Target-dependent effect of phosphorylation on the DNA binding activity of the TAL1/SCL oncoprotein.

Activation of the TAL1 (or SCL) gene, initially identified through its involvement by a recurrent chromosomal translocation, is the most frequent gain-of-function mutation recognized in T-cell acute lymphoblastic leukemia. The translational products of this gene contain the basic domain helix-loop-helix motif characteristic of a family of transcription factors that bind to a consensus nucleotide sequence termed the E-box. Previous work established that the TAL1 proteins are phosphorylated exclusively on serine and identified Ser122 as a substrate for the mitogen-activated protein kinase ERK-1. We provide evidence that an additional serine residue, Ser172, located in a conserved region proximal to the DNA binding domain and sharing homology with a similarly positioned sequence in the HLH oncoprotein LYL1, can be phosphorylated in vitro and in vivo by the catalytic subunit of cAMP-dependent protein kinase. Phosphorylation was found to alter TAL1 DNA binding activity in a target-dependent manner that was influenced by both the specific CANNTG E-box core motif and its flanking sequences. In contrast, the ability of TAL1 to interact with the E2A gene product E12 and its subcellular localization in transfected COS cells were unaffected by Ser172 phosphorylation. These results suggest this serine residue has a regulatory function and indicate a mechanism by which phosphorylation could affect DNA binding site discrimination.

Activation of the TAL1 (or SCL) gene, initially identified through its involvement by a recurrent chromosomal translocation, is the most frequent gain-of-function mutation recognized in T-cell acute lymphoblastic leukemia. The translational products of this gene contain the basic domain helix-loop-helix motif characteristic of a family of transcription factors that bind to a consensus nucleotide sequence termed the E-box. Previous work established that the TAL1 proteins are phosphorylated exclusively on serine and identified Ser 122 as a substrate for the mitogen-activated protein kinase ERK-1. We provide evidence that an additional serine residue, Ser 172 , located in a conserved region proximal to the DNA binding domain and sharing homology with a similarly positioned sequence in the HLH oncoprotein LYL1, can be phosphorylated in vitro and in vivo by the catalytic subunit of cAMP-dependent protein kinase. Phosphorylation was found to alter TAL1 DNA binding activity in a target-dependent manner that was influenced by both the specific CANNTG E-box core motif and its flanking sequences. In contrast, the ability of TAL1 to interact with the E2A gene product E12 and its subcellular localization in transfected COS cells were unaffected by Ser 172 phosphorylation. These results suggest this serine residue has a regulatory function and indicate a mechanism by which phosphorylation could affect DNA binding site discrimination.
The TAL1 (or SCL) gene was initially identified as a result of its involvement in a recurrent chromosomal translocation with the T-cell receptor ␣/␦ locus in T-cell acute lymphoblastic leukemia (1)(2)(3). It was recognized subsequently that its coding region could be juxtaposed to the promoter of the upstream SIL (for SCL interrupting locus) gene as a result of certain interstitial deletions inapparent cytogenetically (4 -6). In still other patients, the gene can be misexpressed in the apparent absence of chromosomal rearrangement (7). In aggregate, ectopic TAL1 expression characterizes up to 60% of individuals with T-cell acute lymphoblastic leukemia, making it the most frequent gain-of-function mutation recognized in this disorder.
TAL1 belongs to the helix-loop-helix (HLH) 1 family of transcription factors (8,9), so-named for the conformation adopted by its defining motif of two amphipathic helices with an intervening loop (10 -12). The HLH domain functions in the formation of protein hetero-and homodimers, whereas an adjacent basic region found in many of the members of this gene family mediates sequence-specific DNA binding (13)(14)(15). As with other tissue-restricted basic domain-HLH (bHLH) proteins, TAL1 binds DNA as a heterodimer with the products of the widely expressed E2A gene (16,17). These complexes recognize a motif with the hexanucleotide core sequence CANNTG termed the E-box (18,19) and function as transcriptional regulators (20).
Within this larger family, distinct subgroups of HLH genes can be recognized on the basis of sequence relatedness, expression pattern, and function. In the best characterized example, the four bHLH genes expressed in developing skeletal muscle share the ability to induce myogenic differentiation when introduced into certain cell lines (21,22) and appear capable of substituting for each other's functions in vivo (23). TAL1 shows greater than 80% sequence identity over its basic and HLH domains with two other genes, LYL1 and TAL2, also involved by chromosomal translocation in T-cell acute lymphoblastic leukemia (24).
The protein products of the TAL1 locus are expressed in a number of cell types during development, becoming essentially restricted to hematopoietic and endothelial cells postnatally (25,26). A requirement for TAL1 in embryonic hematopoiesis was recently demonstrated by targeted mutation of the murine TAL1 coding region (27,28), and evidence also points to its involvement in the terminal, erythropoietin-regulated stages of red cell production (29,30). TAL1 and LYL1 show overlapping expression in several hematopoietic lineages (31) and may, by analogy to the myogenic HLH genes, share some similarities in function.
Up to three proteins can be translated from TAL1 transcripts (32-34), with two potentially made from the full-length message (amino acids 1-330 and 26 -330, numbered according to the mouse protein). The shorter of these, whose synthesis in cells has not been rigorously demonstrated, likely arises by alternative translational initiation (35). A significantly smaller protein (amino acids 176 -330) lacking the amino-terminal half of the molecule but retaining the basic and HLH domains is translated from a series of spliced RNAs that exclude the first coding exon. This species has been detected in leukemia cell lines (33) and differentiating erythroblasts (30).
The TAL1 polypeptides are serine phosphoproteins (32,33), with phosphorylation of Ser 122 by the mitogen-activated protein kinase ERK-1 (34) found to increase the activity of a transcriptional activation domain (36). We provide evidence that an additional serine residue located proximal to the basic region can be phosphorylated in vitro and in vivo by cAMP-de-pendent protein kinase (PKA) and show that its phosphorylation alters TAL1 DNA binding activity in a target-dependent manner.

EXPERIMENTAL PROCEDURES
TAL1 Fusion Proteins-A 640-base pair murine TAL1 cDNA derived by the polymerase chain reaction (25) was subcloned into the EcoRI site of vector pMALc2 (New England Biolabs) and the encoded maltose-binding protein (MBP)-murine TAL1 fusion (MBP-TAL1 164 -330 ) expressed in the DH5␣ strain of Escherichia coli. To express the corresponding Ser 172 to Ala mutant (MBP-TAL1 164 -330(S172A) ), oligonucleotides TATGGGGCAGGCCTCCTCTTCACCCGGTTGTTG and AATTCAACAACCGGGTGAAGAGGAGGCCTGCCCCA were annealed and ligated in a trimolecular reaction to EcoRI-and BamHI-digested pMALc2 and an NdeI-BamHI murine TAL1 cDNA extending from bases 519 to 1003. For preparation of an MBP-TAL1 5-330 fusion protein, a 1.0-kilobase NotI-and BamHI-digested fragment of a full-length murine TAL1 cDNA (provided by Glenn Begley) was subcloned into EcoRI and BamHI-digested pPR1166 (provided by Paul Riggs) after blunting of the NotI site of the insert and the EcoRI site of the vector with the Klenow fragment of DNA polymerase I. Automated nucleotide sequencing of fusion constructs was carried out using dye-labeled dideoxy terminators (Applied Biosystems) to confirm that the malE and TAL1 coding regions were in the same reading frame.
MBP-TAL1 fusion proteins were purified on composite amylose-agarose beads (New England Biolabs) and eluted with maltose as described (37). Purified MBP was provided by Paul Riggs.
Site-directed Mutagenesis-A 1.0-kilobase full-length mouse TAL1 cDNA was subcloned into the BamHI site of vector pALTER-1 (Promega), and oligonucleotide-mediated mutagenesis of Ser 172 to Ala was carried out according to the supplier's instructions and verified by nucleotide sequencing. The mutated cDNA was released from pAL-TER-1 and subcloned into the BamHI site of vector pcDNA I (Invitrogen) for cell transfection.
In Vitro Phosphorylation-Affinity purified fusion proteins (500 ng) were phosphorylated by the addition of 5 units of purified PKA catalytic subunit (provided by Sharron Francis and Jackie Corbin) in a 20-l reaction volume containing 10 mM Tris-HCl, pH 7.4, 2 mM magnesium acetate, 50 M ATP, and 10 Ci of [␥-32 P]ATP (specific activity, 3000 Ci/mmol). The reaction mixture was incubated for 20 min at 30°C, and the radiolabeled proteins were immunoprecipitated with rabbit polyclonal antibody raised against a bacterial expressed TAL1 fusion protein (25), and immunoprecipitates were analyzed by SDS-polyacrylamide gel electrophoresis. For electrophoretic mobility shift assays, 2 g of protein were phosphorylated in a reaction containing 200 M unlabeled ATP and 20 units of PKA. Control reactions lacked only enzyme and were processed in parallel.
To determine the stoichiometry of phosphorylation, 1.57 g of MBP-TAL1 164 -330 fusion protein was incubated with [␥-32 P]ATP and PKA under the conditions described above and fractionated by SDS-PAGE. Protein was visualized by Coomassie Blue staining, and the radioactivity incorporated into a gel slice containing the fusion protein was quantitated in a liquid scintillation counter. The mean number of mol of phosphate incorporated per mol of protein Ϯ S.D. was determined for three independent reactions, taking into account the specific activity of the [␥-32 P]ATP and the efficiency of the counter.
Electrophoretic Mobility Shift Assay-Electrophoretic mobility shift assays (EMSA) of DNA binding activity were carried out with an aminoterminally truncated E12 protein produced by coupled transcription/ translation (TnT system, Promega) of plasmid E12R (10) in rabbit reticulocyte lysates and MBP-TAL1 fusion proteins purified from bacterial sonicates. The indicated proteins were incubated in a buffer containing 20 mM Hepes, pH 7.8, 20% glycerol, 100 mM KCl, 0.5 mM dithiothreitol, 0.2 mM EDTA, and 20 mM MgCl 2 for 10 min. In some cases as noted, anti-TAL1 antibody was added for an additional 5 min. Double-stranded oligonucleotides were labeled with 32 P by T4 polynucleotide kinase and incubated with transcription factor complexes in 30 l of binding buffer containing 100 mM Hepes, pH 7.8, 2.5 mM dithiothreitol, 100 mM KCl, 60 mM MgCl 2 , 1 mM EDTA, and 1 g of poly(dI-dC) for 20 min at room temperature. Protein-DNA complexes were resolved in Tris borate buffer on 5% polyacrylamide gels containing 3% glycerol. Dried gels were exposed to film, and laser scanning densitometry was used to quantitate the relevant retarded complexes on the resulting autoradiographs. The significance of differences in the DNA binding of complexes was determined by a two-tailed t test. Mean values Ϯ S.D. were calculated from four independent experiments.
The sequences of the upper strand of each pair of oligonucleotides used, with the core E-box sequences underlined, are as follows: E2 enhancer, CCTGCAGCAGCTGGCAGGAA; E5 enhancer, GAACCAG-AACACCTGCAGCA; preferred binding site, ACCTGAACAGATGGTC-GGCT; chimeric site with core sequences from preferred binding site and flanking sequences from E2 enhancer, CCTGCAGCAGATGGCA-GGAA; chimeric site with core sequences from E2 enhancer and flanking sequences from preferred binding site, ACCTGAACAG-CTGGTCGGCT.

Metabolic Labeling of Cells and Preparation of Cellular
Extracts-Metabolic labeling of COS cells was carried out 48 h following transfection (38) with the indicated expression vectors. Cells were rinsed once with phosphate-free modified Eagle's medium containing 10% dialyzed fetal bovine serum and incubated for 1 h at 37°C in this medium. [ 32 P]Orthophosphate (ICN Biomedicals) was added (1 mCi/ml) and incubation continued for an additional 3 h. After labeling, the cells were washed with phosphate-buffered saline and lysed directly on plates by addition of radioimmunoprecipitation buffer (150 mM NaCl, 50 mM Tris, pH 8.0, 1.0% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS) containing 1 mM phenylmethylsulfonyl fluoride, 50 g/ml aprotinin, 25 g/ml leupeptin, and 1 mM sodium vanadate. Murine erythroleukemia cells (2.5 ϫ 10 6 cells/ml) were incubated in phosphate-free modified Eagle's medium for 1 h, labeled with 1 mCi/ml [ 32 P]orthophosphate in the presence of 1 mM 8-(4-chlorophenylthio)-adenosine 3Ј:5Јcyclic monophosphate and 1 mM theophylline for 3 h, and processed as described for COS cells.
Immunoprecipitation-TAL1 protein was immunoprecipitated from nuclear extracts using affinity purified rabbit anti-TAL1 antibody as described previously (25,30). Antigen-antibody complexes were collected with protein A-Sepharose and resolved on 8% SDS-polyacrylamide gels. Proteins were transferred to polyvinylidene difluoride (PVDF) membranes and visualized by autoradiography.
Two-dimensional Phosphopeptide Mapping and Phosphoamino Acid Analysis-After immunoprecipitation and fractionation by SDS-PAGE, radiolabeled TAL1 protein was transferred to PVDF membranes and digested in situ with chymotrypsin. Proteolytic fragments were eluted and subjected to two-dimensional phosphopeptide analysis by the method of Boyle et al. (39). Digests were applied to cellulose thin layer electrophoresis plates (EM Science) and fractionated by electrophoresis in pH 1.9 buffer at 1000 V for 20 min. Dried plates were then subjected to ascending thin layer chromatography in a solvent containing 39% butanol, 30% pyridine, and 6% glacial acetic acid. Radiolabeled digestion products were visualized by autoradiography and phosphorimaging (Molecular Dynamics).
Following transfer to PVDF membranes, radiolabeled TAL1 protein was hydrolyzed for phosphoamino acid analysis in 6 M HCl at 110°C for 1 h. Hydrolysates were lyophilized, dissolved in water, lyophilized a second time to dryness, and finally dissolved in pH 1.9 electrophoresis buffer containing phosphoamino acid standards (phosphothreonine, phosphoserine, and phosphotyrosine). Samples were applied to cellulose thin layer electrophoresis plates and electrophoresed in pH 1.9 buffer in one dimension and pH 3.5 buffer in the second dimension, both at 1000 V for 20 min. Dried plates were sprayed with ninhydrin to visualize standards and subjected to autoradiography and phosphorimaging.
Protein Interaction Assay-MBP-TAL1 164 -330 fusion protein treated with or without PKA was incubated for 10 min at room temperature with a radiolabeled E12 polypeptide translated in rabbit reticulocyte lysates in the presence of [ 3 H]leucine (158 Ci/mmol) (Amersham Corp.). Composite amylose-agarose beads were added for an additional 15 min to adsorb heterodimeric complexes through the maltose-binding moiety of the TAL1 fusion protein and then washed with 20 mM Tris-HCL, pH 7.4, 0.2 M NaCl. Bound proteins were released from beads by boiling in loading buffer and resolved on 10% SDS-polyacrylamide gels. Dried gels were subjected to fluorography, and the amount of E12 bound was quantitated by laser scanning densitometry.
Immunofluorescence-COS cells cultured on glass coverslips were transfected by calcium phosphate coprecipitation (38) with an expression vector (pcDNA I) containing a full-length murine TAL1 cDNA or the corresponding Ser 172 to Ala mutant generated by site-directed mutagenesis, with or without an expression vector encoding the ␣ type catalytic subunit of PKA (40). After an additional 48 h in culture, cells were fixed in 2% formaldehyde and permeabilized with 0.1% dimethyl sulfoxide. Coverslips were rinsed in phosphate-buffered saline, incubated with affinity purified anti-TAL1 antibody at a 1:1500 dilution for 1 h, and washed. Following incubation with fluorescein-conjugated goat anti-rabbit antibody, coverslips were washed extensively, mounted on glass slides in aqueous medium (1.2% polyvinyl alcohol, 5% glycerol), and examined with a microscope equipped with epifluorescence.

TAL1 Is a Substrate for Phosphorylation by PKA in Vitro:
Identification of Ser 172 as the Site of Phosphorylation-To identify potential sites of TAL1 protein phosphorylation, bacterial expressed MBP-TAL1 fusion proteins were incubated with purified protein kinases and radiolabeled ATP in in vitro kinase assays. TAL1, but not MBP, sequences were efficiently phosphorylated by the catalytic subunit of PKA, and the site(s) of phosphorylation was localized using a series of amino-terminally truncated TAL1 fusion proteins as substrates to a region encompassing amino acids 164 -255 (not shown). Phosphoamino acid analysis indicated further that PKA-stimulated TAL1 phosphorylation occurred only on serine (not shown). By inspection, a potential PKA recognition sequence (RVKRRP-SPY) resembling, especially, that of the ␤ chain of phosphorylase kinase (41) was noted that contains Ser 172 . This sequence is completely conserved between the chicken (42), mouse (1), and human (2, 43) TAL1 proteins and shows considerable homology to one found in a similar context in another HLH oncoprotein expressed in hematopoietic cells, LYL1 (Fig. 1).
To determine specifically whether Ser 172 was the target of PKA-stimulated phosphorylation, an MBP-TAL1 fusion protein containing this site (MBP-TAL1 164 -330 ) and its corresponding Ser 172 to Ala mutant (MBP-TAL1 164 -330(S172A) ) were incubated with purified catalytic subunit of PKA in the presence of [␥-32 P]ATP. Under the conditions described, the MBP-TAL1 164 -330 fusion was phosphorylated to a stoichiometry of 0.78 Ϯ 0.17 mol of phosphate/mol of protein. The alanine substitution mutant, in contrast, showed no measurable phosphorylation by PKA (Fig. 2), demonstrating the ability of Ser 172 to function as a phosphoacceptor for this kinase in vitro.
Phosphorylation of Ser 172 Has a Target-dependent Effect on TAL1 DNA Binding Activity-Given its proximity to the basic domain, we investigated whether phosphorylation of Ser 172 would affect the binding of TAL1-containing complexes to DNA. Previous work had demonstrated sequence-specific DNA binding activity for hetero-oligomers of TAL1 with E12, E47, and HEB (18), and the ability of the MBP-TAL1 164 -330 fusion protein to bind with E12 to E-box sequences from the E2 and E5 immunoglobulin enhancers and one identified by a polymerase chain reaction-assisted site-selection assay (19) was tested by EMSA. In addition to a complex attributable to the binding of E12 homodimers, particularly prominent with the E5 probe, a less retarded complex dependent on the presence of both TAL1 and E12 protein and specifically disrupted by anti-TAL1 antibody was identified (Fig. 3). Significantly greater binding of this TAL1-E12 heterodimeric complex was found to the preferred sequence, consistent with it being a high affinity site (19) (not shown). No DNA binding was noted with either purified MBP or unprogrammed reticulocyte lysate.
The effect of PKA-mediated TAL1 phosphorylation on the DNA binding activity of TAL1-E12 complexes was further investigated using MBP-TAL1 164 -330 protein as substrate. Phosphorylation resulted in greater than 50% inhibition of binding of MBP-TAL1 164 -330 ⅐E12 complexes to the E2 and E5 enhancers, whereas it had no effect with the preferred site as probe (Figs. 4 and 5A). Site-specific mutagenesis abrogated the effect of phosphorylation on DNA binding activity concomitant with its inhibition of radiophosphate incorporation (Figs. 4 and 5A), demonstrating the action of PKA to be a direct one requiring Ser 172 as phosphoacceptor.
Although it demonstrated the same binding preferences as The dried gel was exposed to film and analyzed by autoradiography. Sizes of protein standards run in parallel are given in kDa.

FIG. 3. DNA binding activity of MBP-TAL1⅐E12 complexes.
Purified MBP-TAL1 164 -330 fusion protein was mixed with in vitro translated E12 protein and tested for DNA binding activity by EMSA using 32 P-labeled double-stranded oligonucleotides corresponding to E-box sequences in the E2 and E5 immunoglobulin enhancers and one identified as a preferred sequence for binding by a polymerase chain reaction-assisted site-selection method. Heterodimeric MBP-TAL1⅐E12 complexes denoted with arrowhead were identified by their dependence on both MBP-TAL1 fusion protein and E12 for formation and their specific abrogation with anti-TAL1 antibody (TAL1 ab). Autoradiographic exposure times varied with different probes. those reported for the native protein (19), the MBP-TAL1 fusion used in these experiments included comparatively little sequence amino-terminal to the phosphorylation site and could, as a result, have been differentially or uniquely sensitive to this effect of phosphorylation. To address this possibility, an MBP fusion containing near full-length TAL1 coding sequence (MBP-TAL1 5-330 ) was purified and tested as described. Ser 172 phosphorylation had the identical effect on this protein as on the MBP-TAL1 164 -330 fusion (not shown), indicating that PKA regulation of TAL1 DNA binding activity is not dependent on protein truncation.
The E-box probes used in these analyses each differed in both their core and flanking nucleotide sequences. To clarify the DNA sequence requirements for the effect of phosphorylation observed, chimeric oligonucleotides containing portions of the E2 and preferred binding sites were synthesized, and the extent of binding of complexes containing PKA-treated versus untreated TAL1 fusion protein was determined. Substitution of the flanking sequences from the E2 enhancer for those in the preferred site had no effect on binding, whereas replacement of the flanking sequences of the E2 E-box with those from the preferred site ablated the effect of phosphorylation (Fig. 5B). Thus, Ser 172 phosphorylation exerts a template-dependent effect on DNA binding through a mechanism involving discrimination of both the CANNTG core motif and its immediate flanking sequences.
Ser 172 Phosphorylation Has No Effect on TAL1 Protein Dimerization-Although the domain involved in protein dimerization is more distal from Ser 172 than that mediating DNA binding (Fig. 1), if phosphorylation were to alter the interaction of TAL1 with its binding partners it could, indirectly, affect DNA binding activity. To investigate this possibility, the extent of binding of [ 3 H]leucine-labeled E12 was measured in a solution interaction assay to MBP-TAL1 164 -330 fusion protein that had been incubated with and without purified catalytic subunit of PKA. No difference was detected (Fig. 6), demonstrating that dimerization of TAL1 with at least one of its known binding partners is unaffected by Ser 172 phosphorylation.
Ser 172 Phosphorylation Has No Effect on Intracellular Local-ization of TAL1 Protein-Given the importance of the basic region for nuclear uptake in addition to DNA binding (44), we also investigated whether Ser 172 phosphorylation could affect the subcellular localization of TAL1 protein. To that end, an expression vector containing a full-length murine TAL1 cDNA was transfected into COS cells with and without one encoding the catalytic subunit of PKA, and the distribution of TAL1 protein was analyzed by indirect immunofluorescence. Consistent with its location in tissue sections (25,26) and with previously published results in this cell line (25,32,44), immunoreactive TAL1 was restricted to the nuclei of transfected cells, where it appeared to be excluded from nucleoli. This pattern was not altered by coexpression of PKA catalytic subunit (Fig.  7), indicating that phosphorylation of Ser 172 , verified by phosphopeptide mapping (see below), has no effect on the protein's intracellular localization. The nuclear uptake of full-length TAL1 proteins containing either Ser 172 to Ala or Ser 172 to Glu substitutions was similarly unaffected (not shown). Ser 172 Acts as a Phosphoacceptor in Vivo-To determine whether Ser 172 could also function as a phosphoacceptor in cells, two-dimensional phosphopeptide mapping was carried FIG. 4. Effect of Ser 172 phosphorylation on TAL1 DNA binding activity. DNA-binding activities of MBP-TAL1⅐E12 heterodimeric complexes (arrowhead) were analyzed by EMSA using double-stranded oligonucleotides containing the E2, E5, and preferred E-box binding sites as probes. To test the effect of phosphorylation on binding, MBP-TAL1 164 -330 fusion protein was incubated with (TAL1 P ) and without (TAL1) purified catalytic subunit of PKA and ATP prior to incubation with in vitro translated E12 protein (E12). Similar studies were also carried out with MBP-TAL1 164 -330(S172A) protein (not shown). (open bars) proteins incubated with or without PKA were mixed with in vitro translated E12 protein and subjected to EMSA with each of the indicated binding sites. Binding was quantitated by densitometric analysis of the retarded TAL1⅐E12 complex on autoradiograms, and the relative binding for complexes containing PKA-treated versus those containing untreated TAL1 protein is expressed as a percentage for each fusion protein and binding site. For the MBP-TAL1 164 -330 fusion, these results provide a graphical representation of the data in Fig. 4. Plotted is the mean percent binding Ϯ S.D. from four independent experiments. The statistical significance of the difference in the means of adjacent conditions is denoted by the symbol, * (p Ͻ 0.001). B, MBP-TAL1 164 -330 protein incubated with or without PKA was mixed with in vitro translated E12 protein and subjected to EMSA with each of the four binding sites illustrated schematically below. Probes consisted of the E2 enhancer, the preferred binding site, a chimeric site containing the core E-box sequences from the preferred site with flanking sequences from the E2 oligonucleotide, and the reciprocal oligonucleotide containing the core E-box sequences of the E2 enhancer with flanking sequences from the preferred binding site. Binding was quantitated by densitometric analysis of the retarded TAL1⅐E12 complex on autoradiograms, and the relative binding for complexes containing PKA-treated versus those containing untreated TAL1 protein was calculated as a percentage for each binding site. Plotted is the mean percent binding Ϯ S.D. from three independent experiments. The statistical significance of the difference in the means of adjacent conditions is denoted by the symbols, * (p Ͻ 0.05) and ** (p Ͻ 0.0001). out on immunoprecipitates from metabolically labeled COS cells engineered to express full-length TAL1 protein or the corresponding Ser 172 to Ala mutant. Labeling of one major and one minor proteolytic fragment likely resulting from incomplete enzymatic digestion was markedly induced by coexpression of the catalytic subunit of PKA (not shown), and the presence of Ser 172 in these fragments was demonstrated by the specific abrogation of their labeling following substitution of alanine for this serine (Fig. 8, A-B). The finding that PKAinduced TAL1 protein phosphorylation was completely inhibited by this mutation provides further confirmation that Ser 172 is the principal site in TAL1 for phosphorylation by this kinase.
A radiolabeled peptide with mobility identical to the major COS cell fragment was noted in TAL1 chymotryptic digests from murine erythroleukemia cells incubated with the cAMP analog 8-(4-chlorophenylthio)-adenosine 3Ј:5Ј-cyclic monophosphate and the phosphodiesterase inhibitor theophylline (Fig.  8C). This result demonstrates that Ser 172 also undergoes phosphorylation in cells in which the protein would normally be expressed. DISCUSSION A fundamental problem in gene regulation is that of how the appropriate binding site on DNA is selected by structurally related members of transcription factor families each able to recognize the same consensus nucleotide sequence. For the bHLH proteins, at least three strategies appear to be used to ensure binding specificity. These include differential heterodimerization (45), interaction with DNA-binding proteins of other classes (46), and discrimination of bases flanking the core E-box motif (47,48).
Although the ability of phosphorylation to modify the binding of a variety of transcription factors to their cognate DNA sequences is well known (reviewed in Ref. 49), it has only recently been appreciated that this action may be target-dependent. For p53 (50,51) and the POU homeodomain protein GHF-1 (Pit-1) (52,53), phosphorylation has been shown to increase or decrease binding in a manner dependent on the specific DNA elements tested. We have found that a conserved serine residue in the oncoprotein TAL1, Ser 172 , can be phosphorylated by PKA and that its phosphorylation regulates DNA binding activity in a template-dependent fashion without effect on protein dimerization or subcellular localization.
Site selection analysis has demonstrated a preference of TAL1-containing complexes for specific sequences flanking, in addition to those comprising, the central CANNTG motif of the E-box, and these same bases appear also to be important for the effect of PKA on TAL1 DNA binding activity. Discrimination of E-box-flanking bases has been similarly noted for E47 and MyoD homodimers (45), and the crystal structures of these bHLH proteins bound to DNA indicate their basic regions can make contact with them (11,12). Although Ser 172 would not be expected to interact with DNA directly, transfer of a charged phosphate group to this residue potentially could induce a conformational change that alters the strength of such interactions.
Our studies suggest that a region of extended homology between TAL1 and another member of the subgroup of HLH oncoproteins activated by chromosomal rearrangement, LYL1, COS cells cultured on coverslips were mock-transfected (top) and transfected with an expression vector for full-length mouse TAL1 cDNA without (middle) and with (bottom) one encoding the catalytic subunit of PKA. Cells were fixed 48 h later with 2% formaldehyde, permeabilized with 0.1% dimethyl sulfoxide, and stained with rabbit anti-TAL1 antibody. TAL1 protein localization was determined by indirect immunofluorescence using fluorescein-conjugated goat anti-rabbit antibody. Fluorescence photomicroscopic images are shown on the left and corresponding phase contrast images on the right. Original magnification, 500 ϫ.
FIG. 8. Two-dimensional phosphopeptide patterns of fulllength TAL1 protein and corresponding Ser 172 to Ala mutant. COS cells coexpressing catalytic subunit of PKA and full-length TAL1 protein (left) or the corresponding Ser 172 to Ala mutant (middle) were metabolically labeled with [ 32 P]orthophosphate. Murine erythroleukemia (MEL) cells (right) cultured with 8-(4-chlorophenylthio)-adenosine 3Ј:5Ј-cyclic monophosphate and theophylline were similarly labeled. TAL1 proteins were immunoprecipitated from cellular extracts with polyclonal anti-TAL1 antibody, resolved on an 8% SDS-polyacrylamide gel, transferred to a PVDF membrane, and subjected to chymotryptic digestion. Radiolabeled proteolytic fragments were subjected to twodimensional phosphopeptide analysis as described under "Experimental Procedures" and visualized by phosphorimaging. Major proteolytic fragment containing Ser 172 is denoted with an arrow. Equivalent amounts of radioactivity were applied to cellulose thin layer chromatography plates for the two COS cell digests.
represents a conserved regulatory site through which phosphorylation could alter protein function. Although not specifically tested, given the extent of homology over these sequences and the similarity in their location, phosphorylation of the corresponding serine residue in LYL1 (Fig. 1) would be predicted to have the same template-dependent effect on DNA binding. In view of the considerable homology in their basic and HLH domains, near-identical preference in DNA binding sites (19,54), and overlapping patterns of expression (31), it is likely the biological consequences of this modification would be similar as well.
Although a requirement for TAL1 in embryonic hematopoiesis was demonstrated by targeted mutation of the mouse gene (27,28) and a role in erythroid differentiation postnatally is suggested by antisense studies (29), little is actually known about the specific functions of this transcription factor, the identity of its molecular targets, and how its activity might be regulated. PKA has been variously proposed to be required for (55,56) and inhibitory to (57) erythroid differentiation, and although its catalytic subunit potently and selectively phosphorylated Ser 172 of TAL1 in vitro, it is not proven this kinase does so uniquely in vivo. Irrespective of the kinase(s) responsible, the demonstration that phosphorylation of this site has a sequence-dependent effect on DNA binding suggests a mechanism by which the affinity of TAL1-containing complexes and, potentially, the sets of target genes activated could be subject to regulation.