365 Casp14 caspase-14 0.029 0.518 1.021 1.432 cystein protease/ apoptotic protein/ ProofsCE maturation Number of genes 29 Mean 0.024 0.268 1.023 1.216 SD 0.164 0.207 0.113 0.176 Number of ALL 14518 14547 14518 14547 Mean of ALL 0:001 0.002 0.999 1.002 SD of ALL 0.10 0.17 Z-score 1.319 8.260 p-value 0.094 1:11 1016

Gene names, gene symbols, entre IDs and their functions are shown. The mean and SD of the gene-level log scale expression ratio (base 2) and expression ratio (GC/ Control) under each experimental condition are shown (0 d, n ¼ 3;2d,n ¼ 4). The mean and SD of log ratios and ratios of overall genes (ALL) are indicated. Z- scores and Z-test p-values were calculated based on the log scale ratio. An asterisk indicates statistical significance in the Z-test ( p < 0:001). corneocyte lipid envelope and thus contributes to mature with threshold values for the expression ratio (power of lamellar membrane formation in SC and to skin barrier base 2 log ratio) of more than 1.1 and less than 0.9 to homeostasis.36–38) An increase of TGase-1 might there- confirm no marked changes of expression level among fore contribute to barrier recovery. However, many other the CE-related genes before the SDS treatment (0 d) genes may also be involved, including those associated (Fig. 3A). Six genes were up-regulated and 3 genes were with the production and assembly of CE components, down-regulated, the mean log ratio being 0.02 (Table 1). and with their anchorage to the intracellular lipid layer. Howevert, many CE-related genes seemed to be up- We also considered that genes related to TJ formation regulated after two SDS treatments (2 d) (Fig. 3B). and TJ protein production in the epidermis might be Among these genes, those for the main components of relevant. CE (loricrin, filaggrin and involucrin) and the gene for We therefore used a microarray assay to examine TGase-1 showed relatively high expression ratios, changes of mRNA expression in the skin of SDS-treated, although the average log ratio for the 29 CE-related GC-fed mice, and obtained normalized gene-level genes was not large (0.27). The result of the Z-test expression ratio data for GC-fed mice against control-fed showed very high statistical significance (a very low p- mice. We used the Ingenuity Pathway analysis software value) (Table 1). Oral Glucosylceramide Uptake Improves Skin Barrier Function 110215-5 Table 2. List of TJ-Related Genes

Log ratio Gene expression Entre Gene Description/ Gene name (log2 (GC/Control)) ratio (GC/Control) gene ID symbol Function 0d 2d 0d 2d 12737 Cldn1 claudin 1 0.089 0.359 1.064 1.282 transmembrane protein 12739 Cldn3 claudin 3 0:342 0.186 0.789 1.138 transmembrane protein 12740 Cldn4 claudin 4 0.055 0.350 1.039 1.275 transmembrane protein 12741 Cldn5 claudin 5 0:093 0:049 0.937 0.967 transmembrane protein 53624 Cldn7 claudin 7 0.000 0.000 1.000 1.000 transmembrane protein 54420 Cldn8 claudin 8 0.031 0.197 1.022 1.146 transmembrane protein 58187 Cldn10a claudin 10A 0.206 0.535 1.153 1.449 transmembrane protein 64945 Cldn12 claudin 12 0:048 0.090 0.967 1.064 transmembrane protein 71908 Cldn23 claudin 23 0.095 0.272 1.068 1.207 transmembrane protein 18260 Ocln occludin 0.256 0.352 1.194 1.276 transmembrane protein Jam1/F11r junction adhesion molecule 0.020 0.219 1.014 1.164 transmembrane protein 16456 1/ 67374 Jam2 junction adhesion molecule 2 0.046 0:256 1.032 0.838 transmembrane protein 83964 Jam3 junction adhesion molecule 3 0.023 0:225 1.016 0.855 transmembrane protein 218518 Marveld2/Tric tricellulin/MARVEL 0.081 0.017 1.058 1.012 transmembrane protein (membrane-associating) domain containing 2 21872 ZO-1/Tjp1 tight junction protein 1/ 0.055 0.228 1.039 1.171 scaffolding/TJ undercoat zona occludence-1 21873 ZO-2/Tjp2 / 0:151 0.223 0.901 1.167 scaffolding/TJ undercoat zona occludence-2 27375 ZO-3/Tjp3 tight junction protein 3/ 0:016 0.355 0.989 1.279 scaffolding/ TJ undercoat Advancezona occludence-3 View 17475 Mupp-1/Mpdz multiple PDZ domain protein 0.099 0:128 1.071 0.915 scaffolding/TJ undercoat 14924 Magi1 membrane associated guanylate kinase, 0.199 0.289 1.148 1.222 scaffolding/TJ undercoat WW and PDZ domain containing 1 18759 Prkci protein kinase C, iota 0.065 0.221 1.046 1.165 aPKC/Par3/Par6 complex/ cell polarity/TJ assembly 18762 Prkcz protein kinase C, zeta 0:026 0.054 0.982 1.038 aPKC/Par3/Par6 complex/ cell polarity/TJ assembly 93742 Par3/Pard3 par-3 (partitioning defective 3) homolog 0:001 0.270 1.000 1.206 aPKC/Par3/Par6 complex/ cell polarity/TJ assembly 56513 Pard6a Mus musculus par-6 homolog alpha 0:059 0:061 0.960 0.958 aPKC/Par3/Par6 complex/ (C. elegans) Proofscell polarity/TJ assembly 58220 Pard6b Mus musculus par-6 homolog beta 0.000 0.000 1.000 1.000 aPKC/Par3/Par6 complex/ cell polarity/TJ assembly 93737 Pard6g Mus musculus par-6 homolog gamma 0.006 0.169 1.004 1.125 aPKC/Par3/Par6 complex/ cell polarity/TJ assembly 224912 Crb3 crumbs homolog 3 0:024 0.184 0.984 1.136 Crb3/Pals/Patj complex/ cell polarity/TJ assembly 56217 Pals1/Mpp5 membrane protein, palmitoylated 5 0.031 0:027 1.022 0.981 Crb3/Pals/Patj complex/ (MAGUKp55 subfamily member 5) cell polarity/TJ assembly 12695 Patj/Inadl InaD-like/PATJ 0.039 0.010 1.027 1.007 Crb3/Pals/Patj complex/ cell polarity/TJ assembly 68188 Sympk symplekin 0:090 0:011 0.939 0.992 nuclear protein/localaize in TJ/ polyadenylation 17356 AF-6/Mllt4 Afadin/myeloid-lymphoid or 0.112 0.288 1.081 1.221 Adherence Junction Formation/ mixed-lineage leukemia; interaction with ZO-1 translocated to, 4

Number of genes 30 Mean 0.022 0.137 1.018 1.109 SD 0.111 0.188 0.077 0.143 Number of ALL 14518 14547 14518 14547 Mean of ALL 0:001 0.002 0.999 1.002 SD of ALL 0.10 0.17 Z-score 1.220 4.266 p-value 0.111 9:93 106

Gene names, gene symbols, entre IDs and their functions are shown. The mean and SD of the gene-level log scale expression ratio (base 2) and expression ratio (GC/ Control) under each experimental condition are shown (0 d, n ¼ 3;2d,n ¼ 4). The mean and SD of log ratios and ratios of overall genes (ALL) are indicated. Z- scores and Z-test p-values are calculated based on the log scale ratio. An asterisk indicates statistical significance in the Z-test ( p < 0:001). 110215-6 R. IDETA et al. It is known that a gene set enrichment analysis simply induce several important genes for CE-forma- (GSEA) and parametric analysis of gene set enrichment tion, but rather induced a broad increase in expression of (PAGE) are useful to evaluate the significant biological the CE-related genes which would be expected to speed changes of predefined gene sets in microarray data.20) up the barrier recovery in barrier-perturbed skin. Both methods are especially useful when the gene Among the 30 TJ-related genes, many appeared to be expression changes in a given array data set is minimal weakly induced, as in the case of the CE-related genes. or moderate, and PAGE is more sensitive compared to However, the increase in genes of the claudin family and GSEA because it uses Z-scores to evaluate the signifi- occludin, the major transmembrane component of TJ, cance of the difference between the mean of the were relatively higher (Fig. 3B). A Z-test result for the expression of the gene set and the mean of the parent 2 d data of the TJ-related gene set showed a significant population.20) We therefore used Z-scores and corre- difference with a p-value of 9:93 106 (Table 2). It sponding p-values to evaluate whether the average log appears that GC administration enhanced TJ formation ratio of the CE-related genes was increased or not. It through a broad increase in the expression of TJ-related seems likely that the oral GC administration did not genes. We performed a TER assay, which is routinely used to evaluate the TJ barrier function (intact TJ shows high 300 TER and impaired TJ shows low TER39)), to confirm these findings experimentally. As shown in Fig. 2, the 250 ** major sphingometabolites of GC, sphingosine and * 40) ) 2 200 phytosphingosine, significantly improved the TER cm

· value in SDS-treated dNHEK cells, as we expected.

Ω 150 In conclusion, we examined the mechanism for the skin barrier improvement induced by dietary GC, and TER ( Advance∆ 100 Viewour results indicate that sphingoid-based metabolites 50 derived from GC enhanced TJ formation by inducing the increased expression of genes related to TJ formation in 0 the epidermis, as well as promoting barrier recovery by Control (-) Control (+) Sphingosine Phytosphingosine inducing an increased expression of those genes related SDS (-) SDS (+) to CE formation. These findings indicate that dietary GC Fig. 2. Effects of the GC Metabolites on SDS-Treated dNHEK. can play an important role in maintaining a healthy skin The TER value for dNHEKs is shown. Control () represents no barrier condition and in promoting recovery from an SDS treatment, while control (+) represents the SDS treatment. impaired skin barrier by enhancing CE and TJ formation Sphingosine and phytosphingosine represent cells treated for 72 h 6 at the gene expression level. We have recently shown with a 10 mM sphingoid base after the SDS treatment. All data are 35) expressed as the mean SD (n ¼ 4). An asterisk indicates statistical the increased expression of Tgm-1 mRNA and significance in Dunnett’s multiple-comparison test ( p < 0:05, claudin-1 and involcrinProofs proteins in GC-fed barrier- p < 0:01). perturbed mice skin after UV exposure (data are not

A B

Fig. 3. Schematic Illustration of the Changes in Expression of CE-Related Genes and TJ-Related Genes on (A) Day 0 and (B) Day 2 Drawn by Using the Pathdesigner Function of IPA. A red color represents genes up-regulated by more than 1.1 times in GC-fed mice against control mice. Green indicates genes down-regulated by below 0.9 times in GC-fed mice against control mice. Oral Glucosylceramide Uptake Improves Skin Barrier Function 110215-7 Table 3. List of AJ-Related Genes

Log ratio Gene expression Entre Gene Description/ Gene name (log2 (GC/Control)) ratio (GC/Control) gene ID symbol Function 0d 2d 0d 2d 12550 Cdh1 cadherin 1/E-cadherin 0.067 0.312 1.048 1.242 AJ formation/adhesion molecule 12560 Cdh3 cadherin 3/P-cadherin 0.189 0.285 1.140 1.219 AJ formation/adhesion molecule 12385 Ctnna1 catenin, alpha 1 0:045 0.083 0.969 1.059 AJ formation/cadherin associate protein 12386 Ctnna2 catenin, alpha 2 0:484 0.000 0.715 1.000 AJ formation/cadherin associate protein 12387 Ctnnb1 catenin, beta 1 0.082 0.166 1.058 1.122 AJ formation/cadherin associate protein 12388 Ctnnd1 catenin, delta 1 0.000 0.039 1.000 1.028 AJ formation/cadherin associate protein 18163 Ctnnd2 catenin, delta 2 0.000 0.000 1.000 1.000 AJ formation/cadherin associate protein 16480 Jup junction plakoglobin 0.096 0.251 1.069 1.190 AJ formation/cadherin associate protein 58235 Pvrl1 poliovirus receptor-related 1/ 0:025 0.366 0.983 1.288 AJ formation/adhesion molecule nectin-1 19294 Pvrl2 poliovirus receptor-related 2/ 0.050 0.170 1.035 1.125 AJ formation/adhesion molecule nectin-2 58998 Pvrl3 poliovirus receptor-related 3/ 0:095 0.089 0.936 1.064 AJ formation/adhesion molecule nectin-3 71740 Pvrl4 poliovirus receptor-related 4/ 0:036 0.282 0.976 1.216 AJ formation/adhesion molecule nectin-4 17356 AF-6/Mllt4 Afadin/myeloid-lymphoid 0.112 0.288 1.081 1.221 AJ Formation/nectin scaffolding/ or mixed-lineage leukemia; interaction with ZO-1 translocated to, 4

mean 0:007 0.179 1.001 1.136 AdvanceSD 0.163 View 0.127 0.102 0.099 Gene names, gene symbols, entre IDs and their functions are shown. The mean and SD of the gene-level log scale expression ratio (base 2) and expression ratio under each experimental condition are shown (0 d, n ¼ 3;2d,n ¼ 4). shown). These results also suggest the relationship 2) Oku H, Wongtangtintharn S, Iwasaki H, Inafuku M, Shimatani between CE- and TJ-related genes in the dietary GC M, and Toda T, Cancer Chemother. Pharmacol., 60, 767–775 effect against skin barrier perturbation. (2007). 3) Pinto MR, Rodrigues ML, Travassos LR, Haido RM, Wait R, Although the involvement of CE- and TJ-related and Barreto-Bergter E, Glycobiology, 12, 251–260 (2002). genes in promoting barrier recovery by dietary GC was 4) Kawano T, Cui J, Koezuka Y, Toura I, Kaneko Y, Motoki K, experimentally confirmed, it is also possible that other Ueno H, NakagawaProofs R, Sato H, Kondo E, Koseki H, and gene groups were involved. We have included afadin in Taniguchi M, Science, 278, 1626–1629 (1997). this study as one of the TJ-related genes, because it 5) Langeveld M and Aerts JM, Prog. Lipid Res., 48, 196–205 directly binds to ZO-1 before the formation of TJ and is (2009). involved in the first step of recruitment for TJ on the 6) Jennemann R, Sandhoff R, Langbein L, Kaden S, Rothermel U, 34) Gallala H, Sandhoff K, Wiegandt H, and Grone HJ, J. Biol. apical side of the cells. Afadin is also essential to the Chem., 282, 3083–3094 (2007). formation of the adherens junction (AJ), and the 7) Chuong CM, Nickoloff BJ, Elias PM, Goldsmith LA, Macher E, formation of AJ is known to facilitate the assembly of Maderson PA, Sundberg JP, Tagami H, Plonka PM, Thestrup- TJ.16) Genes related to AJ formation such as those Pederson K, Bernard BA, Schroder JM, Dotto P, Chang CM, encoding cadherins, catenins and nectins may therefore Williams ML, Feingold KR, King LE, Kligman AM, Rees JL, also be involved in the promotion of barrier recovery by and Christophers E, Exp. Dermatol., 11, 159–187 (2002). Nat. Rev. Mol. Cell Biol. 6 dietary GC. Indeed, weak up-regulation of many AJ- 8) Candi E, Schmidt R, and Melino G, , , 328–340 (2005). related genes was apparent in our microarray results 9) Hitomi K, Eur. J. Dermatol., 15, 313–319 (2005). (Table 3). It therefore appears that dietary GC promoted 10) Kunii T, Hirao T, Kikuchi K, and Tagami H, Br. J. Dermatol., recovery of the skin barrier function by inducing 149, 749–756 (2003). relatively weak increases in expression of many genes 11) Di Nardo A, Wertz P, Giannetti A, and Seidenari S, Acta Derm. related to the formation and maintenance of the skin Venereol., 78, 27–30 (1998). barrier system. Although the changes in expression of 12) Furuse M, Hata M, Furuse K, Yoshida Y, Haratake A, Sugitani Y, Noda T, Kubo A, and Tsukita S, J. Cell Biol., 156, 1099– individual genes were small, such a broadly based action 1111 (2002). might well explain the efficacy of an oral GC intake. 13) Yamamoto T, Saeki Y, Kurasawa M, Kuroda S, Arase S, and Sasaki H, Arch. Dermatol. Res., 300, 517–524 (2008). Acknowledgment 14) Kirschner N, Bohner C, Rachow S, and Brandner JM, Arch. Dermatol. Res., 302, 483–493 (2010). We greatly thank Dr. Jotaro Nakanishi for technical 15) Brandner JM, Eur. J. Pharm. Biopharm., 72, 289–294 (2009). guidance and helpful discussions. 16) Niessen CM, J. Invest. Dermatol., 127, 2525–2532 (2007). 17) Roh MH, Fan S, Liu CJ, and Margolis B, J. Cell Sci., 116, 2895–2906 (2003). References 18) Tsuji K, Mitsutake S, Ishikawa J, Takagi Y, Akiyama M, Shimizu H, Tomiyama T, and Igarashi Y, J. Dermatol. Sci., 44, 1) Chua M, Baldwin TC, Hocking TJ, and Chan K, J. Ethno- 101–107 (2006). pharmacol., 128, 268–278 (2010). 19) Uchiyama T, Nakano Y, Ueda O, Mori H, Nakashima M, Noda 110215-8 R. IDETA et al. A, Ishizaki C, and Mizoguchi M, J. Health Sci., 54, 559–566 Yoneda K, Imamura S, Fujimoto K, and Tsukita S, J. Invest. (2008). Dermatol., 110, 862–866 (1998). 20) Kim SY and Volsky DJ, BMC Bioinformatics, 6, 144 (2005). 32) Tsukita S, Katsuno T, Yamazaki Y, Umeda K, Tamura A, and 21) Rosner BR, ‘‘Fundamentals of Biostatics,’’ Duxbury Press, Tsukita S, Ann. NY Acad. Sci., 1165, 44–52 (2009). Pacific Grove, pp. 300–306 (2000). 33) Kavanagh E, Buchert M, Tsapara A, Choquet A, Balda MS, 22) Steinert PM, Kartasova T, and Marekov LN, J. Biol. Chem., Hollande F, and Matter K, J. Cell Sci., 119, 5098–5105 (2006). 273, 11758–11769 (1998). 34) Ooshio T, Kobayashi R, Ikeda W, Miyata M, Fukumoto Y, 23) Nemes Z and Steinert PM, Exp. Mol. Med., 31, 5–19 (1999). Matsuzawa N, Ogita H, and Takai Y, J. Biol. Chem., 285, 5003– 24) Robinson NA, Lapic S, Welter JF, and Eckert RL, J. Biol. 5012 (2010). Chem., 272, 12035–12046 (1997). 35) Hasegawa T, Shimada H, Ueda O, Matsuoka Y, and Nakashima 25) Steinert PM, J. Cell Biol., 151, F5–F8 (2000). M, Lipids, in press. 26) Yamada H, Takamori K, and Ogawa H, Arch. Dermatol. Res., 36) Marekov LN and Steinert PM, J. Biol. Chem., 273, 17763– 279, 194–197 (1987). 17770 (1998). 27) Nicotera P and Melino G, Nat. Cell Biol., 9, 621–622 (2007). 37) Nemes Z, Marekov LN, Fesus L, and Steinert PM, Proc. Natl. 28) Turksen K and Troy TC, J. Cell Sci., 117, 2435–2447 (2004). Acad. Sci. USA, 96, 8402–8407 (1999). 29) Furuse M, Sasaki H, Fujimoto K, and Tsukita S, J. Cell Biol., 38) Behne M, Uchida Y, Seki T, de Montellano PO, Elias PM, and 143, 391–401 (1998). Holleran WM, J. Invest. Dermatol., 114, 185–192 (2000). 30) Bazzoni G, Martinez-Estrada OM, Orsenigo F, Cordenonsi M, 39) Chen YH, Lu Q, Goodenough DA, and Jeansonne B, Mol. Biol. Citi S, and Dejana E, J. Biol. Chem., 275, 20520–20526 Cell, 13, 1227–1237 (2002). (2000). 40) Ueda O, Uchiyama T, and Nakashima M, Drug Metab. 31) Morita K, Itoh M, Saitou M, Ando-Akatsuka Y, Furuse M, Pharmacokinet., 25, 456–465 (2010). Advance View

Proofs