Anders htromS0, Ulrika PetterssonS, Pierre Chambonn, and John J. VoorheesSII. From the Wepartment of Dermatology, University of Michigan, Ann Arbor, ...
VoI. 269,No. 35,Issue of September 2,pp. 22334-22339,
THEJOURNAL OF BIOLOGICAL CHFMISTRY 0 1994 by The American Society for Biochemistry and Molecular Biology, Inc
I994 Printed in U.S.A.
Retinoic Acid Inductionof Human Cellular Retinoic Acid-binding Protein-I1 Gene Transcription Is Mediated by Retinoic Acid Receptor-RetinoidX Receptor Heterodimers Bound to One Far Upstream Retinoic Acid-responsive Element with 5-Base Pair Spacing* (Received forpublication, April 13, 1994)
Anders htromS0, Ulrika PetterssonS, Pierre Chambonn, and John J. VoorheesSII From the Wepartment of Dermatology, University of Michigan, Ann Arbor, Michigan 48109, the TLaboratoire de Genetique Moleculaire des Eucaryotes du Centre National de la Recherche Scentifique, Unite 184 de Biologie Moleculaire, and the Geneie Genetiaue de l’lnstitut National de la Sante et Reserche Medicale, Institut de ChimieBiologique, Faculte de Medicine, 67085 Strasbourg, Cedex, France
We previouslyclonedthehumancellularretinoic thatactivatetranscriptioninresponseto all-trans-RA and acid-bindingprotein-I1 (CRABPII) gene anddemong-cis-RA, and retinoid X receptors (RXRa,RXRP, and FXR-y) strated a rapid and transient increase in retinoic acid bind and activate transcription in response to 9-cis-RA(15-18). (RAbdependent transcription in cultured human skin RARs can bind cooperatively with RxRs on RAREs to form fibroblasts. To determinewhetherretinoidreceptors heterodimers, andRAR homodimers either do not bindor demcould regulateCRABPII gene transcription, cotransfec- onstrate a weak affinityfor these elements (13-14,18-20). tion experiments were performed.When RARa was co- However, it hasbeen shown thatRXR in the presence of 9 4 s transfected in Cos-1cells with a reporter construct con-RA can form homodimers on a subset of RAREs (21). Natural taining -8.0 kilobases of the upstream region, an 18-foldRAREs are often represented by two direct repeats of the conRA induction was obtained. By deletion analysis, a re- sensus AG(G/T)TCA core motif with different spacing between gion essential for RA induction located approximately the two repeated half-sites. Both spacing and nucleotides sur-5.6 kilobases upstream from the human CRABPIIgene rounding the core motif have been found to be important both start site wasidentified.Sequencingandmutational for binding and transactivation(22-24). analysis identified a direct repeat (GGGTCAttggaAGAll trans-RA, but notScis-RA, can also bind totwo separate GACA) with 5-base pair spacing (DR-5) that is critical for RA-mediated induction of human CRABPII gene tran- butrelated forms of cellular retinoicacid-binding proteins (CRABP) (17, 25). We previously cloned two forms of CRABP scription.This is different fromthemouseCRABPII gene in which two RAREs (DR-1 and DR-2) are required from human skin and demonstrated that CRABPII, but not for full activation. To determine whetherRAR and RXR CRABPI mRNA, is induced by all-trans RA in human skin in (26). In addition, we can bind to the human CRABPII RARE, gel retardation vivo and in cultured human skin fibroblasts of the humanCRABPII gene is assays were performed. In these assays, in vitro trans- demonstrated that transcription induced by all-trans RAin cultured skin lated RARa and RXRa were foundto bind efficiently as rapidly and transiently heterodimers in gel retardation assays; weak binding fibroblasts of and requireson-going protein synthesis (27). RARa homodimerswasobserved.ThesedatademonThe human and mouse CRABPI1 proteins are 93.5% identistrate that thehuman CRABPII gene is regulated by a cal (26, 28), and the gene organization highly is conserved (27, far upstream RARE that most efficiently binds RAR- 29) between these two species. The mouse CRABPII gene is RXR heterodimers. also induced by all-trans-RA and has recently been found to contain two RAR-RxR heterodimer binding sites (RARE1 and RAREB), both within 1200 bp of the upstream region, mediatRetinoids play afundamental role in vertebratedevelopment ing induction of transcription by all-trans-RA and9-cis-RA in and cellular differentiation (1-4). The effect of retinoids on mouse P19 embryonal carcinoma cells(20). Deletion analysis of gene transcriptionis mediated by two receptor families belong- the mouse gene has shown that bothRAREs are required for ing to the superfamily of nuclear receptors(5-14). Retinoic acid full transcriptional induction by RA (20). In the present study, we have, by deletion and mutational (RA)’receptors ( m a , RARP, and RARr) bind t o elements analysis, located a critical RARE far upstream (-5.6 kb) in the * This work was supported in part by the Babcock Dermatological human CRABPII gene that binds RAR-FCSR heterodimers. InEndowment. The costs of publication of this article were defrayed in terestingly, as compared with the mouse gene, the humanuppart by the payment of page charges. This article must therefore be stream region contains by functional analysis, no sequences hereby marked “aduertisement”in accordance with 18 U.S.C. Section analogous to the mouse gene, and has a single site far up1734 solely to indicate this fact. The nucleotide sequence(s)reported in this paper has been submitted stream, which is necessary for RAR-induced transcriptional to the GenBankm/EMBL Data Bank with accession number(s1U09967. regulation. 0 Supported by a Dermatology Foundation Career Development MATERIALS AND METHODS Award. 11 To whom correspondence should be addressed: The University of Receptor Expression Vectors-The human RARa, RARP, and R A R y Michigan Medical Center, Dept. of Dermatology, 1910 A. Alfred Taub- cDNAs in the expression vector pSG5 were as described previously (6, man Health Care Center, Ann Arbor, MI 48109-0314. The abbreviations used are: RA, retinoic acid; CRABP, cellular retinoid X receptor responsive element; CAT, chloramphenicol acetyltransnoic acid-binding protein;RAR, retinoic acid receptor; RARE, retinoic acid receptor responsive element; RXR, retinoid X receptor; RXRE, reti- ferase; kb, kilobaseb); bp, basepaids).
22334
Characterization of a Far UpstreamRARE in the HumanCRABPII Gene
22335
7, 10). Human RXRa was clonedby polymerase chain reaction and subcloned into the expression vector pSVL (Pharmacia Biotech, Inc.) (30). Reporter Constructs in pBLCAT3-A 10-kb fragment of the human CRABPII gene was generated by digesting clone h2.1in A FIX I1(27) at the unique HindIII-XhoI sites and was subcloned into Bluescript SK(Stratagene). This clone was then digested at anAccI site located 108 bp into the first exon (27) and was blunted by Klenow treatment. After digestion with EcoRI, a 2.7-kb fragment was isolated and ligated into Bluescript SK- cut with EcoRI-HincII.This fragment was then isolated from Bluescript SK- by digestion with EcoRI-XhoI and ligated into pBLCAT3 (31) generating the -2.7-kb promoter construct (see Fig. 3 A ) . This construct was then digested with PstI (-1038 in theCRABPII gene " C RXRa RARu RARa RARp RARp RARy RARy (27)),generating a -1.0-kb construct. A-0.5-kb construct was generated + RXRa + RXRa + RXRu by digesting the -2.7-kb construct with BamHI (-460 in the CRABPII gene (27)). The -0.4-kb construct was obtained after digesting the FIG.1.=-dependent transactivation of the humanCRABPII -0.5-kb construct with BamHI and BglII (-415 in the CRABPII gene promoter. Aconstruct containing -8.0 kb of the human CRABPII gene cloned into the reporter plasmid pBLCAT3 (1.2 pg) was cotransfected (27)).The -4.0-kb construct was generated by digesting h2.1 withXbaI (-4.0) and BamHI (-460) and was subcloned into the -0.5-kb construct. into Cos-1 cellswith either carrierDNA as control ( C )or an expression RARa, RARP, RAR-y, or a combination of RARs and The -8.0-kb construct was obtained by digesting h2.1 with XbaI (XbaI vector for RXRa, at -4.0 (27)) and Sal1 (presentA in FIX 11)and was then cloned into the R X R a (0.6 pg each) together with a P-galactosidase expression vector -4.0-kb construct. This fragment was also cloned into Bluescript SK-, (PCH110, 0.6pg). After treatment of cells with ethanol or all-trans-RA generating 4-kb BS. The 4-kb BS clonewas then digested with BamHI- (1 pd for 18 h, CAT activity and P-galactosidase activity was deterXbaI, and a 2.3-kb fragment (see Fig. 3B) was isolated and ligated into mined in cell lysates as described under "Materials and Methods." Data are expressed as -fold induction by all-trans RA relative to cells treated Bluescript SK-, generating 2.3-kb BS.This fragment was used to create with ethanol. Data represent the average of two experiments (not difthe -6.3-kb construct by digestion with HindIII-XbaI and subcloning fering by more than 20%). into the -4.0-kb construct. The -5.8-kb construct was made by digesting a 2.3-kb BS with NarI (located 5 bp 5' of the -5.8 PstI site(see Figs. 3B and 4)) andXbaI, and thenthe 1.8-kbfragment obtained was subcloned CCA-3'; RAREml, 5'-AGCTTGGGGGGTTT'ITGGAAGGACAAG-GT-3' RARE-m2, 5'into the plasmid pSP64 (Promega). The fragment was isolated after and 5'-CTAGACCTTGTCCTTCCAAAAACCCCCCA-3'; digestion with HindIII-XbaI and cloned into the -4.0-kb construct. The AGGCTTGGGGGGTCATTGGAAGAAAAAGGT-3' and 5'-CTAGACCTTTTTCTTCCAATGACCCCCCA-3'; RAREml+2, 5'-AGCTTGGGG-5.8A1.0-kb construct was made by removing a 1012-bpBglII (-415 and GGTTTTTGGAAGAAAAAGGT-3' and 5'-CTAGACCTTTTTCTTCC-1429) fragment from the -5.8-kb construct. The -6.0-,-5.6-, and -5.4-kb constructs were generated by Ex0111 deletions of 4-kb BS. The AAAAACCCCCCA-3';RAREm3, 5"AGCTTGGGAGGTCATTGGAAGplasmid 4-kb BS was digested at a Sac1 site inthe polylinker and at an GTCAAGGT-3' and 5'-CTAGACCTTGACCTTCCAATGACCTCCCA-3'. NdeI site located approximately 500 bp fromthe 5' end of the gene. The The identity of the clones was determined by sequencing. Cell Dansfections and CATAssays-Cos-1 cells were grown in Dulvector including 3.5-kb of the gene was isolated and treated with EzoIII for different amounts of time according to the manufacturer (Erase a becco's modified Eagle's medium containing 10% delipidated fetal calf serum. The day before transfection, cells were plated at approximately Base kit, Promega). After treatment with Klenow, addition of HindIII 40% confluence on 35-mm dishes and were then transfected with 1.2 pg linkers, andtransformation, positive clones weredigested with HindIII of CAT construct, 0.01-0.6 pg of RAR or RXR expressionvectors, and 0.6 and XbaI and sized on an agarose gel. The 2.0-, 1.6-, and 1.4-kb BS pg of the P-galactosidaseexpression vector pCHllO (Pharmacia) using clones were thus identified, and their inserts were subcloned into the HindIII-XbaI sites of the -4.0-kb construct, generating the -6.0-, -5.6-, a calcium phosphate coprecipitation procedure (33). Carrier plasmid (pSP64, Promega) was added, so the total amount ofDNA in each and -5.4-kb constructs. The sequence of the first 673 bp of the -6.0-kb transfection was 3 pg. After overnight incubation with the precipitate, construct was determined on both strands by dideoxychain termination (32) using modifiedT7polymerase (Sequenase, U. S. Biochemical cells werewashed, medium containing all-trans-RAin ethanol or ethanol alone was added, and cells were incubated for an additional 15-24 Carp.). Reporter Constructs in pBLCAT2-The 2300-bp construct (see Fig. h. CAT activity in cell lysates was measured by the xylene extraction method (34) and normalized to P-galactosidase activity, as described 3B) was obtained by digesting 2.3-kb BS with BamHI-XbaI, and the fragment was cloned into pBLCAT2 (31). The 1200-bp construct was previously (35). In vitro Dunslation and Gel Retardation Assays-RARa in pSG5 and made by digesting 2.0-kb BS with NcoI (Fig. 3B) and SpeI (in polylinker), blunted by Klenow treatment and religated. The 1200-bp RXRa in Bluescript SK- werein vitro translated using T7 polymerase in fragment was isolated after digestion with HindIII-BamHI and cloned an in vitro translation system (TNT-Coupled Reticulocyte Lysate Sysinto pBLCAT2. The 395 construct was generated by digesting 2.3-kb BS tem, Promega) according to the manufacturer. Gel retardation assays with PstI, and, after isolation, the 395-bp fragment was ligated into were performed by incubating 3 pl of each receptor or unprogrammed reticulocyte lysate with 32P-labeledKlenow fill in labeled oligonucleopBLCAT2. The direction of the insert was determined by restriction mapping. The constructs, 283, 246, 236, 208, and 151 were obtained tides (RARE or RAREm3) in a total volume of 20 pl containing 20 mM from the 395-bp fragment cloned into Bluescript SK-, using polymerase HEPES pH 7.9, 50 mM KCl, 1 mM dithiothreitol, 2.5 mM MgCl,, 10% chain reaction. Synthetic oligonucleotidescontaining aHindIII site and glycerol, and 1pg of dI-dC, as described previously(36).After 20 min at 25 "C, the samples were analyzed on a 4% nondenaturing polyacryla T3 primer were used. After polymerase chain reaction, fragments were digested with HindIII-XbaI and subcloned into Bluescript SK- for amide gel at 4 "C, using 0.5 x Tris borate EDTA as a running buffer, sequence determination. Clones with correct sequence were then di- after which the gels were dried and autoradiographed. gested with HindIII-XbaI and subcloned into pBLCAT2. 246 mutl and RESULTS 246 mut2 were generated in the same way, except that longer oligonucleotides containing the mutated bases in addition to the HindIII site Danscriptional Regulation of the Human CRABPII Gene by were used. The oligonucleotides used wereas follows, with the HindIII RA-To determine whether the human CRABPII gene was site underlined: 283, 5'-CCTCCCAAGCTTGCCCATCTCTAGCCCTGRARs and RXR, a CAT construct transcriptionally regulated by 3'; 246, 5'-GGAGGGGTAAGCTTGGGGGGTCATTGG-3'; 246 mutl, 5"GGAGGGGTAAGC'M'GGGGGGTI"I'TTGGAAGGACAAGG-3'; 246 containing -8.0 kb of the upstreamregion was transfected into mut2, 5'-GGAGGGGTAAGCTTGGGGGGTCATTGGAAGGAAAAAGG- Cos-1 cells alone or together with 0.6 pg of RARs and R X R a CCCTAG-3'; 236, ~"GGGGTGGGGAAGCTTTGGFL~GGAC-~'; 208, 5'- (Fig. 1).When transfected cells were treated with R A , a low GGCCCTAGCCAAGCTTAGAGGCCAG-3'; 151, 5"CCAGTTCAAGA2.5-fold induction of CAT activity wasobserved when no recepAAGCTTGAGGGTGGGGG-3'. tors were cotransfected (Fig. 1). However, cotransfection with The constructs RARE, RAREml, RAREm2, RAREm3, and an R A R a expressionvector enhanced the induction by allRAREml+2 were generated by annealing oligonucleotidescontaining a cotransfection with 5'-HindIII and a 3'-XbaI overhang and ligated into pBLCAT2. The trans-RA to approximatelyl&fold,and RARP and R A R y resulted inlower induction (Fig. 1).Cotransoligonucleotides wereas follows: RARE, 5"AGCTTGGGGGGTCATTGGAAGGACAAGGT-3' and 5'-CTAGACCTTGTCCTTCCAATGACCCC- fection with R X R a alone did not increase the induction com-
RARu
I
A s.O k b l
CAT
r
-6.3kb -6.0kbl
CAT CAT
-5.8 kbl
10
n i
5
9
CAT
-5.8A1.0kbl -5.6kbl -5.4kb I -4.6kb I -4.0kb L -2.7kb -1.0k
CAT CAT CAT CAT
1-
0.60.050 ..040.30.20.1 pg RECEPTOR PLASMID TRANSFECTED
b m -0.4 k b m
FIG.2. RAR-and ”dependent transactivation of the human CRABPIIpromoter. -8.0-kb CAT (1.2pg)wascotransfectedinto Cos-1 cells witha n expression vectorfor m a , M a ,or a combination of RARa and RXRa (0.05-0.6 pg each) together with a P-galactosidase expression vector (PCH110,0.6 pg). The amountof DNA in each trans- B x -8.0 ” IB B fection was kept constant by addition of carrier DNA. After treatment of ”.” cells with ethanol or all-trans FL4 (1 p ~ for ) 18 h, CAT activity and P-galactosidase activity was determined in cell lysates as described B P P under “Materials and Methods.” Data are expressed as-fold induction -6.3 4.0 &8 66 -5.4 by all-trans FL4relative to cells treated with ethanol. Data represent the 2300 I average of two experiments (not differingby more than 20%).
pared withcontrol and did not enhanceRARa, RARp, or R A R y induction. Cellscotransfected with R x R a andtreatedwith 9-cis-RA (1 PM) gave similar results to all-trans-RA (1 p ~ ) treated cells (data not shown). Both all-trans-- at high (1 PM) concentrations and 9-cis RA (50 nM) stimulated RXRa-dependent transcription over a TRE,-tk-CAT construct (35) in Cos-1 cells (data not shown). Concentration Dependence of RARa and RXRa on Dunsactivation of the Human CRABPII Gene-To determine the concentration dependence of retinoid receptors on human CRABPI1 gene transcription, a titration with RARa, m a , or both was performed. When low concentrations of RARa (0.05-0.25 a maximum 2-fold enhancepg) were used in the transfections, ment in all-trans-RA induction was seen whenR x R a was cotransfected, compared with R A R a alone (Fig. 2). This enhancement was lost when higher concentrations of M a was used, and RxRa alone was found t o be ineffective at all concentrations tested. Similar resultswere also obtained when titration experiments were performed with RARp and R A R y (data not shown). Identification of a Retinoic Acid-responsive Region in theHuman CRABPII Gene-To define the retinoic acid-responsive region in the human CRABPII gene, various 5’ deletions were ligated into the reporter plasmidpBLCAT3 and cotransfected with a RARa expression vector into Cos-1 cells. As can be seen in Fig. 3A, deletions down t o -5.8 kb gave an approximately deletion 20-fold induction by RA. In addition, when an internal between -415 and -1429 was used (-5.881.0 kb), a similar RA induction compared with the -5.8-kb construct was obtained. However, when the upstream region of CRABPII was deleted down to -5.6 kb, a dramatic drop in all-trans-RA induction was seen. To identify the region responsible for R A R a mediated all-trans-RA induction of the human CRABPII gene, different deletions between -4.0 and -6.3 kb were cloned in front of a heterologous tk promoter in pBLCAT2. When constructs containing theregion between -4.0 and -6.3 kb werecotransfected with M a , a 36-fold all-trans-FiA induction ofCAT activity was observed (Fig. 3B). A decrease of induction was observed when this fragment wasdeleted down to -6.0 kb ( 5 ’ )and -4.8 1200 in Fig. 3B). When further deletions were kb (3‘) (construct made, most of the all-trans-RAinduction was lost between constructs 283 and 208 (Fig. 3B). Theregion between thetwo PstI sites was sequenced (Fig. 4)and found to contain threeRARE half-sites (AGGECA) starting at 163, 195, and 309 and one putativedirectrepeat (GGGTCAttggaAGGACA) with 5-bp
5
0
15
10
20
25
FOLD INDUCTION BY RA (1pM) B ““”_
”” ““_
“““_
N
4.8
-4.0
1 kb H FOLD INDUCTION BY RA (1pM)
36
15 11
2 2
FIG.3. Localization of an RA-responsive region in the human CRABPII gene. Different 5’ deletions of the human CRABPII upstream region were cloned into the reporter plasmid pBLCAT3, and constructs containing 3’ deletions were cloned into pBLCAT2 as described under “Materials and Methods.” Cos-l cells were transfected with the pBLCAT3 or pBLCAT2 gene constructs (0.6 pg) and M a (0.6 pg) and were treated as described in the legend to Fig. 1. Data are expressed as -fold induction relative t o the control. A, the approximate 5‘ end of the deletions are indicated in the figure,well as as an internal deletion in the -5.8-kb construct (-5.8A1.0 kb), where sequences between -416 and -1423 were removed. The -fold induction of CAT activity in each bar represents the average of two to five experiments. B, schematic representation of the human CRABPII gene indicating the first exon (EXON) and BamHI (B)and XbaI ( X ) restriction enzyme sites. A region between B (-6.3) and X (-4.0) has been enlarged, and restriction enzyme sitesfor PstI ( P ) ,NcoI ( N ) ,and twoEx0111 deletions at -6.0 and -5.6 are indicated with their approximate location. The -fold induction of CAT activity corresponds to the average valuesof several independent experiments (not differingby more than 20%).
spacing, starting at 242 (Fig. 4). CAT constructs containing deletions fromthe most 5’-PstI sitea t 395 down to 246 resulted in an 11-16-fold all-trans-RA induction ofCAT activity when cotransfected with RARa in Cos-1 cells (Fig. 4). However, when the 246 construct wasdeleted to 236, most of the induction was lost. In addition, when mutations each in of the half-sites were introduced into the 246 construct (246 mutl and 246 mut21, most of the all-trans-RA induction was lost (Fig. 4). The same region was found to be critical for all-trans-RA induction when low amounts of RARa and RxRa were used (data not shown). To determine whether the direct repeat was sufficient to mediate all-trans-RA induction, oligonucleoitides covering this repeat were synthesized and cloned into pBLCAT2 (Fig. 5). When constructs containing the direct repeat and mutations thereof were transfected into Cos-1 cells, a 9.1-fold induction by alltrans-- was obtained with the wild-type RARE (Fig. 5 ) . Most of this induction was lost whenor both eachof the half-sites were mutated ( M E m l , RAREm2, and RAREml+2), anda higher 14.8-fold induction was obtained when the half-sites were mutated to anidealized AGGTCA repeat (RAREm3) (Fig. 5 ) .
Characterization of aFarUpstream -5.4
P
P
I
4.0
'
-6.3 395
X
"
283
TGCCCAGAGAAAACCATTAGGACTAGGTCAGTCTGGGGCCTCCCCCGCCCGCCCATCTCTCA
-236 246 GCCCTGCTATAAGGAGGGGTGGGGGTGGGGGGTCATTGGAGGCCCTAGCCTGGCT A A (246 mut 2) 151 mu1 1) TT 208
2x
8
I
395bp
CTGCAGGGGAAAGGAAATCAGAGCTTCCCACCTCACACATCTTTTTCCCATTTTGTTCATGC
(246
22337
RARE RAREm3
-5.8
B
RARE in theHuman CRABPII Gene a x a
a
+
-k 8 8 ; :
aaa $22 aa xx aa oaaa oaaa
RAGAGGCCAGACTCTGARCTCTTGTATGCCCTTCTGTAATTCCCTGCCAG~~GAGGGGAGG GTGGGGGTGGGGACCAGGAATCTGGGCTCCTGGGGCCCTCCTTTATTCTGCTCAAATCCCCC AACACCTCTGACTATCTCCCGGGAGTCTGACTCACTGGCCCCRAGGAAAGGGCCTGCCGCCG 1
INDUCTION BY RA (1pM)
TCAGTCCTGGGTCTGGTCTGCA~
3
395 283 246 246mull
FOLD x 15 246 mu12 236 11 208 16 151
FOLD x 3 2 2 2
FIG.4. DNA sequence of the 395-bp fragment containing the human CRABPI1 RARE. At the top of the figure is a schematic representation of the region (see Fig. 3B ), with relevant restriction enzyme sites indicated (BamHI ( B ) ,PstI ( P ) ,and XhaI (X)). The sequence is numbered from the most 3'-PstI site, which is located approximately -5.4 kb upstream of the transcription initiation start site. The human CRAJ3PII RARE is shown in boldface type, and arrows indicate the orientation. In addition, three RARE-like motifs are indicated in boldface type. The coordinates for deletions used in CAT assays are given ahove the sequence, and nucleotides differingfrom the 246 deletion in 246 m u t l and 246 mut2 are given helow the sequence. The indicated deletions were cloned into the reporter plasmidpBLCAT2 as described under "Materials andMethods." Cos-1 cells were transfected with constructs (0.6 pg) and RARa (0.6 pg) and treated as described in Fig. 1. Data in the table are expressed as -fold induction relative tothe control and correspond to average values of several independent experiments (not differingby more than 20%).
Fold induction by RA(1 RARE
246 TGGGGGGTCATTGGAAGGACAAGG 223
9.1
RAREml
TGGGGGGTttTTGGAGG
1.7
RAREm2
TGGGGGGTCATTGGAAGaGG
2.3
RAREml+2
TGGGGGGTttTTGGAAGaGG
1.5
RAREm3
TGGGaGGTCATTGGAAGGtCAAGG
14.8
FIG.5. Both half-sites of the human CRABPIIRARE are required for transcriptional activation. Oligonucleotides corresponding to the sequences indicated were cloned into pBLCAT2. Cos-1 cells were transfected with constructs (0.6 pg) Rand A R a (0.6 pg) and treated as described in Fig. 1. Data in the table are expressed as -fold induction by all-trans-RA and represent the average of several independent experiments (not differingby more than 20%).
The Human CRABPII RARE Binds RAR-RXR Heterodimers-To determine whether the direct repeatfound in the human CRABPII gene could bind RAR, RXR, or RAR-RXR dimers, gel retardation assays were performed. When labeled wild-type oligonucleotides (RARE) and oligonucleotides containing a perfect AGGTCA repeat (RAREm3) were incubated with in vitro translated R A R a , a weak specific retarded complex could be observed compared with the control (Fig. 6). The intensity of this bandcould be increased by adding more R A R a (data notshown). No binding of R X R a alone could be observed, and a strong complex could be seen when these oligonucleotides were incubated with RARa and RXRa together. To determine the specificity of the interaction,competition experiments were performed. ColdRARE a t 100-fold excess competed outall RAR-RXR heterodimer binding (Fig. 71, and no or very little competition was observed when cold RAREml, RAREm2, and RAREml+2 (whereeach or both half-sites havebeen mutated) were used a t 100-fold excess. However, cold RAREm3 was able
FIG.6. RXRa enhances b i n d i n g of KARcv to the human CRABPI1 RARE and to an idealized direct repeat (RAREm3).In uitro translated receptors or nonprogrammed reticulocyte lysate (C) were incubated with""P-labeled RARE or RAREm3 (see Fig. 5) and analyzed on a 4% acrylamide gel. Specific receptor-DNA complexes are indicated by an arrowhead. RARa + RXRa e4
+
l-nlv-m
EEEE wwwww ITaaaa I
aaaaa aauaa
FIG.7. Both half-sites of t h e h u m a n CRABPII RARE are required for binding of R A R a - R X R c r heterodimers. In uitro translated receptors were incubatedwlth "P-lnheled RARE alone (-) or with a 100-fold excess of nonlabeled wild-type human CRABPII RARE, RAREml, RAREm2, RAREml+2, or RAREm3 (see Fig. 5) and were analyzed on a 4%~ acrylamide gel. Specific receptor-DNA complexes are indicated by an arrowhead.
to compete out all RAR-RXR binding when used a t a 100-fold excess (Fig. 7). Comparison of the Upstream Regions of Human and Mouse CRkBPII Genes-The mouse CRABPII gene has been shown to contain two RAR-RXR heterodimerbindingsites a t -1161 (DR-2) and -657 (DR-1) that are required for full transcriptional induction by all-trans-RA and 9-cis-RA (20). When the upstream region of the human CRABPII gene was compared with the mouse gene (Fig. 8), we found that the mouse DR-1 core sequence was identical in the human gene but was apparently nonfunctional (Fig. 3A), and no element homologous to the mouse DR-2 could be found, even when up to -1500 bp of the humangene was usedfor comparison(27) (data not shown). An apparently nonfunctional DR-1 RARE a t -457 that we previously reported (27) isalso indicated and is not present in the mouse gene. In addition, thefunctional human CRABPII DR-5
22338
--
Characterization of a Far Upstream RARE in the Human C M P I I Gene
"""""""""""""""-
-5.6Kb GGG
!"-r-mxx"-m:
(Functional) DR-5
r"" """"""""""-
AGG
-457
(Nonfunctional) DR-1 (a)
TCC
LwxzB"Xr-m:
Am-439
DR-l(b) (Nontunctlonal)
I / -1 EXON I hCRABPll I
I
FIG.8. A comparison of location, sequence, and functionality of direct repeats ( D R ) found in the human (27) and mouse (20) CRABPII genes. Schematic representation of the human and mouse CRABPII promoters. The sequence of the functional human CRABPI1 gene DR-5 RARE located at approximately -5.6 kb and a nonfunctional direct repeat (DR-1 (b) at -457 bp are displayed.A comparison is shown between a nonfunctional direct repeat (DR-1 (a))located at -738 bp in the human gene with identical homology to the functional mouse DR-1. In addition, the functional mouse DR-2 at -1161 bp with no homology in the human gene is shown.
that is located approximately at -5.6 kb is alsoindicated (Fig. 8). DISCUSSION
We previously cloned two forms of CRABP from human skin and demonstrated thatCRABPII, but not CRABPI, is strongly induced by all-truns-RA in human skin in vivo and in cultured human skinfibroblasts in vitro (26). We also demonstrated, by nuclear run-on analysis, that the human CRABPII gene is rapidly and transiently transcriptionally regulated by all-trans-RA in cultured human skinfibroblasts (27). To identify cis-acting elements responsible for RA induction, we have, in the present investigation, employed deletion and mutational analysist o identify a critical far upstreamRARE in the human CRABPII gene. Then, by gel retardation analysis, we have characterized the interaction of RAR and RXR with this element. To determine whether retinoid receptors could regulate human CRABPII gene transcription, cotransfection experiments were performed. When Cos-1 cells were transfected with a CAT reporter construct containing -8.0 kb of the upstream region, a1l-trans-R.A (1yM)-dependent transcription by RARs, but not RXRa, was observed (Fig. 1).Then, by deletion analysis, a region essential for RA induction, located approximately -5.6 kb upstream from the human CRABPII gene start site, wasidentified (Figs. 3, A and B , and 4). Sequencing and mutational analysis identified a direct repeat within this region (GGGTCAttggaAGGACA) with 5-bpspacing (DR-5) (Figs. 4 and 5). Both half-sites are important for induction, because mutation in either half-site results in a loss of induction (Figs. 4 and 5 ) . This element has homology to, and the same spacing as, RAREs found in the RARa2, RARp2, and alcohol dehydrogenase 3 (ADH3) promoters (37-44) but different spacing fromthe two RAREs found in themouse CRABPII promoter (20) (Fig. 8). It hasbeen shown that thespacing between repeats, aswell as the natureof bases flanking the repeats, are importantfor determining RAR and RXR binding efficiencies (23). Direct repeats withspacing of 2 and 5 bp most efficiently bind RAR-RXR heterodimers (23-24). In accordance, we find in gel retardation assays that the humanCRABPII RARE most efficiently binds RAR-RXR heterodimers, and a much lower affinity for RAR
homodimers was seen(Fig. 6).No binding of RXRa homodimers was observed, even when high concentrations of RXR were used (Fig. 6) (data notshown). Thus, our studies clearly indicate that thehuman CRABPII RARE binds RAR-RXR heterodimers much more efficiently, compared with RAR homodimers. However, when cotransfection studies were performed, a relatively small effect was seen on all-trans-RA induction when R A R a was cotransfected with R X R a (Fig. 2). It may be that when high amounts of RAR are transfected,RAR homodimers can form in the cell. Consequently, when RARa were transfected at high concentrations, no stimulation by R X R a could be seen (Figs. 1 and 2). However, when lower concentrations of R A R a were used, a stimulation by RXRa could be seen (Fig.2). This stimulation is much smaller than whatwould be expected from the gel retardation experiments, in which a large difference in binding was seen when R X R a and R A R a were used compared with R A R a alone (Fig. 6). It is likely that thisdifference is due to thepresence of endogenous RXRs in Cos-1 cells (23). Future studies usingRAR- and RXR-dominant negative mutants or a yeast system thatis devoid of these receptors may resolve this discrepancy (20, 45). It was recently demonstrated that the mouse CRABPII gene contains two RAR-RXR heterodimer binding sitesRAREl (DR-2) and RARE2 (DR-1) within the first 1200 bp of the upstream region, mediating induction of transcription by alltruns-RA and 94s-RAinmouse P19 embryonal carcinomacells (20). Deletion analysis of the mouse gene demonstrated that both RAREs are required for full transcriptional induction by all-truns-RA and 9-cis-RA. Sequence analysis of the human CRABPII gene revealed that when compared with the mouse gene, it is lacking the most 5"heterodimer binding site (DR-2) and contains a highly conserved repeat (DR-1 (a)) that is identical to the DR-1 found in the mouse gene (Fig. 8). The mouse RAREl is a direct repeat (AGTTCAccAGGTCA) with a 2-bp spacing (DR-2), and RARE2 is a direct repeat (AGGGCAgAGGTCA) with l-bp spacing (DR-1). The human gene also contains a direct repeat with a l-bp spacing (DR-1 (b)), 3' of DR-1 (a), that is not found in the mouse gene. However, when CAT constructs containing up to -2.7 kb of the human CRABPII gene were transfected intodifferent cell lines (including HeLa, F9, CV-1, P19, Cos-1, and skin fibroblasts), together with ex-
Characterization of a FarUpstream RARE in theHumanCRABPII pression vectors for RARs and m a , very little all-trans-RA induction was observed (Fig. 3 A ) (data not shown). In addition, when an internal deletion was made in the -5.8-kb construct that removed sequences between -416 and -1423, including the DR-l(a) and DR-l(b) repeats (Fig. 8), no loss of RA induction was observed (-5.8A1.0 kb, Fig. 3 A ) . This means that both of these repeats do not contribute to RA inducibility and are unlikely to be functional. In conclusion, we have demonstrated that a single far upstream RARE (approximately at -5.6 kb) with 5-bp spacing (DR-5) is critical for all-trans-RA induction of the human CRABPII gene. We have also shown that thiselement strongly binds RAR-FtXR heterodimers in uitro. Moreover there are many similarities as well as differences in regulation of the human in comparison to the mouse CRABPII gene. For example, the humanCRABPII geneis not RA-inducedwhen cells alone in Cos-1 cells (Figs. 1and 21, are transfected with and R X R a strongly activates the mouse CRABPI1 gene in these cells (22). We have also shown that one element (DR-1 (a)) is identical between the mouse and human genes but is apparently not functional in the humangene. However, the possible importance of these differences between the humanand mouse CRABPII genes to in vivo regulation by RA remains unknown. In addition, because CRABPII protein binds all-trans-RA, the direct strong induction via an RARE is compatible with the idea that anincreased amount of CRABPI1functions in thecell to limit the quantity of free all-trans-RA available to bind and activate M s .
m c v
REFERENCES 1. Sporn, M. B., and Roberts, A. B., (1983) Cancer Res. 43,3034-3040 2. Aaselineau, D., Bernard, B. A,, Bailly, C., and Darmon, M. (1989) Deu. Biol. 133,322435 3. Eichele, G. (1989) Dends Genet. 5, 246-251 4. Summerbell, D., and Maden, M. (1990) Dends Neurosci. 13, 142-147 5. Giguere, V., Ong, E.S., Segui, P., and Evans,R. M. (1987) Nature 330,624-629 6. Petkovich, M., Brand, N. J., Krust, A,, and Chambon, P. (1987) Nature 330, 44-50 7. Brand, N., Petkovich M., h s t , A., Chambon, p., deThe,H., M a r c h A., Tiollais, P., and Dejean,A. (1988)Nature 332, 850453 8. Benbrook, D., Lernhardt, E., and Pfahl,M. (1988) Nature 333, 669472 9. Zelent,A,,Krust, A., Petkovich, M., Kastner, P., and Chambon, P. (1989) Nature 339, 714-717 10. &.,&,A,, &stner, p., petkovich, M,, zelent, A,, and Chambon,p, (1989)proc, Natl. Acad. Sci. U. S. A. 8 6 , 53105314 11. Mangelsdorf, D,J., Ong, E.S,, Dyck, J, A,, and Evans,R. M, (1990)Nature 345, 224-229 12. Mangelsdorf, D. J.,Borgmeyer, U., Heyman, R. A,, Zhou, J. Y., Ong, E.s., Oro, A. E., Kakizuka, A., and Evans, R. M. (1992) Genes & Deu. 6, 329-344 13. Leid, M., Kastner, P., Lyons, R., Nakshatri, H., Sauuders, M., Zacbarewski, T., Chen, J.-Y. Staub,A., Garnier,J.-M., Mader, S., andchambou, P. (1992)Cell 68,377-395 14. Yu,V. C., Delsert, C., Andersen, B., Holloway, J. M., Devary, O., Naar, A. M., Kim, S. Y.,Boutin, J.-M., Glass, C. K , and Rosenfeld, M. G. (1991) Cell 67, 1251-1266
Gene
22339
15. Levin, A. A,, Sturzenbecker, L. J., Kazmer, S., Bosakowski, T., Huselton, C., Allenby, G., Speck,J.,Kratzeisen, J., Rosenberger, M.,Lovey, A., and Grippo, J. F. (1992) Nature 356, 359-361 16. Heyman, R. A,, Mangelsdorf,D. J., Dyck, J. A,, Stein, R. B., Eichele, G., Evans, R. M., and Thaller, C. (1992) Cell 68,397-iO6 17. Allenby, G., Bocquel, M.-T., Saunders, M., Kazmer, S., Speck, J., Rosenberger, M., Lovey,A., Kastner, P., Grippo, J. F., Chambon, P., and Levin,A. A. (1993) Proc. Natl. Acad. Sci. U. S. A. 90, 30-34 18. Zhang, X.-K., Hoffmann, B., Tran, P., Graupner, G., and Pfahl, M. (1992) Nature 3 5 6 , 4 4 1 4 6 19. Kliewer, S. A, Umesono, K., Mangelsdorf, D. J.,and Evans,R. M. (1992)Nature 356,446449 20. Durand, B., Saunders, M.,Leroy, P., Leid, M., and Chambon, P. (1992) Cell 71, 73-85 21. Zhang, X.-K., Lehmann, J., Hoffmann, B., Dawson, M. I., Cameron, J., Graupner, G., Hermann, T., Tran, P., and Pfahl, M. (1992) Nature 358, 587-591 22. Nagpal, S., Saunders, M., Kastner, P., Durand, B., Nakshatri, H., and Chambon, P. (1992) Cell 70,1007-1019 23. Mader, S., Leroy, P., Chen, J.-Y., and Chambon, P. (1993) J. Biol. Chem. 268, 591-600 24. Mader, S., Chen, J.-Y., Chen, Z., White, J., Chambon, P., and Gronenmeyer, H. (1993)EMBO J . 12,5029-5041 25. Fogh, K., Voorhees, J. J., and &&om, A. (1993)Arch. Biochem. Biophys. 300, 751-755 26. k t r o m , A,, Tavakkol,A,, Pettersson, U., Cromie, M., Elder, J. T., and Voorhees, J. J. (1991) J. Biol. Chem. 266, 17662-17666 27. Astrom, A,, Pettersson, U., and Voorhees, J. J. (1992) J . Biol. Chem. 267, 25251-25255 28. Giguere, V., Lyn, S., Yip, P., Siu, C.-H., and Amiu, S. (1990) Proc. Natl. Acad. Sci. U. S. A. 87, 6233-6237 29. MacGregor, T. M., Copeland, N. G., Jenkins, N. A., and Giguere, V. (1992) J . Biol. Chem. 267,7777-7783 30. Elder, J. T., Astrom, A,, Pettersson, U., Tavakkol, A., Griffiths, C. E. M., Krust, A., Kastner, P., Chambon, P., and Voorhees, J. J. (1992) J. Znuest. Dermatol. 98,673-679 31. Luckow, B., and Schutz, G. (1987) Nucleic Acids Res. 15, 5490 32. Sanger, F., Nicklen, S., and Coulson, A. R.(1977)Prac.Natl. Acad. Sci.U. S. A. 74,5463-5467 33. Rosenthal, N. (1987) Methods Enzymol. 152, 704-720 34. Seed, B., and Sheen, J.-Y. (1988) Gene (Amst. 67,271-277 35. Astrom, A., Pettersson, U., Krust, A., Chambon, P., and Voorhees, J. J. (1990) Biochem. Biophys. Res. Commun. 173,339-345 36. Hermann, T.,Hoffmann, B., Zhang, X.-K, Tran, P., and Pfahl, M. (1992) Mol. Endocrinol. 6, 1153-1162 37. de The. H.. delMar Vivanco-Ruiz. M.. Tiollais., P... Stunnenbere. H.. and Dejeau, A . (1990) Nature 343, 177-180 38. Hoffmann, B., Lehmann, J. M., Zhang,Hermann, T.,Husmann, M., Graupner, G., and Pfahl, M. (1990) Mol. Endocrinol. 4, 1727-1736 39. Mendelsohn, C., Ruberte, E., LeMeur, M., Morriss-Kay, G., and Chambon, P. (1991) Development 113,723-734 40. Sucov, H. M., Murakami, K. K., and Evans, R. M. (1990) Proc. Natl. h a d . sei. U.S. A . 87, 5392-5396 41. Zelent,A., Mendelsohn, c.3 &stner, p., h s t , A . , Gamier, J.”., Ruffenach, F., Leroy, P., and Chambon, P. (1991) EMBO J. 10, 71-81 42. Leroy, p., h s t , A., Zelent, A., Mendelsohn, c . , Gamier, J.”., Kastner, P., Dierich, A., and Chambon, P. (1991) EMBO J. 10,59-69 43. Leroy, P., Nakshatri, H., and Chambon, P. (1991)Proc. Natl. Acad. Sci. U. S. A. 88,10138-10142 4 4 . Duester, G., Shean, M. L., McBride, M. Stewart, S., and M. J. (1991) Mol. Cell. Biol. 11, 1638-1646 45. Heery, D. M., Zacharewski, T., Pierrat, B., Gronemeyer, H., Chambon, P., and Losson, R. (1993)Aced. Proc. Natl. Sci. U. S. A. 90, 428111285 ~
