| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Endocrinology |
Molecular Endocrinology and Oncology Research Center, Centre Hospitalier Université Laval Research Center (CHUQ) and Laval University, Quebec G1V 4G2, Canada
 |
ABSTRACT |
|---|
 Top ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
-dihydrotestosterone. These steroid hormones interact with the ligand-binding domain of AR and induce conformational changes that allow AR to interact with transcriptional coregulators of which the specific roles in AR-mediated gene regulation remain to be determined (4)
. AR can modulate gene expression directly by interacting with specific elements in the regulatory regions of target genes (5)
or indirectly by activating various growth factor signaling pathways (6)
. Androgens play a determining role in prostate differentiation and development (7) . Androgens have been shown to stimulate the growth of prostate cancer cells both in vitro and in vivo, and they are recognized to play an important role in prostate tumor growth (8) . Since the original discovery of Charles Huggins in 1941, blockade of androgen formation and action has evolved as the cornerstone of the treatment of prostate cancer (9) . Genetic alterations of AR consisting of AR gene amplification (10) and mutations in the AR coding sequence (11) are frequently observed in prostate cancer and are believed to contribute in some manner to cancer progression. In fact, AR mutations are found in several commonly studied human prostate cancer cell lines including LNCaP, CWR22, MDA PCa 2a, and MDA PCa 2b (12, 13, 14, 15) . Given the importance of androgens and the AR in prostate development and cancer, it is important to identify genes that are regulated by androgens in the human prostate.
Transcription factors of which the expression is regulated by androgen in prostate cells are a particularly interesting class of androgen-responsive gene because they control the expression of a specific subset of androgen-responsive genes. Relatively few androgen-responsive transcription factors have been identified to date. One of these, NKX3.1, is a homeobox gene located on chromosome 8p21 that is believed to play an important role in prostate differentiation and the control of prostate cell proliferation (16
, 17)
. Disruption of NKX3.1 in mice results in dysplasia of the prostatic epithelium (18)
, and loss of NKX3.1 expression is observed in a high proportion of hormone-refractory and metastatic prostate tumors (19)
. Androgens also down-regulate the expression of the zinc finger transcription factor early growth response- (20)
and induce the orphan nuclear receptor TR3 (21)
, c-myc, and nm23 (22)
in LNCaP cells. The orphan nuclear receptor TR3 is particularly interesting, because TR3 translocates from the nucleus to mitochondria in response to proapoptotic signals and induces cytochrome c release and apoptosis (23)
. Down-regulation of EGR- by androgens in LNCaP cells is consistent with the growth-stimulatory effect of androgens in prostate cells, because the EGR- protein is nearly identical to transforming growth factor-ß-inducible early gene that has been shown to inhibit cell proliferation (24)
. In fact, EGR- and transforming growth factor-ß-inducible early gene are expressed from alternative promoters of the same gene through the use of alternative first exons (25)
.
More recently, Xu et al. (26) identified a number of putative androgen-responsive genes using serial analysis of gene expression to compare the expression profiles of control and R1881-treated LNCaP cells. Seven of these genes were transcription factors that were not known previously to be androgen-regulated, including KRAB-associated protein 1, prostate-derived Ets factor, mitochondrial transcription factor 1, a sox-like transcription factor, and the orphan nuclear receptor ear-2, all of which were up-regulated by androgens. Two transcription factors, the homeobox protein Hox B13 and the histone acetylase/deacetylase CHD4/Mi-2ß, were down-regulated by R1881. Whereas the androgen regulation of these genes remains to be confirmed, it is of interest to note that the prostate-specific transcription factor PDEF interacts with AR and enhances androgen-induced activation of the PSA promoter (27) .
Our laboratory is interested in identifying androgen-regulated genes in human prostate cancer cells. We hypothesize that these genes may play an important role in androgen-induced prostate development and androgen-dependent prostatic diseases such as prostate cancer. We report here the cDNA cloning, gene structure, expression profile, androgen regulation, and partial characterization of a novel androgen-regulated transcription factor that we have named AIbZIP. AIbZIP is a novel member of the CREB/ATF family of transcription factors that is highly expressed in prostate tumors and of which the expression is up-regulated by androgen in LNCaP cells.
|
MATERIALS AND METHODS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
-32P]dCTP using the DECAprimer II DNA labeling kit (Ambion, Austin, TX), and used to screen 4 x 105 recombinant plaques of a LNCaP cDNA library. After three rounds of screening, positive clones were excised in vivo directly into the phagemid pBK-CMV and sequenced using the T7 sequencing kit (USB Corporation, Cleveland, OH) and Deaza G/A T7 sequencing mixes (Amersham Pharmacia Biotech, Piscataway, NJ) as necessary.
FISH.
To confirm that the BAC clone RP11-422P24 (GenBank accession no. AL358472) contained the AIbZIP gene, we amplified two fragments by PCR using sense primers 5'-GATGGGAGGGAAGCCAAGA-3' and 5'-CTGTTCTGCACAGAACCA-3' in combination with the reverse primer 5'-CAAATGCAGACTAGGCCTC-3'. The BAC clone was then biotinylated by nick translation and hybridized in situ to metaphase chromosomes from normal human lymphocytes (29)
. Chromosomes were counterstained with propidium iodide and 4',6-diamidino-2-phenylindole. After hybridization, the biotinylated probe was detected with avidin-FITC. Images of metaphase preparations were captured digitally using a cooled CCD camera (Biomedical Photometrics, Inc., Waterloo, ON). FISH was performed in Dr. S. Scherer’s laboratory in the Department of Genetics at The Hospital for Sick Children, Toronto, Ontario, Canada.
Cell Lines and Culture Conditions.
All of the cell lines used in these experiments were obtained from American Type Culture Collection (Manassas, VA). The human breast and prostate cancer cell lines and corresponding culture conditions were described previously (28)
. Human kidney 293 cells were cultured in MEM containing 10% FCS, 1% nonessential amino acids, 100 µg/ml streptomycin, and 100 IU/ml penicillin.
Experiments designed to characterize the regulation of AIbZIP expression by androgens in LNCaP cells were performed as described (28)
. In brief, the LNCaP cells used in the time course, R1881 dose-response and cycloheximide/actinomycin D experiments (Fig. 7)
were cultured in RPMI 1640 supplemented with hormone-depleted 0.25% FCS for 6 days before each experiment. Under these conditions, LNCaP cells remain in G1 for 24 to 48 h after addition of 0.1 nM of R1881 (data not shown). The LNCaP cells used in the experiments designed to assess the effects of antiandrogens on AIbZIP expression (Fig. 8)
were cultured for 3 days in hormone-depleted 2% FCS before each experiment. The medium was changed at 2–3 day intervals. At the start of each experiment, the medium was replaced with identical medium containing the indicated concentrations of compounds. Each experiment was performed at least three times, and all of the experiments were performed in LNCaP cells between passage 27 and 37.
|
|
Northern Blotting.
The human multiple tissue blots (2 µg poly(A)+ RNA/lane) were obtained from Clontech Laboratories, whereas total RNA was isolated from cell lines using Tri-Reagent (Molecular Research Center, Inc., Cincinnati, OH). Electrophoresis of cancer cell line RNA samples, Northern blotting, hybridization, and autoradiography were performed as described (28)
. The human tissue and cell line blots were probed with the A25 cDNA probe that was used to screen the LNCaP cDNA library. The human tissue blots were also probed with the full-length AIbZIP cDNA. Radiolabeled cDNA probes for human ß-actin (Clontech Laboratories) and human GAPDH were used as loading controls for the human multiple tissue blots and cell line blots, respectively. A portion of the GAPDH cDNA (nucleotides 172–717 of GenBank accession no. M33197) was amplified by PCR using oligonucleotides 5'-ATTGACCTCAACTACATGGT-3' and 5'-CTTGCCCACAGCCTTGGCAG-3' and radiolabeled with [-32P]dCTP.
RNase Protection Assays.
To confirm that clone A25 corresponded to an androgen-responsive transcript, the plasmid containing the A25 cDNA fragment (pCRII-SSH2-A25/AS) was used to generate a cRNA probe. The plasmid was linearized with EcoRV, and a 434-bp probe (including 135 bp of vector/linker sequences) was synthesized using Sp6 RNA polymerase.
To confirm that AIbZIP corresponded to an androgen-responsive gene, we constructed a plasmid (pBSKS-A25-3'UTR/AS) to generate a probe that would hybridize to a region of the AIbZIP transcript that is not recognized by the A25 probe. A cDNA fragment containing the last 64 bp of the open reading frame and the first 173 bp of the 3'UTR of AIbZIP (nucleotides 1203–1439) was cloned into pBluescript II KS (+/-) by PCR using oligonucleotides 5'-GATGGGAGGGAAGCCAAGA-3' and 5'-CAAATGCAGACTAGGCCTC-3'. The plasmid was linearized with BamHI, and the 318 bp probe (including 81 bp of vector sequences), designated 3'UTR, was produced using T3 RNA polymerase. The 3'UTR probe was used for most of the experiments presented in this report.
A probe designated as E4-7 was designed for the purpose of distinguishing the full-length AIbZIP transcript from putative alternative transcripts that were isolated as cDNA clones. To generate the required probe, a cDNA fragment encompassing nucleotides 610–890 of the full-length AIbZIP cDNA was cloned into pBluescript II KS (+/-) by PCR. The 362 bp E4-7 probe (including 81 bp of vector sequences) was produced using T3 RNA polymerase after linearizing the plasmid (pBSKS-A25-E4-7/AS) with BamHI.
A cRNA probe for ß-actin was used as an internal control in all of the RNase protection experiments. A fragment of the ß-actin cDNA (nucleotides 704–947 of GenBank accession no. X00351) was amplified by PCR and cloned into the XbaI and KpnI sites of pBluescript SK. The plasmid was linearized with DdeI, and a 145-bp cRNA probe that protects a 130-bp fragment of the human ß-actin mRNA (nucleotides 818–947 of the cDNA) was synthesized using T7 RNA polymerase.
The cRNA probes were labeled with [-32P]UTP (800 Ci/mmol) using the Riboprobe in vitro Transcription System (Promega Corp., Madison, WI). RNase protection assays were performed using 10 µg of total RNA using the RPA III kit (Ambion, Inc.) according to the manufacturer’s instructions. Protected fragments were separated by electrophoresis in 6% acrylamide 7 M urea gels. The gels were exposed to Hyperfilm MP (Amersham Pharmacia Biotech, Aylesbury, United Kingdom) and X-ray films were quantitated by scanning densitometry using the Bioimage 110 S (Millipore Corp., Bedford, MA). AIbZIP mRNA levels were corrected for ß-actin mRNA levels, and the AIbZIP:ß-actin ratios in the control cells of each experiment were assigned a value of 1. AIbZIP mRNA levels in treated cells are expressed as AIbZIP:ß-actin ratios relative to those in the corresponding control cells.
Expression Vectors.
To express AIbZIP as a fusion to the NH2 terminus of EGFP, the coding sequence of AIbZIP was amplified from the full-length cDNA by PCR and cloned into the BglII and KpnI sites of pEGFP-N3 (Clontech Laboratories). The fusion protein encoded by plasmid pEGFPN3-AIbZIP contains aa 1–395 of AIbZIP, 10 amino acids (GATGPGSIAT) encoded by vector polylinker DNA, and EGFP. Plasmid AIbZIP (1–290)-EGFP was made by amplifying the first 290 codons of AIbZIP by PCR and subcloning the DNA fragment in-frame with the EGFP open reading frame in plasmid pEGFP-N3.
To express AIbZIP as a fusion to the COOH-terminus of the DNA-binding domain of GAL4, the coding sequence of AIbZIP was amplified from the full-length cDNA by PCR and cloned into the EcoRI and XbaI sites of pM (Clontech Laboratories). The fusion protein encoded by pM-AIbZIP contains aa 1–147 of GAL4, three amino acids (PEF) encoded by vector polylinker DNA, aa 1–395 of AIbZIP, and two residues (SR) that are encoded by the XbaI site that precedes the termination codon. Plasmids expressing GAL4 fused to various deletion mutants of AIbZIP were constructed using conveniently placed restriction sites or PCR. The sequence of each AIbZIP expression plasmid used in our experiments was verified.
AIbZIP Localization Experiments.
For transient expression of EGFP and EGFP fusion proteins, 1 x 105 LNCaP cells were seeded in four-well chamber slides (9 x 18 mm; Nalge Nunc International, Naperville, IL) in RPMI 1640. The cells were transfected with pEGFP-N3, pEGFPN3-AIbZIP, or pEGFPN3-AIbZIP (1–290) (1 µg DNA/well) 24 h after plating using Lipofectin (Life Technologies, Inc., Rockville, MD). After transfection (24 h), the cells were fixed with Formalin solution (Sigma Chemical Co. Diagnostics, St. Louis, MO). Crystal/Mount (Biomeda Corp., Foster City, CA) was used to mount the coverslips to the slides. Fluorescent proteins were detected by fluorescence microscopy, and photographs were taken using a digital camera (SPOP RT Slider; Diagnostic Instruments, Inc., Sterling Heights, MI) mounted directly on a BX-60 microscope (Olympus, Melville, NY). The stability of AIbZIP-EGFP fusion proteins was verified in human kidney 293 cells. The cells were cultured in MEM, and the corresponding expression plasmids were transfected using ExGen 500 (MBI Fermentas, Amherst, NY). The cells were harvested 24 h after transfection.
GAL4-AIbZIP Experiments.
Human kidney 293 cells were seeded in MEM containing 10% FCS at 1.5 x 105 cells/well in 12-well plates (21 mm/well; Becton Dickinson Labware, Bedford, MA) 24 h before transfection. Each well was transfected with 1.5 µg of plasmid DNA comprised of 1.125 µg pM plasmid expressing either the GAL4 DNA-binding domain alone, GAL4 fused to AIbZIP, or GAL4 fused to VP16 (pM3-VP16), and 0.25 µg of a GAL4-responsive luciferase reporter plasmid (pGL3-G5E1BTATA). pRLnull (0.125 µg; Promega) was used as an internal control for transfection efficiency. Transfections were performed using Exgen 500 (MBI Fermentas). After 48 h, the cells were lysed in Passive Lysis Buffer (Promega), and Firefly and Renilla luciferase activities were determined using a Dual-Luciferase Reporter Assay System (Promega) according to the manufacturer’s instructions. Firefly luciferase activity values were normalized using Renilla luciferase activity values. The stability of GAL4-AIbZIP fusion proteins was verified in T-47D cells. The cells were transfected with the corresponding plasmids using FuGENE (Roche, Indianapolis, IN) and harvested 48 h after transfection.
AIbZIP Antibody Preparation.
A synthetic peptide corresponding to aa 379–392 of AIbZIP (KPRPSGRIRSVLHA) was synthesized on an ABI 433A peptide synthesizer using FastMoc chemistry. The peptide was purified on a Vydak 22 x 250 mm C18 reverse-phase high-performance liquid chromatography column using a 0.1% trifluoroacetyl/acetonitrile gradient. Rabbits were immunized with a total of 0.5 mg of peptide per rabbit.
Immunoblotting.
For immunoblotting, cells were harvested in lysis buffer [6 M urea, 20 mM Tris (pH 6.8), 1% SDS, Roche Complete Protease Inhibitor mixture, pepstatin, and DTT] and then sonicated. Protein extracts (30 or 40 µg) were separated by SDS-PAGE on 12% or 13.5% acrylamide gels and transferred to nitrocellulose. EGFP fusion proteins, GAL4 fusion proteins, and -tubulin were detected using the following commercially available primary antibodies: Living Colors full-length A.v. polyclonal antibody (Clontech Laboratories), mouse monoclonal antibody RK5C1 raised against the DNA-binding domain of GAL4 (Santa Cruz Biotechnology, Inc., Santa Cruz, CA), and mouse IgM antitubulin antibody (Santa Cruz Biotechnology). The following peroxidase-conjugated AffiniPure secondary antibodies were obtained from Jackson ImmunoResearch Laboratories Inc. (West Grove, PA): goat antirabbit IgG, goat antimouse IgM, and goat antimouse IgG. Immune complexes were revealed using the Supersignal Ultra Chemiluminescent detection system (Pierce, Rockford, IL).
AIbZIP Immunoblots.
To assess AIbZIP protein levels in R1881-treated LNCaP cells, equal amounts of protein extracts (30 µg/lane) were electrophoresed on 12% SDS-PAGE gels and transferred to nitrocellulose filters. The blots were incubated overnight at 4°C with rabbit polyclonal antibody against AIbZIP (diluted 1:1000 in 5% milk and 1 x Tris-buffered saline; 0.9% NaCl and Tris-HCl 10 mM) followed by a 3-h incubation at 37°C with peroxidase-conjugated AffiniPure goat antimouse IgG (Jackson ImmunoResearch Laboratories Inc.). AIbZIP protein levels were quantified using scanning densitometry. AIbZIP protein levels in R1881-treated cells are expressed relative to the levels determined in control LNCaP cells, which were assigned a value of 1. The AIbZIP levels (relative to control) achieved with 0.01, 0.1, 1, and 10 nM R1881 in two independent experiments were within 4% of the calculated average values.
Prostate Specimens.
The prostate specimens used in our studies were kindly provided by Dr. Jean Emond (Hôtel-Dieu de Lévis, Lévis, Québec, Canada), Dr. Bernard Têtu (Hôtel-Dieu de Québec, Québec City, Québec, Canada), and by Drs. Martin Lemay and Michel Tremblay (Centre Hospitalier de l’Université Laval, Québec City, Québec, Canada). Thirty-seven different prostate specimens were used to evaluate AIbZIP expression by immunohistochemistry and/or in situ hybridization. These tissues consisted of 20 specimens from men undergoing TURP for symptomatic benign prostatic hyperplasia and 17 paraffin blocks of formalin-fixed archival prostatic needle core biopsies displaying infiltrating adenocarcinoma. The TURP specimens were fixed by immersion in 2% glutaraldehyde, 4% formaldehyde, and 3% dextran in 0.05 M phosphate buffer (pH 7.4) for in situ hybridization or in 10% neutral buffered formalin for immunohistochemistry. They were then processed and embedded in paraffin blocks. BT-474 and MDA-MB-231 breast cancer cells were used as controls for the immunohistochemical detection of AIbZIP. The cells were harvested, fixed in 10% neutral buffered formalin, centrifuged in 2% agarose, postfixed in 10% neutral buffered formalin, and embedded in paraffin.
In Situ Hybridization.
Paraffin sections (4 µm) were collected on Superfrost Plus (Fisher) glass slides. The sections were deparaffinized in toluene, rehydrated through graded alcohol, and then treated with proteinase K [1 µg/ml in 100 mM Tris-HCl, 50 mM EDTA, (pH 8.0)] for 30 min at 37°C. Sections were then immersed in prehybridization buffer [50% formamide, 0.8 M NaCl, 120 mM Tris base (pH 8.0), 8 mM EDTA, 200 µg/ml salmon sperm DNA, 200 µg/ml tRNA, and 4% dextran sulfate] for 2 h at room temperature. The [35S]UTP-labeled 318-bp antisense AIbZIP riboprobe was generated by in vitro transcription using T3 RNA polymerase from pBSKS-A25–3'UTR/AS linearized with BamHI. The [35S]UTP-labeled 303-bp sense (control) riboprobe was generated by in vitro transcription using T7 RNA polymerase from pBSKS-A25–3'UTR/AS linearized with EcoRI. Sections were incubated overnight at 53°C in hybridization buffer (prehybridization buffer plus 10 mM DTT and 2 x 106 cpm probe). To remove the excess probe, the sections were washed twice in 50% formamide, 2 x SSC at 58°C for 90 min. The slides were dehydrated in graded alcohol, air-dried, and then dipped in NBT-2 emulsion (Eastman Kodak, Rochester, NY; diluted 1:1 in distilled water) and exposed for 14 days. Slides were developed and counterstained with H&E. The silver grains resulting from hybridization of sense and antisense cRNA probes were counted manually by overlaying photomicrographs with acetate sheets.
Immunohistochemistry.
Immunostaining was performed using Zymed SP kits (San Francisco, CA). Paraffin sections (4 µm) were deparaffinized in toluene and rehydrated through ethanol. Endogenous peroxidase activity was eliminated by preincubation with 3% H2O2 in methanol for 20 min. A microwave retrieval technique using citrate buffer was applied (30)
. After cooling the slides, nonspecific binding sites were blocked using 10% goat serum for 30 min. Sections were incubated with the rabbit polyclonal AIbZIP antibody diluted 1:400 for 2 h at room temperature, washed in PBS buffer, and incubated with biotinylated antirabbit secondary antibody for 12 min and then with streptavidin-peroxidase for another 12 min. Under microscope monitoring, diaminobenzidine was used as the chromogen to visualize the biotin/streptavidin-peroxidase complexes. Counterstaining was performed using #2 Gill’s hematoxylin for 45 s. For the negative control, the preimmune serum was used at the same dilution as the primary antibody.
Nucleotide Sequence Accession No.
The GenBank accession no. of the AIbZIP clone sequence reported here is AF394167.
|
RESULTS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
To obtain a full-length A25 cDNA we screened a LNCaP cDNA library using radiolabeled clone A25 as the probe. Twenty-five cDNA clones were isolated and characterized. The longest cDNA, number A25-15, was 1564 bp in length and contained a 1188-bp open reading frame preceded by a 78-bp 5'-UTR (Fig. 1)
. A termination codon (TGA) is located 30 bp upstream of and in-frame with the first methionine codon of cDNA A25-15, thereby confirming that this cDNA includes the complete protein coding sequence. The 3'-UTR of cDNA A25-15 is 298 bp in length and contains a polyadenylation signal (AAUAAA) 
20 bp from its 3' extremity. Searches of the expressed sequence tag database at National Center for Biotechnology Information revealed several matches to AIbZIP, and a matching contig could be assembled from three partially overlapping ESTs (AL519763, BG699998, and AW270016).
|
. For this reason, the protein encoded by cDNA A25-15 was designated AIbZIP.
|
. It is comprised of a lysine- and arginine-rich basic region separated from the leucine zipper by a 5-amino acid spacer. A bipartite nuclear localization signal (KKVRRKIRNKQSAQDSRRRKKE) is contained within the bZIP domain (aa 223–240). The bZIP domain of AIbZIP is most similar to that of human CREB-H (69% amino acid identity), human CREB3 (62% amino acid identity), and mouse OASIS (54% identity) proteins (32, 33, 34)
. In fact, 17 of 21 amino acids that constitute the basic domain as well as the 5 spacer amino acids present in AIbZIP are identical to those in CREB-H. The leucine zipper of AIbZIP is similar to that of CREB-H. It contains five repeats and consists of five leucine residues and one cysteine residue at the second heptad position (aa 258) of the leucine zipper. CREB3 and OASIS contain tyrosine substitutions at the same position. Twenty-two of the 36 amino acids present between the first and last leucines are identical in AIbZIP and CREB-H. The bZIP domain of AIbZIP shows less homology (26–35% identity) to the bZIP domains of human ATF3, ATF4, and ATF6 proteins.
On the basis of the similarities between the bZIP domains of AIbZIP, CREB-H, and CREB3, we have assigned AIbZIP to the CREB/ATF subfamily of bZIP proteins. In a recent review, Hai and Hartman (35)
classified mammalian ATF/CREB proteins into six subgroups that are represented by CREB, CRE-BP1, ATF3, ATF4, ATF6, and B-ATF. Proteins between subgroups have little in common except for the bZIP domain, whereas the structural similarity between proteins within subgroups extends to regions outside the bZIP domain. A dendogram analysis performed using the amino acid sequences of AIbZIP, CREB-H, CREB3, and those of representative ATF/CREB proteins revealed that AIbZIP, CREB-H, and CREB3 segregate into a distinct subgroup of ATF/CREB proteins (Fig. 2C)
.
Analysis of the protein sequence of AIbZIP (PSORTII server)5 also revealed the presence of two putative transmembrane domains (aa 169–191 and 294–314) flanking the bZIP domain. The COOH-terminal TMD of AIbZIP (TSTCVLILLFSLALIILPSFS) bears significant homology to the TMD of CREB-H (CVAVLLLSFALIILPSI, aa 323–339). The putative TMD located in the NH2-terminal half of AIbZIP (RAGTVAPVPCTTLLPCQTLFLTD) shows little homology to either the COOH-terminal TMD of AIbZIP or that of CREB-H.
Gene Structure and Chromosomal Location of AIbZIP.
To determine the genomic organization of the AIbZIP gene we searched the GenBank genomic sequence database and found that the complete sequence of the AIbZIP cDNA is contained in human chromosome 1 clone RP11-422P24 (GenBank accession no. AL358472). Comparison of the genomic and cDNA sequences allowed us to determine the exon-intron junctions of the AIbZIP gene (Table 1)
. The AIbZIP gene consists of 10 exons ranging from 74 to 493 bp in size. The first exon codes for the 5' UTR of the mRNA, whereas the AIbZIP open reading frame is encoded by exons 2–10. As shown in Fig. 3A
, the AIbZIP gene lies in close proximity to the JTB/PAR gene (36
, 37)
. The two genes are arranged in a head-to-head orientation, and the last 94 bp of AIbZIP exon 10 correspond to the last 94 bp of JTB/PAR exon 5. We used FISH to confirm the chromosomal location of the gene encoding AIbZIP. Hybridization of biotinylated BAC clone RP11-422P24 to metaphase chromosomes confirmed that this BAC maps to 1q21.3 (data not shown).
|
|
. The second variant, designated form 3 (found in three cDNA clones), contained intron 5 sequences. Both variants result in frameshifts that would produce truncated proteins lacking all (form 3) or most (form 2) of the bZIP domain. To determine whether these transcripts were expressed at appreciable levels in LNCaP cells we designed a cRNA probe (E4–7) that would discriminate between the wild-type AIbZIP transcript and the two variants. Neither variant was detected in LNCaP cells (data not shown), and these two clones were not additionally studied. In addition to these cDNAs, some ESTs contained in the GenBank database (BI087317, BG763732, BF347981, and AL528505) are identical to AIbZIP exon 2 except that they contain an additional 241–277 bp DNA at their 5' end, which suggests that alternative transcripts could be expressed from the AIbZIP gene. On the basis of the available genomic data, this alternative exon is predicted to lie 72 bp 5' to exon 1. We did not isolate a matching cDNA in LNCaP cells, but this alternative transcript would not be predicted to alter the open reading frame of AIbZIP unless other exons were also alternatively spliced.
AIbZIP Localization.
To assess the subcellular localization of AIbZIP, the complete open reading frame of AIbZIP was cloned in-frame with sequences encoding EGFP to produce a plasmid expressing AIbZIP as a fusion protein to the NH2 terminus of EGFP (AIbZIP-EGFP). To evaluate the potential role of the COOH-terminal transmembrane domain of AIbZIP we constructed a plasmid that expresses the fusion protein AIbZIP (1–290)-EGFP, which lacks the last 105 amino acids of AIbZIP (Fig. 4A)
. Plasmids expressing EGFP alone, AIbZIP-EGFP, or AIbZIP (1–290)-EGFP were transiently transfected into LNCaP cells. EGFP alone was distributed evenly in the nuclei and cytoplasm of cells, whereas AIbZIP-EGFP was predominantly localized in the cytoplasm in structures that could correspond to the Golgi apparatus (Fig. 4B)
. Interestingly, deletion of the COOH-terminal portion of AIbZIP that includes the putative transmembrane domain produced a fusion protein that localized exclusively in the nuclei of transfected cells. These results indicate that a domain responsible for the cytoplasmic localization of AIbZIP resides within its COOH-terminal region. The putative NH2-terminal transmembrane domain does not appear to play an essential role in the cytoplasmic localization of AIbZIP. Immunoblots performed using an antibody against EGFP confirmed that the expression plasmids produced fusion proteins of the expected sizes (Fig. 4C)
.
|
. Deletion of the first 48 or 98 amino acids of AIbZIP completely abrogated GAL4-AIbZIP-induced transactivation of the GAL4-responsive luciferase reporter. These results suggested that the activation domain of AIbZIP resides within its NH2-terminal portion. In fact, experiments performed using fusion proteins that express the first 71 or 100 amino acids of AIbZIP confirmed that the NH2-terminal region of AIbZIP is sufficient for activation. It is interesting to note that these fusion proteins were 30 times more active than the full-length GAL4-AIbZIP fusion protein. Moreover, fusion proteins containing the first 150–280 amino acids of AIbZIP were less active than the 71- and 100-residue fusion proteins. Immunoblots performed using a GAL4-specific antibody confirmed that the proteins that displayed less transcriptional activity were not expressed at lower levels (Fig. 5B)
. Therefore, we conclude that the transcriptional activity of the AIbZIP activation domain is negatively regulated by the COOH-terminal portion of AIbZIP.
|
, the AIbZIP probe detected a major 1.7-kb band in control LNCaP cells. The size of this transcript is in agreement with the length of the longest AIbZIP cDNA we isolated (1564 bp). A less abundant transcript that migrated slightly above the major band and which could correspond to an alternatively spliced AIbZIP transcript was also detected. Exposure of LNCaP cells to R1881 caused a marked increase in the amount of 1.7-kb mRNA as well as in that of the less abundant longer transcript.
|
. AIbZIP or AIbZIP-related transcripts were not detected in any other human tissue even after prolonged exposure (6 days) of the blots to autoradiography film. An identical tissue distribution pattern was observed using the full-length AIbZIP cDNA as the probe (data not shown). Although some ESTs containing portions of the AIbZIP cDNA have been isolated in nonprostate cDNA libraries, the absence of detectable full-length AIbZIP transcript in tissues other than the prostate suggested that AIbZIP might be a prostate-specific gene.
To additionally test this hypothesis we performed RNase protection assays using the 3'UTR cRNA probe to determine the expression profile of AIbZIP mRNA in human breast and prostate cancer cell lines. In addition to LNCaP cells, AIbZIP mRNA was detected in the androgen-insensitive prostate cancer cell lines DU 145 and PC-3 (Fig. 6C)
. This finding indicates that AIbZIP expression is not exclusively regulated by androgens, because DU 145 and PC-3 cells do not express AR. Interestingly, AIbZIP mRNA was also detected in all of the human breast cancer cell lines tested except for MDA-MB-231 cells. As in prostate cancer cells, AIbZIP expression in breast cancer cells was not strictly dependent on AR expression, because some of the cell lines that express AIbZIP do not express AR. The full-length 1.7-kb AIbZIP transcript was also detected by Northern blot analysis in those cell lines in which AIbZIP mRNA was most abundant (data not shown). We did not verify whether AIbZIP is expressed in normal mammary cells. However, based on these data, it is possible that AIbZIP expression is restricted to hormone-sensitive tissues, i.e., the prostate and mammary gland.
Androgen Regulation of AIbZIP Expression in LNCaP Cells.
To additionally characterize the regulation of AIbZIP expression by androgens we assessed the effect of R1881 on AIbZIP mRNA levels as a function of time and dose. AIbZIP mRNA levels were determined in LNCaP cell cultures between 1 and 72 h after the addition of 0.1 nM of R1881. This dose of R1881 was selected because it is the same dose that was used to treat LNCaP cultures that served for the cDNA subtraction. As shown in Fig. 7A
, AIbZIP mRNA levels (expressed as AIbZIP:ß-actin ratios) did not change markedly up to 12 h after the addition of 0.1 nM of R1881. However, 2–3-fold changes in AIbZIP mRNA levels were observed starting 24 h after the addition of R1881. Similar results were observed in two other experiments.
The induction of AIbZIP expression by R1881 was less rapid than that of other androgen-responsive genes such as NKX3.1 (17)
and PSA (28)
, which suggests that AIbZIP is not an early response gene. Therefore, we examined the effect of the protein synthesis inhibitor cycloheximide and the RNA synthesis inhibitor actinomycin D on R1881-induced AIbZIP expression in LNCaP cells. The effects of actinomycin D and cycloheximide were compared with that of Casodex, an AR antagonist. The cell culture conditions and the dose of R1881 used in this experiment were identical to those used for the cDNA subtraction experiment and the time course experiment. In the experiment shown in Fig. 7B
, R1881 caused a 2-fold increase in AIbZIP mRNA levels at 24 h. This effect was completely blocked by actinomycin D (1 µg/ml), cycloheximide (10 µg/ml), and Casodex (3 x 10-6 M), which suggests that androgens regulate AIbZIP expression indirectly.
The dose of R1881 that we used to isolate androgen-responsive transcripts by cDNA subtraction (0.1 nM) reproducibly caused a 2–3-fold increase in AIbZIP mRNA levels in several independent experiments (Figs. 6A
, and 7, A and B
). To determine whether higher doses of R1881 could produce a correspondingly larger increase in AIbZIP mRNA levels, LNCaP cells were cultured for 24 h in the presence of 0.001–10 nM R1881. As shown in Fig. 7C
, 1 nM of R1881 and 10 nM of R1881 caused 5- and 6-fold increases in AIbZIP mRNA levels, respectively, compared with a 2.3-fold increase in cells exposed to 0.1 nM of R1881.
We performed two additional experiments to examine the effect of R1881 on AIbZIP protein levels in LNCaP cells. The cells were grown under the same culture conditions that were used for the RNA experiments presented in Fig. 7C
. Protein extracts were prepared from cells that were exposed to increasing doses of R1881 for 24 h, and AIbZIP protein levels were determined using a rabbit polyclonal antibody raised against a COOH-terminal peptide of AIbZIP (aa 379–392, KPRPSGRIRSVLHA). This polyclonal antibody is specific for AIbZIP, because it recognizes a protein of the expected size (Mr 43,000) in BT-474 cells, which express high levels of AIbZIP mRNA, whereas no protein was detected in MDA-MB-231 cells that express low levels of AIbZIP mRNA (see Fig. 6C
). Preimmune serum did not detect AIbZIP (data not shown). R1881 elicited nearly identical dose-related increases in AIbZIP protein levels in two independent experiments (Fig. 7D)
. A 2.9-fold (average of two experiments) increase in AIbZIP protein levels was observed at the dose of 1 nM of R1881.
Modulation of AIbZIP Expression by Antiandrogens.
As shown in Fig. 8A
, 3 x 10-6 M Casodex completely reversed up-regulation of AIbZIP mRNA levels induced by doses of R1881 ranging from 0.001 to 0.1 nM. On the other hand, the same dose of Casodex did not completely block the effect of 10 nM of R1881. These results provide additional confirmation that R1881 regulation of AIbZIP is mediated by the AR. It is of interest to note that the AR expressed in LNCaP cells contains a mutation in its hormone-binding domain that alters the ligand-binding specificity of the AR and which imparts "androgen-like" activity to antiandrogens such as hydroxyflutamide (12)
. Because such aberrant behavior can result from specific ligand-induced alterations in receptor conformation, it is of interest to identify genes that may be differentially regulated by androgens and antiandrogens to gain some insight into AR structure-function relationships, AR-coactivator interactions, and target genes. Therefore, LNCaP cells were cultured in the presence of increasing concentrations of either R1881 (0.001–10 nM), hydroxyflutamide (0.001–500 nM), or Casodex (0.001–500 nM) for 48 h. The cells were harvested after 48 h, and RNA was isolated and processed for determination of AIbZIP mRNA levels by RNase protection assay. As expected, R1881 up-regulated AIbZIP mRNA levels, whereas Casodex did not (Fig. 8B)
. Surprisingly, hydroxyflutamide, which up-regulated PSA mRNA levels under these conditions (data not shown), did not markedly increase AIbZIP mRNA levels.
AIbZIP mRNA Expression in Human Prostate.
To evaluate AIbZIP mRNA localization and expression in the prostate we performed in situ hybridization. AIbZIP mRNA was not detected in noncancerous human prostate (Fig. 9, A and B)
. In fact, in normal prostate, the hybridization pattern obtained with the AIbZIP antisense probe was similar to that obtained with the control sense probe. In both cases, scattered silver grains were equally distributed throughout the epithelium, stroma, and lumen. The number of silver grains overlying epithelial cells that were generated by the antisense probe was similar to that obtained with the control sense probe (75 versus 80 in the specimen presented in Fig. 9
). In contrast, hybridization of the antisense probe to sections of prostate adenocarcinoma revealed a more intense and epithelial-specific labeling (Fig. 9C)
. The number of silver grains overlying epithelial cells obtained with the antisense probe was approximately twice that obtained with the control sense probe (574 versus 236 in the specimen presented in Fig. 9
). The results presented in Fig. 9
are representative of experiments performed on 10 sections of normal prostate and 11 sections of cancerous prostate. These results suggested that AIbZIP could be expressed at higher levels in prostate cancer cells compared with normal prostate.
|
, this polyclonal antibody is specific for AIbZIP, because it produced a strong reaction in BT-474 cells, which express high levels of AIbZIP protein, whereas no reaction was detected in MDA-MB-231 cells that do not express AIbZIP protein (see Fig. 7D
). As expected, the preimmune serum did not produce a detectable reaction (Fig. 10B)
.
|
or undetectable (Fig. 11C)
in noncancerous prostate tissue. In contrast, AIbZIP protein levels were appreciably higher in prostatic carcinoma cells compared with the adjacent noncancerous tissue (Fig. 11A
, and compare Fig. 11C
to Fig. 11D
). The immunostaining reaction was primarily localized in the cytoplasm of luminal epithelial cells. In noncancerous cells, the reaction was detected in structures that could correspond to the Golgi apparatus (see inset in Fig. 11A
). In cancerous cells, a more intense reaction was detected in the cytoplasm, especially in the apical portions of these cells. A total of 20 noncancerous TURP specimens and 15 adenocarcinoma-containing needle biopsy specimens were examined. The noncancerous cells present in these specimens exhibited low or undetectable expression of AIbZIP similar to that presented in Fig. 11, A and C
. All 15 of the tumor-containing specimens exhibited an expression pattern similar to that presented in Fig. 11, A and D
.
|
|
DISCUSSION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Several characteristics of AIbZIP are unique. First, the structure of AIbZIP differs from that of most known ATF/CREB proteins. The sequence of the bZIP domain of AIbZIP is most closely related to those of CREB-H (32) and CREB3 (33) , two recently identified and relatively understudied members of the ATF/CREB family. The similarity among AIbZIP, CREB-H, and CREB3 also extends to regions outside the bZIP domain. In fact, a dendogram analysis performed using the amino acid sequences of AIbZIP, CREB-H, CREB3, and those of representative ATF/CREB proteins revealed that AIbZIP, CREB-H, and CREB3 constitute a distinct subgroup of ATF/CREB proteins.
AIbZIP contains an NH2-terminal transcriptional activation domain. Experiments performed using GAL4-AIbZIP fusion proteins revealed that the first 71 amino acids of AIbZIP are sufficient to activate transcription of a GAL4-responsive promoter. This portion of AIbZIP is relatively rich (14 of 71 amino acids; 20%) in acidic amino acids such as aspartic acid and glutamic acid. Similarly, the activation domains of CREB-H and CREB3 have been mapped to the first 141 and 38 amino acids, respectively (32 , 33) . Interestingly, the NH2-terminal domain of AIbZIP was more transcriptionally active than the full-length AIbZIP protein. These results suggest that the activation domain of AIbZIP could be physically masked or regulated in some manner by the COOH-terminal region of the protein. Omori et al. (32) also observed that the COOH-terminus of CREB-H inhibited the transcriptional activity of its NH2-terminal activation domain.
Immunohistochemistry and GFP fusion protein localization experiments revealed that AIbZIP is primarily a cytoplasmic protein. The protein appears to be localized in structures related to the secretory pathway such as the Golgi apparatus. Whereas this finding will need to be confirmed using more precise localization techniques such as electron microscopy, this characteristic of AIbZIP is particularly interesting considering that the closely related CREB-H and CREB3 proteins, as well as ATF6, are associated with the endoplasmic reticulum. Like AIbZIP, all three of the proteins contain COOH-terminal transmembrane domains, which, when deleted, cause the recombinant proteins to localize to the nucleus (32 , 40 , 41) . Despite their highly homologous bZIP and transmembrane domains, a few structural features distinguish AIbZIP from CREB-H. First, CREB-H contains a second COOH-terminal leucine zipper (aa 411–432) anchored by four leucine residues that is not present in AIbZIP. Second, CREB-H contains a COOH-terminal KDEL-like sequence (GDEL; aa 458–461) that can function as an endoplasmic reticulum retrieval sequence (42) . A similar motif is not present in AIbZIP.
Given the fact that AIbZIP possesses the functional domains and activity of a transcription factor, it is highly possible that AIbZIP translocates to the nucleus to regulate gene expression in response to specific though as yet undetermined stimuli. In support of this hypothesis, during the endoplasmic reticulum stress response (unfolded protein response), ATF6 translocates from the endoplasmic reticulum to the nucleus where it activates genes that encode molecular chaperones (43, 44, 45) . Haze et al. (40) found that this process involves cleavage of ATF6 from a mature Mr 90,000 polypeptide to a nuclear-localized Mr 50,000 form of ATF6 that includes the bZIP and activation domains. Thus, it is conceivable that AIbZIP could exist in a transcriptionally inactive form in cytoplasmic compartments and that removal of the COOH-terminal "anchor" would activate AIbZIP.
Another unique feature of AIbZIP is that androgens induce its expression in LNCaP cells. To our knowledge, AIbZIP is the first example of a human CREB/ATF protein of which the expression is regulated by androgen. Most CREB/ATF proteins are regulated post-translationally, primarily through phosphorylation (38) or, as demonstrated recently for ATF6, by proteolysis (40) . Contrary to CREB, AIbZIP does not contain protein kinase A phosphorylation sites nor does it contain a kinase-inducible domain (38) . CREB/ATF proteins may also mediate the effects of androgens in other androgen-sensitive tissues. Kim et al. (46) reported recently that CREB mRNA levels increase in the submandibular glands of rats treated with testosterone. Induction of bZIP proteins by steroid hormones has also been observed in mammary epithelial cells. hXBP-1, a bZIP transcription factor more closely related to jun proteins, was found to be up-regulated by estrogen in MCF-7 human breast cancer cells (47) . Thus, bZIP transcription factors may play an important role in steroid hormone-dependent cellular events in hormone-sensitive tissues.
One particularly interesting characteristic of AIbZIP is its highly restricted expression profile. On the basis of Northern blot analyses, full-length AIbZIP mRNA is only expressed in the human prostate as well as in breast and prostate cancer cell lines. Indeed, the shorter transcript (0.7 kb) expressed in the colon is unlikely to encode full-length AIbZIP. This highly tissue-specific expression profile additionally supports the hypothesis that AIbZIP may mediate steroid-responsive events in these tissues. However, AIbZIP expression is not exclusively responsive to androgen in LNCaP cells, because AIbZIP mRNA was detected in the absence of androgenic stimulation. In preliminary experiments conducted to date, AIbZIP expression was not modulated by steroid hormones in human breast cancer cell lines (data not shown). The highly tissue-specific expression pattern of AIbZIP is reminiscent of that of CREB-H, which is expressed exclusively in the liver (32) .
Probably the most interesting observation is that AIbZIP appears to be expressed at higher levels in prostate cancer cells compared with noncancerous prostate cells. The immunostaining data presented here clearly show that AIbZIP protein levels are low or undetectable in noncancerous prostate cells compared with adjacent cancerous cells in which AIbZIP is uniformly and highly expressed. These findings suggest that AIbZIP may contribute to prostate cancer or that AIbZIP may serve as a good immunohistochemical marker for prostate cancer cells. However, to confirm these hypotheses we will need to quantitate AIbZIP expression in a larger number of tumor specimens that will be collected and processed under standardized conditions. In these future studies it will be of interest to determine whether AIbZIP expression correlates with tumor grade, AR expression, and other clinically relevant end points.
The AIbZIP gene is localized on a segment of chromosome 1 that is amplified in prostate cancer (48 , 49) . Therefore, it would be useful to determine whether AIbZIP overexpression coincides with 1q amplification. In addition, the AIbZIP gene lies in close proximity to the JTB/PAR gene. The fact that these genes are arranged head-to-head, that they share one exon, and that androgens stimulate AIbZIP expression but down-regulate JTB/PAR expression, suggests that regulation of JTB/PAR and AIbZIP expression by androgens could possibly involve common regulatory elements or, alternatively, some form of transcriptional interference. Moreover, the AIbZIP and JTB/PAR genes are located within the fragment of chromosome 1 that harbors the epidermal differentiation complex of genes (50) . Additional studies will be needed to understand the transcriptional regulation of this cluster of androgen-responsive genes.
In summary, we have identified a new androgen-induced transcription factor that is a novel member of the CREB/ATF family of bZIP transcription factors. Additional study of the role of androgen-responsive transcription factors such as AIbZIP should contribute significantly to our understanding of androgen action in the prostate and prostate cancer.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
1 Supported by a fellowship from Le Fonds de Recherche en Santé du Québec (to C. L.). Financial support for this work was provided by Endorecherche. 
2 To whom requests for reprints should be addressed, at Molecular Endocrinology and Oncology Research Center, Centre Hospitalier Université Laval Research Center, Quebec G1V 4G2, Canada. Phone: (418) 654-2296; Fax: (418) 654-2761; E-mail: Claude.Labrie{at}crchul.ulaval.ca
3 The abbreviations used are: AR, androgen receptor; AIbZIP, Androgen-Induced bZIP; CREB, cAMP-responsive element binding protein; ATF, activating transcription factor; TURP, transurethral resection of the prostate; PSA, prostate-specific antigen; FISH, fluorescence in situ hybridization; BAC, bacterial artificial chromosome; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; UTR, untranslated region; EGFP, enhanced green fluorescent protein; aa, amino acid; JTB, jumping translocation breakpoint; PAR, prostate androgen-regulated.
4 Internet address: http://www.ncbi.nlm.nih.gov/.
5 Internet address: http://psort.ims.u-tokyo.ac.jp.
Received 7/ 9/01. Accepted 12/ 3/01.
|
REFERENCES |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, whereas the 5 exons of the JTB/PAR gene are shown as 
. The arrows indicate the direction in which the genes are transcribed, and the 
indicates the shared portion of a common exon. B, schematic representation of the full-length AIbZIP cDNA and of two other cDNAs isolated in LNCaP cells. The exons are numbered from 1 to 10, and the coding region of each cDNA is shown in gray. The form 3 cDNA contains intron 5 (Int5).
