The nuclear factor SPBP contains different functional domains and stimulates the activity of various transcriptional activators.

SPBP (stromelysin-1 platelet-derived growth factor-responsive element binding protein) was originally cloned from a cDNA expression library by virtue of its ability to bind to a platelet-derived growth factor-responsive element in the human stromelysin-1 promoter. A 937-amino acid-long protein was deduced from a 3995-nucleotide murine cDNA sequence. By analyses of both human and murine cDNAs, we now show that SPBP is twice as large as originally found. The human SPBP gene contains six exons and is located on chromosome 22q13.1-13.3. Two isoforms differing in their C termini are expressed due to alternative splicing. PCR analyses of multitissue cDNA panels showed that SPBP is expressed in most tissues except for ovary and prostate. Functional mapping revealed that SPBP is a nuclear, multidomain protein containing an N-terminal region with transactivating ability, a novel type of DNA-binding domain containing an AT hook motif, and a bipartite nuclear localization signal as well as a C-terminal zinc finger domain. This type of zinc finger domain is also found in the trithorax family of chromatin-based transcriptional regulator proteins. Using cotransfection experiments, we find that SPBP enhances the transcriptional activity of various transcription factors such as c-Jun, Ets1, Sp1, and Pax6. Hence, SPBP seems to act as a transcriptional coactivator.

SPBP (stromelysin-1 PDGF 1 -responsive element-binding protein) was originally isolated from a murine gt11 cDNA library as a protein that bound to the stromelysin-1 PDGFresponsive element (SPRE) of the human stromelysin-1 promoter (1). The published cDNA sequence of SPBP is 3995 base pairs (bp) long with an open reading frame (ORF) of 2822 bp encoding a protein of 937 amino acids (1). SPBP has also been called AR1 or TCF20 (2). Stromelysin-1, also known as MMP3, is an extracellular matrix-degrading metalloproteinase that is active against a broad range of substrates. These proteases are invariably up-regulated in epithelial cancers, recognized as targets of oncogenic signal transduction pathways and shown to contribute to tumor invasion and metastasis (3,4). Recently, it was reported that stromelysin-1 actually also promoted mammary carcinogenesis in a mouse model system. Overexpression of stromelysin-1 in the mammary gland of transgenic mice induced all stages of tumor progression from hyperplasia to malignant carcinomas (5). In such experiments, stromelysin-1 seems to act as a natural tumor promoter enhancing cancer susceptibility (6).
The transcriptional activity of the stromelysin-1 promoter is stimulated by PDGF, and SPBP is reported to contribute to this up-regulation (7,8). Interestingly, the expression of SPBP itself was also found to be induced by PDGF or serum (1). The stromelysin-1 promoter contains three elements that are important for induction by mitogenic stimuli. These are an AP-1 element binding the c-Fos/c-Jun transcription factors, two head-to-head PEA3 elements binding Ets family transcription factors, and the SPRE element that binds SPBP (see Ref. 7 and references therein). The signal from the PDGF receptor is dependent on active Ras protein and bifurcates into one pathway that involves Raf-1 and activates the stromelysin-1 promoter via the AP-1 and PEA3 elements. The other pathway involves the atypical protein kinase C isoforms and and acts via the AP-1 and SPRE elements (7,8). It has been proposed that SPBP cooperates with c-Jun to transactivate via the SPRE site, and these two proteins are found to interact in vitro (7).
Transcriptional regulation requires the concerted action of a large number of proteins, many of which act in multiprotein complexes (reviewed in Refs. 9 and 10). These factors, termed transcription factors, transcriptional cofactors, or mediators, seem to be involved at two levels: 1) modulation of chromatin structure making DNA more or less accessible for other transcription factors to bind and 2) recruitment of the general transcriptional machinery to the promoter.
Here we show that SPBP is twice as large as originally described. Due to alternative splicing, two isoforms with different C termini are produced (1983 and 1965 amino acids long for the murine protein). The human and murine proteins show a sequence identity of 92%. The murine gene is located on chromosome 15, while the human is found on chromosome 22 in positions q13.1 to q13. 3. SPBP is expressed in most tissues except ovary and prostate. Functional mapping revealed SPBP to be a nuclear multidomain protein. It contains three nuclear localization signals, a novel type of DNA-binding domain with a single AT-hook, a transactivation domain in the N-terminal end, and in the very C-terminal end an evolutionary conserved zinc finger domain also found in the trithorax family of chromatin-based transcriptional regulator proteins. Interestingly, SPBP has the ability to enhance the transcriptional activity of various transcription factors, such as c-Jun, Ets, Sp1, and Pax6, suggesting that SPBP may be a novel transcriptional coactivator.

MATERIALS AND METHODS
Cloning and Sequencing of Murine and Human cDNAs for SPBP-A mouse brain Marathon-Ready cDNA library (CLONTECH) was used as template for nested 5Ј-and 3Ј-rapid amplification of cDNA ends (RACE) PCR with the primers 5Ј-CTGAACTACCTCTTGACAGCAACGAA-3Ј and 5Ј-TCCGATACCATTACCCATGTGCCAT-3Ј for 3Ј-RACE and 5Ј-TTCATTTCAGGAGCTGTGCTGCTTGA-3Ј and 5Ј-CCAGCTTCTTTC-CTGAAACCCTGATA-3Ј for 5Ј-RACE. The PCRs were performed as recommended by the supplier (CLONTECH) except that TaKaRa Ex Taq polymerase (TaKaRa Shuzo Co. Ltd.) was used. A 2694-bp 5Ј-RACE PCR product containing the missing 5Ј coding region of murine SPBP cDNA was cloned into the SmaI site of pUC18 and sequenced.
To isolate and sequence the remaining 5Ј coding region of human SPBP, PCR was performed on human genomic DNA using the primers 5Ј-CAGTCCTTTCGGGAGCAAAGCAGTTAC-3Ј and 5Ј-ACTACTCA-ACCCAGGATCTGTCAGTCG-3Ј.
DNA sequencing was performed using both manual and automated dideoxysequencing methods. Computer-assisted analyses of DNA and protein sequences were performed using the Wisconsin Genetics Computer Group software package as well as a number of World Wide Web-based sequence analysis tools found at the Expasy molecular biology server, the NCBI, and Sanger Center servers.
Plasmid Constructs-In the following, unless stated otherwise, the numbers given in parentheses in the construct names refer to the part of the murine SPBP amino acid sequence included in the respective constructs. pCIneo-SPBP and pcDNA3-HA-SPBP were constructed for expression of the entire coding region of the SPBP cDNA in mammalian cells. pCIneo-SPBP was constructed in three steps. First, most of the 2.69-kb 5Ј-RACE fragment inserted into the SmaI site of pUC18 was cloned as a XbaI-NheI fragment into the NheI site of pCIneo, giving pCIneo5ЈSPBP. Second, an EcoRI fragment from 5ЈKS-SPBP (1) was ligated into the EcoRI site of pcDNA3-HA-SPBP-(1016 -1965) (1). Finally a 3.3-kb NheI-XbaI fragment from this construct was inserted into the NheI site of pCIneo5ЈSPBP to give pCIneo-SPBP. To construct pcDNA3-HA-SPBP, PCR was first performed using pCIneo-SPBP as the template with the primers 5Ј-CAGTCGTTTCGGGAGCAAAGC-3Ј and 5Ј-GGAATCCTCCTGCTGTGCAAC-3Ј. The resulting PCR product was ligated into the end-filled XhoI site of pcDNA3-HA (1), and the correctly oriented in-frame fusion was verified by DNA sequencing. pcDNA3-HA-SPBP-(1-1864), which expresses SPBP with the Cterminal zinc finger domain deleted, was generated by inserting a 5784-bp-long SmaI-and XbaI-cut PCR fragment obtained from pcDNA3-HA-SPBP, using the primers 5Ј-TCCCCCGGGATGCAGTCG-TTTCGGGAGCAA-3Ј and 5Ј-GAAGGTGGGCCTGAGCTGGAGTGAT-CTAGACG-3Ј, into the EcoRV and XbaI sites of pcDNA3-HA.
All constructs were verified by sequencing, and the expression of correctly sized fusion proteins was confirmed by Western blotting following transfection of HeLa cells.
Western Blot Analyses-HeLa cells were seeded at a density of 4 ϫ 10 5 cells per 10-cm dish and transfected with 1, 5, or 10 g of expression plasmids using the calcium phosphate coprecipitation method. The precipitates were left on the cells for 4 h in medium containing 10% serum. Usually, 0.3 g of pCMV-␤gal (Stratagene) was included in the transfections to allow measurements of ␤-galactosidase activities that were used to normalize for variations in transfection efficiencies between experiments. Extracts were prepared using the Dual-Light luciferase and ␤-galactosidase reporter gene assay system (Tropix Inc.), and ␤-galactosidase activities were determined using a Labsystems Luminoskan RT dual injection luminometer. Following normalization based on measured ␤-galactosidase activities, 5ϫ SDS sample buffer (250 mM Tris-HCl (pH 6.8), 500 mM dithiothreitol, 10% SDS, 0.5% bromphenol blue, 50% glycerol) was added to the extracts, which were heated to 95°C for 5 min and sonicated briefly on ice. Usually, 10 -40 l of lysate was subjected to electrophoresis on 5 or 10% SDS-polyacrylamide gels. When no normalization was needed, cells were harvested directly into 100 l of 2ϫ SDS sample buffer (100 mM Tris-HCl (pH 6.8), 200 mM dithiothreitol, 4% SDS, 0.2% bromphenol blue, 20% glycerol), heated to 95°C for 5 min, and sonicated briefly on ice. 40 l of the extracts were subjected to electrophoresis on 10% SDS-polyacrylamide gels. Blotting onto polyvinylidene difluoride membranes (Millipore Corp.), blocking, incubation with primary antibody, washing, and detection using chemiluminescence were as recently described (11). The following rabbit polyclonal primary antibodies were used: anti-GAL4 DBD (Santa Cruz Biotechnology, Inc., Santa Cruz, CA), anti-GFP (CLONTECH), anti-c-Jun (against amino acids 55-67 of human c-Jun; New England Biolabs), anti-Ets-1 (Santa Cruz Biotechnology), and anti-actin (Sigma). Anti-Ets-1 was used at a dilution of 1:20,000, while the others were diluted 1:1000. For detection of HA epitope-tagged proteins, a mouse monoclonal antiserum from 12CA5 hybridoma cells was used (diluted 1:50). To detect endogenous SPBP, a rabbit polyclonal antibody raised against mouse SPBP (1), diluted 1:2000, was used.
Expression Profiling of Alternatively Spliced Forms of SPBP-In order to determine the expression profile of the two alternative splice forms of SPBP mRNA in different tissues, PCR was performed, using the primers 5Ј-CT(T/C)CCGAAGAATCCACCTCCTAAGAG-3Ј and 5Ј-CCACCTTCTCATCTCCACAGTCTCAC-3Ј flanking the alternatively spliced exon, on a panel of first-strand cDNAs from various human tissues and from mouse embryos at four different stages of development (CLONTECH). The PCR conditions were as described in the manufacturer's protocol with 30 cycles performed on the glyceraldehyde-3-phosphate dehydrogenase control and 38 cycles used for SPBP employing Taq polymerase (Life Sciences).
Fluorescence Microscopy for Subcellular Localization Studies-To determine the subcellular localization of HA-epitope-tagged SPBP HeLa cells were seeded at a density of 8 ϫ 10 3 cells/well in a 24-well culture dish and transfected 24 h later with 0.4 g of pcDNA3-HA-SPBP or the vector control pcDNA3-HA using the calcium phosphate coprecipitation method. Approximately 24 h later, the cells were washed three times with ice-cold phosphate-buffered saline (PBS) and fixed in 4% paraformaldehyde in PBS for 10 min at 4°C. The cells were permeabilized using ice-cold methanol for 5 min at Ϫ20°C and washed in PBS. The aldehyde groups were blocked using 10 mM glycine (pH 8.5) for 5 min at room temperature before blocking in 3% bovine serum albumin in Tris-buffered saline (150 mM Tris-HCl (pH 7.4), 150 mM NaCl) for 45 min at room temperature. HA-SPBP was visualized by fluorescence microscopy (Leitz DMIRB invert microscope equipped for fluorescence and with a Leica DC100 digital camera) after incubation with a monoclonal anti-HA antibody (clone 12CA5; Roche Molecular Biochemicals), diluted 1:100 in 3% bovine serum albumin/PBS, for 1 h at room temperature and the anti-mouse IgG secondary antibody Alexa 594 (Molecular Probes, Inc., Eugene, OR) (2 g/ml in 3% bovine serum albumin/PBS) for 1 h.
For mapping of nuclear localization signals using GFP fusion constructs, HeLa cells were seeded at a density of 4.5 ϫ 10 4 cells per 3-cm well in six-well dishes and transfected 24 h later with 1 g of the different GFP fusion vectors using the calcium phosphate coprecipitation method. The subcellular localization of the GFP fusion proteins was determined by fluorescence microscopy of living cells. For 4Ј,6Јdiamidino-2-phenylindole (DAPI) staining, the cells were washed three times with ice-cold PBS 24 h after transfection, fixed, and permeabilized in 4% paraformaldehyde, 0.01% Triton X-100 for 10 min at 4°C. DNA was stained with 1 g/ml DAPI (Sigma) for 5 min at room temperature.
Protein Expression and Gel Mobility Shift Assays-GST fusion proteins were expressed and purified from E. coli as described previously (12) except that 2% glucose and 0.1 mM ZnCl 2 was added to the growth medium. Gel mobility shift assays were performed mainly as described (13). When cold specific and nonspecific competitor DNA were used, these reagents were added immediately prior to the addition of the labeled probe. DNA-protein complexes were resolved on nondenaturing 5% polyacrylamide gels (29:1) in 1ϫ Tris borate-EDTA run at 230 V and 8°C for 2 h and visualized by autoradiography and phosphorimaging (Molecular Dynamics, Inc., Sunnyvale, CA).
Reporter Gene Assays-HeLa cells were seeded at a density of 4.5 ϫ 10 4 cells/well in six-well dishes and transfected 24 h later using the calcium phosphate coprecipitation method. The luciferase reporter plasmid pG5E1bLuc (0.5 g), containing five GAL4 binding sites upstream of the minimal E1b promoter (14), was cotransfected with 0.5 g of plasmids expressing fusions between the GAL4 DBD and SPBP. To normalize for variations in transfection efficiency, 0.1 g of pCMV-␤gal (Stratagene), which expresses ␤-galactosidase from a CMV promoter, was included in each transfection. Extracts were prepared 24 -48 h following transfection using a Dual-Light luciferase and ␤-galactosidase reporter gene assay system (Tropix) and analyzed in a Labsystems Luminoskan RT dual injection luminometer. All transfection experiments were performed in triplicate and repeated at least three times using different DNA preparations.

Both Murine and Human SPBP cDNAs Encode Two 212-215-kDa Proteins Differing in Their Extreme C Termini Due to
Alternative Splicing-The published cDNA sequence of SPBP is 3995 bp long with a predicted ORF encoding a protein of 937 amino acids with a molecular mass of about 100 kDa (1). During sequencing of a genomic clone of murine SPBP, we discovered that the ORF of SPBP extended much further upstream than originally published. We therefore sequenced the entire original cDNA clone and found sequencing errors both 5Ј and 3Ј to the proposed start codon. Correction of these errors resulted in an extension of the ORF in the 5Ј-end with no in frame stop codons. Also, a different C-terminal sequence resulted following correction of a frameshift error. In order to isolate the remaining upstream coding region, we performed 5Ј-RACE PCR on mouse brain cDNA. Cloning and sequencing of the 2.7-kb RACE PCR product resulted in a 3045-bp extension of the ORF in the 5Ј-end with a putative start codon at nucleotide position 186. The original start codon proposed for SPBP (1) is located at amino acid 1016 in the murine sequence (GenBank TM accession number AY007594) shown in Fig. 1. By performing 3Ј-RACE PCR, we identified two different 3Ј-ends. One contained a 128-bp insertion at position 5879, suggesting splicing of an alternative exon. The inserted exon contained an internal stop codon. Thus, alternative splicing would result in different C termini of the encoded proteins, with one consisting of 1965 amino acids and the other of 1983 amino acids (see Fig.  1).
Data base searches revealed that most of the human SPBP homologue (from nucleotide position 866) was already sequenced as part of a large scale cDNA sequencing project (19). This cDNA clone, KIAA0292 (GenBank TM accession number AB006630) contains the above mentioned alternative exon. In addition, we found parts of the 5Ј-end (bp 1-465) to be sequenced as an EST clone (accession number AA634326). Since these two clones were not overlapping, we isolated and se-quenced the 5Ј-end of human SPBP (bp 1-1015) following PCR on human genomic DNA (GenBank TM accession number AY007595). We used genomic DNA, since most of the coding region of SPBP is contained within the large exon 2 (see below) including the putative start codon. The nucleotide composition of exon 1 makes it unlikely that it is part of the coding sequence. Interestingly, SPBP contains seven regions (see Fig. 1) with strong homology (sequence identity of Ͼ50%) to a large murine protein of unknown function named GT1. The first block of homology encompasses the proposed start codons of the two proteins with the six first residues being identical between SPBP and GT1 (data not shown), supporting the notion that this is indeed the start codon. The overall sequence identity between SPBP and GT1 is 25% (45% similarity). The human and mouse SPBP cDNA sequences encode proteins that show 92% identity and 97% similarity at the amino acid level. For the human proteins, the longest form is only 1960 amino acids.
FIG. 1. Alignment of the deduced amino acid sequences of murine and human SPBP cDNAs. An alternatively spliced version giving a different C-terminal sequence is shown below the sequence alignment (Alt.). The murine and human sequences are 92% identical. If changes to chemically similar residues (indicated by dots in the alignment) are taken into account, the sequences show 97% similarity. Seven regions with 50% or more sequence identity to the GT1 protein (39) are boxed and labeled GT1-A to -G. The proposed start codon for the previously published murine SPBP cDNA-derived amino acid sequence (1) is indicated with a filled square at position 1016 of the murine protein. Open triangles denote exon junctions in the evolutionary conserved zinc finger domain (underlined and designated ZNF2). A minimal DNA-binding domain (Min-DBD) is boxed, and three functionally mapped nuclear localization signals (NLS1 to -3) are indicated by lines above the sequences. Note that the two glutamine-rich stretches in the N-terminal region (denoted Q1 and Q2) are much shorter in the human protein than in the mouse homologue. A glycine-serine-alanine-rich region (GSA) and a serine-rich stretch are overlined. PEST1-PEST3 denote putative PEST motifs determined by computer analysis (20). This is mainly due to the fact that the Q1 and Q2 glutaminerich stretches in the N-terminal part of the proteins are 14 and 16 residues shorter, respectively, in the human proteins than in the murine proteins (Fig. 1). SPBP is a hydrophilic protein especially rich in serine residues (13.6%). It contains putative nuclear localization signals, a positive charge cluster indicative of a DNA-binding domain, and a C-terminal cysteine-and histidine-rich region indicative of a zinc finger domain. Three PEST motifs, suggesting that the protein may be unstable and/or proteolytically processed (20), are also present. The mouse protein contains a putative leucine zipper at positions 1198 -1219 (1). However, in the human protein this possible leucine zipper contains two glycine residues, suggesting that no leucine zipper structure can be established. A strongly predicted coiled-coil region is located within the putative zinc finger domain.
The 1965-amino acid ORF of the short splicing form of murine SPBP encodes a protein with a theoretical molecular mass of approximately 214 kDa. Consistently, HeLa cells transfected with an expression vector for mouse SPBP cDNA express a protein of about 220 kDa, whereas transfection with a vector expressing the originally described SPBP cDNA (1) gave a band of about 110 kDa ( Fig. 2A). We also found that SPBP in vitro translated from cDNA displayed the same apparent molecular weight as the SPBP overexpressed in HeLa cells (data not shown). To determine the expression and size of endogenous SPBP in different cell lines, we subjected total cellular extracts from murine fibroblast (NIH 3T3), human embryonic kidney (HEK 293), human HeLa, and murine pancreatic (InR1-G9) cell lines to Western blot analyses using a polyclonal antibody raised against a region encompassing amino acids 1406 -1913 of murine SPBP (1). We found SPBP to be expressed in all cell lines tested and to be of the same size as SPBP expressed from our cDNA expression vector (Fig. 2B). However, the expression level varied between the different cell lines, with HEK 293 cells containing the highest and NIH 3T3 cells the lowest amounts of endogenous SPBP. Altogether, these results confirm that the isolated cDNAs express full-length SPBP with a molecular weight twice as large as originally described.
The Human SPBP Gene Is Organized into Six Exons and Maps to Chromosome 22q13.1-3-By FISH analyses, human SPBP was found to be located near the end of the long arm of chromosome 22, band positions q13.1 to q13.3 (Fig. 3A). Consistently, the murine gene was mapped to the E band on the long arm of chromosome 15, which is syngeneic to human chromosome 22q13.2-3 (Fig. 3B). The sequence of the euchromatic part of human chromosome 22 was recently reported (21). By searching the data base maintained by the Human Chromosome 22 Sequencing Group at the Sanger Center with the human SPBP cDNA sequence, we were able to elucidate the exon-intron structure of the human gene as shown in Fig. 3C. The gene is oriented toward the centromere and is located at chromosome position 26 Mb. The gene encoding the NADHubiquinone oxidoreductase B14 subunit (NADHB14) is located 57 kb downstream of SPBP, while the closest upstream genes are 228 kb (novel protein CGI-96) and 335 kb (diaphorase) away, toward the telomere. The overlapping sequences from two PAC clones and one BAC clone were aligned to the SPBP cDNA sequence, revealing that the coding region is organized into five exons (exons 2-6) and that the two C-terminal isoforms of SPBP are produced by alternative splicing depending on inclusion or skipping of exon 5. We were unable to isolate human SPBP cDNA sequences by 5Ј-RACE that contained exon 1 sequences. However, we could isolate such murine cDNA sequences. Searching with a 149-bp 5Ј-untranslated region sequence from mouse SPBP cDNA corresponding to exon 1, we identified a highly homologous human sequence (82% sequence identity) with appropriately located splice donor consensus sequence 68.5 kb upstream of exon 2. This most likely corresponds to exon 1 ( Fig. 3C and data not shown). Exon 2 encodes most of the protein (amino acids 1-1823), while four short exons encode the very C-terminal part (amino acids 1824 -1965/1983) (Figs. 1 and 3C). Since exon 5 contains an in frame stop codon, alternative splicing creates two different C-terminal ends of SPBP. The large exon is followed by a very large intron of nearly 30 kb. The sizes of the following introns vary between 1 and 10 kb. Hence, altogether the protein-encoding part of the SPBP gene covers a genomic region of about 70 kb.
SPBP Shares a Conserved Zinc Finger Domain with the Trithorax Group of Chromatin-based Transcriptional Regulator Proteins-Sequence homology searches revealed that the C-terminal of SPBP harbors an 80-amino acid-long Cys-Hisrich cluster constituting a putative zinc finger domain also present in the trithorax family of proteins as well as proteins related to these (Fig. 4A). This domain has been referred to as a ZNF2 domain (22) and shows similarity to the PHD domains (23). However, while the PHD domains represent zinc fingers with a Cys 4 -His-Cys 3 motif, the ZNF2 fingers contain a His-Cys 5 -His-Cys 2 -His motif and may represent a type of "extended PHD domain." The originally described ZNF2 finger contains two Cys residues 23-25 amino acids upstream of the alignment shown in Fig. 4A. These residues are conserved in most of the aligned proteins but are not present in SPBP. Secondary structure predictions of the aligned ZNF2 domains indicate two ␣-helical regions in the first putative zinc finger. One of these ␣ helices is also predicted to have the propensity to form a coiled-coil structure. In Fig. 4B, a phylogram indicates the relationship between the different ZNF2 domains, which show 29 -41% sequence identity to the SPBP ZNF2 domain. Clearly, the ZNF2 domains of SPBP and the Drosophila EST sequence (AA695918) represent more divergent domains than those of the trithorax (Hrx, Trx, and Z81120) and trithorax-related proteins (Alr, Trr, and AC006017). The domain structures of FIG. 2. SPBP is twice as large as originally described. A, the SPBP cDNA encodes a protein of about 220 kDa. Western blot is shown of total cellular extracts from HeLa cells transfected with 10 g of expression vectors for full-length SPBP- , the originally published SPBP- (1016 -1965), or empty expression vector (pcDNA3-HA), respectively. Both proteins were expressed with an N-terminal HA epitope tag, and the proteins were resolved by SDS-PAGE (5%), transferred to a polyvinylidene difluoride membrane, and detected by chemiluminescence using the 12CA5 monoclonal anti-HA antibody. The migration of molecular weight standards is shown to the left. The asterisk indicates a protein unspecifically recognized by the anti-HA-antibody. B, endogenous SPBP and SPBP expressed from the cloned cDNA comigrate in SDS-PAGE. The upper panel shows a Western blot of a nuclear extract isolated from HEK 293 cells transfected with the HA-SPBP expression vector and of total cellular extracts isolated from untransfected NIH 3T3, HEK 293, HeLa, and InR1-G9 cells, respectively. The extracts were resolved on SDS-PAGE (5%) and transferred to a polyvinylidene difluoride membrane, and SPBP was detected by chemiluminescence using a polyclonal antibody raised against SPBP (1). The lower panel shows the same membrane stripped (in 0.2 M NaOH for 5 min) and reprobed with an actin-specific antibody. The migration of molecular weight standards is indicated to the left.
the different proteins harboring a ZNF2 domain are shown in Fig. 4C. All of the other proteins except for SPBP and AA695918 contain a C-terminal SET domain. In addition to the proteins shown here, we found Cys-His-rich regions with similarity to ZNF2 in predicted proteins from Saccharomyces and Arabidopsis and in a novel family of human transcription fac-  Fugufish trithorax (F-hrx), the trithorax-related proteins Alr (human) and Trr (Drosophila), a protein of unknown function from C. elegans (GenBank TM number Z81120), and predicted proteins from human (GenBank TM number AC006017) and Drosophila (GenBank TM number AA695918). Cys and His residues that may serve as zinc ligands are indicated by asterisks above the alignment and in single letter code below the alignment. Other conserved residues as well as conserved hydrophobic (#) or hydrophilic ($) residues are also indicated below the alignment. Two long ␣-helices (open boxes) and a ␤ sheet structure (closed box) predicted by the JPRED server are indicated along with a coiled-coil region predicted by Lupas' method (41). B, a phylogram of the ZNF2 domains aligned in A made using the Growtree program of the GCG software package. The sequence identities between the different ZNF2 domain sequences and the SPBP ZNF2 domain are given in parentheses. C, schematic representation of the domain structure of proteins harboring the ZNF2 domain. In addition, all of these proteins contain regions that are rich in specific amino acids such as proline, glutamine, threonine, and serine. AT-hook motifs are only found in trithorax proteins and in SPBP. ATA1 and ATA2 represent regions of homology unique to trithorax and trithorax-related proteins. ZNF1 is a Cys-His-rich domain related to the conserved ZNF2 domain. PHD and SET domains are also indicated. tors (AF10, AF17, and BR140). However, these domains are C-terminally truncated relative to the ZNF2 domain defined here, giving a conserved His-Cys 5 -His-Cys motif.
Differential Expression of the Two Alternative Forms of SPBP in Different Tissues-To determine the expression patterns of the two alternatively spliced forms of SPBP in different tissues, we performed PCR on a panel of first-strand cDNAs, from various human tissues and from mouse embryos at four different stages, using primers flanking the alternative exon. We found SPBP to be expressed in all tissues tested, except ovary and prostate with very low expression in pancreas and skeletal muscle (Fig. 5A, lanes 11 and 15). However, both the expression level and the distribution of the two different splice forms of SPBP varied in the different tissues. In brain, heart, and testis only the transcript including the alternative exon is expressed, and this form predominates in liver and kidney, while mainly the shorter transcript is present in the lung (Fig.  5A). For thymus, intestine, leukocytes, colon, and spleen, similar amounts of the two splice forms are expressed. Hence, the long splice form of SPBP is expressed in all 14 positive tissues, while the short form is expressed in 7-9 of these tissues. In mouse embryos, however, the short form of SPBP seems to be exclusively expressed at 7 and 11 days of development. The long form is found only at very low levels in 15-and 17-day embryos (Fig. 5B). Inspection of the entry Hs.201668 at the UniGene collection at the NCBI revealed 56 human SPBP ESTs derived from cDNA libraries prepared from brain, stomach, colon, foreskin, kidney, lung, pancreas, testis, thyroid, tonsils, uterus, liver, and spleen as well as germ cells. Together with our data, this suggests that SPBP is expressed in 19 different tissues and is thus almost ubiquitously expressed.
Three Different Regions of SPBP Determine Nuclear Localization-To identify nuclear localization signal(s) in SPBP, GFP was fused to different regions of the SPBP protein (Fig.  6A). Full-length SPBP localizes exclusively to the nuclei of HeLa cells transfected with the HA-SPBP-(1-1965) expression vector (Fig. 6B). However, the nucleoli are not stained, suggesting that SPBP is excluded from these nuclear structures. Transient transfections of HeLa cells with the different GFP fusions revealed that three regions of SPBP can independently direct the fusion proteins to the nucleus (Fig. 6B). One of these regions, residues 1602-1629 (NLS2), contains a bipartite nuclear localization signal predicted by computer analyses. The regions from amino acid 1280 to 1298 (NLS1) and from amino acid 1812 to 1819 (NLS3) both directed the fusion proteins exclusively to the nucleus. Comparison of these sequences (underlined in Fig.  1) with consensus sequences for basic type, classical NLSs (24) suggests that NLS1 and NLS3 are monopartite, arginine-rich NLSs, while NLS2 is a bipartite NLS.
SPBP Contains a DNA-binding Domain with a Single AThook Motif-SPBP was originally isolated from an expression library by virtue of its ability to bind to a concatenated SPRE probe (1). The first clone obtained corresponded to amino acids 1406 -1901 of murine SPBP. Hence, this region must contain DNA binding activity. In order to functionally map the DNAbinding domain(s), different regions were fused to GST and expressed in E. coli. The DNA binding activities of the purified GST fusion proteins were analyzed by gel mobility shift assays (GMSA) using an oligonucleotide probe with a single SPRE motif. We found that the region between amino acids 1535 and 1731 contained specific DNA binding activity. Further deletions showed the 50 N-terminal amino acids (1535-1584, denoted Min-DBD) to bind to SPRE with much higher affinity than the region from 1564 to 1656 (data not shown). However, titration and competition analyses showed that highest affinity and specificity were obtained with the GST fusion containing the entire region from 1535 to 1731 (denoted Max-DBD). Interestingly, GMSA with the Min-DBD GST fusion protein resulted in two retarded complexes (Fig. 7A, lane 1), while Max-DBD gave only one complex (Fig. 7A, lane 9). This could suggest that SPBP has the ability to bind as a dimer/multimer. To test this hypothesis, GMSA with oligonucleotides containing mutated SPRE motifs were performed. The SPRE motif is composed of two partially overlapping palindromic sequences (Pal-1 and Pal-2 in Fig. 7D). Pal-2 (ACTAGT) was previously shown to be important for binding of SPBP to SPRE (8). Thus, two mutant SPRE probes disrupting either Pal-1 (SPRE-m1) or Pal-2 (SPRE-m2) were made. GMSA with Max-DBD and the mutated probes showed that the DNA-protein complexes migrated longer in the gel than with the wild-type SPRE probe (Fig. 7A,  compare lane 9 with lanes 12 and 14). In addition, the intensities of the bands are much weaker. These results indicate that Max-DBD binds as a dimer and that each monomer may recognize different parts (half-sites) of the binding sequence. When one of the two palindromes is mutated, Max-DBD cannot bind as a dimer but binds weakly as a monomer. This assumption is strengthened by the observation of two retarded complexes when Min-DBD is used in GMSA with the wild type SPRE probe. Here, the lower band has highest intensity ( Fig.   FIG. 5. The expression level and distribution of the two different SPBP mRNA species vary in different tissues. A, PCR was performed on a panel of first strand cDNAs prepared from different human tissues (CLONTECH). 10% of the PCR products obtained from each reaction following 38 cycles were separated on a 2% agarose gel and visualized by ethidium bromide staining. The different SPBP transcripts give PCR products of 658 or 531 bp depending on the presence or absence of exon 5. As a control of the cDNA from each tissue, PCRs (30 cycles) of the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (G3PDH) were performed in parallel. The amounts of the glyceraldehyde-3-phosphate dehydrogenase PCR products are nearly equivalent in the different tissues, except in testis, where the amount is lower. Hence, the amounts of the SPBP PCR products most probably reflect the relative expression levels of SPBP mRNA in these tissues. The results shown are representative of three independently performed PCRs. B, in mouse embryos, the short form lacking exon 5 is exclusively expressed at 7 and 11 days of development with very low expression of the long form in 15-and 17-day-old embryos.
7A, lanes 1 and 3), indicating that this region mainly binds as a monomer but is also able to bind as a dimer when the protein concentration is increased (Fig. 7B, lanes 1 and 2). Min-DBD is unable to bind to the SPRE probes containing mutated Pal-1 or -2 (Fig. 7A, lanes 4 and 6). This indicates that Min-DBD requires both palindromes, each constituting a half-site, to be intact in order to bind to SPRE. To ensure that the Min-and Max-DBD GST fusion proteins bound specifically to the SPRE element, competition experiments with a cold oligonucleotide containing an unrelated binding sequence (a GC-box) (Fig. 7A,  lanes 3 and 11) and binding reactions using this unrelated oligonucleotide as the probe (Fig. 7A, lanes 8 and 16) were performed. Large amounts of poly(dA-dT) were also unable to compete for binding to the AT-rich SPRE motif, showing that the binding is sequence-specific and not simply dependent on AT-rich DNA (data not shown).
To further examine the possible dimerization upon DNA binding, GMSA with increasing amounts of Min-DBD or a mix of Min-DBD and Max-DBD was performed. Doubling the protein concentration resulted in a shift from a weak, quickly migrating complex to a strong, slowly migrating complex (Fig.  7B, lanes 1 and 2). Hence, Min-DBD seems to bind cooperatively to SPRE. Mixing of Min-and Max-DBD fusion proteins resulted in increased intensity of the slowly migrating complex (Fig. 7B, lanes 3 and 4). Max-DBD alone gives rise to only this complex (Fig. 7B, lane 5). Together, these results suggest that SPBP binds to SPRE as a dimer, as indicated to the left of Fig.  7, A and B. Insertion of 5 bp (SPRE-m3) or 10 bp (SPRE-m4) between the two palindromic elements in SPRE showed that the binding affinity decreased when the distance between the two elements was increased (Fig. 7B, lanes 6 and 7). Hence, the spacing between Pal-1 and -2 seems to be more important than their orientation relative to each other. However, as long as both elements are present, a dimeric binding seems to be preferred, since no quickly migrating complexes were observed as with SPRE-m1 and SPRE-m2 (Fig. 7A, lanes 12 and 14). By searching GenBank TM , we found that a partial cDNA clone of the chicken homologue of SPBP corresponding to amino acids 1520 -1731 of murine SPBP had been isolated in a screen for proteins binding to a sequence motif in the chicken ␦1-crystal- HeLa cells were transfected with expression vectors for the different GFP fusions (1 g) and analyzed by fluorescence microscopy 24 h post-transfection. For DAPI staining, the cells were fixed using 4% paraformaldehyde and stained with 1 g/ml DAPI for 5 min before fluorescence microscopy. C, Western blot of the different GFP fusion proteins, showing that they are all expressed and of the expected sizes. Total cellular extracts of HeLa cells transfected with the indicated GFP fusion expression vectors (10 g) were resolved on SDS-PAGE (10%), transferred to a polyvinylidene difluoride membrane, and immunoblotted with an antibody to GFP. The expression level of the different constructs varies, and many of the lanes contain degradation products detected by the antibody. The migration of molecular weight standards is shown to the right. lin enhancer core (25). This SPBP binding sequence does not contain the palindromic element. Instead it contains five Ts and an AT-rich region separated by CTG (Fig. 7D). GMSA with a probe containing this ␦-crystallin binding element showed that Max-DBD also bound as a dimer to this sequence (Fig. 7B,  lane 8), suggesting binding to each of the AT-rich regions. However, the affinity for the ␦-crystallin binding element is significantly lower than that for the SPRE. Hence, Pal-2 is important for high affinity binding. This is consistent with previous results emphasizing the importance of Pal-2 for binding of SPBP-(1016 -1965) to SPRE (1,8) and with our competition analyses showing that SPRE-m1 (containing an intact Pal-2) is a better competitor than SPRE-m2 (containing an intact Pal-1) (Fig. 7C). In conclusion, our binding studies suggest that SPBP binds as a dimer to SPRE in a cooperative manner. The Min-DBD region (amino acids 1535-1584) is important for specific binding. The region C-terminal of Min-DBD up to amino acid 1731 is important for dimerization and high affinity binding. Together, these two regions form the Max-DBD (amino acids 1535-1731) whose dimerization seems to be dependent on two specific recognition elements in close proximity on DNA (Pal-1 and Pal-2 in SPRE).
SPBP contains a single AT-hook as part of the Min-DBD ( Fig. 1 and Fig. 7E). The motif contains a lysine, instead of the most frequently found arginine residue, N-terminal to the absolutely conserved GRP tripeptide characteristic of AT-hooks (26). The GRP motif is followed by a cluster of basic and polar residues and a glycine in the second position C-terminal to GRP. This is characteristic of type I AT-hooks (27). Type I AT-hooks are shown to display high affinity DNA-binding activity (28). There are several examples of proteins containing a single AT-hook and additional larger DNA-binding domains (27). Some AT-hook motifs from different nuclear proteins are aligned to the SPBP AT-hook in Fig. 7E. Deletion of the AThook from Min-DBD (GST-SPBP-(1535-1566)) completely abolished DNA binding (data not shown). The high affinity binding observed with Max-DBD suggests the presence of an additional DNA-binding domain C-terminal to the AT-hook that cooperates with, or is dependent on, the AT-hook for DNA recognition and binding.
A Transactivating Domain Is Located in the Very N-terminal Part of SPBP-In order to identify putative transactivating domains, we fused different regions of murine SPBP to the  1-3) or mutated (lanes 4 -7) SPRE motifs. SPRE-m1 has mutated the most 5Ј palindrome, while SPRE-m2 has mutations in the most 3Ј palindrome (Pal-1 and -2 in D, respectively). The two retarded complexes observed indicate that SPBP may bind as a dimer/multimer to SPRE. Similarly, Max-DBD was incubated with the labeled wild-type (lanes 9 -11) and mutated (lanes 12 and 13 and lanes 14 and 15) SPRE oligonucleotides. The slowly migrating protein-DNA complex in lanes 9 and 11 indicates that SPBP may bind as a dimer to wild-type SPRE, while the faster migrating complexes in lanes 12 and 14 suggest that SPBP binds as a monomer when one of the palindromes of the binding site is mutated. Competition experiments with cold oligonucleotides (1.0 g) containing SPRE (lanes 2 and 10), the mutated forms of SPRE (lanes 5, 7, 13, and  15), or a GC-box sequence (lanes 3 and 11) show that the binding to SPRE, SPRE-m1, and SPRE-m2 is specific. This is further confirmed in lanes 8 and 16, showing no retarded complexes when the fusion proteins are incubated with a [␥-32 P]ATP-labeled probe containing an unrelated binding site (a GC-box). The complexes were separated on a 5% poly-acrylamide gel, dried, and visualized by autoradiography. B, SPBP binds cooperatively as a dimer on SPRE. Increasing amounts (1.5 and 3 g) of Min-DBD, denoted Min (lanes 1 and 2)  The complexes were separated on a 5% polyacrylamide gel, dried, and visualized by autoradiography. C, the 3Ј palindrome, Pal-2, has higher affinity for SPBP than the AT-rich Pal-1. 1 g of Max-DBD was incubated with the labeled SPRE oligonucleotide (lane 1) and increasing amounts (1, 2, and 3 g) of the cold competitor SPRE-m1 (lanes 2-4, respectively) or increasing amounts (1, 2, and 3 g) of the cold competitor SPRE-m2 (lanes 5-7, respectively). The complexes were separated on a 5% polyacrylamide gel, dried, and visualized by autoradiography. D, alignment of the different oligonucleotides used in GMSA. The two palindromic motifs (Pal-1 and -2) are indicated. Substitution mutations are shown in boldface type, while insertions are boxed. E, SPBP contains a type 1 AT-hook. The AT-hook sequence of SPBP is aligned to similar sequences from the Drosophila chromatin remodeling protein ISWI, human trithorax homologue 2 (HRX2), human HMGI-C and methyl CpG-binding protein 2 (MeCP2), ARS-binding protein 2 from Schizosaccharomyces pombe (ARS BP2), chloroplast DNA-binding protein PD3 from Pisum sativum, and AT-hook protein 1 from Arabidopsis thaliana (AT-H P1).
DNA-binding domain of the yeast transcription factor GAL4 (Fig. 8A). These GAL4 fusions were cotransfected into HeLa cells with a luciferase reporter plasmid containing five GAL4binding sites upstream of the E1b TATA promoter (14). The results show that SPBP contains a transactivation domain from amino acid 1 to 355 while the entire region from amino acid 463 to the C terminus showed no activity (Fig. 8B). The two glutamine-rich stretches (Q1 and Q2) alone give a 4-and 6-fold induction of the reporter gene. However, together they give a synergistic effect, with 75-fold induction of transcription. The region N-terminal to the glutamine-rich stretches gave a nearly 20-fold induction of transcription alone. Together with the glutamine-rich stretches, a transcriptional activation of about 200-fold was achieved. Inclusion of the region C-terminal to the glutamine-rich stretches did not result in any further activation. Instead, the transcriptional activation seemed significantly reduced. However, as revealed by immunoblotting of total cellular extracts from HeLa cells transfected with the different GAL4-constructs, this is simply due to a lower expression level of these fusion proteins (Fig. 8C).
SPBP Stimulates the Transcriptional Activity of Different Transcription Factors-SPBP contains several features similar to large nuclear proteins acting as coactivators, corepressors, or subunits of chromatin remodeling complexes such as the large size and presence of an AT-hook, zinc finger motifs, and polyglutamine stretches. Previous work showed that SPBP-(1016 -1965) could cooperate with c-Jun to stimulate transcription from a SPRE-containing promoter (7). Thus, SPBP may have the potential to act as a coactivator. In order to test this assumption, we performed transient cotransfections of mouse SPBP together with the transcription factors c-Jun, Sp1, Pax6, Ets-1, and E2F-1 in HeLa cells using different luciferase reporter vectors containing promoters being activated by the different transcription factors used. Interestingly, SPBP enhanced the transcriptional activation potential of all of these factors except for E2F-1 (Fig. 9A). This enhancement was most potent for c-Jun and Ets-1. The transcriptional coactivator CREB-binding protein enhances the transcriptional activation of Ets-1 (29) and c-Jun (30). Transient cotransfections in HeLa cells using either a CREB-binding protein or a SPBP expression plasmid together with Ets-1 or c-Jun expression plasmids showed that CREB-binding protein and SPBP enhanced the transcriptional activity of Ets-1 and c-Jun to a similar level (data not shown). Western blots following cotransfection of HeLa cells with different amounts of the expression plasmid HA-SPBP-  together with fixed amounts of Pax6, c-Jun, or Ets-1 expression plasmids showed that the observed superactivation was not due to a SPBP-induced increase in the protein concentration of the transcription factors.
To determine if the C-terminal zinc finger domain could be important for the transcriptional enhancement, we performed cotransfections in HeLa cells with c-Jun, Ets-1, Sp1, and Pax6 expression plasmids together with an expression plasmid encoding SPBP lacking the the zinc finger domain (SPBP-(1-1864)). As shown in The gray ovals indicate the GAL4 DBD. B, the two polyglutamine stretches (Q1 and Q2) act synergistically, but full activation requires also the N-terminal region. The different GAL4 fusions (0.5 g) were cotransfected into HeLa cells with a reporter plasmid (0.5 g) containing five GAL4-binding sites upstream of the E1b TATA promoter. To normalize for variations in transfection efficiencies, 0.1 g of pCMV-␤gal was included in each transfection to allow measurement of ␤-galactosidase activities. The luciferase activity was standardized to the ␤-galactosidase activity, and the activity of GAL4-DBD alone was set to 1. The results from an experiment performed in triplicate that is representative of three independent experiments are shown. S.D. values varied between 1 and 15%. C, Western blot of total cellular extracts from HeLa cells transfected with the different GAL4 fusion plasmids. The proteins were separated on SDS-PAGE (10%), transferred to a nitrocellulose membrane, and detected by immunoblotting with an antibody to the GAL4-DBD. Loading of the extracts was standardized according to the transfection efficiency determined by measuring ␤-galactosidase activities obtained from cotransfected pCMV-␤gal.
SPBP-(1016 -1965) enhanced the transcriptional activation of all of these factors except for Sp1, but not as potently as the full-length protein (Fig. 9B). Importantly, the SPBP-(1016 -1965) is more efficiently expressed than the full-length protein.
Hence, the N-terminal part of SPBP, showing transactivating activity in the context of the GAL4 DBD, contributes to the superactivation of the transcription factors observed here. DISCUSSION We show that SPBP, originally described as a 110-kDa novel transcription factor acting to regulate stromelysin-1 transcription, is actually a 220-kDa nuclear factor that is also able to function as a transcriptional coactivator. SPBP is expressed in most tissues, except for prostate and ovary. Two different isoforms, differing in their C termini, are produced by alternative splicing. The splicing pattern varies between different tissues and during embryonic development. The significance of the different isoforms is currently unknown. However, the differential expression of the two splice forms in the various tissues analyzed suggests that they may serve different roles or functions. Since SPBP is expressed in most tissues while stromelysin-1 expression is more restricted, SPBP may have a more general role in transcriptional regulation than only to control stromelysin-1 expression.
Except for the evolutionary conserved ZNF2 zinc finger domain, we have been unable to detect sequences in the sequenced genomes of Saccharomyces cerevisiae, Caenorhabditis elegans, or Drosophila that could represent direct homologues of SPBP. Thus, SPBP may be a vertebrate-specific protein. By FISH analysis, the human gene encoding SPBP was located at 22q13.1-3, and the murine gene was localized to the syntenic chromosome 15E region. Consistently, while this work was in progress Rajadhyaksha et al. (2) reported that AR1 or TCF20 (other names for SPBP) is located at 22q13.3. We find SPBP to be located at position 26 Mb by alignment to the sequence data from chromosome 22q (21). Several tumor forms have been associated with deletions and other lesions in this part of chromosome 22 including gliomas, meningiomas, and ovarian and colon cancers. However, the mapping data are too imprecise to say anything about a possible involvement of SPBP. SPBP was originally isolated by virtue of its ability to bind specifically to a concatenated SPRE probe (1). We found the region from amino acid 1535 to 1584 to contain a minimal DNA binding domain (Min-DBD), but both the affinity and the specificity of the binding increased when the region was extended C-terminally to amino acid 1731 to give the maximal DNA binding domain (Max-DBD). The SPRE consists of an AT-rich palindrome (Pal-1), which partially overlaps with another palindrome (Pal-2). These two palindromes may constitute halfsites. Our results from GMSA with the Min-and Max-DBDs are compatible with the notion that SPBP binds as a dimer or multimer to the SPRE motif when the complete binding site is present and as a monomer when Pal-1 or Pal-2 is mutated. This notion is supported by the occurrence of a differently migrating complex when the Min-and Max-DBDs were mixed compared with the complexes observed when these two domains were assayed separately. Interestingly, the affinity of DNA binding increased significantly when the protein concentration was raised above a certain level. This is indicative of cooperativity in the DNA binding or dimerization. A cooperative, dimeric/ multimeric DNA binding by SPBP is also supported by the hyperbolic activation kinetics of a SPRE reporter upon cotransfection of increasing amounts of SPBP-(1016 -1965) (7). The Min-DBD contains a single AT-hook motif. Deletion of the AT-hook from the Min-DBD abolished DNA binding. The region C-terminal to the AT-hook, constituting the rest of the Max-DBD, does not show any sequence homology to any known classes of DBDs. By itself, this region is unable to bind to the SPRE motif, suggesting that the Min-DBD with an intact single AT-hook motif is essential for DNA binding. Multiple and single AT-hooks motifs are found in several multidomain pro- FIG. 9. SPBP enhances the transcriptional activity of various transcription factors. A, the transcriptional activity of Sp1, c-Jun, Ets-1, and Pax6 was stimulated 3-6-fold upon coexpression with murine SPBP. Luciferase reporter vectors (0.5 g) with promoters responsive to the different transcription activators assayed here were transfected into HeLa cells together with expression vectors for c-Jun (0.5 g), Ets-1 (0.1 g), Pax6 (0.5 g), or Sp1 (0.5 g) and HA-SPBP-(1-1965) (1.0 g), as indicated in the figure. Luciferase activities were normalized to ␤-galactosidase activities derived from the cotransfected vector pCMV-␤gal to account for variations in transfection efficiencies. The background activities measured from the reporters upon cotransfection with empty expression vectors were set to 1 for each of the transcriptional activators assayed. All transfection experiments were performed in triplicate. The results of a typical experiment, representative of more than three other independent experiments performed with different preparations of the plasmids, are shown. B, the C-terminal zinc finger domain of SPBP is required for stimulation of the transcriptional activity of Sp1 and c-Jun. Similar experiments as in A were performed except that the ability of full-length SPBP (amino acids  or the originally described SPBP (amino acids 1016 -1965) or SPBP lacking the C-terminal zinc finger domain (amino acids 1-1864) to enhance transcriptional activation was compared. The reporter gene activities measured for cells transfected with expression vectors for Sp1, c-Jun, Ets-1, or Pax6 without any SPBP expression vector were set to 1. The data represent the mean of four independent experiments performed in triplicate. teins that associate with chromatin (27). Interestingly, among these are members of the trithorax group of regulatory genes that also contain zinc finger domains (ZNF and PHD) with homology to the zinc finger domain of SPBP. Trithorax proteins are involved in chromatin decondensation (31), and their AThooks have been implicated in several chromosomal translocations resulting in different lymphoid leukemias (32)(33)(34). The type 1 AT-hook motifs such as the one found in SPBP seem to be auxiliary elements necessary for cooperation with other DNA binding activities in the same or different proteins (27). This is consistent with our finding that SPBP has a Min-DBD with the AT-hook and a larger domain C-terminal to it that together with the Min-DBD constitutes the Max-DBD. AThooks are known to bind to AT-rich sequences in the minor groove of DNA. In doing so they often change the DNA conformation, making DNA more or less accessible for other DNA binding factors, depending on the sequence context. This suggests that the Min-DBD of SPBP recognizes the AT-rich Pal-1 of the SPRE motif, binds in the minor groove, and thereby makes the Pal-2 sequence more accessible for binding by another SPBP molecule.
A dual picture of the function(s) of SPBP is emerging. On one hand, two previous reports suggest that SPBP acts as a classical sequence-specific transcriptional activator. SPBP-(1016 -1965) and c-Jun were found to cooperate to transactivate a promoter under SPRE control (1,7). On the other hand, we find that SPBP may act as a coactivator for structurally and functionally rather different transcription factors binding to distinct target sequences in promoters/enhancers. However, this is not inconsistent with the previously reported collaboration between c-Jun and SPBP-(1016 -1965). In fact, it was found that c-Jun and SPBP bound to each other in vitro and that c-Jun could activate the reporter alone but that this activation was markedly increased upon coexpression of SPBP (7). This is in complete agreement with our finding of SPBP as a potent coactivator of c-Jun. Since SPBP coactivated such diverse transcription factors as Sp1, Pax6, c-Jun, and Ets-1, the mechanism(s) involved may be relatively general targeting common components. The finding that the zinc finger domain was necessary for stimulation of the c-Jun and Sp1 activity but not for the stimulation of the Ets-1 and Pax6 activity indicates that different parts of SPBP are involved, depending on the specific transcription factor. At what level does SPBP act to stimulate the transcription mediated by these factors? We found that SPBP does not act by increasing translation or increasing protein stability. The fact that SPBP did not enhance the transcriptional activation potential of E2F-1, together with the nonstimulatory effect of SPBP alone, suggests that SPBP does not act at the levels of pre-mRNA processing or mRNA stability of the luciferase reporter gene transcript. Thus, most likely, SPBP acts on the transcriptional initiation step, either at the chromatin/DNA level or in the recruitment of necessary factors to the initiation site. Several AT-hook-containing proteins play important roles in modulating chromatin structure and act as transcriptional cofactors (see Refs. 35-38 and references therein). AT-hook proteins may act both as coactivators or corepressors. In addition to changing the DNA conformation, the stimulatory effect of some AT-hook-containing proteins is also dependent on protein-protein contacts (35). Thus, SPBP may stimulate the activity of other transcription factors both by changing the DNA conformation, making it more accessible, and by protein-protein contacts recruiting the transcription factors to DNA. In this context, it is noteworthy that the ZNF2 or extended PHD domain that SPBP shares with the trithorax family of chromatin-based transcriptional regulatory proteins is most likely a protein-protein interaction domain. We have found that a GST fusion of the ZNF2 domain of SPBP binds to in vitro translated full-length SPBP. Furthermore, in a yeast two-hybrid screen with the ZNF2 domain as bait, we isolated a part of murine GT1 corresponding to the GT1-F homology region of SPBP. 4 In future studies, it will be important to interfere with or knock out the endogenous SPBP to determine what physiological processes may depend on active SPBP. Our attempts to inhibit the expression of endogenous SPBP using antisense RNA expression were unsuccessful. Thus, other strategies like microinjection of neutralizing antibodies (not presently available) or overexpression of dominant negative mutants and the production of knock-out mice are needed to study the effects on transcription and phenotypic changes associated with blockade of endogenous SPBP.