Binding Site Selection Research Articles

Imino proton exchange provides an 1H-NMR footprint of protein-DNA interactions: general strategy and application to the SRY HMG box.

A novel 1H nuclear magnetic resonance (NMR) strategy for "footprinting" specific protein-DNA target sites is demonstrated. Relative rates of site-specifc imino-proton exchange in the free and bound DNA duplex are determined by use of laminar-shifted shaped pulses in NOESY spectra. 2D exchange crosspeaks between imino (omega2 dimension) (omega1 dimension) resonances in principle provide site-specific probes of protein binding. Chemical exchange is distinguished from nuclear Overhauser enhancements(NOEs) to bound water by use of ROESY spectroscopy. This strategy is illustrated in 1H-NMR studies of the SRY high-mobility group (HMG) box, the Y-chromosome-encoded "master switch" for testis determination in man. In a specific complex between the protein and a 15-basepair DNA site, imino-proton exchange was observed to be damped selectively within the six basepair subsite 5'-ATTGTT, previously identified by random binding-site selection as an optimal SRY target sequence. The extent of damping correlates with sites of protein-DNA contacts in the minor groove but not with the magnitude of 1H-NMR complexation shifts. SRY binding has recently been shown to introduce significant distortions in DNA structure. The DNA is sharply bent and underwound; the minor groove is widened and major groove compressed. Our results demonstrate that despite such distortions base pairing is stably maintained. Protein binding in the DNA minor groove shields DNA imino protons from exchange with solvent.

The Monomer-Binding Orphan Receptor Rev-Erb Represses Transcription as a Dimer on a Novel Direct Repeat

Rev-Erb is an orphan nuclear receptor which binds as a monomer to the thyroid/retinoic acid receptor half-site AGGTCA flanked 5' by an A/T-rich sequence, referred to here as a Rev monomer site. Fusion of Rev-Erb to the DNA binding domain of yeast GAL4 strongly repressed basal transcription of a GAL4-luciferase reporter gene as a result of the presence of a C-terminal domain containing both the hinge and heptad repeat regions. Nevertheless, wild-type Rev-Erb did not repress basal transcription from the Rev monomer binding site. Therefore, a DNA binding site selection strategy was devised to test the hypothesis that Rev-Erb may function on a different site as a dimer. This approach identified sequences containing two Rev monomer sites arranged as direct repeats with the AGGTCA motifs separated by 2 bp (Rev-DR2). Remarkably, Rev-Erb bound as a homodimer to Rev-DR2 but not to other direct repeats or to a standard DR2 sequence. The DNA binding domain contained all of the determinants for Rev-DR2-specific homodimerization. Rev-Erb bound cooperatively as a homodimer to Rev-DR2, and this interaction was 5 to 10 times more stable than Rev-Erb monomer binding to the Rev monomer site. Functionally, Rev-Erb markedly repressed the basal activity of a variety of promoters with a strong Rev-DR2 specificity. The C terminus was required for this repression, consistent with the GAL4 results. However, the Rev-DR2 specificity did not require the C terminus in vivo, since fusion of C-terminally truncated Rev-Erb to a heterologous transactivation domain created a transcriptional activator specific for Rev-DR2. In addition to idealized Rev-DR2 sites, Rev-Erb also repressed basal as well as retinoic acid-induced transcription from a naturally occurring Rev-DR2 in the CRBPI gene. Thus, although Rev-Erb is distinguished from other thyroid/steroid receptor superfamily members by its ability to bind DNA as a monomer, it functions as a homodimer to repress transcription of genes containing a novel DR2 element.

Identification of DNA sequences that bind retinoid X receptor-1,25(OH) 2D 3-receptor heterodimers with high affinity

Vitamin D 3 receptors (VDR) bind as heterodimers with retinoid X receptors (RXR) to vitamin D response elements (VDRE) and transactivate gene expression in a 1,25(OH) 2D 3-dependent manner. These elements are tandem direct repeats (DRs) of the hexamer RGGTCA separated by three nucleotides (DR3). We determined whether this DR3 was the optimal and/or only recognition sequence, by PCR-mediated binding site selection with reticulocyte lysate-expressed hVDR and mRXRα, and a pool of random sequences. We derived a consensus binding site for RXR-VDR heterodimers, RGGTCANN R RG T TCAB , and analyzed 10 of the 45 sequences selected by EMSA, methylation interference and transfection experiments: all the sequences were specific and acted as positive VDREs; the underlined purine of the spacer interacted with the heterodimer; the mutation of the third T in the second motif to a G did not influence VDRE activity. Thus, the selectivity of vitamin D pathway involving heterodimerization rather than VDR-homodimerization is not due to internal sequence variations. Except for mouse osteopontin VDRE, the natural VDREs would be efficient, only when helped by adjacent sequences and/or transactivators other than VDR and RXR.

Selection of High-Affinity Binding Sites for Sequence-Specific, DNA Binding Proteins from Random Sequence Oligonucleotides

We describe a rapid and sensitive method to isolate sets of high-affinity binding sites for sequence-specific DNA binding proteins. The DNA binding domain of the protein is expressed in Escherichia coli as a fusion with glutathione S-transferase (GST). Binding reactions are set up with total soluble extract from induced bacteria and a double-stranded oligonucleotide for which the central 32 bp have been randomized. To ensure stringent conditions, binding is done in the presence of high levels of poly(dI:C). The GST fusion protein is recovered by the addition of glutathione-Sepharose. Following extensive washing of the Sepharose beads, the bound oligonucleotides are rescued by polymerase chain reaction amplification. The amplified material is used in the next cycle of selection and amplification. Approximately five cycles are needed to obtain a pure population of high-affinity sites, which are then cloned and sequenced. This procedure should be applicable to any sequence-specific DNA binding protein for which the cDNA is available and which can be expressed in bacteria in a functional form.

E-Box Variants Direct Formation of Distinct Complexes with the Basic Helix-Loop-Helix Protein ALF1

The murine transcription factor ALF1 belongs to the class of basic helix-loop-helix proteins specific for the NCAGNTGN-version of the E-box. Binding of homodimeric ALF1 to variants of this motif was studied by a combination of binding site selection technology and DNA modification interference analysis. The results showed that substitutions at the non-conserved positions in the E-box sequence could cause profound alterations in the patterns of specific contacts at the protein-DNA interface. Thus, both the overall extent of the binding region and the backbone phosphate contact pattern differed markedly between closely related E-boxes with similar affinities for ALF1. The identity of the base at the inner N was an important determinant of contact pattern specification. The E-box variants differed in their ability to mediate ALF1 dependent transcriptional activationin vivo. We discuss the possibility that adaptability in basic helix-loop-helix protein-DNA interactions can result in complexes with different functional properties.

41] Selection of protein binding sites from random nucleic acid sequences

This chapter describes the selection of protein binding sites from random nucleic acid sequences. A variety of strategies have been developed by which specific DNA or RNA sequences that can be bound by a protein, or by a protein complex, can be isolated by selecting these molecules in vitro from libraries of random sequences. In principle, these protocols make it possible to evaluate the capability of any protein to recognize specific nucleic acid sequences, thus providing information which can be invaluable for investigations of how that protein functions. These techniques have a number of experimental applications that can be useful for study of various proteins that have been implicated in oncogenesis. Either a biological or a biochemical selection scheme can be employed to isolate protein binding sites from a random sequence nucleic acid library, and each strategy can potentially yield different information. The power of current random sequence selection techniques derives from a central concept: that the selection procedure itself is coupled with a step in which the selected molecules are amplified by the polymerase chain reaction (PCR). The amplification step makes it possible to perform multiple sequential rounds of selection and amplification to isolate specific binding sequences.

Mutation of the Oct-1 POU-specific recognition helix leads to altered DNA binding and influences enhancement of adenovirus DNA replication.

To assess which residues of Oct-1 POU-specific (POUs) are important for DNA recognition and stimulation of adenovirus DNA replication we have mutated 10 residues of the POUs helix-turn-helix motif implicated in DNA contact. Seven of these turned out to have reduced DNA binding affinity. Of these, three alanine substituted proteins were found to have a changed specificity using a binding site selection procedure. Mutation of the first residue in the recognition helix, Gln44, to alanine led to a loss of specificity for the first two bases, TA, of the wild-type recognition site TATGC(A/T)AAT. Instead of the A, a T was selected, suggesting a new contact and a novel specificity. A change in specificity was also observed for the T45A mutant, which could bind to TATAC(A/T)AAT, a site hardly recognized by the wild-type protein. Mutation of residue Arg49 led to a relaxed specificity for three consecutive bases, TGC. This residue, which is critical for high affinity binding, is absent from the structurally homologous lambdoid helix-turn-helix motifs. Employing a reconstituted system all but two mutants could stimulate adenovirus DNA replication upon saturation. Mutation of residues Gln27 and Arg49 impairs the ability of the Oct-1 POU domain protein to enhance replication, with a concomitant loss of DNA contacts. Since the POU domain binds the precursor terminal protein-DNA polymerase complex and guides it to the origin, lack of stimulation may be caused by incorrect targetting of the DNA polymerase due to loss of specificity.

Identification of deoxyribonucleic acid sequences that bind retinoid-X receptor-gamma with high affinity

The retinoid-X receptor (RXR) family (-alpha, -beta and -gamma) forms homodimers that bind to a number of retinoid-X response elements and trans-activate gene expression in a retinoid-dependent manner. Although, the RXRs are known to bind tandem direct repeats (DR) of the hexamer, RGGTCA, separated by 1 nucleotide, it is not known whether these represent the optimal and/or only recognition sequences. We, therefore, used a nonbiased strategy to identify sequences that efficiently bound RXR gamma, an isoform preferentially expressed in cardiac and skeletal muscle tissue. We performed binding site selection with bacterially expressed RXR gamma bound to glutathione-agarose and a pool of random sequences to derive a consensus DNA-binding site for RXR gamma. We analyzed a total of 41 individually selected oligonucleotides and found that RXR gamma bound with high affinity to motifs that were accommodated by the consensus AARGRNCAAAGGTCAA/cR. We observed that the majority of the sequences that formed complexes with RXR gamma in electrophoretic mobility shift analysis were DR-1 motifs; however, DR- motifs separated by 2, 4, and 8 nucleotides and a palindrome-0 motif were also demonstrated to interact with RXR gamma. Mutagenesis of the derived sequences indicated that both RGGTCA motifs were required for high affinity binding to RXR gamma. These derived sequences conferred appropriate 9-cis- and all-trans-retinoic acid (RA) responses to a thymidine kinase promoter. Furthermore, supershift experiments with a RXR antibody verified that these sequences specifically interacted with RXR in nuclear extracts derived from C2C12 muscle cells. In conclusion, this study rigorously defines the range of DR motifs that can recognize RXR and regulate gene expression in a RA-dependent fashion. The derived consensus accommodates retinoid-X response elements that have been identified in a diverse range of genes trans-activated by 9-cis-RA via the RXR family.

Identification of deoxyribonucleic acid sequences that bind retinoid-X receptor-gamma with high affinity.

The retinoid-X receptor (RXR) family (-alpha, -beta and -gamma) forms homodimers that bind to a number of retinoid-X response elements and trans-activate gene expression in a retinoid-dependent manner. Although, the RXRs are known to bind tandem direct repeats (DR) of the hexamer, RGGTCA, separated by 1 nucleotide, it is not known whether these represent the optimal and/or only recognition sequences. We, therefore, used a nonbiased strategy to identify sequences that efficiently bound RXR gamma, an isoform preferentially expressed in cardiac and skeletal muscle tissue. We performed binding site selection with bacterially expressed RXR gamma bound to glutathione-agarose and a pool of random sequences to derive a consensus DNA-binding site for RXR gamma. We analyzed a total of 41 individually selected oligonucleotides and found that RXR gamma bound with high affinity to motifs that were accommodated by the consensus AARGRNCAAAGGTCAA/cR. We observed that the majority of the sequences that formed complexes with RXR gamma in electrophoretic mobility shift analysis were DR-1 motifs; however, DR- motifs separated by 2, 4, and 8 nucleotides and a palindrome-0 motif were also demonstrated to interact with RXR gamma. Mutagenesis of the derived sequences indicated that both RGGTCA motifs were required for high affinity binding to RXR gamma. These derived sequences conferred appropriate 9-cis- and all-trans-retinoic acid (RA) responses to a thymidine kinase promoter. Furthermore, supershift experiments with a RXR antibody verified that these sequences specifically interacted with RXR in nuclear extracts derived from C2C12 muscle cells. In conclusion, this study rigorously defines the range of DR motifs that can recognize RXR and regulate gene expression in a RA-dependent fashion. The derived consensus accommodates retinoid-X response elements that have been identified in a diverse range of genes trans-activated by 9-cis-RA via the RXR family.

Selection of DNA binding sites for zinc fingers using rationally randomized DNA reveals coded interactions.

In the preceding paper [Choo, Y. & Klug, A. (1994) Proc. Natl. Acad. Sci. USA 91, 11163-11167], we showed how selections from a library of zinc fingers displayed on phage yielded fingers able to bind to a number of DNA triplets. Here, we describe a technique to deal efficiently with the converse problem--namely, the selection of a DNA binding site for a given zinc finger. This is done by screening against libraries of DNA triplet binding sites randomized in two positions but having one base fixed in the third position. The technique is applied here to determine the specificity of fingers previously selected by phage display. We find that some of these fingers are able to specify a unique base in each position of the cognate triplet. This is further illustrated by examples of fingers which can discriminate between closely related triplets as measured by their respective equilibrium dissociation constants. Comparing the amino acid sequences of fingers which specify a particular base in a triplet, we infer that in most instances, sequence-specific binding of zinc fingers to DNA can be achieved by using a small set of amino acid-nucleotide base contacts amenable to a code.

Functional selection and characterization of DNA binding sites for trp repressor of Escherichia coli.

trp repressor of Escherichia coli controls transcription initiation in operons involved in tryptophan biosynthesis by binding to operator sequences within the regulated promoters. Naturally occurring operators are homologous over an 18-base pair region and display dyad symmetry. We have examined the sequence determinants of a repressor binding site using a functional selection/polymerase chain reaction (PCR) amplification strategy. A trp repressor affinity column was generated and used to select binding-competent DNAs from a randomized pool of synthetic double-stranded DNA. DNAs that showed tryptophan-dependent high-affinity binding were eluted by addition of the tryptophan analog beta-indole acrylic acid and amplified by PCR. Following iterative cycles of affinity chromatography and PCR, the selected DNAs were cloned and sequenced. The CTAG tetranucleotide, present in the consensus sequence of all natural operators, was found in all selected DNAs. Mapping experiments utilizing the repressor affinity column showed the CTAG motif to be a critical determinant for repressor binding. Quantitative electrophoretic mobility shift assays with purified trp repressor revealed that although some of the DNAs were bound by one repressor dimer, others were bound by two repressor dimers with cooperativity. Measured binding constants ranged from 0.035 to 0.5 nM for the selected DNAs, compared with 0.1 nM for the trp operator.

DNA-binding and transcriptional regulatory properties of hepatic leukemia factor (HLF) and the t(17;19) acute lymphoblastic leukemia chimera E2A-HLF.

The t(17;19) translocation in acute lymphoblastic leukemias results in creation of E2A-hepatic leukemia factor (HLF) chimeric proteins that contain the DNA-binding and protein dimerization domains of the basic leucine zipper (bZIP) protein HLF fused to a portion of E2A proteins with transcriptional activation properties. An in vitro binding site selection procedure was used to determine DNA sequences preferentially bound by wild-type HLF and chimeric E2A-HLF proteins isolated from various t(17;19)-bearing leukemias. All were found to selectively bind the consensus sequence 5'-GTTACGTAAT-3' with high affinity. Wild-type and chimeric HLF proteins also bound closely related sites identified previously for bZIP proteins of both the proline- and acidic amino acid-rich (PAR) and C/EBP subfamilies; however, E2A-HLF proteins were significantly less tolerant of certain deviations from the HLF consensus binding site. These differences were directly attributable to loss of an HLF ancillary DNA-binding domain in all E2A-HLF chimeras and were further exacerbated by a zipper mutation in one isolate. Both wild-type and chimeric HLF proteins displayed transcriptional activator properties in lymphoid and nonlymphoid cells on reporter genes containing HLF or C/EBP consensus binding sites. But on reporter genes with nonoptimal binding sites, their transcriptional properties diverged and E2A-HLF competitively inhibited activation by wild-type PAR proteins. These findings establish a spectrum of binding site-specific transcriptional properties for E2A-HLF which may preferentially activate expression of select subordinate genes as a homodimer and potentially antagonize expression of others through heteromeric interactions.

DNA-binding and transcriptional regulatory properties of hepatic leukemia factor (HLF) and the t(17;19) acute lymphoblastic leukemia chimera E2A-HLF.

The t(17;19) translocation in acute lymphoblastic leukemias results in creation of E2A-hepatic leukemia factor (HLF) chimeric proteins that contain the DNA-binding and protein dimerization domains of the basic leucine zipper (bZIP) protein HLF fused to a portion of E2A proteins with transcriptional activation properties. An in vitro binding site selection procedure was used to determine DNA sequences preferentially bound by wild-type HLF and chimeric E2A-HLF proteins isolated from various t(17;19)-bearing leukemias. All were found to selectively bind the consensus sequence 5'-GTTACGTAAT-3' with high affinity. Wild-type and chimeric HLF proteins also bound closely related sites identified previously for bZIP proteins of both the proline- and acidic amino acid-rich (PAR) and C/EBP subfamilies; however, E2A-HLF proteins were significantly less tolerant of certain deviations from the HLF consensus binding site. These differences were directly attributable to loss of an HLF ancillary DNA-binding domain in all E2A-HLF chimeras and were further exacerbated by a zipper mutation in one isolate. Both wild-type and chimeric HLF proteins displayed transcriptional activator properties in lymphoid and nonlymphoid cells on reporter genes containing HLF or C/EBP consensus binding sites. But on reporter genes with nonoptimal binding sites, their transcriptional properties diverged and E2A-HLF competitively inhibited activation by wild-type PAR proteins. These findings establish a spectrum of binding site-specific transcriptional properties for E2A-HLF which may preferentially activate expression of select subordinate genes as a homodimer and potentially antagonize expression of others through heteromeric interactions.

Isoform-specific amino-terminal domains dictate DNA-binding properties of ROR alpha, a novel family of orphan hormone nuclear receptors.

Three isoforms of a novel member of the steroid hormone nuclear receptor superfamily related to the retinoic acid receptors have been identified. The three isoforms, referred to as ROR alpha 1, ROR alpha 2, and ROR alpha 3, share common DNA- and putative ligand-binding domains but are characterized by distinct amino-terminal domains generated by alternative RNA processing. An exon encoding a functionally important subregion of the amino-terminal domain of the ROR alpha 2 isoform resides on the opposite strand of a cytochrome c-processed pseudogene. Binding site selection using in vitro-synthesized proteins reveals that the ROR alpha 1 and ROR alpha 2 isoforms bind DNA as monomers to hormone response elements composed of a 6-bp AT-rich sequence preceding a half-site core motif PuGGTCA (RORE). However, ROR alpha 1 and ROR alpha 2 display different binding specificities: ROR alpha 1 binds to and constitutively activates transcription from a large subset of ROREs, whereas ROR alpha 2 recognizes ROREs with strict specificity and displays weaker transcriptional activity. The differential DNA-binding activity of each isoform maps to their respective amino-terminal domains. Whereas truncation of the amino-terminal domain diminishes the ability of ROR alpha 1 to bind DNA, a similar deletion relaxes ROR alpha 2-binding specificity to that displayed by ROR alpha 1. Remarkably, transfer of the entire amino-terminal region of ROR alpha 1 or amino-terminal deletion of ROR alpha 2 confers RORE-binding specificities to heterologous receptors. These results demonstrate that the amino-terminal domain and the zinc finger region work in concert to confer high affinity and specific DNA-binding properties to the ROR isoforms and suggest a novel strategy to control DNA-binding activity of nuclear receptors.

HEN1 encodes a 20-kilodalton phosphoprotein that binds an extended E-box motif as a homodimer.

HEN1 and HEN2 encode neuron-specific polypeptides that contain the basic helix-loop-helix (bHLH) motif, a protein dimerization and DNA-binding domain common to several known transcription factors. We now describe characteristics of the HEN1 gene product that are consistent with its postulated role as a transcription factor that functions during development of the mammalian nervous system. Thus, transcription of the HEN1 gene is activated upon the induction of neural differentiation in PC12 cells by nerve growth factor. HEN1 encodes a 20-kDa polypeptide (pp20HEN1) that is phosphorylated exclusively at serine residues and forms dimeric bHLH complexes either by self-association or by heterologous interaction with the E2A gene products (E12 or E47). The resultant HEN1/HEN1 homodimers and HEN1/E2A heterodimers bind DNA in a sequence-specific manner. Moreover, a binding site selection procedure revealed that HEN1-HEN1 homodimers preferentially recognize E-box motifs represented by an 18-bp consensus sequence (GGGNCG CAGCTGCGNCCC). The E-box half-site recognized by HEN1 polypeptides (GGGNCGCAG) is distinct from those of other known bHLH proteins, suggesting that HEN1 binds, an regulates the transcription of, a unique subset of target genes during neural development.

HEN1 encodes a 20-kilodalton phosphoprotein that binds an extended E-box motif as a homodimer.

HEN1 and HEN2 encode neuron-specific polypeptides that contain the basic helix-loop-helix (bHLH) motif, a protein dimerization and DNA-binding domain common to several known transcription factors. We now describe characteristics of the HEN1 gene product that are consistent with its postulated role as a transcription factor that functions during development of the mammalian nervous system. Thus, transcription of the HEN1 gene is activated upon the induction of neural differentiation in PC12 cells by nerve growth factor. HEN1 encodes a 20-kDa polypeptide (pp20HEN1) that is phosphorylated exclusively at serine residues and forms dimeric bHLH complexes either by self-association or by heterologous interaction with the E2A gene products (E12 or E47). The resultant HEN1/HEN1 homodimers and HEN1/E2A heterodimers bind DNA in a sequence-specific manner. Moreover, a binding site selection procedure revealed that HEN1-HEN1 homodimers preferentially recognize E-box motifs represented by an 18-bp consensus sequence (GGGNCG CAGCTGCGNCCC). The E-box half-site recognized by HEN1 polypeptides (GGGNCGCAG) is distinct from those of other known bHLH proteins, suggesting that HEN1 binds, an regulates the transcription of, a unique subset of target genes during neural development.

Preferred sequences for DNA recognition by the TAL1 helix-loop-helix proteins.

Tumor-specific activation of the TAL1 gene is the most common genetic alteration seen in patients with T-cell acute lymphoblastic leukemia. The TAL1 gene products contain the basic helix-loop-helix (bHLH) domain, a protein dimerization and DNA-binding motif common to several known transcription factors. A binding-site selection procedure has now been used to evaluate the DNA recognition properties of TAL1. These studies demonstrate that TAL1 polypeptides do not have intrinsic DNA-binding activity, presumably because of their inability to form bHLH homodimers. However, TAL1 readily interacts with any of the known class A bHLH proteins (E12, E47, E2-2, and HEB) to form heterodimers that bind DNA in a sequence-specific manner. The TAL1 heterodimers preferentially recognize a subset of E-box elements (CANNTG) that can be represented by the consensus sequence AACAGATGGT. This consensus is composed of half-sites for recognition by the participating class A bHLH polypeptide (AACAG) and the TAL1 polypeptide (ATGGT). TAL1 heterodimers with DNA-binding activity are readily detected in nuclear extracts of Jurkat, a leukemic cell line derived from a patient with T-cell acute lymphoblastic leukemia. Hence, TAL1 is likely to bind and regulate the transcription of a unique subset of subordinate target genes, some of which may mediate the malignant function of TAL1 during T-cell leukemogenesis.

Preferred sequences for DNA recognition by the TAL1 helix-loop-helix proteins.

Tumor-specific activation of the TAL1 gene is the most common genetic alteration seen in patients with T-cell acute lymphoblastic leukemia. The TAL1 gene products contain the basic helix-loop-helix (bHLH) domain, a protein dimerization and DNA-binding motif common to several known transcription factors. A binding-site selection procedure has now been used to evaluate the DNA recognition properties of TAL1. These studies demonstrate that TAL1 polypeptides do not have intrinsic DNA-binding activity, presumably because of their inability to form bHLH homodimers. However, TAL1 readily interacts with any of the known class A bHLH proteins (E12, E47, E2-2, and HEB) to form heterodimers that bind DNA in a sequence-specific manner. The TAL1 heterodimers preferentially recognize a subset of E-box elements (CANNTG) that can be represented by the consensus sequence AACAGATGGT. This consensus is composed of half-sites for recognition by the participating class A bHLH polypeptide (AACAG) and the TAL1 polypeptide (ATGGT). TAL1 heterodimers with DNA-binding activity are readily detected in nuclear extracts of Jurkat, a leukemic cell line derived from a patient with T-cell acute lymphoblastic leukemia. Hence, TAL1 is likely to bind and regulate the transcription of a unique subset of subordinate target genes, some of which may mediate the malignant function of TAL1 during T-cell leukemogenesis.

A new bipartite DNA-binding domain: cooperative interaction between the cut repeat and homeo domain of the cut homeo proteins.

The recently cloned Clox (Cut-like homeo box) and CDP (CCAAT displacement protein), two mammalian counterparts of the Drosophila Cut homeo protein, correspond to alternatively spliced products of the same gene (mClox, for mammalian Cut-like homeo box). Although these proteins reportedly bind to apparently unrelated DNA sequences, we show by in vitro selection of optimal binding sites that both Clox and CDP have the same preferred DNA-binding specificity. The palindromic consensus target sequence, 5'-(t/a)(a/t)tATCGATTAt(t/c)(t/g)(t/a)-3', contains a bona fide homeo domain binding motif (ATTA). In addition, 37% of the in vitro-selected sequences have a CCAAT box, the canonical target for members of the family of CCAAT-binding factors. A characteristic feature of the cut homeo proteins is the presence of three evolutionarily conserved 73-amino-acid repeats of unknown function, the so-called cut repeats. We present evidence that the cut repeat II binds to mClox consensus targets independently of the DNA-binding activity of the homeo domain. In vitro selection of binding sites shows that the optimal targets for the cut repeat II contain one or more CCAAT boxes and, like the homeo domain, an ATTA core. These results indicate that the DNA-binding activity of the second cut repeat can account for the suggested role of CDP mClox as CCAAT displacement protein, a putative repressor of gene expression. We also report that the mClox homeo domain and cut repeat II interact in vitro in the absence of DNA. This interaction, which greatly enhances the DNA-binding activity of the binary complex, is specific to the cut homeo proteins. No cooperativity was observed between the cut repeat II and the homeo domains of Oct-1 and Gtx. Furthermore, the Drosophila cut repeat II, which does not appear to bind to DNA, also enhances the DNA-binding activity of the mClox homeo domain. Thus, the bifunctional cut repeat II, which defines a new family of bipartite DNA-binding proteins, is likely to play an important role in the function of the cut homeo proteins.

Cooperative binding of Ets-1 and core binding factor to DNA.

Two phorbol ester-inducible elements (beta E2 and beta E3) within the human T-cell receptor beta gene enhancer each contain consensus binding sites for the Ets and core binding factor (CBF) transcription factor families. Recombinant Ets-1 and purified CBF bound individually to beta E2 and beta E3, in which the Ets and core sites are directly adjacent. In this report, we show that CBF and Ets-1 bind together to beta E2 and beta E3 and that Ets-1-CBF-DNA complexes are favored over the binding of either protein alone to beta E2. Formation of Ets-1-CBF-DNA complexes increased the affinity of Ets-1-DNA interactions and decreased the rate of dissociation of CBF from DNA. Ets-1-CBF-DNA complexes were not observed when either the Ets or core site was mutated. The spatial requirements for the cooperative interaction of Ets-1 and CBF were analyzed by oligonucleotide mutagenesis and binding site selection experiments. Core and Ets sites were coselected, and there appeared to be little constraint on the relative orientation and spacing of the two sites. These results demonstrate that CBF and Ets-1 form a high-affinity DNA-binding complex when both of their cognate sites are present and that the relative spacing and orientation of the two sites are unimportant. Ets and core sites are found in several T-cell-specific enhancers, suggesting that this interaction is of general importance in T-cell-specific transcription.