Biochemical Journal

Research article

Transcriptional regulation of the TGF1 promoter by androgen receptor

Wei Qi, Shen Gao, Zhengxin Wang

Abstract

TGF (transforming growth factor)-β1 is a multifunctional cytokine that influences homoeostatic processes of various tissues. TGF-β1 expression is inhibited by androgens in the prostate gland, whereas its expression is enhanced by androgens in highly metastatic prostate cancer cells. Here, we examined regulation of human TGF1 promoter activity by androgen in prostate cancer cells. The full-length (−3363 to +110) promoter showed a high level of activity in response to androgen in PC3mm2 cells expressing AR (androgen receptor). Further deletion analysis revealed three distal and three proximal AREs (androgen-response elements) in the promoter. Gel-shift and footprint assays show that these AREs physically interact with the DNA-binding domain of AR. Chromatin immunoprecipitation assays revealed the androgen-dependent recruitment of AR to the ARE-containing regions of the TGF1 gene. More importantly, a negative ARE was detected in the TGF1 promoter. Both positive and negative AREs are functional in the androgen-regulated transcription of the TGF1 promoter. These findings imply that androgen signalling may positively or negatively regulate TGF-β1 expression in response to various signals or under different environmental conditions.

  • androgen receptor
  • androgen response element
  • promoter
  • transforming growth factor β1 (TGF-β1)
  • transcriptional regulation

INTRODUCTION

The AR (androgen receptor) controls physiological processes through activation and repression of specific target genes. The ligand-activated AR binds to the ARE (androgen-response element) located within target genes [13]. A few functional AREs have been characterized and studied in detail, including those in the human PSA (prostate-specific antigen) gene [46], the rat C3(1) gene (which encodes the prostatic binding protein) [7], the probasin genes [810], the mouse Slp (sex limited protein) gene [11,12] and the prostate stem cell antigen gene [13]. The androgen-induced repression of gene expression was found to be mediated by negative AREs. Regulation via negative AREs has been reported in the Maspin [14], prostate-specific membrane antigen [15,16] and PSA [17] genes.

TGF (transforming growth factor)-β1 inhibits the proliferation of prostate epithelial cells in the presence of physiological levels of androgens in vitro and in vivo [1820]. Prostate tumours have been found to express higher levels of TGF-β1 than the normal prostate gland [21], and increasing levels of this growth factor are associated with cancer progression [22]. In line with this, serum levels of TGF-β1 correlate with tumour burden, metastasis, and serum PSA in prostate cancer patients [22,23]. TGF-β1 has been shown to promote prostate cancer growth by inhibiting immune responses against the tumour and stimulating angiogenesis and metastasis [23]. These observations seem to contradict indications that TGF-β1 is a negative regulator of prostate epithelial growth [1820]. Further study revealed, however, that prostate cancer cells are less sensitive to TGF-β1 growth inhibition than normal prostatic epithelial cells because of the loss of TGF-β1 receptor expression in cancer cells [2427]. TGF-β1 expression was suppressed by androgens in the prostate gland [28,29], whereas it was activated by androgens in hepatocellular carcinoma and prostate cancer cells [30]. A putative androgen-response sequence half-site was identified in the rat TGF1 promoter [30], but the mechanism for the opposing responses of the TGF1 gene to androgens in normal prostate and prostate cancer has not been defined.

The objective of the present study was to investigate how androgens regulate TGF1 expression in prostate cancer cells and to identify functional AREs in the TGF1 promoter. Using a series of TGF1 promoter deletion constructs, we identified six positive AREs and one negative ARE in the TGF1 promoter. We also present in vivo and in vitro evidence that androgens can directly regulate TGF-β1 expression through these AREs. This finding suggests that activation or repression of TGF-β1 expression by androgens is mediated through positive or negative AREs in its promoter.

EXPERIMENTAL PROCEDURES

Construction of TGF1 promoter reporter plasmids

A BAC (bacterial artificial chromosome) clone (RP11-662N17) that contains the human TGF1 locus was purchased from CHORI Bacpac Resources (Oakland, CA, U.S.A.). PCR was performed using the BAC DNA as a template to amplify human TGF1 promoter sequences (−3363 to +110). The PCR product was subcloned into the pGL3 luciferase reporter construct (Promega). Luciferase reporter plasmids containing various deletion mutants of the TGF1 promoter were constructed similarly. Point mutants of TGF1 AREs were generated using the QuikChange® Site-Directed Mutagenesis kit (Stratagene). All constructs were confirmed by mapping restriction enzyme sites and by DNA sequencing analysis.

Cell cultures

LNCaP, LNCaP-LN2, PC3mm2 and HT human prostate carcinoma cells were cultured in RPMI 1640 medium (Cellgro; Mediatech) with 5% (v/v) fetal bovine serum (HyClone). Phenol Red-free RPMI 1640 medium (Invitrogen) supplemented with 5% (v/v) serum (treated with dextran-coated charcoal) was used when the androgen was applied. R1881, a synthetic androgen, was purchased from New England Nuclear and dissolved in ethanol.

Transient transfection and luciferase assay

All transfection assays were performed according to the manufacturer's protocol for Lipofectamine™ or Lipofectamine™ 2000 (Invitrogen). Briefly, PC3mm2 cells (3×104) were seeded in 24-well plates 24 h before transfection. The total amounts of DNA for each transfection reaction were adjusted to 70 fmoles by addition of the pcDNA3.1 plasmid. The pRL-CMV (2.5 ng) vector that expresses the Renilla luciferase was included in each transfection reaction. The transfected cells were incubated in the absence or presence of 10 nM androgen (R1881) for 36–48 h and then harvested for the Dual-Luciferase Reporter Assay (Promega). In brief, the cells were washed with ice-cold PBS twice, and subjected to lysis in passive lysis buffer (Promega) for 15 min at room temperature (23 °C) with constant shaking. The lysates (10 μl) were used for the luciferase activity assay. The relative luciferase activity was eqivalent to the firefly luciferase activity/the Renilla luciferase activity. All transfection experiments and luciferase assays were carried out in triplicate.

ChIP (chromatin immunoprecipitation) assay

LNCaP cells were grown in the absence or presence of 10 nM androgen (R1881), cultured cells were fixed with 1% formaldehyde, and the resulting cross-linked chromatin was sheared by sonication to approx. 1 kb (Figure 1C) or 400 bp (Figure 5) fragments. These short chromatin fragments were immunoprecipitated by antigen affinity-purified AR-specific antibodies; the immunoprecipitated chromatin contained the enriched AR DNA-binding sites. The protein–DNA cross-links were reversed. Immunopurified DNA (2.5 μl) was used for the PCR reaction (30 cycles, annealing at 50 °C and extension at 72 °C for 20 s).

Figure 1 AR directly regulates transcription of the TGF1 gene

(A) Androgens repressed TGF1 gene expression in the mouse prostate gland. C57BL male mice were castrated and injected with androgen or ethanol (control). After 48 h, RNAs were isolated from the prostate, and expression of TGF1 and NKX3.1 was detected using real-time RT–PCR. The fold change was the relative quantification of TGF-β1 or NKX3.1 compared with β-actin. The values represent means±S.D. (n=3). (B) Androgen enhanced TGF1 gene expression in LNCaP-LN2 cells. LNCaP or LNCaP-LN2 cells were grown in the absence or presence of 10 nM R1881 for 48 h. Cells were then harvested for analysis of gene expression by real-time RT–PCR. The fold change was the relative quantification in the presence compared with the absence of androgen. Values represent means±S.D. (n=3). (C) ChIP assay identified the androgen-dependent interaction between the TGF1 gene and AR in vivo. LNCaP cells were grown in the absence (lanes 2, 4 and 6) or presence (lanes 1, 3 and 5) of 10 nM R1881. Immunoprecipitation was performed with antigen-purified anti-AR antibodies. The cross-links in the immunoprecipitated protein–DNA complexes were reversed, and the resulting DNA was amplified by a PCR reaction with three specific primers derived from promoter regions of β-actin (lanes 1 and 2), PSA (lanes 3 and 4) or TGF1 (lanes 5 and 6). The same set of PCR reactions was performed with chromatin DNA used for immunoprecipitation (top panel). For TGF1, the primer pair GGATTAAGCCTTCTCCGCCTGG with CTGGGCCACCGTCCTCATCTCG amplifies a 212 bp product from −629 to −417 upstream of the TGF1 gene. For PSA, the forward primer sequence was TCTGCCTTTGTCCGCTAGAT and the reverse primer sequence was AACCTTCATTCCCCAGGACT, which amplifies a 211 bp product from −250 to −39 upstream of the PSA gene. For β-actin, the forward primer sequence was TCCTCCTCTTCCTCAATCTCG and the reverse primer sequence was AAGGCAACTTTCGGAACGG, which amplifies a 145 bp product from −1118 to −974 of the β-actin gene (the A of the ATG translation start codon was arbitrarily given the number +1). (D) Activation of the TGF1 promoter by AR. PC3mm2 cells were transfected with 100 ng of the reporter plasmids containing various truncations of the TGF1 promoter, with or without 150 ng of pcDNA-AR. Transfected cells were grown in the absence or presence of 10 nM R1881 for 48 h. Cells were then harvested for luciferase assay. Values represent means±S.D. (n=3).

Gel-shift assay

The electrophoretic mobility shift assay was performed as described previously [31]. Oligos derived from TGF1 AREs were synthesized, annealed and labelled with [γ-32P]ATP by T4 polynucleotide kinase. In the gel-shift assays, each 20 ml reaction contained 20 mM Hepes, pH 7.9, 70 mM KCl, 1 μg of poly(dI-dC), 1 mM dithiothreitol, 0.1% NP-40 (Nonidet P40), 0.1 mg/ml BSA (bovine serum albumin) and various ARDBD (AR DNA-binding domain) proteins. The reaction mixture was incubated for 20 min at room temperature; the binding reaction was initiated by addition of the labelled probes (10000 c.p.m.) and the mixture was incubated for an additional 30 min at room temperature. The reaction mixture was loaded directly on to a 4% (37.5:1, acrylamide:bisacrylamide) non-denaturing polyacrylamide gel with 0.25×TBE [Tris/borate/EDTA (1×TBE =45 mM Tris/borate and 1 mM EDTA)] buffer and run at 150 V for 2 h at room temperature.

DNase I footprinting

DNA templates used for footprinting on the transcribed and non-transcribed strands of the TGF1 promoter were prepared by isolating the Pst1-XhoI fragments (−2466 to −2283 and −943 to −694) from pGL3-TGF-β1 and end-labelling with [32P] by a fill-in reaction with the Klenow enzyme. A 30 μl reaction contained 2 fmol of the 32P-labelled DNA fragment, 0.5 ng of poly[dI-dC], 5 μg of BSA, 10 mM Hepes, pH 7.9, 60 mM KCl, 0.2 mM EDTA, 4 mM MgCl2, 10% glycerol and indicated amounts of purified ARDBD. The reaction mixture was incubated for 20 min at room temperature, and then 50 μl of Ca/Mg solution (5 mM CaCl2 and 10 mM MgCl2) was added to each tube. DNase I (1 μl, 0.4 ng/ml) was added to each tube and digestion was continued at room temperature for 2 min. The reaction was terminated by adding 100 μl of stop solution (0.2 M NaCl, 30 mM EDTA, 1% sodium dodecyl sulfate and 0.1 mg/ml tRNA) and chilled on ice. After phenol/chloroform extraction and ethanol precipitation, the DNA was resuspended in 10 μl of gel loading buffer (90% formamide, 0.02% Bromophenol Blue and xylene, and 1×TBE) and loaded on an 8% polyacrylamide-urea sequencing gel. The gel was dried under vacuum and exposed to an X-ray film at −80 °C.

RNA isolation and real-time RT–PCR (reverse transcription–PCR)

Total RNAs were isolated from cultured cells using the TRIzol reagent and reverse-transcribed using the Reaction Ready First Strand cDNA Synthesis Kit (SuperArray Bioscience). The cDNA products were PCR-amplified (40 cycles of 30 s at 94 °C; 20 s at 55 °C; 30 s at 72 °C) with the RT2 real-time SYBR green PCR master mix and the gene-specific primer sets for human Nkx3.1 (PPH02267A), TGF-β1 (PPH00508A) and β-actin (PPH00073A) genes (SuperArray Bioscience) by SmartCycler II (Cepheid). Raw data processing and quantification were performed with the SmartCycler software (version 2.0C). The 2−ΔΔCT method was used to determine the relative quantification (in the presence of ligand compared with in the absence of ligand) of target gene expression [32].

Castration and androgen administration

C57BL mice (n=10) were castrated at 8 weeks of age under anaesthesia in accordance with the protocol approved by the International Animal Care and Use Committee. To castrate mice, mice were anaesthetized by intraperitoneal injection of Nembutal (45 mg/kg weight). Hairs over the surgical region were cut and the region was disinfected with 10% iodine and then with 70% ethanol. A 0.5 cm incision was made and surgical string was used to perform the ligation. Testis were then removed and the wound was closed with wound clips. The would clips were removed 5 days later. Seven days after the operation, the mice were subcutaneously injected with testosterone enanthate (3.6 ng/g body weight). The prostate glands were dissected at 0 (n=5) and 48 (n=5) h after testosterone replacement. Total RNA was isolated from prostate glands using the Trizol method and subjected to real-time RT–PCR analysis.

FLP (flipase)-mediated integration of the TGF1 promoter

The Wt (wild-type) and Mt (mutant) TGF1 promoters (−3363 to +110) were subcloned into pCM-luc. PCM-TGF-β1-luc and pcDNA5/FRT were co-transfected into HT1080 cells (HT55), which contain an FRT (FLP-recombinase target) site. The transfected cells were selected with zeocin for 2 weeks and the resistant colonies were picked up and expanded. Whole-cell lysates were prepared from cell lines and submitted for luciferase assay. HT55-TGF-β1 cells (5×104) were seeded in 24-well plates and transfected with 100 ng of the AR-expressing construct (pcDNA-AR). The transfected cells were incubated in the absence or presence of 10 nM R1881 for 36–48 h and assayed for luciferase activity.

RESULTS

AR directly targets the TGF1 gene

Expression of the TGF1 gene is induced by castration [29], suggesting that androgens suppress its expression. To investigate regulation of the TGF1 gene by androgens, we administered testosterone enanthate to castrated C57BL mice. Total RNAs were isolated from mouse prostate glands 48 h after androgen replacement. Changes in TGF1 gene expression induced by the androgen were then analysed by real-time RT–PCR. Expression of the TGF1 gene was inhibited by androgen replacement in the prostate (Figure 1A). In contrast, androgen replacement dramatically induced expression of the NKX3.1 gene, whose expression is positively regulated by the androgen [33]. Similarly, TGF1 expression was inhibited slightly by the androgen in the LNCaP prostate cancer cell line (Figure 1B). However, the androgen enhanced expression of the TGF1 gene in a variant of LNCaP (LNCaP-LN2) that was selected for increased metastatic potential by orthotopic “recycling” in mice [34]. The cells that were isolated by this procedure displayed not only increased metastasis to regional lymph nodes but also androgen-independent growth in vitro and in vivo [34,35]. In contrast, the androgen-induced expression of the NKX3.1 gene in LNCaP and LNCaP-LN2 did not change (Figure 1B).

In the ChIP assay, AR bound to the PSA and TGF1 promoter regions in the presence of androgen (Figure 1C, compare lane 3 with lane 4 and lane 5 with lane 6). The products amplified by PCR at the same time as the β-actin promoter, which served as a negative control, were not changed in response to the addition of androgen (Figure 1C, lanes 1 and 2). As AR directly targets the PSA promoter [36], these results indicate that AR also directly targets the TGF1 promoter.

TGF1 promoter contains both positive and negative AREs

To map AREs in the TGF1 gene, a genomic DNA fragment (3473 bp) that spans the 5′-flanking region of the human TGF1 gene was isolated from the BAC clone and was cloned into the pGL3 basic vector upstream of the luciferase gene. Transient transfection of this reporter plasmid into PC3mm2 cells resulted in an approx. 5-fold higher level of luciferase activity in the presence of androgen than in the absence of androgen (Figure 1D). The relative luciferase activities of the empty pGL3 luciferase reporter were not significantly different between cells cultured in the absence of the androgen and those cultured in the presence of the androgen. These results suggest that the cloned TGF1 genomic fragment contained functional AREs, which we then further mapped using a collection of deletions (Figure 1D). The deletion of nucleotides to −2112 or −999 decreased the response of the luciferase reporter to the androgen. Further deletion of nucleotides to −784 resulted in a slightly negative response to androgen. The deletion of nucleotides to −684, −575 or −394 resulted in positive responses to androgen. Further analysis with additional deletions indicated that there are positive AREs located within the regions from −2466 to −2283 and from −341 to +1, and a negative ARE within the region from −943 to −694 of the TGF1 promoter.

Indeed, the positive AR response was observed when the regions of −2466 to −2283 and −341 to +110 were subcloned into luciferase reporter constructs (Figures 2A and 2B). Further deletion analysis and alignments with the ARE consensus sequence [37] revealed three putative AREs in the proximal region and three within the distal region (Figure 2D). Mutation on any single putative ARE (Mt1, Mt2 or Mt3) in the distal region (−2466 to −2283) decreased the response to androgen (Figure 2A), whereas mutations on all three putative distal AREs (Mt1,2,3) completely abolished the response to the androgen. Mutation on the ARE1 (Mt1) or ARE2 (Mt2) in the proximal region (−341 to +110) decreased the response to androgen (Figure 2B), but mutation on the proximal ARE3 (Mt3) completely abolished the response to androgen. These results suggest that these putative AREs are functional.

Figure 2 The TGF1 promoter contains both positive and negative AREs

(A) TGF1 contains three positive AREs in the distal region. PC3mm2 cells were transfected with 150 ng of pcDNA-AR and 100 ng of the reporter plasmid containing the Wt or Mt (Mt1, Mt2, Mt3, and Mt1,2,3) TGF1 promoters (−2466 to −2283) and grown in the absence or presence of 10 nM R1881 for 36–48 h. Cells were then harvested for luciferase assay. Values represent means±S.D. (n=3). (B) TGF1 contains three positive AREs in the proximal region. PC3mm2 cells were transfected with 150 ng of pcDNA-AR and 100 ng of the reporter plasmid containing the Wt or Mt (Mt1, Mt2, Mt3, Mt1,3) TGF1 promoters (−341 to +110) and grown in the absence or presence of 10 nM R1881 for 36–48 h. Cells were then harvested for luciferase assay. Values represent means±S.D. (n=3). (C) The TGF1 promoter contains a negative ARE. PC3mm2 cells were transfected with 150 ng of pcDNA-AR and 100 ng of the reporter plasmid with or without insertion of the TGF1 promoter in both orientations (−943/−694 and −943/−694R) or scramble control (SC) DNA sequence (243 bp) derived from the human SPDEF (GeneID: 25803) promoter (−3018 to −2775) and grown in the absence or presence of 10 nM R1881 for 36–48 h. Cells were then harvested for luciferase assay. Values represent means±S.D. (n=3). (D) The nucleotide sequences of the AREs in the TGF1 promoter and the alignments of the AREs with the AR-binding consensus sequence are shown. The point mutations used in transient transfection and gel-shift assays are underlined.

To assay the negative ARE of the TGF1 promoter, we constructed a luciferase reporter (pGL3-4×ARE-E4-Luc) (Figure 2C), which includes four tandem AREs derived from the PSA gene upstream E4 core promoter with a pGL3-basic backbone. When the pGL3-4×ARE-E4-luc reporter was co-transfected with expression plasmid for AR into PC3mm2 cells, AR activated the reporter approx. 36-fold in the presence of the androgen (Figure 2C). When the TGF1 promoter region from −943 to −694 was inserted into the luciferase reporter between 4×ARE and E4 in both orientations, the androgen-driven activation of the luciferase activity was almost completely abolished, indicating that there is a negative ARE in this region. As a negative control, a 243 bp DNA fragment (SC) derived from the human SPDEF (GeneID: 25803) promoter (−3018 to −2775) inserted into the luciferase reporter between 4×ARE and E4 did not alter the androgen-driven transcription.

ARDBD directly interacts with the TGF1 AREs

To test whether ARDBD would interact with the AREs in the TGF1 promoter, we performed a DNase I footprinting assay. Strong footprints were observed over the region −2343 to −2311 (containing distal ARE1) (Figure 3A, lanes 5 and 6) and the region −2421 to −2341 (containing distal ARE3) (lane 11). Essentially no protection was seen over the distal ARE2-containing region (−2376 to −2354) (Figure 3A, lanes 3–6). Consistent with these observations, ARDBD bound strongly to the distal ARE1 and weakly to the distal ARE2 and ARE3 in the gel-shift assay (Figure 3B, lanes 4, 8 and 12). Mutations on the conserved nucleotides in the distal AREs (Figure 2D) abolished the ARDBD–ARE interactions (Figure 3B, lanes 2, 6 and 10). Similarly, ARDBD interacted with the proximal ARE1 and ARE3 in the gel-shift assay (Figure 3C, lanes 2 and 10), but no ARDBD–ARE2 interaction was detected (lanes 6). Thus, three distal AREs and two proximal AREs in the TGF-β1 promoter directly interact with ARDBD.

Figure 3 ARDBD directly interacts with AREs in the TGF1 promoter

(A) DNase I footprinting of ARDBD on the TGF1 promoter. The DNA templates spanning −2466 to −2283 (lanes 1–11) and −943 to −694 (lane 12–16) of the TGF1 promoter were prepared as described in the Experimental procedures section. No protein (lanes 2, 8 and 14) or increasing amounts of the recombinant ARDBD (5, 10 or 20 ng for lanes 3–5 or 9–11; 20 ng for lane 15; 40 ng for lanes 6 and 16) were included in the binding reaction. The protected regions are marked. Sequencing reactions are based on Maxim and Gilbert chemistry to prepare the G and G/A footprinting markers (lanes 1, 7, 12 and 13). (B) Binding of TGF1 distal AREs to the ARDBD. The different 32P-labelled ARE sites in the TGF1 promoter were incubated with 0 (lanes 1, 3, 5, 7, 9 and 11) or 20 (lanes 2, 4, 6, 8, 10 and 12) ng of the recombinant ARDBD. (C) Binding of TGF1 proximal AREs to the ARDBD. The different 32P-labelled ARE sites in the TGF1 promoter were incubated with 0 (lanes 1, 3, 5, 7, 9, 11, 16 and 19) or 20 (lanes 2, 4, 6, 8, 10 and 12) ng of the recombinant ARDBD.

Similar analyses were performed to investigate whether ARDBD would interact with the negative AREs in the TGF-β1 promoter. No protection was observed over the negative ARE region (−694 to −943) in the footprinting assay (Figure 3A, lanes 12–16). Consistent with this observation, ARDBD failed to bind to the negative ARE in the gel-shift assay (results not shown).

Comparison of the ARE-containing regions in human TGF1 promoter with that in mouse and rat

Comparison of the ARE-containing regions in the human TGF1 promoter with published sequences for mouse (GenBank, L42456) and rat (GenBank, AF249327) TGF1 genes was carried out using MacVector 10.0 (MacVector). Distal ARE3 is highly conserved but distal ARE1 and ARE2 are less conserved among the three species (Figure 4B). All three AREs in the proximal region are highly conserved among the three species (Figure 4C). The negative ARE region is also conserved (Figure 4D, 249 bases, 45% identity). A previous study had identified an androgen-response sequence half site (−1932 to −1927) in the rat TGF1 promoter, which is conserved in the mouse TGF1 promoter [30]. This androgen-response sequence is not localized in the distal androgen-response region identified here.

Figure 4 The human TGF1 promoter contains six positive and one negative ARE

(A) The identified distal (dAREs) and proximal (pAREs) and the negative (nARE) AREs in the human TGF1 promoter. (B) Nucleotide sequence alignment of the regions that contain distal AREs in the human (−2435/−2322), mouse (−2143/−1994) and rat (−3007/−2868) TGF1 promoters. The identified positive AREs in the human TGF-β1 promoter are marked by lines. (C) Nucleotide sequence alignment of the regions that contain proximal AREs in the human (−280/−14), mouse (+31/+282) and rat (−844/−591) TGF1 promoters. The identified positive AREs in the human TGF-β1 promoter are marked by lines. (D) Nucleotide sequence alignment of the regions that contain the negative ARE in the human (−943/−694), mouse (−575/−324) and rat (−1449/−1198) TGF1 promoters.

AR is recruited onto the distal and proximal ARE-containing regions in the TGF1 gene

To investigate whether AR was recruited on to the positive and negative ARE-containing regions in vivo, we performed the ChIP assay with smaller genomic fragments (∼400 bp) (Figure 5A, lanes 1 and 2) and using primer pairs specific for the distal, proximal, and negative ARE-containing regions. AR bound to the distal and proximal ARE-containing regions in the presence of the androgen (Figure 5B, lanes 2 and 3, third panel). In contrast, no androgen-dependent recruitment of AR to the negative ARE-containing region was seen (Figure 5B, lane 1, third panel). As a negative control, the immunoprecipitation was performed at the same time with purified IgG (Sigma–Aldrich) (Figure 5B, fourth and fifth panels). The products amplified by PCR at the same time from the β-actin promoter, which served as another negative control, were not changed in response to the addition of androgen (Figure 5B, lane 4). These results indicate that AR directly interacted with the distal and proximal ARE-containing regions in the TGF1 promoter in the presence of androgen.

Figure 5 ChIP assay reveals the androgen-dependent recruitment of AR to the distal and proximal ARE-containing regions in the TGF1 gene

(A) LNCaP cells were grown in the absence (lane 1) or presence (lane 2) of 10 nM R1881 and fixed with 1% formaldehyde. The resulting cross-linked chromatin was sheared by sonication to approx. 400 bp fragments. The protein–DNA cross-links were reversed and DNA fragments were analysed by agarose gel electrophoresis with ethidium bromide staining. Lane 3 is the DNA marker (Invitrogen). (B) The protein–DNA cross-links obtained from LNCaP cells grown in the absence (second and fourth panels) or presence (third and fifth panels) of 10 nM R1881 were immunoprecipitated with antigen-purified anti-AR antibody (second and third panels) or IgG (Sigma-Aldrich) (fourth and fifth panels). The cross-links in the immunoprecipitated protein–DNA complexes were reversed, and the resulting DNA was amplified by a PCR reaction with three specific primer pairs derived from the negative (lane 1), proximal (lane 2), and distal (lane 3) ARE-containing regions in TGF1. As a negative control, a PCR reaction was performed with a primer pair derived from β-actin (lane 4). The same set of PCR reactions was performed with chromatin DNA used for immunoprecipitation (top panel). For TGF1, the primer pairs CTCATTCTGCTCTAGCCATGC with CTAGCCAGAGGGAACAGCAAGT, CCCTATCTGTAAATTGGGGACA with GGTAGGAGAAGAGGGTCTGTCA, and GGAGCGGAGGAAGGAGTC with CTCTTCTCCCGACCAGCTC amplify 119, 213 and 131 bp products from −2427 to −2308 (distal ARE-containing region), −913 to −702 (negative ARE-containing region), and −159 to −28 (proximal ARE-containing region) upstream of the TGF1 gene respectively.

Positive and negative AREs in the TGF-β1 promoter are functional

Given the fact that multiple positive AREs and a negative ARE are present in the promoter of the TGF1 gene (Figure 4A), it is likely that regulation of TGF-β1 expression in response to androgens is complicated. To investigate the role of these identified AREs in their native promoter environment, we made point mutations on the positive AREs and deleted the negative ARE in the TGF1 promoter (−3363 to +110). Seven TGF1 promoter mutants were generated (Figure 6A). Point mutations on three distal AREs or on two proximal AREs (Figure 2D) and deletion of the negative ARE (−943 to −694) were generated separately or in combinations. The Wt and Mt TGF1 promoters were subcloned into the luciferase reporter (pGL3). These constructs were transiently transfected into PC3mm2 cells with the AR-expressing construct (pcDNA-AR). The transfected cells were grown in the absence or presence of androgen and submitted to the luciferase assay. Mutations on the distal AREs (M1) slightly decreased the response of the TGF1 promoter to the androgen (Figure 6B). In contrast, mutations on the proximal AREs (M2) completely abolished this response. Mutations on both distal and proximal AREs (M3) turned the positive response to the AR to a negative response. On the other hand, deletion of the negative ARE (M4) enhanced the positive response of the TGF1 promoter to the androgen. Further mutation on the distal AREs (M5) slightly decreased the positive response of the TGF1 promoter to the androgen, and mutation on the proximal AREs (M6) or on both distal and proximal AREs (M7) abolished both positive and negative responses of the TGF1 promoter to the androgen. These results indicate that the identified positive AREs and the negative ARE are functional in the TGF1 promoter.

Figure 6 The identified AREs are functional in the TGF1 promoter

(A) Diagram shows the mutations on the TGF1 promoter. (B) Mutations on the AREs affect the response of the TGF1 promoter to AR. PC3mm2 cells were transfected with 150 ng of pcDNA-AR and 100 ng of the reporter plasmid containing the Wt or Mt (M1–M7) TGF1 promoters and grown in the absence or presence of 10 nM R1881 for 36–48 h. Cells were then harvested for luciferase assay. (C) An illustration of the major features of the Flp-In system used to integrate TGF1 promoter into the chromosome of HT cells. (D) Mutations on the AREs affect the response of the integrated TGF1 promoter to AR. HT cells containing the vector, Wt or Mt (M3 and M4) TGF1 promoter were transfected with 100 ng of pcDNA-AR and grown in the absence or presence of 10 nM R1881 for 36–48 h. Cells were then harvested for the luciferase assay. Values represent means±S.D. (n=3).

Positive and negative AREs are functional in the integrated TGF1 promoter

The transfected promoter-reporter assay uses many copies of the construct in a single transfected cell, and the promoters, poorly integrated into the chromatin, may limit the corresponding transcription factors with respect to the exogenous construct. Yan et al. [38,39] developed a novel assay system to study the transcriptional regulation of promoters in vivo in mammalian cells. This system uses a recombinase (Flp) and site-specific recombination to facilitate integration of the promoter-luciferase sequence into a specific site in the genome of mammalian cells. Thus the activity of the integrated promoter closely mirrors expression of the endogenous gene in response to physiological cues.

We obtained the HT1080 cell line (HT55), which contains one FRT site, and integrated the Wt or Mt TGF-β1 promoter (−3363 to +110) into the FRT site in these cells (Figure 6C). As a negative control, the empty vector (pCM-luc), that did not contain the TGF-β1 promoter sequence, was also integrated into the same FRT site in HT55 cells. To determine whether the genome-integrated TGF1 promoter/luciferase reporters were active, the HT55-luc (lacking any TGF-β1 promoter sequence) and HT55-TGF-β1 (Wt, M3 or M4)-luc cell lines were analysed for luciferase expression. Although the luciferase activity in the HT55-luc cell line was barely detectable, HT55-TGF-β1 (Wt, M3 or M4)-luc showed robust (>50-fold) luciferase activity (results not shown), indicating that the integrated TGF1 promoters were active. The high luciferase activity was maintained in these cell lines over 10 passages, which contrasts sharply with that for extrachromosomal reporter plasmids, whose activity is rapidly lost.

We then determined whether the activity of the integrated TGF1 promoters was responsive to the androgen. Trace amounts of AR was detected in the HT cells (results not shown). We transiently transfected pcDNA-AR into these cells. The transfected cells were grown in the presence or absence of androgen for 24 h and the luciferase activity was measured. Variable responses of the integrated TGF1 promoter in the HT55-TGF-β1-luc cell line (Wt) to the androgen were observed. This might be due to the fact that the TGF1 promoter contains both positive and negative AREs and may reflect different cell densities or growth stages used for the assay. The result in Figure 6(D) shows addition of the androgen very slightly enhanced the activity of the integrated TGF1 promoter (Wt). Deletion of the negative ARE (M4), however, resulted in a stronger positive response of the integrated TGF1 promoter to the androgen. Mutation of the positive AREs (M3) turned the androgen response of the integrated TGF1 promoter to negative. These results imply that both positive and negative AREs are functional in the integrated TGF1 promoter.

DISCUSSION

Previous studies suggested that the TGF1 gene was negatively regulated in the prostate gland and positively regulated in prostate cancer cells by androgens. In the present study, we demonstrate that the TGF1 promoter contains multiple positive AREs and a negative ARE. The positive AREs interact with AR in vitro and in vivo, and both positive and negative AREs are functional in the synthetic promoters as well as in the integrated TGF1 promoter. These findings indicate that regulation of TGF-β1 expression in response to the androgen is complicated and could be neutral, positive or negative depending on cell types or cell environments.

Positive AREs in the TGF1 promoter

The following lines of evidence show that TGF-β1 expression could be positively regulated by androgens through the proximal and distal AREs in the TGF1 promoter. First, TGF1 gene expression was enhanced by the androgen in LNCaP-LN2 cells (Figure 1B). Secondly, the cloned full-length TGF1 promoter displayed a positive response to the androgen in PC3mm2 cells (Figure 1D). Thirdly, the distal region (−2466 to −2283) and proximal region (−341 to +110) responded to the androgen independently (Figures 2A and 2B). Fourthly, three distal AREs and two proximal AREs interacted with AR in gel-shift and footprint assays (Figure 3), and mutation on the conserved nucleotides in these AREs abolished the ARE–AR interaction (Figure 3) as well as the response of luciferase reporters to the androgen (Figures 2A and 2B). Most importantly, mutations on these AREs affected the response of the TGF1 promoter to the androgen (Figure 6).

The palindromic androgen-response consensus sequence (RGAACANGNTGTNCT) is involved in AR binding and positive androgen regulation [37]. The proximal ARE3 is more closely related to the consensus ARE and interacts more strongly with AR than the proximal ARE1 and ARE2. Consistent with the interacting data, mutation on the proximal ARE3, but not on proximal ARE1 or ARE2, completely abolished the response of the TGF1 proximal promoter (−341 to +110) to androgen. Mutation on the proximal AREs completely abolished the positive response of the TGF1 promoter, indicating that the proximal AREs play a critical role in the androgen-driven expression of the TGF1 promoter. Although the AR-binding affinity of the three distal AREs was significantly different in DNase I footprint and gel-shift assays, mutation on a single ARE decreased, but did not abolish, the response of the TGF1 promoter to androgen. Mutation on all three distal AREs completely abolished the response of the distal enhancer (−2466 to −2283) to androgen, but not that of the full-length TGF1 promoter. The organization of distal and proximal AREs in the TGF1 promoter is reminiscent of that in the PSA and probasin genes [4].

Negative ARE in the TGF1 promoter

Gene regulation via negative AREs has been described for PSA, Maspin and prostate-specific membrane antigen genes [1417]. Our analysis revealed a negative ARE in the TGF1 promoter. The negative ARE strongly inhibited androgen-driven transcription in the synthetic promoter (Figure 2C), whereas deletion of the negative ARE augmented the positive response of the TGF1 promoter to the androgen (Figures 6B and 6D). Previous studies identified one negative ARE (XBE) in the PSA gene [17] and one in the Mapsin gene [14]. XBE contains the AR-binding site overlapped with the NF-κB (nuclear factor-κB)-binding site. AR competed with NF-κB for a DNA-binding site to negatively regulate PSA gene expression. The mechanism for the negative ARE in the Maspin gene is unknown. We did not find any DNA sequence that was significantly similar to the XBE or the Maspin negative ARE in the negative ARE region of TGF1 and failed to detect ARDBD interacting with the negative ARE. More interesting were the findings that the minimal sequence required for maximal inhibition activity is approx. 250 bp and that further deletion from this region gradually led to loss of its activity (results not shown). These observations suggest that the negative ARE in the TGF1 gene is novel and the repression by androgen is indirect. Further mapping and analysis of this negative ARE is underway.

Expression of the TGF1 gene is negatively regulated by androgen in the normal prostate and in prostate cancer LNCaP cells (Figures 1A and 1B). TGF-β1 expression was enhanced by the androgen, however, in a metastatic variant (LNCaP-LN2) of LNCaP cells. TGF1 promoter activity was enhanced in the highly metastatic prostate cancer (PC3mm2) cells. These observations suggest that there is a gradual loss of negative response to androgens and gain of positive response to androgens of the TGF1 gene during prostate tumourigenesis and prostate cancer progression. Consistent with this possibility is the observation that prostate tumours express higher levels of TGF-β1 than normal prostate tissue [21] and that increasing levels of this growth factor are associated with prostate cancer progression [22]. Identification of positive and negative AREs in the TGF1 gene would provide a basis for understanding this transition and the roles of AR and TGF-β1 in prostate tumourigenesis and progression.

Acknowledgments

We thank Kathryn Hale for critical editorial review, prior to submission, and Dr Douglas Boyd (Department of Cancer Biology, The University of Texas M. D. Anderson Cancer Center, TX, U.S.A.) for the FRT-HT cell line. This work was supported by a grant (1R01 DK065156 01) to Z. W. from the National Institute of Diabetes and Digestive and Kidney Diseases and the National Institutes of Health.

Abbreviations: AR, androgen receptor; ARDBD, AR DNA-binding domain; ARE, androgen-response element; BAC, bacterial artificial chromosome; ChIP, chromatin immunoprecipitation; FLP, flipase; FRT, FLP-recombinase target; Mt, mutant; PSA, prostate-specific antigen; RT–PCR, reverse transcription–PCR; TGF, transforming growth factor; Wt, wild-type

References

View Abstract