Retinol Dehydrogenase 10 but Not Retinol/SterolDehydrogenase(s) Regulates the Expression of RetinoicAcid-responsive Genes in Human Transgenic Skin Raft Culture*□S

Received for publication, September 1, 2010, and in revised form, February 18, 2011 Published, JBC Papers in Press, February 23, 2011, DOI 10.1074/jbc.M110.181065

Seung-Ah Lee1, Olga V. Belyaeva1, Lizhi Wu, and Natalia Y. Kedishvili2

From the Department of Biochemistry and Molecular Genetics, Schools of Medicine and Dentistry, University of Alabama atBirmingham, Birmingham, Alabama 35294

Retinoic acid is essential for skin growth and differentiation,and its concentration in skin is controlled tightly. In humans,four different members of the short-chain dehydrogenase/re-ductase (SDR) superfamily of proteinswereproposed to catalyzethe rate-limiting step in the biosynthesis of retinoic acid (theoxidation of retinol to retinaldehyde). Epidermis contains atleast three of these enzymes, but their relative importance forretinoic acid biosynthesis and regulation of gene expressionduring growth and differentiation of epidermis is not known.Here, we investigated the effect of the four human SDRs on ret-inoic acid biosynthesis, and their impact on growth and differ-entiation of keratinocytes using organotypic skin raft culturemodel of human epidermis. The results of this study demon-strate that ectopic expression of retinol dehydrogenase 10(RDH10, SDR16C4) in skin rafts dramatically increases prolif-eration and inhibits differentiation of keratinocytes, consistentwith the increased steady-state levels of retinoic acid and acti-vation of retinoic acid-inducible genes in RDH10 rafts. In con-trast, SDRswith dual retinol/sterol substrate specificity, namelyretinol dehydrogenase 4 (RoDH4, SDR9C8), RoDH-like 3�-hy-droxysteroid dehydrogenase (RL-HSD, SDR9C6), andRDH-likeSDR (RDHL, SDR9C4) do not affect the expression of retinoicacid-inducible genes but alter the expression levels of severalcomponents of extracellular matrix. These results reveal essen-tial differences in the metabolic contribution of RDH10 versusretinol/sterol dehydrogenases to retinoic acid biosynthesisand provide the first evidence that non-retinoid metabolicproducts of retinol/sterol dehydrogenases affect gene expres-sion in human epidermis.

All-trans-retinoic acid is a small lipophilic molecule derivedfrom vitamin A that regulates the expression of �530 differentgenes in various types of cells and tissues through binding to

nuclear transcription factors (retinoic acid receptors (RARs))3(1). Skin is one of the best characterized targets of retinoidaction (2, 3) and contains all of the major components of reti-noid signaling andmetabolicmachinery such as retinoid recep-tors RAR� and RXR� (4); cellular retinoic acid binding protein(5) and cellular retinol binding protein (6, 7); retinaldehydedehydrogenase (8); lecithin retinol acyltransferase (9); andcytochrome P450 (10). Keratinocytes can synthesize retinoicacid in situ from plasma-derived retinol (11–13); the reportedconcentration of retinoic acid in the epidermal cells is very low(�20 nM), and it is controlled strictly (14). Concentrations thatexceed the optimal range suppress differentiation and promotehyperproliferation, whereas concentrations below this rangelead to formation of orthokeratotic epithelium (15).Retinoic acid is synthesized from retinol in two steps; first,

retinol is reversibly oxidized to retinaldehyde, and then retin-aldehyde is oxidized irreversibly to retinoic acid. The oxidationof retinol to retinaldehyde is the rate-limiting step in retinoicacid biosynthesis (16). Recent studies suggested that this step iscatalyzed by the members of the short-chain dehydrogenase/reductase (SDR) superfamily of proteins (17) (for SDR nomen-clature, see Ref. 18). In humans, four different SDRswere impli-cated in the biosynthesis of retinoic acid. Three of theseenzymes, namely retinol dehydrogenase 4 (RoDH4, SDR9C8),RoDH-like 3�-hydroxysteroid dehydrogenase (RL-HSD,SDR9C6), and RDH-like SDR (RDHL, also known as DHRS9,SDR9C4) share significant sequence similarity with oneanother and belong to the same branch of the SDRphylogenetictree (18, 19). Besides the retinol dehydrogenase activity, allthree of these human enzymes exhibit high activity toward3�-hydroxysteroids and were proposed to catalyze the backconversion of inactive 5�-androstane-3�,17�-diol to thepotent androgen dihydrotestosterone (20) and to oxidize andinactivate the bioactive neurosteroid allopregnanolone (21). Inaddition, RL-HSD was shown to exhibit a 3(�3�)-hydroxy-steroid epimerase activity, converting 3�-hydroxysteroids into3�-hydroxysteroids (22). The fourth SDR enzyme that wasshown to catalyze the oxidation of retinol for retinoic acid bio-synthesis is retinol dehydrogenase 10 (RDH10, SDR16C4) (23,

* This work was supported by National Institute on Alcohol Abuse and Alco-holism Grant AA12153 (to N. Y. K.). The University of Alabama-BirminghamTargeted Metabolomics and Proteomics Laboratory is supported by theNational Institutes of Health Grants RR19231, P30 CA13148, P50 AT00477,U54 CA100949, P30 AR50948, and P30 DK079337.

□S The on-line version of this article (available at http://www.jbc.org) containssupplemental Fig. 1.

1 Both authors contributed equally to this work.2 To whom correspondence should be addressed: Dept. of Biochemistry and

Molecular Genetics, Schools of Medicine and Dentistry, University of Ala-bama-Birmingham, 720 20th St. South, 440B Kaul Genetics Bldg., Birming-ham, AL 35294. Tel.: 205-996-4023; Fax: 205-934-0758; E-mail: nkedishvili@uab.edu.

3 The abbreviations used are: RAR, retinoic acid receptor; RDH or RoDH,retinol dehydrogenase; HSD, hydroxysteroid dehydrogenase; RL-HSD,RoDH-like 3�-hydroxysteroid dehydrogenase; RDHL, retinol dehydro-genase-like); RSDH, retinol/sterol dehydrogenase; SDR, short-chaindehydrogenase/reductase; PHK, primary human keratinocyte; ADT,androsterone; ALLO, allopregnanolone.

THE JOURNAL OF BIOLOGICAL CHEMISTRY VOL. 286, NO. 15, pp. 13550 –13560, April 15, 2011© 2011 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in the U.S.A.

13550 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 286 • NUMBER 15 • APRIL 15, 2011

by guest on September 10, 2020

http://ww

w.jbc.org/

Dow

nloaded from

24). RDH10 shares little similarity with the retinol/sterol dehy-drogenases (abbreviated here as RSDHs) described above andbelongs to a different branch of the SDR phylogenetic tree. It isnot yet known whether RDH10 is active toward hydroxy-steroids or any other substrates besides retinoids.Data from this and other laboratories (25–27) indicate that

the human epidermis contains at least three of the retinoid-active SDRs. However, their relative roles in the biosynthesis ofretinoic acid and regulation of gene expression are not known.In part, this is due to technical difficulties associated with ana-lyzing functions of enzymes that are expressed at very low levelsin human cell lines and tissues and their generally low enzy-matic activity. Furthermore, the use of mouse models has beenlimited by the redundancy of RoDH-like SDRhomologs inmiceand the lack of orthologs for some of the human genes (19). Inthis study, we took advantage of the ex vivo human organotypicskin culture to examine the contribution of human SDRs toretinoic acid biosynthesis and their impact on gene expression.Human organotypic skin culture is very similar to human

skin in its morphology and metabolism because human fore-skin keratinocytes are grown at the liquid-air interface, a systemthat recreates fully differentiated squamous epithelium (28).Keratinocytes placed on top of the collagen bed receive mois-ture and nutrients through the support matrix, growingupwards and forming a “raft” culture. The keratinocytes in thisraft culture proliferate, stratify, differentiate, and form layersjust like normal skin. Importantly, this model recreates thecomplex process of epidermal differentiation that involves thetemporal and spatial regulation of a large number of key mole-cules (29); gene expression pattern in raft cultures is very simi-lar to that seen in whole foreskin (30, 31). The genetic make-upof skin raft culture can be manipulated using retrovirus-medi-ated gene expression. This aspect of skin raft culturemodel wasutilized in the present study to investigate the individual con-tribution of the four human SDR enzymes to the biosynthesis ofretinoic acid and regulation of gene expression during thegrowth and differentiation of human epidermis. The results ofthis study reveal important differences in the physiologicalroles of the four human SDRs.

EXPERIMENTAL PROCEDURES

Retroviral Vectors—The cDNAs encoding human RDH10,RoDH4, RL-HSD, and RDHL were PCR-amplified from clonesdescribed previously using the primers containing BamHI andEcoRI restriction sites. Sequences of primers are as follows:RDH10, 5�-GTG GAT CCA TGA ACA TCG TGG TGG AGTTCT TCG-3� (forward) and 5�-GAG AAT TCA GAT TCTTAGATT CCA TTT TTTGCT TCA T-3� (reverse); RL-HSD,5�-ATCGGATCCATGTGGCTCTACCTGGCAGCCT-3�(forward) and 5�-ATC GAA TTC TTA GAC TGC CTG GGCTGG TTT GG-3� (reverse); RDHL, 5�-ATC GGA TCC ATGCTC TTT TGGGTG CTAGGC CT-3� (forward) and 5�-ATCGAA TTC TCA CAC TGC CTT GGG ATT AGC C-3�(reverse); and RoDH4, 5�-ATC GGA TCC ATG TGG CTCTAC CTG GCG GTT TTC-3� (forward) and 5�-ATC GAATTC TCA TAG AGC CTT GGC CGG GCT TG-3� (reverse).The PCR products were digested with BamHI and EcoRI endo-nucleases and cloned into the corresponding sites of Moloney

murine leukemia retroviral vector pBabe puro under the con-trol of the retroviral long terminal repeat promoter. All of theconstructs were verified by sequencing.Production of Retroviruses—Retroviral vectors were intro-

duced into an ecotropic cell line Bosc23 by electroporation.Culture media from electroporated cells containing ecotropicretroviruses were used to infect GP�envAM12 cells (ATCC,Manassas, VA), an NIH3T3-derived amphotropic packagingcell line. Twenty-four hours after infection, the cells wereplaced into selection medium containing 1 �g/ml puromycinfor 6 days. The resistant cells were allowed to reach confluenceto obtain high titer retroviruses. The retrovirus-containingmedia from the producer cells was used to infect primaryhuman keratinocytes (PHKs) (32).Preparation of Transgenic Organotypic Skin Rafts—Neonatal

foreskins were obtained from theNewbornNursery of the Uni-versity of Alabama at BirminghamHospital in compliance withUniversity of Alabama at Birmingham Institutional ReviewBoard regulations. Epidermal raft cultures were prepared asdescribed previously (32). Briefly, PHKs were isolated fromfreshly collected neonatal foreskins and cultured in DermaLifecalcium-free medium (Lifeline Cell Technology, Walkersville,MD). Freshly collected retrovirus-containing media from pro-ducer cells were used to infect PHKs. Infected PHKs wereselectedwith 1.5�g/ml puromycin for 2 days and then culturedin DermaLife calcium-free medium until confluency. Retrovi-rus-transduced PHKs (4 � 105 cells/ml) were seeded onto adermal equivalent consisting of collagen with embedded Swiss3T3 J2 fibroblasts (kindly provided by Dr. Louise T. Chow,Department of Biochemistry and Molecular Genetics, Univer-sity of Alabama, Birmingham). After 3 days, skin equivalentswere lifted onto stainless steel grids and cultured at the medi-um-air interface using raft culturemedium prepared fromDul-becco’s modified Eagle’s medium, Ham’s F12 medium, andbovine fetal serum, which was supplemented with choleratoxin, insulin, apo-transferrin, hydrocortisone-21, and humanepidermal growth factor as described previously (32). The raftcultures were allowed to stratify and differentiate for 10–11days, whereupon they were harvested for analysis. Twelvehours before harvest, the medium was supplemented withBrdU (50 �g/ml) to mark cells in S phase. The rafts were har-vested, fixed in 10% buffered formalin, and embedded in paraf-fin. Alternatively, epithelial tissues were separated manuallyfrom the collagen bed and used for protein extraction.Immunohistochemistry andH&E Staining—Paraffin-embed-

ded skin rafts were cut into 5-�m sections, mounted on Super-frost/Plus slides (Fisher Scientific, Pittsburgh, PA), and thendeparaffinized and rehydrated by a series of washes in contain-ers with decreasing concentrations of ethanol (95, 85, 70, 50,and 30%). For H&E staining, the sections were incubated withWeigert’s hematoxylin (Poly Scientific, Bay Shore, NY), dehy-drated through graded ethanol, restained with eosin (FisherScientific), and mounted with Permount (Fisher Scientific).For immunohistochemical staining, the sections were pre-

treated with sodium citrate buffer (pH 6.0) at 95 °C for 10 minto unmask antigens, followed by incubation in 3% hydrogenperoxide for 20 min to block endogenous peroxidase activity.The slides were then washed in PBS (pH 7.2) and incubated

Functional Analysis of Retinoid-active SDRs in Human Skin

APRIL 15, 2011 • VOLUME 286 • NUMBER 15 JOURNAL OF BIOLOGICAL CHEMISTRY 13551

by guest on September 10, 2020

http://ww

w.jbc.org/

Dow

nloaded from

with primary antibodies as follows: a 1:100 dilution of BrdUantibodies (Invitrogen); and a 1:100 dilution of filaggrin anti-bodies (Leica Microsytems, Inc. Bannockburn, IL). After incu-bation with primary antibodies at 4 °C overnight, samples wererinsed with PBS and reincubated with a 1:50 dilution of biotin-ylated secondary antibodies (anti-rabbit, anti-mouse immuno-globulins) from the SuperSensitive Link-Label-IHC detectionkit manufactured by Biogenex (San Ramon, CA) for 30 min atroom temperature. After washing with PBS several times, thesections were incubated for 30 min with a 1:50 dilution ofstreptavidin-conjugated horseradish peroxidase followed byincubation with 3,3�-diaminobenzidine tetrahydrochloridefrom Turbbo DAB kit (Innovex Biosciences, Richmond, CA) asthe chromogenic substrate. Sections were counterstained withWeigert’s hematoxylin (Poly Scientific), dehydrated throughgraded ethanol, cleared in xylene, andmounted with Permount(Fisher Scientific). All sections were analyzed at a 20� magni-fication using AxioImager A2 microscope equipped with anAxioCam camera andAxioVision image capture software (CarlZeiss MicroImaging, Inc., Thornwood, NY).Western Blotting—The epidermis from raft cultures was

peeled off collagen beds and homogenized on ice in PBS with20% glycerol containing a mixture of protease inhibitors: apro-tinin (1�g/ml), leupeptin (1�g/ml), pepstatin A (1�g/ml), andphenylmethylsulfonyl fluoride (50�g/ml). Lysates were clearedby centrifugation at 16,100 � g for 15 min at 4 °C. Protein con-centrations were quantified using Bio-Rad DC protein assay(Bio-Rad). Samples were resolved by electrophoresis in 15%SDS-PAGE and transferred to a nitrocellulose membrane forsubsequent probing with polyclonal antibodies against RoDH4(1:5000 dilution) (33), RL-HSD (1:5000 dilution) (34), RDHL(1:5000 dilution) (21), or with affinity-purified polyclonal anti-body against RDH10 (1:500 dilution) (24) in conjunction withthe appropriate HRP-conjugated secondary antibody and ECL-enhanced chemiluminescence detection system (GE Health-care). Gel loading was determined by reprobing withHRP-con-jugated monoclonal �-actin antibody (1:5000; Sigma-Aldrich).Activity Assays—Organotypic cultures of PHKs expressing

RDH10, RoDH4, RL-HSD, and RDHL were prepared essen-tially as described above with minor modifications. Retrovirus-transduced PHKs (4 � 105 cells/ml) were seeded onto CostarTranswells (Corning Life Sciences, Lowell, MA), which wereplaced into 24-well plates. At confluency, transwells containingPHKs were transferred onto dermal equivalents, and skin raftswere cultured as described above. For analysis of retinoidmetabolism, culture medium was supplemented with 2 �M all-trans-retinol under reduced light 24 h before harvest. Raftswere peeled off transwells and homogenized in PBS on ice. Cul-ture media were collected separately. All manipulations weredone under reduced light. Retinoidswere extracted into hexaneand separated by reversed phase HPLC using aWaters Allianceseparation module and 2996 Photodiode array detector (35).Chromatographic peaks were identified by comparing reten-tion times and spectra against retinoid standards andquantifiedas described previously (36). Total retinoids were normalizedby tissue weight and protein amount.For steroid activity assays, commercially available radiola-

beled steroids (PerkinElmer Life Sciences; �40–60 Ci/mmol

each) were diluted with cold steroids (Steraloids, Inc. Newport,RI) dissolved inMe2SO. For analysis of activities in raft cultures,steroid stock solutions were added to skin raft culture mediumcontaining fetal bovine serum. Steroid reference standardswere generated by incubating Sf9 microsomes expressing RL-HSDwith androsterone (ADT) or allopregnanolone (ALLO) inphosphate buffer (90mMKH2PO4, 40mMKCl, pH7.4) contain-ing fatty acid-free BSA equimolar to steroids as described pre-viously (34).After a 24-h incubation, media and cells were collected sep-

arately as described above and extracted with 7 volumes ofdichloromethane. The organic layer was evaporated under astream of nitrogen and dissolved in 50 �l of fresh dichloro-methane. Steroids were separated by development in toluene:acetone (4:1) on silica gel TLC plates (Sigma-Aldrich, St. Louis,MO). TLC plates containing 3H-labeled steroids were exposedto a PhosphorImager tritium screen (GEHealthcare) overnight,and the intensity of the bandswas calculated using ImageQuantprogram (version 5.0). Products of each reaction were identi-fied by comparison to reference steroids (34).LC/MS-MRMAnalysis—Sampleswere separated by gradient

reversed phase high performance liquid chromatography. Thegradientwas generatedwith a Shimadzu ProminenceUltra FastLiquid Chromatography system consisting of a vacuum degas-ser, binary pump, and a temperature-controlled autosampler.The injection volume was 80 �l, and the samples were main-tained in the autosampler at 4 °C. The separation was per-formed on a Develosil RP Aqueous 2.00 mm � 150 mm (5 �)column with a 2.00 mm � 30 mm guard column. Mobile phaseA was 0.1% formic acid in water; mobile phase B was acetoni-trile � 0.1% formic acid. The gradient was as follows: 0–3 min,70% B; 4–6 min, 95% B; and 6.5–10 min, 70% B. The flow ratewas 0.75 ml/min.Data were acquired with an Applied Biosystems API-4000

triple-quadrupole mass spectrometer equipped with atmo-spheric pressure chemical ionization operating in positive ionmode. The mass spectrometer and HPLC were controlled withAnalyst (version 1.4.2) in multiple reaction monitoring (MRM)mode. The mass transitions used to monitor all-trans-retinoicacid werem/z 301 (M � H)� to them/z 123, 105, and 81 frag-ments. The dwell time was 40 ms for all mass transitions. Theatmospheric pressure chemical ionization-optimized condi-tions include the following: curtain gas, 10; collision gas, 12;nebulizer gas, 70; nebulizer current, 4; and source temperature,350. All gases were nitrogen.q-PCR—RNA from rafts was isolated using TRIzol reagent

(Invitrogen). First-strand cDNA was synthesized using Super-Script III kit (Invitrogen). cDNAwas purified using aMiniPrepDNA purification kit (Qiagen). Real-time PCR was performedin a LightCycler� 480 instrument (Roche Applied Science)using LightCycler� 480 SYBR Green I Master Mix (RocheApplied Science) with 0.5�M primers and 2.5–25 ng of purifiedcDNA per reaction. Sequences of the primers are available byrequest. Levels of transcripts were determined using relativequantification method (37). Three empty vector-transducedand three SDR-expressing rafts were included in each q-PCRexperiment. To evaluate the significance of differences inexpression levels of each transcript between control and SDR-

Functional Analysis of Retinoid-active SDRs in Human Skin

13552 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 286 • NUMBER 15 • APRIL 15, 2011

by guest on September 10, 2020

http://ww

w.jbc.org/

Dow

nloaded from

expressing rafts, an unpaired t test was performed, and the two-tailed p value was determined using GraphPad InStat (version3.00; GraphPad Software, San Diego California).

RESULTS

Expression of Human SDRs in Transgenic Skin Rafts—Previ-ous studies have indicated that the retinoid-active SDRs areexpressed naturally in human skin (25–27), suggesting that skinorganotypic culture is a physiologically relevantmodel for analyz-ing functions of these enzymes. We have confirmed the presenceof RoDH4, RDHL, and RDH10 transcripts in human organotypicraft cultures by RT-PCR (supplemental Fig. 1). However, the pro-tein levels ofmost of these enzymes were below detection limit byWestern blotting (Fig. 1). Therefore, we examined their functionsby increasing the levels of individual proteins through retrovirus-mediated ectopic expression of the corresponding cDNAs. PHKsfrom neonatal foreskin infected with retroviruses were platedon a collagen-fibroblast matrix and then lifted to the air-liquidinterface to generate stratified and differentiated epithelium.To confirm the ectopic expression of the SDRs, epidermis wasseparated manually from collagen beds and processed forimmunoblotting. As shown in Fig. 1, protein levels of RDH10,RoDH4, and RL-HSD, which were below detection limit byavailable antisera in skin rafts infected with empty virus (Fig. 1,A–C), increased significantly following transduction with thecorresponding recombinant retroviruses. The levels of RDHLprotein, which was visible in mock-transduced rafts, increasedfurther in transgenic skin infected with RDHL-expressing ret-rovirus (Fig. 1D). This analysis confirmed successful expressionof the SDR proteins in transgenic skin.Histological Analysis of Skin Rafts—As PHKs proliferated

and formed stratified epithelium, significant differences were

noted in the appearance of rafts expressing RDH10 comparedwith other rafts. RDH10 rafts appeared to be softer and moretransparent than mock-transduced rafts or rafts expressingRoDH4, RL-HSD, or RDHL. To evaluate the morphology oftransgenic epidermis, rafts were sectioned and stained withH&E.This analysis revealed striking differences in the histologyof RDH10 rafts compared with other rafts (Fig. 2). RoDH4, RL-HSD, RDHL, and mock-transduced rafts had a well developedhistological appearance with clearly visible spinous, granular,and cornified layers. In contrast, the granular layer in RDH10rafts was developed poorly, and the cornified layer was muchthinner than in control cultures or completely absent. Extend-ing the time of RDH10 rafts in culture from11 to 16 days did notproduce a more differentiated epidermis (data not shown),indicating that the differentiation was not simply delayed butwas profoundly impaired. Immunohistochemical staining forBrdU incorporation revealed increased proliferation of basalcells in RDH10 rafts (Fig. 2). Furthermore, the expression ofkeratinocyte differentiation marker filaggrin, which was de-tected in a continuous band in the granular layer of RSDH raftsandmock-transduced rafts, was severely reduced or completelyabsent in RDH10 rafts (Fig. 2), indicating an overall inhibitionof keratinocyte differentiation and stratification/cornificationof the epidermis.Analysis of Retinoic Acid-responsive Genes—Retinoic acid is

known to induce proliferation of basal keratinocytes and toinhibit their differentiation (2, 3). To determine whether theabnormal differentiation pattern in RDH10 rafts was associatedwith up-regulation of retinoic acid-responsive genes, we car-ried out quantitative RT-PCR analysis of several well knowntargets of retinoic acid such as STRA6, RAR�, lecithin retinolacyltransferase, CRBP1, CRABP2, DHRS3, and CYP26A1. Thisanalysis revealed significant up-regulation of these genes inRDH10 rafts (Fig. 3A). A corresponding increase in protein lev-els was confirmed by Western blot analysis for CRBP1 andDHRS3 (Fig. 3B). In contrast, the retinoic acid-responsive geneswere not up-regulated in skin rafts expressing RoDH4,RL-HSD, or RDHL (Fig. 3A). This result suggested that anincrease in the steady-state levels of retinoic acid occurred inRDH10 rafts, but not in the rafts expressing RoDH4, RL-HSD,or RDHL enzymes.Analysis of Retinoic Acid Production—To examine retinoic

acid biosynthesis in RDH10 rafts versus mock-infected rafts,epidermis was grown in transwell inserts to exclude the contri-bution of collagen-embedded fibroblasts to the metabolism ofretinol. Prior to the addition of retinol, transwell inserts withstratified epidermiswere lifted off collagen beds and placed intomultiwell culture plates containing medium supplementedwith retinol or vehicle. After an overnight incubation, the raftswere peeled off transwells, homogenized, and extracted forHPLC analysis following the protocol designed tomaximize theyield of retinoic acid (38). This analysis revealed that RDH10rafts that were grown in regular culture medium and did notreceive additional retinol contained a peak that was not detect-able in control rafts (Fig. 4A) and had a retention time andspectral properties identical to those of all-trans-retinoic acid(Fig. 4, B, peak 2 and inset 2, and C). Another peak detected inRDH10 rafts but not in control rafts had spectral properties of

FIGURE 1. Western blot analysis of SDR expression in skin rafts. A, threeskin rafts were peeled off collagen beds, combined, and homogenized usinga Dounce homogenizer. The microsomal fraction was isolated by differentialcentrifugation and used for Western blotting (100 �g). The loading control forthe microsomal fraction was P450 reductase. Total homogenate of HEK293cells expressing recombinant RDH10 (RDH10 st.) was used as a molecularweight marker for RDH10 (30 �g). B–D, total homogenates prepared fromsingle rafts were used for Western blotting (50 �g); the loading control was�-actin. Microsomes from Sf9 cells (0.3– 0.5 �g) expressing correspondingSDRs were used as markers of the molecular masses.

Functional Analysis of Retinoid-active SDRs in Human Skin

APRIL 15, 2011 • VOLUME 286 • NUMBER 15 JOURNAL OF BIOLOGICAL CHEMISTRY 13553

by guest on September 10, 2020

http://ww

w.jbc.org/

Dow

nloaded from

all-trans-3,4-didehydro-retinoic acid (Fig. 4B, peak 1 and inset1) (39). The presence of all-trans-3,4-didehydro-retinoic acid inskin epidermis is consistent with the previous reports (40–42);however, a conclusive identification of this peak was not possi-ble in the absence of commercially available all-trans-3.4-dide-hydro-retinoic acid standard.

The identity of the peak that had the retention time and spec-tral properties of all-trans-retinoic acid was examined furtherby LC/MS-MRM analysis. Three fragment ions that had thehighest signal to noise ratio in MRM analysis were selected tomonitor all-trans-retinoic acid. As shown in Fig. 4D, the threemass transitions eluted at the same retention time and in

FIGURE 2. Analysis of skin rafts morphology. Skin rafts embedded in paraffin were sectioned and stained with H&E for analysis of skin morphology, with BrdUantibodies for analysis of cell proliferation, and with filaggrin antibodies for analysis of cell differentiation as described under “Experimental Procedures.”

Functional Analysis of Retinoid-active SDRs in Human Skin

13554 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 286 • NUMBER 15 • APRIL 15, 2011

by guest on September 10, 2020

http://ww

w.jbc.org/

Dow

nloaded from

approximately the same ratio for both the retinoic acid stan-dard and the RDH10 skin raft sample, confirming the presenceof all-trans-retinoic acid in RDH10 rafts.The amount of endogenous all-trans-retinoic acid in RDH10

rafts grown in regular medium was estimated to be �0.15pmol/mg cellular protein. An overnight incubation inmedium supplemented with 2 �M retinol resulted in furtherincrease in the amount of retinoic acid in RDH10 rafts (up to0.52 pmol/mg cellular protein) but not in control rafts. Thisend point activity assay demonstrated that RDH10 was cat-alytically active toward retinol and that overexpression of

RDH10 resulted in the overall increase in the steady-statelevels of retinoic acid, consistent with the up-regulation ofretinoic acid-responsive genes.Activity Assays with 3�-Hydroxysteroids—To exclude the

possibility that the lack of up-regulation of retinoic acid targetgenes in RSDH rafts was due to the lack of catalytic activity ofrecombinant RoDH4, RL-HSD, and RDHL proteins, we incu-bated the respective skin rafts with 3�-hydroxysteroids, thealternative substrates recognized by these enzymes. ADT wasadded to the media of RoDH4 and RL-HSD rafts, as this isthe preferred 3�-hydroxysteroid substrate for RoDH4 and RL-HSD enzymes (33, 34), whereas RDHL rafts were incubatedwith ALLO, which is a better substrate for RDHL (21). RDH10rafts were included in this experiment to determine whetherRDH10 exhibits a 3�-hydroxysteroid dehydrogenase activitysimilarly to the three other SDRs.TLC analysis of the reaction products demonstrated that

RDH10 was inactive toward androsterone, but skin raftsexpressing RoDH4 and RL-HSD efficiently converted ADT to5�-androstanedione (5�-dione) and epiandrosterone (Fig. 5).The formation of epiandrosterone in RL-HSD skin rafts was inagreement with the previously described 3(�3�)-hydroxy-steroid epimerase activity of RL-HSD (22). As expected, RoDH4converted ADT to 5�-dione, which was further reduced to epi-androsterone by the endogenous 3�-ketosteroid reductaseactivity described previously (22). RDHL rafts showed in-creased conversion of ALLO to 5�-dihydroprogesterone (12%versus 9% in mock-transduced rafts) and epiallopregnanolone(18% versus 15% conversion), also consistent with the previousreport (21). These experiments demonstrated that all three ret-inol/sterol dehydrogenases expressed in skin rafts were catalyt-ically active.Gene Expression and Morphology of Rafts Treated with

Retinol—RoDH4 and RL-HSD both exhibit higher Km valuesfor all-trans-retinol than RDH10 (24, 34, 43). To test a possibil-ity that RoDH4, RL-HSD, and RDHL could contribute to reti-noic acid production at a higher concentration of retinol thanthat provided by the standardmedium,we grew skin rafts trans-duced with each of the four SDRs as well as control rafts inmedium supplementedwith 2�M retinol. Interestingly, supple-mentation with retinol resulted in epithelial morphology verysimilar to that observed for RDH10 rafts that were grown inregular medium (Fig. 6), suggesting that the additional retinolwas utilized efficiently by the endogenous retinol dehydroge-nase(s) present in the original PHKs, resulting in increased lev-els of retinoic acid. This was further confirmed by q-PCRanalysis, which showed significant induction of retinoic acid-responsive genes in all rafts grown in the presence of additionalretinol including rafts infected with empty virus (Fig. 7, upperpanel). However, importantly, the rafts expressing RoDH4, RL-HSD, or RDHL did not show any greater up-regulation of reti-noic acid-responsive genes than mock-transduced rafts (datanot shown), as would be expected if these enzymes could func-tion as retinol dehydrogenases at higher concentrations of ret-inol. On the other hand, RDH10 rafts that were grown in thepresence of additional retinol showed further increase in RAR�expression compared withmock-transduced rafts (Fig. 7, lowerpanel). This suggested that, although most of the retinoic acid-

FIGURE 3. q-PCR analysis of retinoic acid-inducible gene expression.A, q-PCR analysis was performed on RNA isolated from three individual rafts.Control groups are represented by white bars and experimental groups arerepresented by gray bars. Error bars represent standard error of the mean(S.E.). Levels of transcripts were determined using the relative quantificationmethod (24). The two-tailed p value was determined using GraphPad InStat(version 3.00; GraphPad Software, San Diego, CA). Statistically significantchanges are indicated by *, p � 0.05); **, p � 0.01; or ***, p value � 0.0001.STRA6, stimulated by retinoic acid gene 6; RAR�, retinoic acid receptor �;LRAT, lecithin retinol acyltransferase; CRBP1, cellular retinol binding proteintype 1; CRABP2, cellular retinoic acid binding protein type 2; DHRS3, dehydro-genase/reductase member 3 (also known as retSDR1); CYP26B1, cyto-chrome P450 26B1; CYP26A1, cytochrome P450 26A1; RAR�, retinoic acidreceptor gamma; PDK1, 3-phosphoinositide-dependent kinase 1. B, Westernblot analysis of CRBP1 and DHRS3 expression in skin rafts transduced withempty virus (mock) and RDH10 retrovirus (RDH10). Cytosolic fractions wereused for the detection of CRBP1, and microsomal fractions were used fordetection of DHRS3 with �-actin as the loading control for cytosol, and cyto-chrome P450 reductase was used for microsomes essentially as described inthe legend to Fig. 1. Rabbit polyclonal antiserum custom made against theCRBP1-intein fusion protein by Cocalico Biologicals, Inc. (Reamstown, PA) wasused at a 1:1000 dilution. Rabbit polyclonal antibodies against human DHRS3from Proteintech Group, Inc. (catalog no. 15393-1-AP; Chicago, IL) were usedat a 1:2000 dilution.

Functional Analysis of Retinoid-active SDRs in Human Skin

APRIL 15, 2011 • VOLUME 286 • NUMBER 15 JOURNAL OF BIOLOGICAL CHEMISTRY 13555

by guest on September 10, 2020

http://ww

w.jbc.org/

Dow

nloaded from

responsive genes reached their maximum expression levels inrafts treated with retinol, RAR� was not yet fully activated, andoverexpression of RDH10 resulted in further increase in thelevel of retinoic acid and, consequently, in the expression levelof RAR�. This result demonstrated that while RDH10 raftsresponded to retinol supplementation by fully activatingRAR�,neither RoDH4, nor RL-HSD, nor RDHL rafts displayed activa-tion of retinoic acid-responsive genes over that seen in mock-transduced rafts even when provided with a higher concentra-tion of retinol.Identification of Novel Gene Targets for Retinol/Sterol Dehy-

drogenase and RDH10—Having established that the retinol/sterol dehydrogenases are catalytically active in skin rafts butare not involved in the biosynthesis of retinoic acid, we exam-ined whether their activities have an impact on the expressionof genes other than those regulated by retinoic acid.Microarraydata analysis revealed that the expression of a number of geneswas altered in skin rafts expressing RoDH4, RL-HSD, or RDHLthat were grown in regular skin raft culture medium. To con-firm these findings, we carried out q-PCR analysis of severalidentified genes. As shown in Fig. 8, protease inhibitor Kazaltype 6 (SPINK6), fibroblast activation protein (FAP), and uro-plakin 1B (UPK1B) were up-regulated, whereas corneodesmo-sin (CDSN) was slightly down-regulated in RoDH4 rafts. RDHLrafts displayed statistically significant down-regulation offibronectin 1 (FN1), CDSN, and anterior gradient homolog 2(AGR2) genes. The AGR2 gene was also down-regulated in RL-HSD rafts. This analysis revealed some overlap in the genestargeted by the RSDHs, consistent with their overlapping sub-strate specificity.Interestingly, several novel genes were identified as retinoic

acid targets in RDH10 rafts. Mucin 21 (MUC21) and �-ami-nobutyric acid (GABA) receptor subunit � (GABRP) were verystrong responders with 40-fold and 28-fold increases in the lev-els of mRNA, respectively. FN1, AGR2, and periostin (POSTN)were up-regulated 2–4-fold, whereas SPINK6 and solute car-rier family 7, (cationic amino acid transporter, y� system)member 11 (SLC7A11) were down-regulated �2-fold. Theoppositechanges in theexpressionofSPINK6andAGR2observedfor RDH10 rafts compared with rafts expressing RoDH4 orRDHL/RL-HSD further emphasized the differences in the physi-ological impact of RDH10 versus RSDHs. Together, these datasuggested that RDH10 and RSDHs have different metabolic sub-strates and different sets of target genes in human skin rafts.

DISCUSSION

Human epidermis contains several retinoid-active SDRs(25–27); however, their relative roles in the regulation of reti-noic acid-responsive genes in skin remain unknown. Here, forthe first time, we compared the contribution of individualhuman SDRs to retinoic acid biosynthesis and regulation ofgene expression during proliferation, stratification, and differ-entiation of organotypic skin raft cultures.FIGURE 4. Analysis of retinoic acid production in RDH10 skin rafts. Rafts

infected with empty virus (A) or with RDH10 expression construct (B) wereincubated with 2 �M retinol for 24 h. Three rafts of each kind were combined,and retinoids were extracted for reversed phase HPLC analysis. Peak 2 in B wasidentified as all-trans-retinoic acid (atRA) based on the elution time of all-trans-retinoic acid standard (C) and the absorption spectrum of the peak (B, inset 2)characteristic of all-trans-retinoic acid. Spectral analysis of peak 1 (B, inset 1)suggests that this peak represents all-trans-3,4-didehydro-retinoic acid.

D, MRM chromatograms of standard all-trans-retinoic acid solution (upperpanel) and of RDH10 skin raft extract (lower panel). The arrows indicate thepeaks corresponding to specific fragment ions from mass transitions m/z 301/123, m/z 301/105, and m/z 301/81. Cps, counts per second; AU, absorbanceunits.

Functional Analysis of Retinoid-active SDRs in Human Skin

13556 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 286 • NUMBER 15 • APRIL 15, 2011

by guest on September 10, 2020

http://ww

w.jbc.org/

Dow

nloaded from

The results of this study show that all four SDRs ectopicallyexpressed in skin rafts are catalytically active, but RDH10 has avery different effect on skinmorphology thanRoDH4, RL-HSD,or RDHL. Skin rafts ectopically expressing RDH10 exhibit bothgreater proliferation of basal cells, as indicated by the increasedincorporation of BrdU and severe inhibition of terminal differ-entiation, evidenced by the reduced expression of filaggrin (themarker of terminally differentiated keratinocytes). Increased

proliferation of basal cells and inhibition of terminal differenti-ation are the hallmark features of the effects of retinoids onepithelial morphology (2, 3), and they are consistent with thehigher steady-state levels of retinoic acid detected in RDH10skin rafts.In normal human skin, the level of retinoic acid is undetect-

able or detectable only at trace amounts although topical appli-cation of retinol produces such biological effects as epidermal

FIGURE 5. Analysis of steroid metabolism by SDR skin rafts. Rafts expressing RDH10, RL-HSD, RDHL, or mock-transfected with pBabe vector (Mock) wereincubated with tritiated ADT or ALLO for 3–24 h. Steroids were extracted and separated by TLC. The plates were exposed to PhosphorImager tritium screenovernight. Numbers indicate individual skin raft samples. For a positive control, 1 �g of RDHL-expressing Sf9 microsomes (ms) was incubated with 1 �M ALLOfor 30 min at 37 °C. ADT, androsterone (3�-hydroxy-5�-androstan-17-one); 5�-Dione, 5�-androstanedione (5�-androstan-3,17-dione); Epi-ADT, epiandros-terone (3�-hydroxy-5�-androstan-17-one; 5�-DHP, 5�-dihydroprogesterone (5�-pregnane-3,20-one; ALLO, allopregnanolone (3�-hydroxy-5�-pregnan-20-one); Epi-ALLO, epiallopregnanolone (3�-hydroxy-5�-pregnan-20-one).

FIGURE 6. Effect of retinol on skin raft morphology. Skin rafts were grown for 11 days in the absence (control) or presence of additional 2 �M all-trans-retinol(retinol). Sections of paraffin-embedded rafts were analyzed by H&E staining and immunostaining for BrdU and filaggrin as described in the legend to Fig. 2.

Functional Analysis of Retinoid-active SDRs in Human Skin

APRIL 15, 2011 • VOLUME 286 • NUMBER 15 JOURNAL OF BIOLOGICAL CHEMISTRY 13557

by guest on September 10, 2020

http://ww

w.jbc.org/

Dow

nloaded from

thickening and an increase in expression of cellular retinoicacid binding protein type II (14). Thus, the conversion of retinolto retinoic acid in human skin is regulated tightly (44). Our datasuggest that RDH10 is expressed in human skin at a very lowlevel, and the results of this study highlight the importance ofthe low level of RDH10 by demonstrating that overexpressionof RDH10 results in significant up-regulation of retinoic acidtarget genes and disruption of normal epithelial growth anddifferentiation. It is important to note that this phenotype isobserved in regular raft culture medium, without supplemen-tation with retinol, and it is consistent with the high affinity ofRDH10 for retinol as a substrate (Km � 0.035 �M) (24). Thus, itis clear that the amount of RDH10 in skin rafts controls theamount of retinoic acid produced from the physiologically rel-evant levels of retinol.RoDH4 and RL-HSD have lower affinities for retinol (Km

values of 1.1 �M and 3.2 �M, respectively) than RDH10, whichmight indicate that these enzymes function as retinol dehydro-genases at higher levels of retinol. However, supplementationof the growing RoDH4, RL-HSD, or RDHL skin rafts with 2 �M

retinol does not lead to a greater up-regulation of retinoic acid-inducible genes in these rafts compared with that observed inmock-infected rafts, indicating that even at a higher retinol

concentration, these enzymes have no effect on retinoic acidbiosynthesis in the reconstructed epidermis. Together, theseobservations suggest that out of four candidate SDRs onlyRDH10 can function as a retinol dehydrogenase in human skinin vivo.Although the RSDHs do not increase the expression of reti-

noic acid target genes, elevated levels of these proteins in skinrafts grown in regularmedium do affect the expression levels ofother genes. Here, we report for the first time that the activitiesof RoDH4, RL-HSD, and RDHL induce changes in the mRNAlevels of SPINK6 (serine protease inhibitor Kazal type 6), FAP(fibroblast activation protein), UPK1B (uroplakin 1B), FN1(fibronectin 1), AGR2 (anterior gradient homolog 2) andCDSN(corneodesmosin). The relatively small changes in the expres-sion levels of these genes could be due to limited availability ofSDR substrates in skin raft culture medium.Changes in some of these transcripts are specific for certain

SDRs, but other genes are affected similarly in at least two of theSDR-expressing rafts, suggesting an overlap in their gene tar-gets. For example, the 3-fold up-regulation of FAP (also calledFAP� or seprase) appears to be specific for RoDH4 rafts. FAP isa cell surface serine protease that has emerged as a marker ofreactive fibroblasts in tumors butmay also play a role as a tumorsuppressor (reviewed in Ref. 45).Another gene that is up-regulated specifically in RoDH4 rafts

is UPK1B (uroplakin 1B). UPK1B codes for a structural protein

FIGURE 7. Effect of retinol on RA-responsive gene expression in mock-transduced versus RDH10 rafts. Mock-transduced or RDH10-expressingrafts were grown in the presence of additional 2 �M retinol. Retinoic acid-inducible gene expression was analyzed by q-PCR as described in the legendto Fig. 3. Gene expression levels were compared between mock-transducedrafts grown in the absence (Mock, white bars) or presence of retinol (ROL)(Mock � ROL, gray bars) (upper panel); and between mock-transduced raftsgrown in the presence of retinol (Mock � ROL, white bars) and RDH10 raftsgrown in the presence of retinol (RDH10 � ROL, gray bars). Statistically signif-icant changes are indicated by *, p � 0.05. Error bars represent standard errorof the mean (S.E.). RALDH2, retinaldehyde dehydrogenase 2; STRA6, stimu-lated by retinoic acid gene 6; RAR�, retinoic acid receptor �; LRAT, lecithinretinol acyltransferase; CRBP1, cellular retinol binding protein type 1; CRABP2,cellular retinoic acid binding protein type 2; DHRS3, dehydrogenase/reduc-tase member 3 (also known as retSDR1); CYP26B1, cytochrome P450 26B1;CYP26A1, cytochrome P450 26A1; RAR�, retinoic acid receptor gamma; PDK1,3-phosphoinositide-dependent kinase 1. Note that there is a further increasein the expression of RAR� in RDH10 rafts after treatment with retinol. In con-trast, skin rafts expressing RoDH4, RL-HSD, or RDHL did not exhibited increasein retinoic acid-responsive genes greater than that observed in mock-trans-duced rafts (not shown).

FIGURE 8. Quantitative RT-PCR analysis of gene expression in SDR skinrafts. q-PCR analysis was performed as described in the legend to Fig. 3.Control groups are represented by white bars, and experimental groups arerepresented by gray bars. Statistically significant changes are indicated by *, pvalue � 0.05 or **, p value � 0.01. Error bars represent standard error of themean (S.E.). IGFL1, IGF-like family member 1; SPINK6, serine peptidase inhibi-tor, Kazal type 6; THBS1, thrombospondin 1; FAP, fibroblast activation protein�; POSTN, periostin, osteoblast-specific factor; MUC21, mucin 21; UPK1B, uro-plakin 1B; TGFBI, transforming growth factor, �-induced, 68 kDa; GABRP,�-aminobutyric acid (GABA) A receptor �; FN1, fibronectin 1; TGFB1, trans-forming growth factor, �1; AGR2, anterior gradient homolog 2 (Xenopus lae-vis); TOB2, transducer of ERBB2, 2; CDSN, corneodesmosin; SLC7A11, solutecarrier family 7, (cationic amino acid transporter, y� system) member 11.

Functional Analysis of Retinoid-active SDRs in Human Skin

13558 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 286 • NUMBER 15 • APRIL 15, 2011

by guest on September 10, 2020

http://ww

w.jbc.org/

Dow

nloaded from

that is a terminal differentiation component of the mammalianbladdermembrane and is the only uroplakin that is expressed intissues other than the urothelium (reviewed in Ref. 46).SPINK6 is significantly up-regulated in RODH4 rafts but

exhibits similar upward tendency in RL-HSD and RDHL rafts.SPINK6 is induced during keratinocyte differentiation and isthought to function as the selective inhibitor of kallikrein-re-lated peptidases, which play a central role in skin desquamationby cleaving corneodesmosomes (47).Examples of gene expression changes that overlap between at

least two of RSDH rafts are the down-regulation of AGR2 inrafts expressing RDHL and RL-HSD and down-regulation ofCDSN in RoDH4 and RDHL rafts. AGR2 (also known as AG2,HAG-2, or GOB-4) is a protein disulfide isomerase that wasshown to be essential for production of mucus (48). Interest-ingly, in contrast to RSDH rafts, this gene was up-regulated inRDH10 rafts.CDSN is an extracellular component of corneodesmosomes

that is specific to desmosomes of epithelia undergoing cornifi-cation. CDSN displays adhesive properties and is proteolyzedsequentially as corneocytes migrate toward the skin surface.Deletion of Cdsn gene in mice results in desmosomal breaks atthe interface between the living and cornified layers (49, 50).Thus, non-retinoid products of RoDH4, RDHL, andRL-HSD

activities appear to regulate directly or indirectly several com-ponents of the extracellular matrix important for structuralintegrity, adhesion, and movement of cells. In general, raftsexpressing RSDHs seem to exhibit up-regulation of genes asso-ciated with terminal differentiation of keratinocytes (e.g.SPINK6 andUPK1B) as opposed to RDH10 rafts which exhibitup-regulation of genes associated with increased proliferation.Interestingly, microarray analysis of gene expression in

RDH10 skin rafts revealed a number of previously unrecog-nized genes that respond toRDH10 activity and, presumably, toretinoic acid. Mucin 21 (MUC21, also known as epiglycanin)was identified as one of the strongest responders with a 40-foldincrease in expression.Mucin 21 was the first mucin associatedwith the malignant behavior of carcinoma cells and was latershown to prevent cell adhesion (51, 52). The up-regulation ofMUC21 is consistentwith the fragile nature and abnormalmor-phological appearance of RDH10 rafts.The second gene highly up-regulated in RDH10 rafts is

GABRP (peripheral type GABA receptor subunit �). AlthoughGABA is known to function primarily as an inhibitory neu-rotransmitter in the mature central nervous system, GABAreceptors are also present in nonneural tissues, including can-cer. GABRP is abundant in the uterus and was detected inbreast (53, 54) and lung (55) but rarely in the brain. In agree-ment with our detection of GABRP expression in skin raft cul-ture, keratinocyte growth factor and an air-liquid interfaceculture system were shown to significantly up-regulate theexpression of GABRPmRNA and protein (55). GABRP appearsto mediate the growth-promoting effect of GABA (56). This isconsistent with the retinoic acid-induced increase in GABRPexpression and proliferation of basal keratinocytes.Among the genes negatively regulated by RDH10 activity are

SPINK6, which is up-regulated in RSDH rafts, and solute car-rier family 7 (cationic amino acid transporter, y� system),

member 11 (SLC7A11 or xCT). The SLC7A11 gene encodes thexCT subunit of the heterodimeric x(c)(�) cystine/glutamateantiporter that was implicated in GSH-based chemoresistancebecause it mediates cellular uptake of cystine/cysteine for sus-tenance of intracellular GSH levels (57, 58). In skin, SLC7A11 isrequired for melanogenesis (reviewed in Ref. 59). In contrast toRDH10 rafts, expression of SLC7A11 does not change signifi-cantly in RSDH rafts.Thus, analysis of gene expression in RDH10 versus RSDH

rafts revealed a number of important differences. First, the ret-inoic acid-inducible genes are up-regulated in RDH10 rafts butnot in RSDH rafts, indicating that RDH10 but not RSDH isinvolved in retinoic acid biosynthesis in human epidermis. Sec-ond, a number of genes exhibit opposite changes in RDH10versus RSDH rafts, suggesting that they are regulated by differ-ent biologically active metabolites. Finally, some genes exhibitsimilar changes in RSDH rafts, suggesting a potential overlap inthe metabolic products of different RSDHs, consistent withtheir overlapping substrate specificity. Together, these resultsdemonstrate that RDH10, but not RSDHs, functions as a bonafide retinol dehydrogenase in a growing and differentiatinghumanorgan and provide the first evidence that the activities ofRSDHs regulate the expression of non-retinoid target genes inhuman epidermis. Identification of the extracellular matrixcomponents as themajor targets of RSDHs in skin should facil-itate the search for the true physiological substrates of theseenzymes.

Acknowledgments—We thank Dr. Louise Chow (University of Ala-bama School of Medicine, Birmingham, AL) and members of theChow laboratory, especially NickGenovese, Hsu-Kun (Wayne)Wang,andEun-YoungKho for generously sharing expertise in preparation ofhuman skin rafts. We also thank Doyle RayMoore, 2nd, at Universityof Alabama at Birmingham Targeted Metabolomics & ProteomicsLaboratory for help with LC/MS-MRM analysis of retinoic acid.

REFERENCES1. Mangelsdorf, D., Umesono, K., and Evans, R. M. (1994) in The Retinoids:

Biology, Chemistry, andMedicine (Sporn,M. B., Roberts, A. B., and Good-man, D. S., eds), pp. 319–350, Raven Press, New York

2. Gudas, L. J., Sporn,M. B., and Roberts, A. (1994) in The Retinoids: Biology,Chemistry, and Medicine (Sporn, M. B., Roberts, A. B., and Goodman,D. S., eds), pp. 443–520, Raven Press, New York

3. Fisher, G. J., and Voorhees, J. J. (1996) FASEB J. 10, 1002–10134. Kang, S. (2005) Cutis 75, 10–135. Chatellard-Gruaz, D., Randolph, R. K., Hagens, G., Saurat, J. H., and Sieg-

enthaler, G. (1998) J. Lipid Res. 39, 1421–14296. Busch, C., Siegenthaler, G., Vahlquist, A., Nordlinder, H., Sundelin, J.,

Saksena, P., and Eriksson, U. (1992) J. Invest. Dermatol. 99, 795–8027. Karlsson, T., Virtanen, M., Sirsjo, A., Rollman, O., Vahlquist, A., and

Torma, H. (2002) Exp. Dermatol. 11, 143–1528. Cheung, C., Smith, C. K., Hoog, J. O., andHotchkiss, S. A. (1999) Biochem.

Biophys. Res. Commun. 261, 100–1079. Baron, J. M., Heise, R., Blaner, W. S., Neis, M., Joussen, S., Dreuw, A.,

Marquardt, Y., Saurat, J. H., Merk, H. F., Bickers, D. R., and Jugert, F. K.(2005) J. Invest. Dermatol. 125, 143–153

10. Pavez Lorie, E., Chamcheu, J. C., Vahlquist, A., andTorma, H. (2009)Arch.Dermatol. Res. 301, 475–485

11. Torma, H., Rollman, O., and Vahlquist, A. (1999) Arch. Dermatol. Res.291, 339–345

12. Randolph, R. K., and Simon, M. (1993) J. Biol. Chem. 268, 9198–9205

Functional Analysis of Retinoid-active SDRs in Human Skin

APRIL 15, 2011 • VOLUME 286 • NUMBER 15 JOURNAL OF BIOLOGICAL CHEMISTRY 13559

by guest on September 10, 2020

http://ww

w.jbc.org/

Dow

nloaded from

13. Siegenthaler, G., Saurat, J. H., and Ponec, M. (1990) Biochem. J. 268,371–378

14. Kang, S., Duell, E. A., Fisher, G. J., Datta, S. C., Wang, Z. Q., Reddy, A. P.,Tavakkol, A., Yi, J. Y., Griffiths, C. E., Elder, J. T., et al. (1995) J. Invest.Dermatol. 105, 549–556

15. Asselineau, D., Bernard, B. A., Bailly, C., and Darmon,M. (1989)Dev. Biol.133, 322–335

16. Napoli, J. L., and Race, K. R. (1987) Arch. Biochem. Biophys. 255, 95–10117. Pares, X., Farres, J., Kedishvili, N., and Duester, G. (2008) Cell. Mol. Life

Sci. 65, 3936–394918. Kallberg, Y., Oppermann, U., and Persson, B. (2010) FEBS J. 277,

2375–238619. Belyaeva, O. V., and Kedishvili, N. Y. (2006) Genomics 88, 820–83020. Biswas, M. G., and Russell, D. W. (1997) J. Biol. Chem. 272, 15959–1596621. Chetyrkin, S. V., Belyaeva, O. V., Gough, W. H., and Kedishvili, N. Y.

(2001) J. Biol. Chem. 276, 22278–2228622. Belyaeva, O. V., Chetyrkin, S. V., Clark, A. L., Kostereva, N. V., SantaCruz,

K. S., Chronwall, B. M., and Kedishvili, N. Y. (2007) Endocrinology 148,2148–2156

23. Wu, B. X., Chen, Y., Chen, Y., Fan, J., Rohrer, B., Crouch, R. K., and Ma,J. X. (2002) Invest. Ophthalmol. Vis. Sci. 43, 3365–3372

24. Belyaeva, O. V., Johnson, M. P., and Kedishvili, N. Y. (2008) J. Biol. Chem.283, 20299–20308

25. Markova,N.G., Pinkas-Sarafova, A., Karaman-Jurukovska,N., Jurukovski,V., and Simon, M. (2003)Mol. Genet. Metab. 78, 119–135

26. Jurukovski, V., Markova, N. G., Karaman-Jurukovska, N., Randolph, R. K.,Su, J., Napoli, J. L., and Simon, M. (1999)Mol. Genet. Metab. 67, 62–73

27. Karlsson, T., Vahlquist, A., Kedishvili, N., and Torma, H. (2003) Biochem.Biophys. Res. Commun. 303, 273–278

28. Banerjee, N. S., Chow, L. T., and Broker, T. R. (2005) in Human Papillo-maviruses:Methods and Protocols,Methods inMolecularMedicine (Davy,C., and Doorbar, J., eds), pp. 187–202, Humana Press, Totowa, NJ

29. Fuchs, E., and Raghavan, S. (2002) Nat. Rev. Genet. 3, 199–20930. Wilson, J. L., Dollard, S. C., Chow, L. T., and Broker, T. R. (1992) Cell

Growth Differ. 3, 471–48331. Bernard, F. X., Pedretti, N., Rosdy, M., and Deguercy, A. (2002) Exp. Der-

matol. 11, 59–7432. Cheng, S., Schmidt-Grimminger, D. C., Murant, T., Broker, T. R., and

Chow, L. T. (1995) Genes Dev. 9, 2335–234933. Gough, W. H., VanOoteghem, S., Sint, T., and Kedishvili, N. Y. (1998)

J. Biol. Chem. 273, 19778–1978534. Chetyrkin, S. V., Hu, J., Gough, W. H., Dumaual, N., and Kedishvili, N. Y.

(2001) Arch. Biochem. Biophys. 386, 1–1035. Molotkov, A., Ghyselinck, N. B., Chambon, P., and Duester, G. (2004)

Biochem. J. 383, 295–30236. Belyaeva, O. V., Korkina, O. V., Stetsenko, A. V., Kim, T., Nelson, P. S., and

Kedishvili, N. Y. (2005) Biochemistry 44, 7035–704737. Pfaffl, M. W. (2001) Nucleic Acids Res. 29, e4538. Napoli, J. L., and Horst, R. L. (1998)Methods Mol. Biol. 89, 29–40

39. Hoover, F., Gundersen, T. E., Ulven, S. M., Michaille, J. J., Blanchet, S.,Blomhoff, R., and Glover, J. C. (2001) J. Comp. Neurol. 436, 324–335

40. Karlsson, T., Vahlquist, A., and Torma, H. (2010) J. Dermatol. Sci. 57,207–213

41. Andersson, E., Bjorklind, C., Torma, H., and Vahlquist, A. (1994) Biochim.Biophys. Acta 1224, 349–354

42. Sani, B. P., Venepally, P. R., and Levin, A. A. (1997) Biochem. Pharmacol.53, 1049–1053

43. Lapshina, E. A., Belyaeva, O. V., Chumakova, O. V., and Kedishvili, N. Y.(2003) Biochemistry 42, 776–784

44. Tang, X. H., Vivero, M., and Gudas, L. J. (2008) Exp. Cell Res. 314, 38–5145. Pure, E. (2009) Expert Opin. Ther. Targets. 13, 967–97346. Wu, X. R., Kong, X. P., Pellicer, A., Kreibich, G., and Sun, T. T. (2009)

Kidney Int. 75, 1153–116547. Meyer-Hoffert, U., Wu, Z., Kantyka, T., Fischer, J., Latendorf, T., Hans-

mann, B., Bartels, J., He, Y., Glaser, R., and Schroder, J. M. (2010) J. Biol.Chem. 285, 32174–32181

48. Park, S. W., Zhen, G., Verhaeghe, C., Nakagami, Y., Nguyenvu, L. T.,Barczak, A. J., Killeen, N., and Erle, D. J. (2009) Proc. Natl. Acad. Sci. U.S.A.106, 6950–6955

49. Leclerc, E. A., Huchenq, A., Mattiuzzo, N. R., Metzger, D., Chambon, P.,Ghyselinck, N. B., Serre, G., Jonca, N., and Guerrin, M. (2009) J. Cell Sci.122, 2699–2709

50. Matsumoto, M., Zhou, Y., Matsuo, S., Nakanishi, H., Hirose, K., Oura, H.,Arase, S., Ishida-Yamamoto, A., Bando, Y., Izumi, K., Kiyonari, H., Os-hima, N., Nakayama, R.,Matsushima, A., Hirota, F.,Mouri, Y., Kuroda, N.,Sano, S., and Chaplin, D. D. (2008) Proc. Natl. Acad. Sci. U.S.A. 105,6720–6724

51. Yi, Y., Kamata-Sakurai, M., Denda-Nagai, K., Itoh, T., Okada, K., Ishii-Schrade, K., Iguchi, A., Sugiura, D., and Irimura, T. (2010) J. Biol. Chem.285, 21233–21240

52. Itoh, Y., Kamata-Sakurai, M., Denda-Nagai, K., Nagai, S., Tsuiji, M., Ishii-Schrade, K., Okada, K., Goto, A., Fukayama, M., and Irimura, T. (2008)Glycobiology 18, 74–83

53. Hedblom, E., and Kirkness, E. F. (1997) J. Biol. Chem. 272, 15346–1535054. Zafrakas, M., Chorovicer, M., Klaman, I., Kristiansen, G., Wild, P. J.,

Heindrichs, U., Knuchel, R., and Dahl, E. (2006) Int. J. Cancer 118,1453–1459

55. Jin, N., Narasaraju, T., Kolliputi, N., Chen, J., and Liu, L. (2005)Cell TissueRes. 321, 173–183

56. Takehara, A., Hosokawa, M., Eguchi, H., Ohigashi, H., Ishikawa, O., Na-kamura, Y., and Nakagawa, H. (2007) Cancer Res. 67, 9704–9712

57. Chintala, S., Li, W., Lamoreux, M. L., Ito, S., Wakamatsu, K., Sviderskaya,E. V., Bennett, D. C., Park, Y. M., Gahl, W. A., Huizing, M., Spritz, R. A.,Ben, S., Novak, E. K., Tan, J., and Swank, R. T. (2005) Proc. Natl. Acad. Sci.U.S.A. 102, 10964–10969

58. Sato, H., Tamba, M., Kuriyama-Matsumura, K., Okuno, S., and Bannai, S.(2000) Antioxid. Redox. Signal. 2, 665–671

59. Lo, M., Wang, Y. Z., and Gout, P. W. (2008) J. Cell. Physiol. 215, 593–602

Functional Analysis of Retinoid-active SDRs in Human Skin

13560 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 286 • NUMBER 15 • APRIL 15, 2011

by guest on September 10, 2020

http://ww

w.jbc.org/

Dow

nloaded from

Seung-Ah Lee, Olga V. Belyaeva, Lizhi Wu and Natalia Y. KedishviliCulture

Expression of Retinoic Acid-responsive Genes in Human Transgenic Skin Raft Retinol Dehydrogenase 10 but Not Retinol/Sterol Dehydrogenase(s) Regulates the

doi: 10.1074/jbc.M110.181065 originally published online February 23, 20112011, 286:13550-13560.J. Biol. Chem. 

  10.1074/jbc.M110.181065Access the most updated version of this article at doi:

 Alerts:

  When a correction for this article is posted• 

When this article is cited• 

to choose from all of JBC's e-mail alertsClick here

Supplemental material:

  http://www.jbc.org/content/suppl/2011/02/23/M110.181065.DC1

  http://www.jbc.org/content/286/15/13550.full.html#ref-list-1

This article cites 56 references, 18 of which can be accessed free at

by guest on September 10, 2020

http://ww

w.jbc.org/

Dow

nloaded from