Isolation and Characterization of the Promoter of the Human Prostate Cancer-specific DD3 Gene*

Recently, we have described a novel gene, DD3, which is one of the most prostate cancer-specific genes described to date (Bussemakers, M. J. G., van Bokhoven, A., Verhaegh, G. W., Smit, F. P., Karthaus, H. F. M., Schalken, J. A., Debruyne, F. M. J., Ru, N., and Isaacs, W. B. (1999) Cancer Res. 59, 5975–5979). The prostate cancer-specific expression of DD3 indicates that the DD3 gene promoter is a promising tool for the treatment of prostate cancer. To identify the promoter elements that are responsible for the prostate cancer-specific expression of DD3, we have isolated and characterized the DD3 promoter. Sequence analysis of the DD3 5*-flanking region was performed and several promoter-human growth hormone reporter constructs were prepared, which were transiently transfected in the DD3-positive cell line LNCaP and several DD3-negative cell lines. Using a 500-base pair DD3 promoter construct, we could detect promoter activity in LNCaP cells, which was not affected by increasing the size of the constructs. Truncated constructs, however, showed an increased transcriptional activity, suggesting the presence of a silencer that negatively regulates the expression of DD3. DNase-I footprint analysis, using nuclear extracts from LNCaP cells, revealed the presence of three DNase-Iprotected areas within the DD3 proximal promoter. We show that the high mobility group I(Y) protein binds to one of the DNase-I-protected areas and recruits another, yet unidentified, protein to the DD3 promoter in LNCaP cells.

Prostate cancer is the most commonly diagnosed malignancy and the second leading cause of cancer-related deaths in the Western male population (1). When this carcinoma has locally or distantly spread, no curative therapy can be offered. Because there is no effective treatment available for patients with advanced and/or hormone-refractory prostate cancer, there is an urgent need to develop new approaches to treat patients with progressive prostate cancer. A better understanding of the molecular changes associated with the onset and progression of prostate cancer may provide a rational basis for the development of new treatment modalities. For example, gene therapy using prostate-specific gene promoters, i.e. linking up prostatespecific promoter sequences to genes that suppress tumor cell growth, induce apoptosis, and/or kill tumor cells, may provide a new way to attack this mordacious disease (2,3).
A number of human genes have been identified that are specifically expressed in the human prostate, including prostate-specific antigen (PSA) 1 (e.g. Ref. 4), prostatic acid phosphatase (e.g. Ref. 5), human kallikrein 2 (e.g. Ref. 6), prostatespecific membrane antigen (7,8), prostate-specific transglutaminase (9), and prostate stem cell antigen (10). The promoter sequences responsible for the prostate-specific expression of these genes have been cloned, and the unraveling of their transcriptional regulation is ongoing and will provide prostate-specific promoter fragments that can activate therapeutic agents selectively in prostatic (cancer) cells.
The PSA gene promoter has been most extensively studied and revealed the existence of a proximal prostate-specific promoter with an upstream prostate-specific enhancer that are both required for high, androgen-regulated activation of PSA expression (11)(12)(13)(14)(15). The PSA enhancer-promoter was linked up to the HSV-tk gene, encoding a prodrug-converting enzyme, and was delivered by the human adenovirus into prostatic tumor cells growing subcutaneously in nude mice. As a result of the PSA promoter-driven HSV-tk expression, prostate tumor cell growth was significantly suppressed and life span of the animals was prolonged (16,17). This proof of principle opens the way for the application of promoter-based gene therapy for prostate cancer patients.
Recently, we have cloned another human prostate-specific expressed gene, DD3 (18). In addition to its prostate-specific mRNA expression, DD3 was shown to be highly overexpressed in prostatic tumors. These data indicated that DD3 is one of the most prostate cancer-specific genes described. The prostatespecific DD3 expression and the sharp up-regulation in prostate cancer, suggest a unique transcriptional regulation. Consequently, the DD3 gene promoter becomes an interesting candidate to be used for the application of promoter-based gene therapy of prostate cancer patients.
In this paper we describe the isolation of the DD3 5Ј-flanking sequences, and the identification of the proximal promoter that is required for the prostate-specific expression of DD3 mRNA. Furthermore, nuclear factor binding-sites in the DD3 promoter are identified, that may be important for the transcriptional regulation of the DD3 gene.

MATERIALS AND METHODS
Isolation and Sequence Analysis of DD3 Promoter Clones-Genomic clones containing the DD3 gene were described previously (18). FIX-ME4 DNA, containing the DD3 exon 1-flanking region, was endonuclease digested and subcloned in plasmid vectors pGEM-3Zf(ϩ) or pT 2 . Double-stranded plasmid DNA was isolated by standard procedures (19), and sequenced using the Thermo Sequenase cycle sequencing kit (Amersham Pharmacia Biotech) and Texas Red-labeled universal primers. Sequencing products were separated and analyzed using the Vistra DNA sequencer 725.
RNase Protection Assay-A DNA fragment encompassing the presumed transcription start site was used as a probe for the RNase protection assay. A 517-bp Sau3AI DNA fragment from clone pME4.6 was ligated into the BamHI site of M13mp18. Single-stranded M13 DNA was isolated, annealed with the DD3 exon 1-specific oligonucleotide BUS21 (5Ј-CCTTCCCACCATGCAGATCTTCCTGGTCTCCCTCG-GCTGCAGCCACACAA-3Ј), and extended using [␣-32 P]dATP and Klenow DNA polymerase. The radiolabeled probe was linearized with HincII and purified from a denaturing 5% polyacrylamide-urea gel. All reactions were performed according to standard protocols (19).
The probe (10 5 cpm) was annealed overnight with 40 g of total RNA at 30°C in 30 l of hybridization buffer (40 mM PIPES, pH 6.7, 1 mM EDTA, 0.5 M NaCl, 80% formamide). DNA-RNA hybrids were digested with 300 units of S1 nuclease (Amersham Pharmacia Biotech) in 300 l of S1 buffer (0.28 M NaCl, 0.05 M NaAc, pH 4.5, 4.5 mM ZnSO 4 , supplemented with 20 g/ml single-stranded herring sperm DNA) for 60 min at 30°C or 37°C. Digestions were stopped by the addition of stop buffer (4 M NH 4 Ac, 50 mM EDTA, 50 g/ml tRNA). Protected DNA fragments were resolved by 6% polyacrylamide-urea gel electrophoresis. A DD3 genomic sequence ladder generated with the BUS21 primer was used as size marker. Autoradiography was performed as described above.
Cell Culture-The following cell lines were used in transient transfection and/or electrophoretic mobility shift assays (EMSA): the human prostate adenocarcinoma cell lines LNCaP, PC-346C (kindly provided by Dr. W. van Weerden, Dept. of Urology, Erasmus University, Rotterdam, The Netherlands), and TSU-pr1; the human bladder cancer cell line SW800; the human colon carcinoma cell line HT-29; the renal cell carcinoma cell line SKRC-7; and the vulval epidermoid cancer cell line A431. All cells were grown in RPMI 1640 medium, supplemented with 10% fetal calf serum (Life Technologies, Inc.) in an atmosphere of 5% CO 2 and 37°C.
Production of Promoter Constructs-The promoterless plasmid p0GH (Nichols Institute) was used for cloning DD3 promoter fragments into the polylinker upstream of the human growth hormone gene. DD3 promoter fragments were produced by PCR using 5Ј-HindIII-tagged and 3Ј-BamHI-tagged primers, using normal human genomic DNA as a template. Subsequently, the fragments were cleaved with HindIII and BamHI and cloned into p0GH, using standard procedures (19). Mutant promoter constructs were generated using the "GeneEditor" in vitro site-directed mutagenesis system (Promega). Because the hGH secretion capacity of different cell lines may vary, for each cell line the activity of the DD3-hGH constructs was compared with that of plasmid pTKGH containing the human growth hormone gene driven by the constitutive promoter of the HSV-tk gene (20).
Transient Transfections-LNCaP cells were seeded at a density of 1 ϫ 10 6 cells per 10-cm dish 2 days prior to transfection, and TSU-pr1, HT-29, A431, or SW800 cells were seeded at a density of 5 ϫ 10 5 cells per 10-cm dish 1 day before transfection. For each transfection, 3 g of the appropriate DD3-hGH construct and 2.3 g of pCH110 (internal marker (21)) were complexed with Fugene-6 reagent (Roche Molecular Biochemicals) in serum-free medium for 15 min at room temperature. The Fugene-6⅐DNA complexes were added to the cell cultures, and cells were grown for an additional 72 h. Transfections were performed at least three times in duplicate.
Human Growth Hormone Assay-After transfection, medium was collected and stored at Ϫ20°C until use. Human growth hormone secretion in the medium was determined using the two-site fluoroimmunometric Delfia hGH assay kit (Wallac Oy, Turku, Finland) according to the manufacturer's instructions. hGH values were normalized to the ␤-galactosidase activities measured in the corresponding cell extracts. hGH values and relative induction values are expressed as mean and standard error of the mean (S.E.).
DNase-I Footprinting Analysis-Nuclear protein extracts for footprinting were prepared as described (22,23). Single strand end-labeled DNA probes were generated by first labeling the 5Ј-end of DD3 promoter-specific primers with [␥-32 P]ATP (ϳ3000 Ci/mmol, Amersham Pharmacia Biotech), followed by PCR amplification of DNA fragments with one labeled and one unlabeled primer (as indicated below in Fig. 5A).
Electrophoretic Mobility Shift Assay-Double-stranded oligonucleotides of DNase-I-protected DD3 promoter sequences were synthesized and end-labeled with [␥-32 P]ATP (ϳ3000 Ci/mmol, Amersham Pharmacia Biotech). Binding assays contained 32 P-labeled oligonucleotide (0.25 ng), 15 g of nuclear extract or 10 ng of recombinant HMG-I protein (provided by Dr. R. Reeves), 10 g of bovine serum albumin, 4 mM DTT, 10 M Zn 2 Cl, and 1.5 g of herring sperm DNA, and were adjusted to a final volume of 30 l with binding buffer (20 mM HEPES, pH 7.6, 50 mM NaCl, 10% glycerol, 0.1 mM EDTA). The antibody against the human HMG-I(Y) protein was provided by Dr. A. Fusco (Istituto Nazionale dei Tumori, Naples, Italy). Binding reactions were incubated for 30 min at room temperature. A 15-l aliquot of each reaction was loaded onto a 4% non-denaturing polyacrylamide gel and run in TBE buffer at 120 V for approximately 3 h. Autoradiography was performed as described above.
Data Base Accession Number-The nucleotide sequence for the DD3 promoter region has been deposited into the GenBank data base under accession number AF279290.

Structure of the 5Ј-Flanking Region of the Human DD3
Gene-Genomic clones, FIX-ME3, -ME4, and -IH1, containing the 5Ј-end of the human DD3 cDNA were described previously (18). Lambda phage DNA was subcloned in plasmid vectors for DNA sequence analysis. In Fig. 1A, a restriction map is shown of the resulting clones pFS28, pGV61, and pME4.6, containing DD3 exon 1 and its 5Ј-flanking sequences. The nucleotide sequence of the 5Ј-flanking region was determined (Fig. 1B) (Gen-Bank accession number AF279290).
Comparison of the DD3 5Ј-flanking nucleotide sequence with sequences in the non-redundant nucleotide data bases and the eukaryotic promoter data base, using BLAST (24), revealed no homology to any gene or promoter sequences described. Iden-tification of potential transcription factor binding sites by the MatInspector program, using the TRANSFAC 3.5 matrices (25), revealed no canonical binding sites at consensus positions, i.e. no initiator nor TATA boxes could be identified. Only a single CAAT element at a non-consensus position (Ϫ374 to Ϫ378) was found. Other potential transcription factor binding sites are shown in Fig. 1B.
Determination of the DD3 Transcription Start Site-The isolated human DD3 cDNAs were shown to possess different 5Ј-ends, with respect to their length (18, and data not shown). To precisely define the DD3 transcription start site, we have performed primer extension analysis and RNase protection assays, using total RNA from human prostate cancer tissue and normal human liver and lung tissue as a negative control.
For the primer extension analysis, DD3 exon 1 (BUS2)-and exon 3 (BUS7)-specific oligonucleotides were used as a primer. The use of the exon 3-specific primer, BUS7, is validated, because exon 2 is present in only a minority of transcripts (Ͻ5%) due to alternative splicing (18). Two prostate-specific extension products could be identified, as shown in Fig. 2A  (lanes 4, 5, 7, 8, and 9). The transcription start site deduced from the longest extension product was designated ϩ1, and consequently, the smaller fragment was initiated from an alternative start site at position ϩ34.
RNase protection analysis, using S1 nuclease, revealed two prostate-specific S1-protected fragments (Fig. 2B, lanes 1-4). The most abundant product was initiated from position ϩ3 with respect to the transcription start site identified by primer extension analysis (position ϩ1). The minor S1-protected fragment was initiated from an alternative start site at position ϩ12. The identified major and minor transcription start sites are shown above the DD3 nucleotide sequence in Fig. 2C.
Transcriptional Activity of the Human DD3 Promoter-To demonstrate promoter activity, the DD3 5Ј-flanking region was cloned upstream of the human growth hormone (hGH) reporter gene (construct pDDGH-1.9, position Ϫ433 to ϩ62). When pD-DGH-1.9 was transfected into LNCaP cells, a human prostate carcinoma cell line expressing DD3 mRNA (18), weak promoter activity was seen (Fig. 3), i.e. hGH production was about 20% of the HSV-tk-driven hGH production. Promoters are known to function unidirectionally. Therefore, the pDDGH-1.9 promoter sequences were cloned in the reverse orientation upstream of the hGH gene (pDDGH-2.1, position ϩ62 to Ϫ433). This pD-DGH-2.1 promoter construct was inactive in LNCaP cells, moreover, the hGH production was below that found in cells transfected with the promoterless p0GH construct. Sequences upstream of the 500-bp DD3 promoter (Ϫ433 to ϩ62) had no effect on the DD3 promoter activity (Fig. 3, construct pDDGH-1.5, position Ϫ1900 to ϩ62).
To investigate whether regions within the 500-bp DD3 promoter contributed to DD3 promoter activity, a series of 5Јdeletion constructs were generated. Transfection of the deletion constructs pDDGH-1.10, -1.11 and -1.12 into LNCaP cells resulted in an increased promoter activity, compared with the activity observed in pDDGH-1.9 (Fig. 3). Construct pDDGH-1.12 (position Ϫ152 to ϩ62) displayed the highest DD3 promoter activity of all constructs tested. Shortening the latter construct led to a decreased promoter activity (pDDGH-1.16) and a complete loss of activity in construct pDDGH-1.18 (position Ϫ41 to ϩ62).
Cell Type Specificity of the DD3 Promoter-To define the specificity of the observed DD3 promoter activity, promoter constructs pDDGH-1.9, displaying basal activity, and pDDGH-1.12, displaying maximum activity, were transfected into cell lines of different tissue origin. DD3 promoter activity of pD-DGH-1.9 was found in LNCaP cells (Fig. 4), but not in A431 (vulval carcinoma), HT-29 (colon carcinoma), SKRC-7 (renal cell carcinoma), and SW800 (bladder carcinoma) cells. Importantly, this promoter construct is also silent in prostate carcinoma cell lines that do not express DD3 mRNA (TSU-pr1, Fig.  4, and PC-3, data not shown). The increased promoter activity of the truncated promoter construct pDDGH-1.12, however, was also observed in the DD3-negative cell lines, although the maximum promoter activity was significantly lower than in LNCaP cells.
In Vitro DNase I Footprint Analysis of the Human DD3 Promoter-DNase-I footprinting analysis was performed using nuclear extracts (NE) of LNCaP (DD3 mRNA-positive) and TSU-pr1 (DD3-negative) cells. Single-strand end-labeled probes, covering the DD3 promoter region from position Ϫ433 to ϩ62 (Fig. 5A), were used to identify binding of nuclear factors to the DD3 promoter. Three DNase-I-protected regions  6) were used for primer extension using oligonucleotides BUS7 (lanes 1-5) and BUS2 (lanes 6 -11) as primers as indicated under "Materials and Methods." Primer extension products are indicated by arrows (4, BUS2, and 4//-, BUS7 products). A DNA sequence ladder of the DD3 genomic clone pME1.5S3 primed with BUS7 was used as a size marker. B, for RNase (S1 nuclease) protection analysis, a 252-bp radiolabeled single-stranded DNA probe was synthesized from plasmid pME4.6, containing exon 1 of DD3 and its 5Ј-flanking region, using BUS21 as a primer. The HincII-digested probe was hybridized to 40 g of total RNA from prostatic adenocarcinoma (lanes 1-4), liver (lane 5), or lung (lane 6) tissue or tRNA (lane 7) and treated with S1 nuclease at 37°C (lanes 1-7) or at 30°C (data not shown). The arrows mark protected fragments. A DNA sequence ladder of the DD3 genomic clone pME4.6 primed with BUS21 was used as a size marker. C, nucleotide sequence surrounding the DD3 transcription start site. Primer extension products are indicated by triangles (closed triangle, major start site; open triangle, minor site), and S1 nuclease protected fragments by diamonds (closed diamond, major start site; open diamond, minor site). or footprints (FPs) could be identified within the DD3 5Ј-flanking region: FP1 Ϫ29 to Ϫ63, FP2 Ϫ173 to Ϫ201, and FP3 Ϫ281 to Ϫ307 (Fig. 5B). The protection pattern in FP1 was identical using NEs from LNCaP and TSU-pr1 cells. A qualitative different protection pattern was observed in FP2, i.e. NEs of TSU-pr1 cells gave protection from position Ϫ173 to Ϫ201, whereas a NE of LNCaP cells protected the same DNA probe from position Ϫ182 to Ϫ198. FP3 was only observed using NEs of LNCaP cells. All three footprints were confirmed using probes that were end-labeled at the complementary DNA strand (data not shown). Furthermore, all DNase-I footprints could be prevented by the addition of a 250-fold excess of unlabeled specific DNA, but not by a 250-fold excess of unlabeled unrelated DNA fragments (Fig. 5C (FP2) and data not shown (FP1 and FP3)). These data show that at least three nuclear protein(s) (complexes) sequence-specifically bind to the DD3 proximal promoter (Fig. 5D).
Comparative EMSA of Footprinted Regions within the Human DD3 Promoter-To further substantiate the sequencespecific binding of nuclear factors to the DD3 promoter, and to discriminate between DD3-specific and nonspecific interactions, DNA bandshift assays were performed. As shown in Fig.  6A, three protein-DNA complexes (indicated I, II, and III) were formed when NEs from LNCaP and PC-346C cells were incubated with an oligonucleotide enclosing FP1. PC-346C is another androgen-dependent, DD3-expressing, prostate cancer cell line described recently (26). The slowest migrating complex (I) was more pronounced with PC-346C nuclear extracts than with LNCaP extracts. Furthermore, this complex was not formed using NEs from TSU, DU-145, or A431 cells, which do not express DD3 mRNA. This suggests that, although FP1 was identical with NEs from both LNCaP and TSU-pr1 cells, a DD3-specific nuclear factor interacts with proteins in the FP1 region.
When an oligonucleotide enclosing FP2 (position Ϫ212 to Ϫ172) was used as a probe, a single specific bandshift was observed (Fig. 6B). The addition of NEs of LNCaP and PC-346C cells to the binding reactions resulted in a slower migrating complex compared with the bandshift using TSU and other NEs (Fig. 6B and data not shown). Protein-DNA complex formation could be prevented by the addition of an excess of unlabeled specific probe, but not by an excess of unlabeled unrelated oligonucleotides (data not shown), confirming the specificity of DNA binding. The DD3 promoter region Ϫ211 to Ϫ159 was found to be AϩT-rich (Fig. 1B), and therefore, we investigated whether the high mobility group proteins I and Y (HMG-I(Y)), which preferentially bind to the minor groove of AϩT-rich sequences (27), are involved in DD3 promoter binding. When recombinant HMG-I was incubated with probe 2, a similar bandshift was observed to that of TSU-pr1 extracts. The inclusion of HMG-I(Y)-specific antibodies to the binding assay, abolished the protein-DNA interactions observed using LNCaP and PC-346C extracts (Fig. 6B). The inclusion of the HMG-I(Y) antibody in the TSU-pr1 binding reaction diminished HMG-I(Y) binding to the FP2 oligonucleotide by 75%, as determined by densitometry (Fig. 6B). Other, non-HMG-I(Y) IgG molecules had no effect on the formation of protein-DNA interactions (data not shown). These data indicate that HMG-I(Y) interacts with the DD3 promoter at the FP2 region. In addition, another unidentified protein interacts with the HMG-I(Y)-DD3 promoter complex in the DD3-expressing cell lines LNCaP and PC-346C.
Despite the observed DD3-specific FP3 (Fig. 5B), no difference in binding of LNCaP and TSU-pr1 nuclear factors to the oligonucleotide enclosing FP3 (position Ϫ279 to Ϫ311) could be observed by EMSA so far (data not shown).
Site-directed Mutagenesis of Factor Binding Sites-To identify the potential significance of the three identified DNAbinding sites for the promoter activity, several base substitution mutants in the reporter constructs pDDGH-1.9, pDDGH-1.10, and pDDGH-1.12 were created. Base substitutions were introduced in those motifs that were predicted by the MatInspector program, using the TRANSFAC 3.5 matrices (25), with a core similarity of 1.00 and a matrix similarity of over 0.90 (see Fig. 1B). The positions and nature of the base substitutions are shown in Fig. 7A. The examined mutations in the C/EBP␤, CCAAT, and Th1/E47 sites (i.e. FP3 and upstream region) did not affect the promoter activity when compared with the wildtype pDDGH-1.9 reporter construct (Fig. 7B). Scrambling the AϩT-rich sequence, present in FP2, resulted in a 50% reduced transcriptional activity (Fig. 7B, construct 1.10/51 versus pD-DGH-1.10), indicating a relevant role of the HMG-I(Y) binding site in the transcriptional activation of DD3. Base substitutions in the FP1 region most clearly reduced promoter activity (Fig.  7B). Mutations in the NF-1 sequence motif reduced promoter activity of the 1.9/60a and 1.12/60a constructs to 27% and 18%, respectively, and mutations in the E-box motif reduced transcription from reporter constructs 1.9/60c and 1.12/60c to 43% and 67%, respectively. These data show that the cis-acting elements present in FP1 and FP2 are functionally involved in the initiation of DD3 transcription.

DISCUSSION
The DD3 gene was previously shown to be highly overexpressed in the majority of prostatic adenocarcinomas (18). These data suggested that DD3 mRNA expression is regulated by a unique prostate cancer-specific transcriptional mechanism. In this paper we have described the cloning and the initial characterization of the prostate cancer-specific DD3 gene promoter.
Nucleotide sequence analysis did not reveal any obvious promoter elements. No known initiator motif, no TATA-box, no CAAT-box, and no GC-rich regions were found at consensus positions within the DD3 promoter. These canonical transcription factor binding sites are involved in the positioning of the transcription initiation complex. Interestingly, many TATAless promoters initiate transcription at multiple start sites (reviewed in Ref. 28). Some TATA-less promoters, on the other hand, initiate transcription from one or a few clustered start sites. These transcription initiation sites are encompassed by an initiator element, which contains all the necessary information for correct initiation of transcription (reviewed in Ref. 28). The DD3 promoter initiates transcription mainly at a single FIG. 4. Cell-type specificity of DD3 promoter activity. The pD-DGH-1.9 and pDDGH-1.12 DD3-promoter constructs were transiently transfected into LNCaP (L), TSU-pr1 (T), A431 (A), SKRC-7 (SK), HT-29 (H), and SW800 (SW) cells. DD3 promoter activity, i.e. hGH production, was determined as described as described in Fig. 3. site but so far lacks characterized initiator elements. A few promoters have been described that, like the DD3 promoter, are TATA-less and initiator-less and are not GC-rich (Ref. 29, and references therein). No further similarities to these gene promoters could be identified. Therefore, novel and tissue-specific cis-acting elements and trans-acting transcription factors might define the specific and characteristic expression of DD3.
We have shown that the 500-bp DD3 promoter region, encompassing the transcription start site, contains promoter ac-tivity upon transfection of reporter constructs into the DD3expressing prostate cancer cell line LNCaP. The promoter activity increases 3-fold upon shortening of the 500-bp promoter fragment (Fig. 3, construct pDDGH-1.12 versus -1.9), indicating the presence of a silencer in the DD3 proximal promoter. The activity of the pDDGH-1.12 DD3 promoter construct was restricted to the DD3-positive cell line LNCaP, as DD3-negative prostatic and non-prostatic cell lines only marginally initiated transcription from the DD3 promoter. These data suggest that the DD3 gene promoter is tissue and cell-type specific and, therefore, is a genuine prostate cancer-specific promoter. The absolute promoter activities of the DD3 promoter constructs tested, however, are rather low compared with the HSV-tk promoter activity. This correlates with the low level of endogenous DD3 mRNA expression observed in LNCaP cells, in contrast to the high DD3 expression in prostate cancer cells. Until now, we have found no other prostatic carcinoma cell lines that display high(er) DD3 expression levels; the expression of DD3 in the PC-346C cell line equals that of LNCaP cells.
On the basis of the deletion analysis of the DD3 promoter (see above), the Ϫ433 to ϩ62 region seems to contain most of the cis-acting elements necessary for DD3 expression. Therefore, we have analyzed the binding of nuclear factors to the DD3 promoter region by DNase-I footprint analysis and EMSA. DNase-I footprinting revealed three footprinted regions within the first 500 bp of the DD3 promoter. The first footprint, FP1, was found both in DD3-positive and DD3-negative cells. Nevertheless, a DD3-specific protein-DNA interaction was found in this region by EMSA (Fig. 6A). This DD3-specific factor might represent one or more of the components of the transcription initiation complex, because the protein-DNA interaction occurs in the region (Ϫ29 to Ϫ63) where the transcription initiation complex is supposed to accumulate. Site-directed mutagenesis revealed that the NF-1 and E-box sequence motifs in FP1 are necessary for the activation of DD3 transcription. E-boxes are recognized by transcription factors belonging to the basic helixloop-helix (bHLH) family, several of which are involved in tissue differentiation (reviewed in Ref. 30). However, antibodies against the bHLH E2A proteins did not affect the bandshifts observed with the FP1 oligonucleotide (data not shown). In addition, co-transfection of the HLH Id proteins, known antagonists of bHLH proteins (31), together with the DD3 promoter constructs did not affect the activity of the latter constructs (data not shown). The most pronounced effect on DD3 transcription was observed using constructs mutated in the NF-1 motif (position Ϫ58 to Ϫ62). Nevertheless, antibodies against the human NF-1 protein family, did not supershift or abolish any of the bandshifts shown in Fig. 6A (data not shown). These data indicate that the bHLH and NF-1 proteins are most probably not involved in the binding and regulation of DD3 transcription. Other factors might recognize and bind the NF-1 and E-box motif present in the DD3 promoter.
Another footprint, FP2, was found further upstream at position Ϫ173 to Ϫ201 in the DD3 promoter. Mutation of this AϩT-rich region, resulted in a decreased transcription rate, suggesting a positive role for this element in DD3 transcription. The high mobility group proteins I and Y (HMG-I(Y)) preferentially bind to the minor groove of AϩT-rich sequences (27). In addition, the expression of HMG-I(Y) was shown to be up-regulated in prostate cancer cells and a significant correlation with tumor grade and stage was found (32). Therefore, we investigated whether HMG-I(Y) binds to the FP2 region in the DD3 promoter, and as such might be involved in the transcriptional regulation of DD3. Bandshift analysis using HMG-I(Y)specific antibodies revealed that HMG-I(Y) interacts with DD3 promoter sequences in all prostate cancer cell lines studied. In addition, in the DD3-expressing cell lines LNCaP and PC-346C another unidentified factor is recruited to the HMG-I(Y)-DNA complex. Alternatively, a larger DNA-binding protein, antigenically similar to HMG-I(Y), may bind to the DD3 promoter instead of HMG-I(Y) itself. The DNA bending properties of HMG-I(Y) suggest that binding of HMG-I(Y) or HMG-I(Y)-like proteins could activate transcription initiation by altering the chromatin structure and/or by creating recognition sites for other trans-acting factors. DNA affinity purification methods will be approached to identify the protein(s) present in the slow migrating protein-DNA complex at FP2 of the DD3 promoter.
In addition to HMG-I(Y), the second footprint harbors a very FIG. 7. Site-directed mutagenesis of the human DD3 minimal promoter. A, DD3 minimal promoter constructs (pD-DGH-1.9, -1.10, and -1.12) are shown as lines with numbers corresponding to their positions relative to the transcription start site. Mutant constructs are shown below. The position of the mutations with respect to the transcription start site and the substituted bases are indicated to the left of each construct. AϩT3 N x , construct 1.10/51 contains a large substitution of the wild-type AϩT-rich sequence by 5Ј-CTCGAGCAGTCGACTAGCTCAGTACA-TAGCTCAGTAAGATCGTCTTTCAGAG-CT-3Ј. B, promoter mutants were transiently transfected into LNCaP cells and hGH production, and promoter activities were determined as described in Fig. 3. strong consensus motif for the FREAC-7/FKHL11 protein. FREAC-7 belongs to the forkhead/winged helix family of transcription factors (33). Several of the forkhead-related factors are tissue-specifically expressed and regulate cell-specific transcription, e.g. freac-6/HFH-3 is expressed in the distal tubules of the kidney (34) and HNF-3 in hepatocytes and respiratory and intestinal epithelia (35). No expression of freac-7 was reported in any adult human tissue (33). Therefore, despite the presence of a strong FREAC-7 consensus binding site in the DD3 promoter, the role of FREAC-7 in the regulation of DD3 expression seems speculative.
The putative silencer-containing region, position Ϫ132 to Ϫ366, was shown to lack promoter activity (Fig. 3, construct pDDGH-1.13). However, the cloning of fragments of the putative silencer (Ϫ132 to Ϫ254, Ϫ132 to Ϫ366, and Ϫ132 to Ϫ433) in front of the HSV-tk promoter did not reduce the HSV-tk promoter activity in transient transfection assays in LNCaP and TSU-pr1 cells (data not shown). This might indicate that, in contrast to the data from the promoter assays (Fig. 3), the DD3 promoter region from position Ϫ132 to Ϫ433 does not harbor a functional silencer. On the other hand, the (DD3specific) silencer may not repress transcription of a heterologous promoter. Alternatively, multiple repressor binding sites may be required to repress transcription from such promoters, as has been demonstrated for other silencer elements (36,37).
Our data suggest that the binding of a prostate-specific repressor (complex) to the proximal DD3 promoter tightly regulates DD3 transcription. Such repressor complex may bind and inhibit the transcription initiation complex, the latter which most probably harbors one or more prostate-specific activators (like complex I in Fig. 6A). For re-activation of transcription in vitro (LNCaP cells) or up-regulation of transcription in vivo (prostate cancer), specific mechanisms are required to overcome repression, such as inactivation of the repressor by mutation, transcriptional inactivation, binding of an inhibitor, or otherwise. Further studies are required to identify the nature of the trans-acting factors that are involved in the transcriptional regulation of the DD3 gene in vitro.