The interaction of GATA-binding proteins and basal transcription factors with GATA box-containing core promoters. A model of tissue-specific gene expression.

The core promoters of the rat platelet factor 4 (PF4), mouse erythropoietin and chicken beta globin genes contain a GATA motif in place of the consensus TATAAA site. In the case of the PF4 gene, this site has been shown to play a critical role in restricting transcription to the megakaryocyte lineage. In order to understand the mechanism of tissue specificity, we investigated the function of the GATA box-containing promoters in vitro. Our studies show that the TATA-binding protein of TFIID is required for initiation of transcription from the GATA box-containing promoters. GATA-1 interacts with the core promoter GATA motif and inhibits generation of preinitiation complexes. The functional significance of the inhibition of preinitiation complexes is supported by in vitro transcription assays in which transcription from the PF4 and erythropoietin core promoters is suppressed by GATA-1. We also demonstrate that GATA-2 inhibits initiation of transcription from the PF4 core promoter. Based on these results, we propose a model in which repression of PF4 expression in nonmegakaryocytes is mediated, in part, by competition between GATA-binding proteins and basal factors for the core promoter.

row is limited to megakaryocytes, suggesting that this region contains the tissue-specific regulatory elements (Ravid et al., 1991a). In transient transfection assays using rat bone marrow, deletion analysis of the PF4 promoter has led to the identification of 5' upstream sequences that regulate transcription in the megakaryocyte and inhibit expression in other hematopoietic cell types (Ravid et a l . , 1991b). Surprisingly, a single point mutation of a GATA motif (-31 to -28) to the canonical TATA sequence leads to low level expression in non-megakaryocyte cells (Ravid et al., 1991b). In view of the critical role of this motif in directing tissue-specific expression, we asked how initiation of transcription from the PF4 promoter differs from TATA-containing promoters.
Initiation of transcription from all class I1 promoters is believed to be mediated by the assembly of a common set of general factors on the core promoter. Usually, this process begins with the binding of TFIID to the consensus TATAAA and is followed by the recruitment of other general factors, leading to the formation of a multi-protein initiation complex (Sawadogo and Roeder, 1985;Buratowski et al., 1989). The TATAAA consensus sequence is not only important as a recognition site for TFIID, it also plays a role in defining an accurate start site and determining the direction of transcription. The sequence of events leading to initiation of transcription from the minority of promoters that lack a defined TATA box is less clear. In some cases, TFIID may bind specifically to the -30 site, despite the absence of the consensus TATAAA Wiley et al., 1992). In other instances, an initiator element may play an important role in localizing the start of transcription (Smale and Baltimore, 1989). In TATA-less promoters that contain a n upstream SP1-binding site, TFIID appears to be associated with the promoter via protein-protein interactions involving a tethering factor (Pugh and Tijan, 1989). Thus, the above data suggest that the only difference between promoters that contain a TATA box and those that lack this sequence is the mechanism by which TFIID is recruited to the preinitiation complex.
The rat PF4 core promoter, as well as the mouse erythropoietin and chicken p globin promoters not only lack the consensus TATAAA sequence, but in its place contain a GATA motif. The GATA element was initially identified as an upstream cis regulatory site in erythroid cell-specific promoters (Evans et al., 1988;Wall et al., 1988;Martin et al., 1989). Studies of nuclear proteins binding to this site led to the cloning of GATA-1, a transcription factor restricted in tissue distribution to erythroid, megakaryocyte, and mast cell lineages (Tsai et al., 1989;Evans and Felsenfeld, 1989). Subsequent studies have led to the isolation of related proteins (GATA-'2 to GATA-4) that share a highly conserved zinc finger DNA-binding domain but differ widely in tissue distribution (Yamamoto et al., 1990;Lee et al., 1991;Ho et al., 1991;Arceci et al., 1993). GATA factors bind to aa4

An in Vitro
Study of GATA Box-containing Core Promoters a common upstream consensus site T/A(GATA)AIG and activate transcription in cotransfection assays. More importantly, targeted mutagenesis in mouse ES cells has established that GATA-1 expression plays a critical role in the normal differentiation of erythroid cells (Pevney et al., 1991). In contrast to the regulatory function of the upstream GATA element, the role of the core promoter GATA motif is unknown. The importance of this site for tissue-specific expression of PF4 raises the interesting issue of how GATA-binding proteins and the basal transcription apparatus interact within this region. Although others have shown that GATA-1 competes with yeast TATAbinding protein (TBP) for binding to the chicken p globin promoter, the functional significance of this competition has not been elucidated (Fong and Emerson, 1992). In the present study, we show that GATA proteins inhibit the assembly of the basal transcription apparatus as well as initiation of transcription from the PF4 and erythropoietin promoters. Based on these results, we propose a model to explain the mechanism by which the PF4 core promoter helps to establish tissue-specific expression of this megakaryocyte protein.

MATERJALS AND METHODS
Plasmids and Cells-Plasmids containing core promoter sequences were constructed with double-stranded oligonucleotides containing a 5' EcoRI site linked to -50 to +3 of the rat PF4 and mouse erythropoietin genes and -51 -+3 of the p globin gene (Fig. la). These fragments were inserted between the EcoRI and SmaI sites of PUC19 and the resultant constructs were used to generate probes for electrophoretic mobility shiR assays. As a group, these plasmids were designated p(core)PUC19. In vitro transcription templates were constructed by inserting the above promoter fragments between the EcoRI and blunt-ended Sac1 sites of the G-less cassette template (Sawadogo and Roeder, 1985). The construction of pAML(2OO) has been described elsewhere (Buratowski et al., 1988). To obtain the GATA-1-expressing plasmid pETHisGATA-1, the cDNA-containing XhoI fragment of PXMcGATA (Martin and Orkin, 1990) was blunt ended and inserted into the blunt-ended NcoI site of PET-15b (Novagen). HeLa S3 (ATCC CCL 2.2) and HEL (ATCC TIB 180) cells were grown, and nuclear extracts were prepared as previously described (Dignam et al., 1983;Shapiro et al., 1988).
Purification of Basal Panscription Factors and GATA Protein-Recombinant mouse GATA-1 was expressed in Escherichia coli (BL21(DE3)) carrying the plasmid pETHisGATA-1. An overnight 5-ml culture was diluted into 500 ml of LB medium with ampicillin and grown at 37 "C. When theAsoo reached 0.5-0.7,0.5 m~ isopropyl-l-thio-P-D-galactopyranoside was added and the culture was incubated for an additional 2 h. The bacteria were then centrifuged and the pellet was resuspended in 10 ml of binding buffer (8 M urea, 0.1 M NaHPO,, Tris-HCl (pH 8.0), and 20 m~ '2-mercaptoethanol). The bacteria were sonicated and then incubated at room temperature for 1 h. Following centrifugation, the supernatant was passed through a 45-pm filter and then applied to a 2-ml nickel-NTAcolumn that was preequilibrated with binding buffer. The column was washed with 10 ml ofbinding buffer (pH 6.3), and GATA-1 was then eluted with binding buffer (pH 4.5). Fractions containing GATA-1 were identified by their ability to specifically bind to a probe containing a GATA motif in a mobility shift assay. Western blot assays employing SDS-polyacrylamide gel electrophoresis revealed two GATA-1 bands (48 and 43 m a ) , which were responsible for greater than 50% of the Coomassie-stainable protein. Recombinant GATA-1 was renatured by dialyzing the fractions against buffer D (20 m~ Hepes (pH 7.9), 0.1% Nonidet P-40, 0.5 m~ phenylmethylsulfonyl fluoride, 2 m~ dithiothreitol, 300 m~ KCl, 10% glycerol, 100 p M ZnSO,) containing decreasing amounts of urea. The final dialysis was against buffer D without urea and with 100 m~ KCl. Histidine-tagged human GATA-2, truncated to include the zinc finger DNA-binding domains (amino acids 284-406), was kindly provided by S. Orkin. HeLa cell TFIIA was prepared by chromatography of HeLa whole cell extract using phosphocellulose and DEAE-Sepharose matrices (Samuels et al., 1982). Human TFIIB was expressed in E. coli and isolated as described previously (Ha et al., 1991). Histidine-tagged human TBP was expressed in bacteria and purified by nickel-NTA chromatography (Parvin et al., 1992). The TBP preparation was approximately 10% pure as estimated by Coomassie-stained SDS gels. The subunits of TFIIE were expressed and purified as outlined previously (Peterson et al., 1991). A HeLa cell fraction containing TFIIF, TFIIH, and TFIIE (referred t o as TFIIF(E/H)) was purified by phosphocellulose and DEAE-Sepharose chromatography, followed by gel filtration as described in a prior communication (Parvin et al., 1992). Chinese hamster ovary cell amantinresistant RNA polymerase I1 was prepared as previously outlined (Carthew et al., 1988). The modulatory 700-kDa activity was a HeLa cell fraction purified by phosphocellulose and DEAE-Sepharose chromatography and gel filtration (Parvin et al., 1992).
Electrophoretic Mobility Shift Assay-A 95-bp DNA fragment containing the -50 to +3 region of the PF4 or erythropoietin promoters and a 96-bp DNA fragment containing the -51 to +3 region of the p globin promoter were prepared by EcoRI and Hind111 digestion of p(core) PUC19. These fragments were labeled by end filling with Klenow polymerase in the presence of [32PldATP (3000 CUmmol, Du Pont NEN). Binding reactions contained 0.2 ng of probe, 50 ng of TBP, 0.5 pl of TFIIA fraction, 70 ng of recombinant TFIIB, 1.0 pl of RNA polymerase I1 fraction, 1.0 pl of TFIIF(E/H) fraction andlor 250 ng of GATA-1,50 m~ Tris-HC1 (pH 7.51, 1 m M dithiothreitol, 10% glycerol, 5 n" MgCI,, 60 m M KCl, 17 pg.h/ml poly(dG.dC), 0.15 mg/ml bovine serum albumin in a total volume of 15 p1. The reactions were assembled on ice and then incubated at 30 "C for 30 min. The DNA-protein complexes were separated by electrophoresis on a 4% nondenaturing polyacrylamide gel containing 25 m~ Tris, 200 m~ glycine, 1 n" EDTA.
Transcription products as well as DNA-protein complexes were quantitated by a phosphorimager (Molecular Dynamics, Sunnyvale, CA) with ImageQuant 2.0 software.

RESULTS
Basal Factors Bind to the GATA Box a n d Are Inhibited by GATA-1-The binding of highly purified proteins to the core promoter templates was examined in an electrophoretic mobility shift assay. The four templates used are depicted in Fig  1. GATA-1 was added to reaction mixtures at the same levels as in Fig.  1 simultaneously with basal factors (lanes 2 and 5) or following a 15-min preincubation of probe with basal factors (lanes 3 and 6 ) . All mutant probe (Fig. lb, lane 7). Thus, the preinitiation cornreactions were incubated a t 30 "C for a total of 30 min. The basal factor, plexes form on both the GATA and TATA box-containing pro-GATA-1, and supershift complexes are labeled as in Fig. 2. moters, whereas GATA-1 binds only to those core promoters that contain a GATA motif. In order to study the simultaneous interaction of basal factors and GATA-1 with the GATA motif, these proteins were added together in the binding reactions. Inclusion of GATA-1 resulted in complete inhibition of the DAB and DABPolF complexes on the erythropoietin promoter (Fig. 2, lane 6) and significant inhibition of basal factor complex formation on the PF4 promoter (Fig. 2, lane 2). In contrast, DAB complex formation on the / 3 globin and PF4T-31 promoters was minimally inhibited. Formation of the DABPolF complex on both GATA-and TATAcontaining promoters was significantly inhibited (Fig. 2, lanes  2,4,6, and 8). However, when lower concentrations of GATA-1 were included in the reaction mix, the DABPolF complex on PF4T-31 was competed less efficiently compared with the PF4, erythropoietin and p globin promoters (data not shown). The template-nonspecific inhibition of DABPolF preinitiation complex formation in the presence of GATA-1 is consistent with nonspecific inhibition of in vitro transcription by GATA-1 (see below). Although the underlying mechanism remains to be clarified, it may be the result of a protein-protein interaction between GATA-1 and one of the basal transcription factors.

G G A C T G G G C T G G C A G T G A A G A T~C G T G T C T A G T C A C A G G A G C C
The inclusion of GATA-1 in the binding reactions with PF4 probe resulted in a new slowly migrating complex (Fig. 2, asterisk). This may represent a nonspecific complex or a supershift of the DAB complex. A similar complex also formed on the TATA-containing MLP promoter (data not shown), suggesting that the slowly migrating species represents a protein-protein interaction rather than the co-binding of GATA-1 and basal factors to the PF4 promoter.
A Committed Preinitiation Complex Is Resistant to Inhibition by GATA-1-The order of addition of basal factors and GATA-1 to the binding reaction was varied to determine whether GATA-1 could disrupt a preformed preinitiation complex. When TBP, TFIIA, and GATA-1 were added to the reaction mix together, GATA-1 competed for binding with the DA complex (Fig.   3, lane 2 1. When the probe was incubated with TBP and TFIIA alone, subsequent addition of GATA-1 also resulted in inhibition of the DA complex (Fig. 3, lane 3). However, on dark exposure of this autoradiogram, the preformed DA complex was competed less efficiently (data not shown). Similarly, late addition of GATA-1 to a reaction mixture that contained TBP, TFIIA, TFIIB, RNA polymerase 11, and TFIIF(E/H) fraction also resulted in less inhibition of the DAB and DABPolF complexes (Fig. 3, lane 6).

The Basal Factors Have Lower Affinity for the GATA Box
Compared with the TATA Box-The relative affinity of TBP for GATA box-containing promoters was determined by incubating a radiolabeled probe containing the adenovirus major late promoter (MLP) core promoter with TBP and TFIIA in the presence or absence of excess unlabeled GATA box-containing probe. The MLP contains a TATAAA consensus sequence that efficiently binds the DA complex   (Fig.   4, lane 1 ). The addition of excess unlabeled core promoter DNA to the reaction resulted in varying degrees of competition of the DA complex (Fig. 4, lanes 2-16). In reactions that contained a 200 molar excess of cold probe (Fig. 4, lanes 2,5,8,11, and 141, the activity of the DA complex relative to the control (Fig. 4, lane 1) was as follows: MLP, 2%; PF4T-31, 4%; /3 globin, 17%; PF4, 30%; erythropoietin, 84%. These values provide an estimate of the relative affnity of the DA complex for the promoters: MLP > PF4T-31 > p globin > PF4 > erythropoietin. A similar spectrum of affinities of the DAB and DABPolF complexes for the core promoters was observed (data not shown).

Initiation of IFanscription from PF4 and Erythropoietin Is
Inhibited by GATA-1 and GATA-2-In order to determine the functional significance of the protein interactions described above, we measured the in vitro transcriptional activity of GATA box-containing promoters in the presence or absence of GATA protein. Recombinant GATA-1 protein has been shown to stimulate in vitro transcription from templates that contain upstream GATA motifs (Kim et al., 1990). Transcription from PF4, erythropoietin, and p globin was reconstituted with basal factors provided either as HeLa nuclear extract (Fig. 5, lanes 1 4 1 , nuclear extract from a human erythroleukemia cell line (HEL) (Fig. 5, lanes 5-81, or as a set of highly purified proteins known to be sufficient for initiation of transcription from TATAcontaining promoters (Fig. 7, lanes 1 4 ) . HeLa cells are known to contain GATA-2, while HEL cells contain both GATA-1 and GATA-2. The basal level of transcription from GATA box-containing promoters was lower as a group when compared to PF4T-31. Within the group, there was a spectrum of activity with p globin > PF4 > erythropoietin. These differences were consistent whether the source of basal factors was HeLa cell, HEL cell, or purified proteins, suggesting that a common set of basal factors is sufficient for initiation of transcription from the GATA box-containing promoters. As expected, transcription did not occur with the PF4 promoter when TBP was omitted from the reaction mix (Fig. 6, lanes 1 and 3 1. The importance of TBP in transcriptional initiation is further supported by the observation that activity in vitro correlated with the relative affinity of the DA complex for the promoter (compare Figs. 5 and 4). Finally, GATA-1 is not able to replace TBP in the initiation of transcription (Fig. 6, lane 3 ).
In the presence of GATA-1, transcription from wild-type and mutant PF4T.31 promoters is partially suppressed in reactions containing TBP or TFIID. In multiple experiments using different preparations of GATA-1, the magnitude of inhibition of transcription from the GATA-containing PF4 promoter was 2-4fold greater than the extent of suppression of transcription from the PF4T-31 promoter in reactions containing TBP (Fig. 7, lanes  1,2,5, and 6). A similar phenomenon was observed with the erythropoietin promoter as compared to the TATA box-containing MLP (Fig. 7, lanes 3 and 7). Prolonged exposure of the autoradiograph showed that transcription from the erythropoietin promoter in the presence of GATA-1 was undetectable (data not shown). In contrast, transcription from the p globin promoter was minimally inhibited by addition of GATA-1 to the reaction mix (Fig. 7, lanes 4 and 8). These differences in the extent of competition appear to be due to the degree of inhibition of basal factor complex formation by GATA-1 (see Fig. 2), which is consistent with the relative affinities of GATA-1 and basal factors for the core promoters (see Fig. 4). Thus, GATA-1 mainly inhibits initiation of transcription from GATA box-containing promoters by steric hindrance of basal factor complex formation but to a lesser extent also suppresses DABPolF preinitiation complex formation via protein-protein interactions. I t is of interest that GATA-2 seems to exhibit less template-nonspecific inhibition of transcription. The addition of GATA-2 to a reconstituted transcription reaction produces significant suppression of transcription from the PF4 promoter (Fig. 8, lanes 1 and2) but essentially no change in transcription from PF4T-31 (Fig. 8, lanes 3 and 4 ) . These variations in the inhibitory effects of GATA-1 and GATA-2 are due either to the different structures of the two proteins or the use of a slightly truncated form of GATA-2. Finally, in transcription reactions containing TFIID instead of TBP, the addition of GATA-1 resulted in strong inhibition from all test templates, while addition of GATA-2 resulted in specific inhibition of transcription from PF4 (data not shown).

DISCUSSION
In the present study, we demonstrate that GATA-1 competes  ofTBP (lanes 2 and 4 ) , or 200 ng of GATA-1 (lanes 3 and 4). The transcripts intermediate in size between the test and MLP transcripts represent nonspecific internal initiations or premature termination of transcription from the test template.  1 and 2; PF&.31r lanes 3 and 4 ) , 100 ng MLP, and basal factors as described in Fig. 7. Approximately 100 ng of GATA-2 was included in the reaction mixtures (lanes 2 and 4 ) . globin core promoters. As a result, initiation of in vitro transcription from a GATA box-containing core promoter may be inhibited by GATA protein. This mechanism of transcriptional repression may have biologic importance. For example, in transient transfections using rat bone marrow, a single point mutation of G-31 to the consensus TATAAA in the PF4 promoter results in expression in the non-megakaryocytic cells (Ravid et al., 1991b). If the increased transcription in these cell types was due merely to the introduction of a consensus TATA motif, the mutation should also have resulted in higher levels of expression in megakaryocytes. Since this was not observed (Ravid et al., 1991b), we believe the results to be consistent with an inhibitory effect of the binding of a GATA fador to the -31 GATA site. It is difficult to associate this biologic effect with a single transacting factor since more than one GATA binding protein binds to the same consensus GATA motif in vitro (Dorfinan et al., 1992). However, given the limited expression of GATA-1, we suggest that a more widely distributed GATA protein such as GATA-2 may be responsible for transcriptional repression of PF4. In support of this possibility, we show that GATA-2 also inhibits initiation of transcription from PF4 in vitro.
Steric interference of preinitiation complex formation has been described for other proteins, including the Engrailed homeodomain, the Drosophila P-element transposase, as well as the BPV-1 E2-transactivating proteins, and may constitute a general mechanism of negative gene regulation (Dostatni et al., 1991;Ohkuma et al., 1990;Kaufman and Rio, 1991). In each case, the effect is mediated by specific transcription factors that have a dual function depending on the location of its cognate site relative to the TBP-binding site. In upstream promoter regions these factors function as transcriptional activators while in the vicinity of the TATA box, they inhibit initiation of transcription by interfering with the assembly of basal factor complexes. This is the first study to identify GATA-binding proteins as members of this group of transcriptional repressors. The minimum distance between the GATA motif and the TBPbinding site that determines whether GATA-1 functions as an activator or repressor of transcription has not been established. However, we note that the TATA-less glycophorin B promoter, which contains a GATA motif a t -40, is activated by GATA-1 in a cotransfection assay (Rahuel et al., 1992). Based upon our results, we suggest that the above GATA motif must be sufflciently upstream of the TBP-binding site to preclude steric interference of preinitiation complex formation.
Hematopoietic bone marrow cells, which normally express PF4 and @ globin, also synthesize GATA-binding proteins. The mechanism by which these cells circumvent the inhibitory effect of GATA factors has not been established. It has been postulated that an adaptor protein mediates preferential binding of basal factors to the chicken @ globin promoter, allowing for appropriate expression in a GATA-1-containing environment (Fong and Emerson, 1992). However, the proposed adaptor protein appears to be TFIIA.2 We believe that formation of In a mobility shif€ assay, Fong and Emerson (1992) showed that the addition of a calf thymus DNA flow-through fraction of erythroid nuclear extract resulted in a supershift of the TBP-p globin promoter complex. Since the supershifted complex was competed poorly by GATA-1 compared with the TBP-promoter complex, the authors speculated that the flow-through fraction contained an 'adaptor" that promoted preferential binding of TBP in the presence of GATA-1 (Fong and Emerson, 1992). We repeated these experiments with HeLa cell extract and found that addition of the calf thymus DNA flow-through fraction to binding reactions resulted in a supershift of TBP-p globin promoter that comigrated with the DA complex. Subsequent addition of TFIIB to the reaction mix resulted in a more slowly migrating complex that comigrated with the DAB complex (data not shown). Finally, formation of the TBP-adaptor complex was not dependent on the presence of a GATA motif; a similar supershitt formed on promoters that contained a TATA site (Fong and Emerson, 1992; data not shown). In conclusion, we propose that the activity in the calf thymus flow-through fraction represents TFIIA. preinitiation complexes in the presence of GATA binding proteins is more likely to be mediated by interaction of basal factors with tissue-specific transcription factors bound to upstream promoter sequences. It is noteworthy that the PF4 promoter contains a GATA-binding site at -134 (data not shown). Deletion of this region results in reduced expression of a reporter gene in megakaryocytes, suggesting that the upstream GATAmotif acts as a positive regulatory element (Ravid et al., 1991b). An exciting possibility is that GATA-binding proteins serve to positively and negatively regulate PF4 expression at different stages of megakaryocyte development. In early progenitors cells, for example, GATAfactor may bind to the -31 GATA site and repress PF4 gene expression. During later stages of development, GATA protein or a co-factor may bind to upstream sites, resulting in derepression and activation of transcription.
We have provided evidence that GATA box-containing core promoters exhibit a lower basal level of in vitro transcription as compared to those with a consensus TATAAA site. It is not surprising that the GATAA(A) motif functions as a weak TATA box, since most point mutations of the consensus TATAAA site lead to reduced in vitro activity (Wobbe and Struhl, 1990;Chen and Struhl, 1988). However, in transient transfection assays, a point mutation in the chicken p globin promoter of GATAAA to TATAAA leads to loss of activity in erythroid cells (Fong and Emerson, 1992). These results demonstrate that the core promoter GATA motif is critical for efficient p globin expression, but they do not prove that this effect is mediated by GATA factors. In a similar type of experiment, a point mutation in the rat PF4 core promoter of GATAAA to TATAAA does not significantly alter activity in megakaryocytes, suggesting that the wild type PF4 GATAAA sequence functions as an efficient TATA box (Ravid et al., 1991b). The comparable activity of the GATAand mutant TATA-containing promoters also implies that binding of GATA protein to the core promoter is not essential for transcriptional activation.
Previous studies comparing the effect of various mutations in the consensus TATAAA of the yeast his3 promoter have also documented discrepancies between promoter activity in vitro and in vivo (Harbury and Struhl, 1989;Wobbe and Struhl, 1990). For example, in vitro transcription from the mutant TATTTA is 10-fold weaker as compared with the wild type TATAAA, while in vivo activity of the mutant promoter coupled to the GAL4 enhancer is comparable to wild type (Harbury and Struhl, 1989;Wobbe and Struhl, 1990). Together, these findings indicate that the weak in vitro activity of certain TATA-less core promoters, including those with a GATA motif, may be compensated by upstream promoter sequences.
In summary, we have demonstrated that the GATA box is a recognition site for basal factors and GATA-binding proteins. The GATA-binding proteins inhibit transcription from the PF4 and erythropoietin promoters by interfering with formation of preinitiation complexes and may represent an important mechanism for repressing transcription in cell types that do not express these genes. The initiation of transcription in cells that normally express these genes as well as GATA-binding proteins is likely to occur via preferential binding of basal factors to the core promoter secondary to interaction with tissue-specific transcription factors bound to upstream promoter sequences.