Competitive Electrophoretic Mobility Shift Assays Research Articles

Demonstration of an E-box and its CNS-related binding factors for transcriptional regulation of the mouse type 1 inositol 1,4,5-trisphosphate receptor gene.

The type 1 inositol 1,4,5-trisphosphate receptor (IP3R1) is expressed abundantly in the CNS, such as in cerebellar Purkinje cells and the hippocampus. We established a tissue-specific cell-free transcription system and studied regulatory properties of the 5' upstream region of the IP3R1 gene by use of this system. Deletion analyses of the promoter revealed several cis elements that function significantly in brain nuclear extracts. Among those elements, sequences from -398 to -295 showed the most predominant cerebellum-specific positive function. Footprint analyses demonstrated a factor-binding region from -334 to -318, termed box-I, that contained an E-box consensus sequence. Electrophoretic mobility shift assay revealed CNS-related basic helix-loop-helix proteins for the box-I. Mutational studies using the function assay and competitive electrophoretic mobility shift assays demonstrated a good correlation between the box-I-binding factors and the activated transcription. Box-I-binding factors were present abundantly in adult mouse CNS, whereas their existence was restricted in embryonic and nonneural tissues. Transient chloramphenicol acetyltransferase assay for the IP3R1 promoter revealed the requirement of box-I in Neuro2a neuroblastoma cells. In the postnatal CNS, multiple basic helix-loop-helix factors are expressed abundantly, some of which are suggested to activate IP3R1 gene expression in the mammalian CNS.

A GT Box Element Is Essential for Basal and Cyclic Adenosine 3',5'-Monophosphate Regulation of the Human Surfactant Protein A2 Gene in Alveolar Type II Cells: Evidence for the Binding of Lung Nuclear Factors Distinct from Sp1

The gene encoding surfactant protein-A (SP-A) is developmentally regulated in type II cells of the fetal lung. In humans there are two SP-A genes, SP-A1 and SP-A2. The SP-A2 gene is more highly regulated by cAMP and during fetal development than SP-A1. In earlier studies we determined that 296 bp of sequence flanking the 5'-end of the SP-A2 gene is sufficient to mediate high basal and cAMP-inducible reporter gene expression in primary cultures of transfected type II cells, suggesting that this region contains important cis-acting elements involved in tissue-specific and hormonal regulation of SP-A2 promoter activity. We also observed that mutagenesis of a cAMP response element (CRE)-like sequence at -242 bp (CRE(SP-A2)) greatly reduced basal and cAMP-stimulated expression in transfected type II cells. In the present study, we identified a GT box (GGGGTGGGG) at -61 bp of SP-A2 5'-flanking sequence that is highly conserved among the SP-A genes of different species. In type II cell transfection studies, we found that mutagenesis of the GT box of SP-A2 markedly reduced basal and abolished cAMP-induced reporter gene expression. Thus, CRE(SP-A2) and the GT box cooperatively interact to mediate basal and cAMP induction of SP-A2 promoter activity in type II cells. By electrophoretic mobility shift assays (EMSA), it was observed that nuclear proteins isolated from primary cultures of type II cells bound the GT box as five specific complexes. By contrast, nuclear proteins isolated from lung fibroblasts displayed notably reduced binding activity. Competition and supershift EMSA indicate that the ubiquitously expressed transcription factor Sp1, a GC box-binding protein of approximately 100 kDa, is a component of the complex of proteins that bind the GT box of SP-A2. The finding that only two of the five GT box-binding complexes were supershifted by incubation with Sp1 antibody suggests that a factor(s) in type II cell nuclear extracts that is distinct from Sp1 also interacts with the GT box. By UV cross-linking and SDS-PAGE/EMSA analysis, we have identified a approximately 55-kDa GT box-binding factor in type II cell nuclear proteins that preferentially binds the GT box of SP-A2 over the consensus Sp1 GC box sequence. This 55-kDa factor was able to bind the GT box independently of Sp1.

Purification and Characterization of a Rat Liver Protein That Recognizes CCAAT-Homologous Sequences of the Metallothionein Promoter andtrans-Activates This Promoter

C′BP-1, a protein that binds to the MRE-c′ region (-135 to -110) of the mouse metallothionein-I (MT-I) gene in metal-independent manner, was purified from rat liver nuclear extract by ion exchange and affinity chromatography. Analysis by SDS–PAGE, UV crosslinking, and glycerol gradient sedimentation, taken together, showed that C′BP-1 is a dimer of the 34-kDa polypeptides. Affinity-purified C′BP-1 could significantly stimulate transcription from mouse MT-I gene promoter. DNase I footprinting with the purified protein identified two binding sites for C′BP-1 located at positions -135 to -100 and -210 to -175 with respect to the start site of MT-I gene transcription. Both C′BP-1 binding sequences were found to contain imperfect dyad of the CCAAT homology. C′BP-1 was shown to make critical contacts with the CCAAT homology by methylation interference analysis and competition electrophoretic mobility shift assay with mutants harboring alterations in the CCAAT homology. An antibody that specifically recognizes C/EBPδ partially supershifted C′BP-1/MRE-c′ complex, suggesting that C′BP-1 is identical to C/EBPδ or is closely related to C/EBPδ.

A functional Sp1 binding site is essential for the activity of the adult liver-specific human insulin-like growth factor II promoter.

The human gene encoding insulin-like growth factor II contains four promoters (P1-P4) that are differentially activated in various tissues during development. Expression of insulin-like growth factor II in adult liver tissue is directed by P1, which is activated by liver-enriched members of the CCAAT/enhancer binding protein family of transcription factors. In the present report we show that the region around -48 relative to the transcription start site contains a high affinity Sp1 binding site. This was demonstrated by electrophoretic mobility shift assays using nuclear extracts from Hep3B hepatoma cells and with specific antibodies directed against Sp1. Competition electrophoretic mobility shift assays revealed that the Sp1 binding site of P1 and a consensus Sp1 binding site bind Sp1 with comparable efficiencies. Mutation of the Sp1 binding site results in an 85% decrease in P1 promoter activity in transient transfection assays using two different cell lines, COS-7 and Hep3B. Investigation of P1 mutants in which the spacing of the Sp1 binding site and the transcription start site was increased showed that the role of the Sp1 binding site in regulation of P1 is position dependent. Interestingly, the Sp1-responsive element cannot be exchanged by a functional TATA box. Activation of P1 by transactivators CCAAT/enhancer binding protein-beta and hepatocyte nuclear factor-3beta is strongly impaired after mutation of the Sp1 binding site. These results demonstrate that the specific presence of a binding site for the ubiquitously expressed transcription factor Sp1 is of eminent importance for efficient activation of P1 by liver-enriched transactivators.

Analyses of cis-acting and trans-acting elements that are crucial to sustain pregnancy-specific glycoprotein gene expression in different cell types.

Pregnancy-specific beta 1 glycoprotein genes (PSG) are mainly expressed during human placental development, though their expression has been reported in other normal and pathological tissues, e.g. hydatidiform mole (HM), of distinct origins. However, the molecular components implicated in the regulation of PSG are not well understood. To identify some of the regulatory elements involved in the transcriptional control of PSG expression, the DNA-protein interactions and the basal activities of the TATA-box-less PSG5 promoter were determined in different tissues and cell types. In DNAse-I protection assays, DNA-binding proteins from human term placenta (HTP) protected a region of 27 bp located from nucleotides --150 to --124, overlapping the farthest 5' upstream cap site and resembling an initiator-like element. In electrophoretic mobility shift assays (EMSA), three complexes were detected using nuclear extracts from HTP and an oligonucleotide containing the 27-bp motif. In situ ultraviolet crosslinking analysis of the specific complexes revealed that two proteins of 78.0 kDa and 53.0 kDa are involved in such interactions, in accordance with the bands of 80.0 kDa and 57.5 kDa observed by Southwestern blotting. Competitive EMSA using mutant oligonucleotides with the substitution of 5'ACCCAT3' by 5'GATATC3' within the 27-bp motif revealed that this sequence is fundamental for the formation of the specific DNA-protein complexes. We show in transient transfection experiments performed in HeLa, COS-7 and JEG-3 cells, that such mutation completely abolished the transcriptional activity of the PSG5 promoter, independently of the cell type. Moreover, this mutation disrupted the formation of the specific DNA-protein complexes which were essentially the same as those displayed by HTP. We also determined the binding activities of nucleoproteins derived from placental tissues in earlier developmental and pathological stages, i.e. first trimester placenta (1-TRIM) and HM, respectively, showing that the DNA-binding patterns were different from each other and distinct from those elicited by HTP. Our results indicate that the cis-acting and trans-acting elements analyzed are indispensable to support PSG5 promoter activity in cell lines which do or do not produce PSG. In addition, these elements appear to play a role in the mechanisms involved in PSG basal expression during placental development and differentiation.

Characterization of the cyclic adenosine 3',5'-monophosphate response element of the rabbit surfactant protein-A gene: evidence for transactivators distinct from CREB/ATF family members.

Surfactant protein-A (SP-A) gene transcription in fetal lung explants is stimulated by factors that increase intracellular cAMP. In transfected type II cells, expression of fusion genes containing 991 bp of 5'-flanking DNA from the rabbit SP-A gene linked to the human GH gene as reporter is stimulated more than 20-fold by cAMP. Mutagenesis of a putative cAMP responsive element (CRE) located -261 bp upstream of the SP-A transcription initiation site to a sequence known not to bind the transcription factor CRE-binding protein (CREB) caused a marked decrease in basal and cAMP-inducible reporter gene expression. This element, termed CREsp-a (TGACCTCA), differs by one nucleotide from a palindromic CRE (CREpal, TGACGTCA), which is known to bind CREB as a homodimer. In the present study, we found that mutagenesis of CREsp-a to CREpal also caused a marked decrease in basal and cAMP-induced fusion gene expression. The findings of competitive electrophoretic mobility shift assays (EMSA) using fetal rabbit lung nuclear extracts suggest that different protein complexes bind CREsp-a and CREpal. By UV cross-linking analysis, an approximately 43-kilodalton protein complex was found to interact both with CREsp-a and CREpal; however, purified CREB was ineffective in binding CREsp-a but did bind CREpal. In EMSA using fetal rabbit lung nuclear proteins, antibodies directed against CREB, CRE modulator (CREM), and activating transcription factor-1 (ATF-1) failed to supershift the complex of proteins bound to CREsp-a; whereas, a supershift was evident using CREpal as a probe. Moreover, in competition EMSA using radiolabeled CREsp-a and fetal rabbit lung nuclear proteins, a purified basic leucine zipper (bLZ) polypeptide failed to compete for binding. By contrast, the bLZ polypeptide competed effectively with CREpal for lung nuclear protein binding. This finding suggests that leucine zipper transcription factors do not bind CREsp-a. Additionally, expression of a CREsp-a:HIS3 fusion gene in yeast was unaffected either by CREB or bLZ polypeptides fused to the GAL4 activation domain. By contrast, HIS3 expression was markedly induced both by CREB and bLZ fusion proteins in a CREpal:HIS3 yeast strain. By competition EMSA using mutagenized CREsp-a oligonucleotides, the critical protein-binding nucleotides in CREsp-a were found to constitute a hexameric element, TGACCT, which corresponds to a binding site for members of the steroid receptor superfamily. Since the TGACCT motif is present in the SP-A gene as a single site, we propose that a unique orphan member of the steroid receptor superfamily may bind to this element.

Identification of an Essential Cis-Element Near the Transcription Start Site for Transcriptional Activation of the Proliferating Cell Nuclear Antigen Gene

Interleukin 2 (IL-2) stimulates T lymphocyte proliferation and induces the expression of proliferating cell nuclear antigen (PCNA), a processivity factor for DNA polymerase delta. Previously, deletion analysis suggested cis-element(s) in the proximal region of the PCNA promoter (-40 to +143) are required for IL-2 induction in cloned T lymphocytes. The sequence 5'-TTGCGGGC-3' located at +10 to +17 is similar to the E2F consensus binding site and is required for optimal PCNA promoter activity. In IL-2-stimulated T cells, nuclear proteins are induced to bind to this sequence as demonstrated using electrophoretic mobility shift assay (EMSA), competition EMSA, and methylation interference analysis. A 180-kDa polypeptide was detected by UV cross-linking to bind specifically to the PCNA E2F-like sequence. Our data indicate that the protein bound to the PCNA E2F-like site is not one of the transcription factor E2F proteins. Our results demonstrate that the E2F-like sequence and the protein(s) binding to it are required for optimal PCNA promoter activity and IL-2 induction of PCNA expression.

Transcriptional control of muscle plasticity: Differential regulation of troponin I genes by electrical activity

Plasticity of the skeletal muscle phenotype can result from the selective repression and activation of gene expression in response to innervation patterns. Motoneurons, eliciting different patterns of depolarization, regulate the contractile properties of the myofibers they innervate by selectively activating expression of genes encoding fiber-type-specific (fast vs. slow) contractile proteins. We have analyzed the regulation of the troponin I slow (Tnls) and fast (Tnlf) genes as a model to study the molecular mechanisms regulating fiber-type plasticity. We found that expression of the two Tnl isoforms is downregulated by denervation. Moreover, Tnl expression is upregulated by specific patterns of electrical activity (10 Hz vs. 100 Hz) used to depolarize muscle. We previously isolated the rat Tnls gene and demonstrated that regulatory sequences reside in its upstream region and second intron [Banerjee-Basu S, Buonanno A (1993), Mol Cell Biol 12:5024–5032]. Using transgenic mice, we show that the upstream region of the Tnls gene extending from -949 to +50 is sufficient to confer transcription specifically in slow-twitch muscles. Serial deletions of the Tnls upstream and intronic regions were generated in a CAT reporter vector to delineate transcriptional regulatory elements in transiently transfected Sol8 myotubes. Sequences necessary to confer the highest levels of Tnls transcription mapped to the upstream region between -0.95 and -0.72 kb, and to a 56 bp sequence located in the second intron. Comparison of the rat sequence between -0.95 and -0.72 to the human Tnls gene identified a highly homologous region of 128 bp that we named the Tnl SURE (slow upstream regulatory element). Alignment of these two SURE sequences with the quail Tnl fast intronic regulatory element identified common motifs, namely, two A/T-rich sequences (A/T1 and A/T2) with homology to homeotic protein and MEF2 binding sites, a CACC box, an E box, and a novel motif (GCAGGCA) that we denoted the CAGG box. Mutation of either the A/T2 site, E box, or CAGG box practically abolish the SURE function in transfected myotubes; mutation of the A/T1 and CACC sites has a lesser effect. Using competitive electrophoretic mobility shift assays with nuclear extracts derived from Sol8 myotubes, we demonstrate specific binding to these motifs. The A/T1 and A/T2 sites are shown to form different complexes. The A/T2 site, which bares extensive homology to a MEF2 site, forms complexes that are super shifted by MEF2A antisera and that are competed by a consensus MEF2 site present in the MCK enhancer. Our results demonstrate that the linear arrangement of DNA sequence motifs is conserved in the regulatory elements of the Tnl slow and fast genes and suggest that the interaction of multiple protein-DNA complexes are necessary for enhancer function. © 1996 Wiley-Liss, Inc. 1 This article is a US Government work and, as such, is in the public domain in the United States of America.

Negative regulation of the major human AP-endonuclease, a multifunctional protein.

Abasic sites in DNA are generated either spontaneously or after removal of altered bases during the base excision repair process. These as well as 3' damaged ends of DNA at single-strand breaks induced by reactive oxygen species are repaired by AP-endonucleases. The major human AP-endonuclease (named APE-1) has two unrelated activities. It may function as an activator of c-Fos and c-Jun transcription factors and as a repressor of the parathyroid hormone (PTH) gene by binding to the negative Ca(2+)-response elements (nCaRE) in its promoter. Preliminary studies indicate that the h-APE-1 gene is highly regulated. Analysis of its promoter activity by transient expression of the luciferase reporter gene in human, HeLa and TK6 cells suggested the presence of a negative regulatory element in the promoter. Two nCaRE-like sequences were identified in the promoter segment responsible for inhibiting reporter gene expression. Competitive electrophoretic mobility shift assay with HeLa nuclear extract indicated that the nCaRE sequences of the APE-1 and PTH genes are recognized by the APE-1 polypeptide. These results suggest that the APE-1 gene may be down-regulated by its own product.

Human herpesvirus 6B origin-binding protein: DNA-binding domain and consensus binding sequence.

We previously demonstrated by a DNA-binding assay that the human herpesvirus 6B (HHV-6B) replication origin has a structure similar to those of alphaherpesviruses, although the HHV-6B and herpes simplex virus type 1 (HSV-1) origin-binding proteins (OBPs) and origins are not interchangeable. Here we describe additional properties of the interaction between HHV-6B OBP and the HHV-6B origin. Competitive electrophoretic mobility shift assays (EMSAs) with DNA duplexes containing single-base alterations allowed deduction of a consensus DNA sequence for HHV-6B-specific OBP binding, YGWYCWCCY, where Y is T or C and W is T or A, while that for HSV-1-specific binding was reported to be YGYTCGCACT. By EMSA, the HHV-6B OBP DNA-binding domain was mapped to a segment containing amino acids 482 to 770. However, in Southwestern (protein-DNA) blotting, the region sufficient for the DNA binding encompassed only amino acids 657 to 770. Similarly, Southwestern blotting showed that amino acids 689 to 851 of HSV-1 OBP had HSV-1 origin-binding activity, although this region was insufficient for origin binding in the EMSA. Although the longer DNA-binding domains identified by EMSA have marginal overall homology among HHV-6B and alphaherpesvirus OBP homologs, the smaller regions sufficient for the binding observed by Southwestern blotting have significant similarity. From these results, we propose a hypothesis that the DNA-binding domain of herpesvirus OBPs consists of two subdomains, one containing a conserved motif that contacts DNA directly, and another, less well conserved, that may modulate either the conformation or accessibility of the binding domain.

The RNA Polymerase I Promoter-activating Factor CPBF Is Functionally and Immunologically Related to the Basic Helix-Loop-Helix-Zipper DNA-binding Protein USF

Analysis of the core promoter sequence of the mammalian ribosomal RNA genes revealed an E-box-like sequence (CACGCTG) to which the upstream stimulatory factor (USF) binds. Because the core promoter binding factor (CPBF) binds specifically to the core promoter sequence of the ribosomal RNA gene (Liu, Z., and Jacob, S. T. (1994) J. Biol. Chem. 269, 16618-16626) and resembles USF in some respects, we explored the potential relationship between USF and CPBF. Competition electrophoretic mobility shift assay using labeled core promoter probe and several unlabeled competitor oligonucleotides showed that USF can indeed bind to the core promoter and that only those oligonucleotides with an E-box sequence could compete in the promoter-protein complex formation characteristic of CPBF. This complex formation was thermostable, a unique property of USF. Furthermore, antibodies raised against USF cross-reacted with the 44-kDa component of rat CPBF. To prove further the relationship between CPBF and USF, we examined the effects of the unlabeled USF and ribosomal gene core promoter oligonucleotides on polymerase (pol) I and pol II transcription. Both oligonucleotides inhibited rDNA transcription as well as transcription from the adenovirus major late promoter. Only the oligonucleotides that contain the E-box sequence competed in the transcription. These data indicate that the promoter sequence of mammalian ribosomal RNA gene contains an USF-binding site, that the 44-kDa polypeptide of CPBF is related to the 44/43-kDa polypeptide of human USF, and that USF or USF-related protein can transactivate both pol I and pol II promoters.

The c-myc promoter binding protein (MBP-1) and TBP bind simultaneously in the minor groove of the c-myc P2 promoter.

The c-myc promoter binding protein (MBP-1) is a DNA binding protein which negatively regulates the expression of the human c-myc gene. MBP-1 binds to a sequence which overlaps the binding site for the general transcription factor TBP, within the c-myc P2 promoter region. Since TBP binds in the minor groove, MBP-1 might inhibit c-myc transcription by preventing the formation of a functional preinitiation complex. In support of this hypothesis, we have demonstrated that MPB-1 is a minor groove binding protein. In order to characterize MBP-1 binding, we substituted A-T base pairs in the MBP-1 binding site with I-C base pairs, which changes the major groove surface without altering the minor groove surface. This substitution did not inhibit the sequence-specific binding of MBP-1 and TBP. On the other hand, G-C to I-C substitution within the MBP-1 binding site alters the minor groove and prevents MBP-1 binding. Competitive electrophoretic mobility shift assays were used to show that berenil, distamycin, and mithramycin, all of which bind in the minor groove, compete with MBP-1 for binding to the MPB-1 binding site. These minor groove binding ligands also effectively inhibit the simultaneous DNA binding activity of both MBP-1 and TBP. We conclude that both MBP-1 and TBP can bind simultaneously in the minor groove of the TATA motif on the c-myc P2 promoter. This suggests that MBP-1 may negatively regulate c-myc gene expression by preventing efficient transcription initiation.

Characterization of a cis-acting regulatory element which silences expression of the class II-A beta gene in epithelium.

Class II major histocompatibility complex (MHC) genes encode for alpha/beta chain pairs that are constitutively expressed principally on mature B cells and dendritic cells in mice. These gene products are easily induced on macrophages with cytokines, and may also aberrantly appear on the surface of epithelium during immune injury. The appearance of class II determinants in parenchymal tissue potentially renders these somatic cells capable of antigen presentation to circulating CD4+ T lymphocytes, and their absence may be protective for normal tissues expressing self-antigens. The low surface class II expression observed on parenchymal cells generally correlates with low levels of mRNA, suggesting that transcription rate is a major element in class II regulation. To understand the transcriptional mechanism maintaining low basal surface expression of class II in somatic cells, we transiently transfected mini-gene reporter constructs to study the regulation of the murine A beta promoter in a cultured renal epithelial cell line. We describe here a negative cis-acting regulatory region located between -552 and -489 bp upstream of the A beta cap site that silences the transcriptional activity of the A beta promoter in epithelial cells in an orientation-dependent manner, and is also able to silence a heterologous promoter. This region is not active in class II-expressing B cells (BAL-17) in culture, but is functional in two other murine class II-negative cell lines, fibroblasts and thymoma T cells. Using competition electrophoretic mobility shift assays, we have localized the core protein binding site within this region to an 8-10-bp response element, designated A beta NRE, at -543 to -534 bp. A nuclear extract from BAL-17 cells does not bind to this element. Mutation of this site abrogates the transcriptional silencing activity of the region. We conclude that the transcription of class II-A beta in parenchymal cells, and some lymphocytes, can be actively repressed by an upstream silencing element.

Characterization of a protein that interacts with the rat ribosomal gene promoter and modulates RNA polymerase I transcription

An RNA polymerase I core promoter binding factor (CPBF) was purified to apparent homogeneity from rat adenocarcinoma ascites cells by chromatographic fractionation on a series of columns including an oligodeoxynucleotide affinity column. The final preparation contained two polypeptides with molecular masses of 44,000 and 39,000 daltons. The binding of the factor to the promoter was demonstrated by Southwestern blotting, UV cross-linking and electrophoretic mobility shift assay. The specificity of its binding to the core promoter was confirmed by competitive electrophoretic mobility shift assay using several unlabeled oligo probes in the assay. The addition of increasing amounts of purified CPBF to the in vitro transcription reaction that contains a limiting quantity of the factor resulted in dramatic stimulation of RNA polymerase I (pol I) transcription of rat ribosomal RNA gene. The transcription stimulatory activity associated with the purified CPBF fractions co-purified with the core promoter binding activity in an electrophoretic mobility shift assay. Finally, in a reconstitution transcription system which is devoid of the factor and is incapable of ribosomal gene transcription, purified CPBF could trans-activate the pol I promoter. These data indicate that CPBF is a novel pol I promoter binding factor required for ribosomal gene transcription.

Regulatory sequences and protein-binding sites involved in the expression of the rat plasminogen activator inhibitor-1 gene.

Plasminogen activator inhibitor-1 (PAI-1) is a serine protease inhibitor that inhibits both tissue-type and urokinase-type plasminogen activators. Expression of PAI-1 is regulated by growth factors, cytokines, and hormones. To determine the molecular mechanisms involved in the basal expression of the rat PAI-1 gene, we have analyzed the cis-acting sequences and the trans-acting factors involved in the transcription of this gene in the HTC rat hepatoma cell line. DNase I protection analyses revealed eight regions within the first 764 base pairs of 5'-flanking sequence that interact specifically with HTC cell nuclear proteins. The proteins that bind to five of the eight footprinted sites were identified as PEA3-, Sp1-, and CTF/NF-1-like proteins using competition electrophoretic mobility shift assays. The expression of fusion genes containing progressive 5' deletions of the rat PAI-1 promoter linked to the chloramphenicol acetyltransferase reporter gene were analyzed in transient transfection experiments in HTC cells. These studies demonstrated the Sp1 and CTF/NF-1 sites to be important for transcriptional activation. Two of the footprinted sites contain the sequence 5'-TTTGn(n)TCAAT-3' and were shown in competition electrophoretic mobility shift assays to bind the same or related protein(s). Sequences containing these sites, from -764 to -628 base pairs, and from -266 to -188 base pairs, were identified in functional studies as repressor elements of transcription.