The Tal1 oncoprotein inhibits E47-mediated transcription. Mechanism of inhibition.

The Tal1 oncogene is a class II basic helix-loop-helix (bHLH) transcription factor, overexpressed in as much as 60% of T cell acute lymphoblastic leukemia cases. Like other class II bHLH proteins, Tal1 can heterodimerize with the class I bHLH proteins, such as E47, and bind to a DNA recognition sequence termed E box. Therefore, it is believed that the oncogenic capacity of Tal1 lies in its ability, as a heterodimer with E47, to activate aberrantly a set of "leukemogenic" genes in T cells. However, compared with E47 homodimers, Tal1/E47 heterodimers are very poor transactivators. Thus the effect of Tal1 is actually to inhibit E47 homodimer activity. Here we propose that the transforming properties of Tal1 are the result of its ability to inhibit E47 activity. We address the mechanism of Tal1 inhibition and demonstrate that Tal1/E47 heterodimers cannot activate transcription because their respective activation domains are incompatible. Furthermore, we present data showing that Tal1 can inhibit E47-mediated activation of the CIP1 gene. Finally, we demonstrate that Tal1 inhibits E47 activity in leukemic T cells.

The Tal1 oncogene is a class II basic helix-loop-helix (bHLH) transcription factor, overexpressed in as much as 60% of T cell acute lymphoblastic leukemia cases. Like other class II bHLH proteins, Tal1 can heterodimerize with the class I bHLH proteins, such as E47, and bind to a DNA recognition sequence termed E box. Therefore, it is believed that the oncogenic capacity of Tal1 lies in its ability, as a heterodimer with E47, to activate aberrantly a set of "leukemogenic" genes in T cells. However, compared with E47 homodimers, Tal1/ E47 heterodimers are very poor transactivators. Thus the effect of Tal1 is actually to inhibit E47 homodimer activity. Here we propose that the transforming properties of Tal1 are the result of its ability to inhibit E47 activity. We address the mechanism of Tal1 inhibition and demonstrate that Tal1/E47 heterodimers cannot activate transcription because their respective activation domains are incompatible. Furthermore, we present data showing that Tal1 can inhibit E47-mediated activation of the CIP1 gene. Finally, we demonstrate that Tal1 inhibits E47 activity in leukemic T cells.
Members of the basic helix-loop-helix (bHLH) 1 family of transcription factors have been shown to play important roles in cell type determination and differentiation (1,2). They have been grouped into two classes depending on their expression pattern as either ubiquitous or tissue-specific (3). Class I ubiquitous bHLH proteins can form either homodimers or heterodimers with other class I proteins and with the class II tissue-specific bHLH proteins. Both homodimers and heterodimers have been shown to bind to a canonical DNA sequence, termed E box, located in the regulatory elements of various cell type-specific genes, and concomitantly activate gene transcription. In addition to their roles in cell development, four members of this family, E2A, Lyl1, Tal1, and Tal2, have also been implicated in oncogenesis (4 -8).
Tal1, encoding a class II bHLH protein, was originally identified by virtue of its translocation and overexpression in 3% of T cell acute lymphoblastic leukemia cases (4,5). Subsequent analyses have shown that Tal1 is overexpressed in as much as 60% of T cell acute lymphoblastic leukemia cases through either a specific, interstitial chromosomal deletion event or through other undetectable alterations in DNA (9,10). However, Tal1 is not normally expressed in T cells; rather it is expressed in the developing hematopoietic system, endothelial cells, and the developing brain (11,12). Gene knockout experiments in mice have shown that Tal1 plays a crucial role in primitive erythropoiesis (13). Subsequent rescue experiments with chimeras have demonstrated that Tal1 is required for the development of all hematopoietic lineages (14). In conjunction, cell culture experiments have suggested that Tal1 may also play a role in terminal erythroid differentiation, possibly in cooperation with the zinc finger proteins LMO2 and GATA-1 (15,16).
Recent experiments in transgenic mice have demonstrated the oncogenic potential of Tal1, for when Tal1 expression is directed to the T cell compartment through the promoter of the lymphoid cell kinase gene, these mice develop T cell lymphoma (17,18). However, the exact mechanism of Tal1-mediated leukemogenesis remains unknown. Because Tal1 is capable of heterodimerizing with the ubiquitous bHLH E proteins (E12, E47, E2-2, HEB) and binding to DNA, it is thought that Tal1 aberrantly activates genes in T cells, whose expression then leads ultimately to leukemogenesis (19,20). However, Tal1/ E47 heterodimers are less than 10% as active as E47 homodimers in their ability to activate a reporter gene driven by E box elements (21). Thus, the effect of Tal1 is actually to inhibit the transactivation potential of E47 homodimers. Therefore, an alternative mechanism may be that Tal1, instead of acting as a transcriptional activator, promotes oncogenesis by acting as an inhibitor of E47 activity.
The purpose of this study was to address mechanistically the following question: If Tal1/E47 heterodimers can bind to DNA with the same or higher affinities than those of E47 homodimers (19), why do they activate transcription so poorly? To address this question, fusion proteins of Tal1 and E47 were constructed. Specifically, we asked which regions of E47 can turn Tal1 into an activator, and which regions of Tal1 can turn E47 into an inhibitor. Furthermore, in addition to artificial reporters, we also tested the effect of Tal1 on the native promoter of an E47-activated gene, which could potentially be a target for Tal1 inhibition. Thus far, E47 has been shown to regulate the expression of B cell-specific genes, such as the immunoglobulin and terminal deoxynucleotide transferase genes (22,23). Obviously in the context of T cell leukemia, these are not relevant targets for Tal1. Recent experiments have shown that E47 plays a general role in negatively regulating the proliferation of cells (24) and that this effect may be mediated by the ability of E47 to activate expression of the gene for CIP1 (p21, WAF1, Sdi1), an inhibitor of cyclin-dependent kinases (25)(26)(27)(28). To determine if CIP1 may be a potential target for negative regulation by Tal1, we therefore asked if Tal1 can inhibit the ability of E47 to transactivate the CIP1 gene.

EXPERIMENTAL PROCEDURES
Plasmids-The pTal1-myc (hereafter called pTal1), TalM1/pBS, and pfLUC constructs were as described (29 -31). The p21-LUC construct (a gift from X.-F. Wang, Duke University) contains the 2.4-kilobase HindIII fragment of the p21 promoter inserted into the pGL2-basic vector (Promega) (32). The pCMV-LacZ construct was made by inserting an HindIII-XbaI fragment from LacZ/pcDNAI (gift from D. Levy, New York University Medical Center) into pcDNA3 (Invitrogen). The CIP1/pBKS plasmid was a gift from M. Garabedian (New York University Medical Center). E47/pcDNA3 was constructed by a two-step cloning procedure. An EcoRI-XhoI fragment, spanning amino acid residues 1-372 from a full-length E47 cDNA clone, was first ligated into pcDNA3. The remaining 3Ј-end of the cDNA was then added as an XhoI fragment. The pEbox-LUC construct, containing five copies of the optimal binding site for Tal1/E47 heterodimers, was generated by annealing and ligating the two phosphorylated oligonucleotides: 5Ј-TCGAC-GAACAGATGGTGTC-3Ј and 5Ј-TCGAGACACCATCTGTTCG-3Ј. After digesting the ligated oligonucleotides with SalI, concatemers of 60 -150 base pairs containing direct repeats of the E boxes were purified by electrophoresis on a 1.5% agarose gel and subsequently cloned into the SalI site of pfLUC. A clone containing five copies of the E box was selected by restriction analysis.
Construction of E47-Tal1 Fusion Constructs-To generate the pE-T/1 construct, a PCR product was made using the fusion primer (E-T/1, 5Ј-AAGCTGCTCAATGAGCGGAACCTG-3Ј; Tal1 and E47 sequences are in bold and plain text, respectively) and the SP6 sequencing primer together with E47/pcDNA3 as a template. This product was then allowed to anneal to the TalM1/pBS plasmid and amplified using a Tal1 primer, binding to a region upstream of the internal SacI site (5Ј-GTGAACGGGGCCTTTGCC-3Ј) and the SP6 sequencing primer. The resulting PCR product was digested with SacI and XbaI and used for a three-part ligation into pcDNA3 along with an HindIII-SacI fragment from pTal1.
E47-Tal1 fusion constructs, pE-T/2, pE-T/4, pE-T/5, and pE-T/3bm, were made by a two-step PCR method essentially as described by Vitola et al. (33). As shown in Scheme 1, complementary primers were designed spanning the fusion region such that each fusion primer was comprised of sequences one half from E47 and the other half from Tal1. These fusion primers in conjunction with various 5Ј-and 3Ј-primers were employed to generate the first two PCR products using either E47/pcDNA3 or TalM1/pBS as template. The two products were then denatured, annealed, and used as a template for the final PCR (PCR3) using the previous 5Ј-and 3Ј-primers. This final PCR product was then subcloned into the appropriate vector. PCR fusion primers were as follows: E-T/2, 5Ј-ACGCCGCACAACGTCCCTCAGGTC-3Ј; E-T/4, 5Ј-GATGGAGATTACTGAGAAAGACCTGA-3Ј; and E-T/5, 5Ј-GCAGCA-GGTGCGAGACCAGGAGGAGG-3Ј.
For the pE-T/2 construct, the NOT primer (5Ј-TGGGCTAGGGC-GAGCAGG-3Ј, nucleotides 1359 -1376 (6), upstream of the internal NotI site of E47) and the T3 sequencing primer (Promega) were used as 5Ј-and 3Ј-primers, respectively, along with the fusion primers, in the two-step PCRs with E47/pcDNA3 (5Ј reaction) and TalM1/pBS (3Ј reaction) plasmids as initial templates. The final PCR product was first subcloned into pGEM-T (Promega), subsequently digested with NotI and SpeI, and inserted into the NotI and XbaI sites of E47/pcDNA3. For pE-T/4, the T7 and SP6 sequencing primers (Promega) were used as the 5Ј-and 3Ј-primers in conjunction with the appropriate fusion primers in the two-step PCRs using TalM1/pBS (5Ј reaction) and E47/pcDNA3 (3Ј reaction) as initial templates. The final PCR product was digested with NotI and XbaI and inserted into the NotI and XbaI sites of pTal1. For the pE-T/5 construct, the 5Ј-and 3Ј-primers were the NOT primer and the T3 sequencing primer, and the templates were E47/pcDNA3 and TalM1/pBS, respectively. The final PCR product was subcloned into pGEM-T, digested with NotI and SpeI, and inserted into the NotI and XbaI sites of E47/pcDNA3.
To generate pE-T/3, a three-part ligation was performed by inserting the NotI-SacI fragment from pE-T/2 and the SacI-XbaI fragment from pE-T/1 into the NotI and XbaI sites of E47/pcDNA3. pE-T/6 was generated through a three-part ligation by inserting the HindIII-PstI fragment from pE-T/4 and the PstI-ApaI fragment from pE-T/5 into the HindIII and ApaI sites of pcDNA3. pE-T/3bm was constructed also using the two-step PCR method. Oligonucleotides to the basic region of Tal1 containing the appropriate mutations (5Ј-AAAGTTGTGGGTGG-TATCTTCACC-3Ј; mutated codons are underlined) were used together with the NOT and SP6 oligonucleotides as 5Ј-and 3Ј-primers, respectively, together with pE-T/3 as the template. The final PCR product was subcloned into pGEM-T, digested with NotI and XbaI, and inserted into NotI and XbaI sites of E47/pcDNA3. pE-T/2bm was generated through a three-part ligation by inserting the SacI-XbaI fragment from pTal1 and the NotI-SacI fragment from pE-T/3bm into the NotI and XbaI sites of E47/pcDNA3. The pE-T/2ADm plasmid was created by replacing the EcoRI-NotI fragment of pE-T/2 with an EcoRI-NotI fragment from an E47 mutant construct that had activation domains I and II deleted. 2 To generate the pdE-T/4 construct, PCR was carried out using a Tal1 primer (5Ј-GTCAGGTGATG-GAGATTACTGAG-3Ј) and the SP6 sequencing primer, together with pE-T/4 as the template. The PCR product was digested with PvuII and XbaI and inserted into pcDNA3 along with the HindIII-PvuII fragment from pTal1.
Cell Culture-293T and HeLa cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum and glutamine, penicillin, and streptomycin. Jurkat cells were grown in RPMI medium with 10% fetal calf serum and glutamine, penicillin, and streptomycin.
Transfections, Luciferase Assays, and ␤-Galactosidase Assays-293T and HeLa cells were transfected by the CaPO 4 precipitation method (34). Jurkat cells were transfected by electroporation essentially as described by Hsu et al. (35). Transfected cells were lysed and assayed for ␤-galactosidase activity using the Galacton kit according to the manufacturer's protocol (Tropix). The same cell lysates were used to measure luciferase activity using the luciferin substrate (Promega). For both assays, luminescence was determined on a luminometer (Berthoid).
Northern Analyses-Total cellular RNA was isolated using TRIZOL reagent (Life Technologies), and 20 g of each sample was loaded onto a 1% agarose gel containing 1 ϫ MOPS and 0.68 M formaldehyde. After electrophoresis at 5 V/cm, the gel was rinsed for 45 min in 20 ϫ SSC at room temperature and transferred to a nylon membrane (Zetabind) overnight in 20 ϫ SSC. The blot was hybridized using QUIKHYB (Stratagene) with a CIP1 cDNA probe. The DNA used for the probe was prepared by digesting CIP1/pBKS with XhoI to generate a full-length CIP1 cDNA fragment that was subsequently gel purified. The DNA was labeled using a kit from Boehringer Mannheim. The same membrane was then stripped and probed with a human glyceraldehyde-3-phosphate dehydrogenase cDNA probe. Membranes were exposed to a Phos-phorImager and quantitated.
Antibodies and Western Analyses-The anti-Tal1 immune serum was as described (30). Anti-CIP1 and anti-E47 immune sera were from Transduction Laboratories and Santa Cruz Biotechnology, respectively. Monoclonal antibody 9E10 was a gift from P. Cowin (New York University Medical Center). Transfected 293T or HeLa cells were harvested and lysed in 10 times the volume of the cell pellet with SDS-polyacrylamide gel loading buffer. Samples were boiled and separated on 12% SDS-polyacrylamide gel, and the gel was electroblotted overnight in transfer buffer (25 mM Tris, 190 mM glycine, and 20% methanol) onto a nitrocellulose membrane (Protran). The membrane was blocked in 5% non-fat milk in TBST (10 mM Tris 8.0, 150 mM NaCl, 0.05% Tween 20) for at least 2 h and blotted subsequently with primary antibody for at least 1 h. The blot was then washed three times with TBST and incubated with a secondary antibody conjugated to horseradish peroxidase (Promega) for 1 h. The membrane was washed with TBST three times and detected by enhanced chemiluminescence (Amersham).

Expression of Fusion Constructs-Although
Tal1/E47 heterodimers can bind to DNA with equal or higher affinities than E47 homodimers in vitro, they activate transcription much poorly relative to E47 homodimers. Consequently, Tal1 acts as an inhibitor of E47-activated gene expression. To understand why Tal1 has such an inhibitory effect, fusion proteins containing various regions of Tal1 and E47 were constructed. We asked which domains can be interchanged between Tal1 and E47 so that the effect of Tal1 on E47 activation is not inhibition but activation. As a corollary, we asked which portions of Tal1 can confer inhibitory properties on E47. The schematic of the fusion proteins constructed is shown in Fig. 1A. The E47 and Tal1 proteins were divided into three domains: the NH 2 terminus, the bHLH region, and the COOH terminus. The NH 2 terminus of E47 has been shown to contain two activation domains (36 -38), whereas the NH 2 terminus of Tal1 contains one such domain (39). In both proteins these domains have been identified in the context of fusion proteins with the DNA binding domain of the GAL4 protein. The bHLH region is responsible for the dimerization and DNA binding properties of E47 and Tal1 (3,40). The function of the COOH terminus is less well defined.
To determine the expression levels of the fusion proteins, the constructs were transfected into HeLa cells, and cell lysates were analyzed using Western blots. Because no single antibody could detect all of the fusion proteins, Western analyses were performed using the same blot or cell lysates with antibodies against Tal1, the myc tag (9E10), or E47 (Fig. 1B). Because E-T/2 and E-T/5 contain epitopes reacting to both anti-Tal1 and anti-E47 antibodies and because Tal1 reacts to both the anti-Tal1 and 9E10 antibodies, it was possible to estimate roughly the levels of the fusion proteins expressed. All of the fusion proteins appeared to be expressed at comparable levels, except for E-T/1, which seemed to be expressed at a lower level.
The NH 2 Terminus of Tal1 Possesses Transactivating Ability-To test first whether the fusion proteins could activate transcription on their own, the constructs were examined for their ability to activate a luciferase reporter construct, driven by five copies of the E box sequence optimal for binding to Tal1/E47 heterodimers. The plasmids expressing the fusion proteins were cotransfected into HeLa cells along with the reporter plasmid, and luciferase activity was measured 2 days later. Fig. 2A shows the activities of the constructs relative to E47 homodimers. E47 homodimers routinely gave 200 -400 fold-activation over reporter alone. As expected, those constructs containing the bHLH domain of Tal1 could not activate by themselves because the bHLH region of Tal1 cannot mediate homodimer formation (40), which is a prerequisite for DNA binding. However, constructs containing the bHLH domain of E47, such as E-T/4 and E-T/6, were able to activate transcription. Because E-T/4 contains the NH 2 terminus of Tal1 fused with the bHLH domain and COOH terminus of E47 and be- Western analyses were carried out using antibodies as indicated next to each blot. The anti-Tal1 immune serum was directed against the COOH terminus of Tal1. Monoclonal antibody 9E10 recognizes the myc epitope, which was fused to the NH 2 terminus of Tal1. The anti-E47 immune serum reacts to the NH 2 terminus of E47. The relevant protein in each lane is indicated with an arrow. cause these two domains do not appear to have any transactivating function, the ability of E-T/4 to activate transcription may be attributed to the NH 2 terminus of Tal1. This result is in accordance with previous findings that the NH 2 terminus of Tal1, when fused to the DNA binding domain of Gal4, can act as a transcriptional activator (39). To test this hypothesis, the putative activation domain defined by Sanchez-Garcia and Rabbitts (39) was deleted in our construct dE-T/4. The transactivating activity of dE-T/4 was reduced by about 60% compared with E-T/4, suggesting that the NH 2 terminus of Tal1 is indeed able to activate transcription. However, whether this transactivating function of Tal1 is of physiological significance remains to be determined. It is also interesting to note that construct E-T/5, which differs from wild type E47 only at the COOH terminus, possesses only 30% of the transcriptional activity of wild type E47 homodimers. As this construct is expressed as well as the wild type E47, this result would suggest that the COOH terminus of Tal1 may inhibit the transactivating potential of E47. A similar phenomenon was also observed by Hofmann and Cole (41).
The NH 2 Terminus of Tal1 Is Incompatible with the Transactivating Activity of E47-We next asked what effect these fusion proteins had on transcriptional activation by E47. Equal amounts of E47 and the various fusion constructs were cotransfected with the E box reporter construct into HeLa cells. The luciferase activities in the transfected cells are shown in Fig.  2B. Compared with wild type E47, cotransfection of Tal1 led to a 14-fold reduction of activity as observed previously (21). Similarly, the E-T/1 construct inhibited reporter activation by 3-fold. Although the reduced level of E-T/1 expression relative to wild type Tal1 may explain this weaker inhibition, it may also be because of the lack of the COOH terminus of Tal1.
When the NH 2 terminus of Tal1 was replaced with that of E47 as in constructs E-T/2 and E-T/3, the resulting fusion proteins activated transcription even more effectively than wild type E47. The higher transcriptional activities of E-T/2 and E-T/3 may result from the presence of the bHLH domain of Tal1 because the interaction between Tal1/E47 heterodimers is tighter than between E47 homodimers (21). 3 In addition, Tal1/ E47 heterodimers may possess higher affinities to the E box sequence than E47 homodimers, thus leading to greater activation. Activation by E-T/2 and E-T/3 was dependent on their ability to bind DNA because mutations in the basic amino acids critical for DNA binding (21) (arginines at amino acid positions 188 and 189 were changed to glycines) completely abolished their activities, as demonstrated using constructs E-T/2bm and E-T/3bm. Furthermore, this activation was also dependent on the NH 2 -terminal activation domains from E47, since deletion of the two activation domains in construct E-T/2ADm abrogated the transcriptional activation.
Most importantly, these results suggest that Tal1/E47 heterodimers are capable of activating transcription if Tal1 is provided with suitable transactivation domains such as those from E47. In contrast, fusion proteins E-T/4 and E-T/6, when complexed with E47, inhibited activation by about 6.5-fold compared with E47 homodimers (Fig. 2B), even though the NH 2 terminus of Tal1 contains a potential activation domain ( Fig.  2A). Thus it appears that the activation domain of Tal1 is not compatible with those of E47. As expected, construct E-T/5,

Inhibition of E47 by Tal1
which is similar to E-T/6 except that it contains the NH 2 terminus of E47, results in a 4-fold higher transcriptional activity than E-T/6 when both are complexed with E47, again suggesting that the activation domain of Tal1 is not compatible with those of E47. However, it should be noted that this activity of E-T/5 is still only 60% of E47 homodimers. Similar to the lower activity of E-T/5 homodimers relative to E47 homodimers (Fig. 2B), the Tal1 COOH terminus appears to inhibit the transcriptional activation of E-T/5 when complexed as het-erodimers with E47. The COOH terminus of Tal1 may also be responsible for the lower activity of E-T/2 heterodimers compared with E-T/3 heterodimers (Fig. 2B). Taken together, analyses of these fusion proteins lead us to conclude that the lack of transactivation domains compatible with E47 may be the primary reason for the inhibitory effect of Tal1 on E47-activated gene expression. To a lesser extent, the COOH terminus of Tal1 may also contribute to the inhibition.
Tal1 Inhibits E47-mediated Activation of the CIP1 Gene-Because Tal1 is able to interfere with the transcriptional activation mediated by E47, it is possible that Tal1 exerts its oncogenic activity by inhibiting the expression of genes activated by E47. Because E47 has been shown to possess properties of a growth inhibitor, these genes may be involved directly in growth suppression. Recently, data from our laboratory have demonstrated that E47 can activate the transcription of CIP1, a gene encoding a universal inhibitor of the cyclin-dependent kinases (28). Cotransfection of E47 with a luciferase reporter gene driven by the promoter of the CIP1 gene (p21-LUC) resulted in a 20-fold activation of luciferase expression (Fig. 3A).
To determine if Tal1 could inhibit this activation, Tal1 was cotransfected along with E47 and p21-LUC into HeLa cells, and luciferase activity was measured subsequently. As shown in Fig. 3A, Tal1 inhibited this activation dramatically in a dose-dependent manner. Stoichiometric amounts of Tal1 reduced the activation by as much as 90%. This inhibitory effect of Tal1 on E47-mediated activation CIP1 expression was as strong, if not stronger than, the effect of Id1, a known inhibitor of E47 (42,43). 3 In contrast, the E-T/2 fusion protein, which forms heterodimers with E47 and activates the E box luciferase reporter (Fig. 2B), did not inhibit E47-activated CIP1 expression.
We next asked if Tal1 could inhibit E47-activated expression of the endogenous CIP1 gene. Using Northern blots, we found that E47 enhanced the level of CIP1 mRNA by more than 3-fold in transiently transfected 293T cells (Fig. 3B). Cotransfection of E47 with Tal1 reduced the level of CIP1 mRNA by 50%. Similarly, the level of CIP1 protein was elevated significantly in the 293T cells transfected with E47 (Fig. 3C). Cotransfection of E47 with Tal1 brought the level of CIP1 protein down to background levels. In contrast, when the E-T/2 fusion protein was coexpressed with E47, both the transcript and protein levels of CIP1 were increased, suggesting that heterodimers of E-T/2 with E47 are functionally equivalent to E47 homodimers. From these results we conclude that Tal1 is able to inhibit E47-activated CIP1 expression. However, whether this inhibition occurs in a physiological setting remains to be determined. Nevertheless, it is important that on the native promoter of an E47-activated gene, as opposed to artificial E box reporter constructs, Tal1 inhibits E47 activity.
Tal1 Inhibits E47 Activity in T Cells-Because Tal1 is found to be expressed aberrantly in the T cells derived from T cell acute lymphoblastic leukemia patients, we tested if Tal1 in such T cells could inhibit the transcriptional activity of E47. We utilized the Jurkat T cell line, which was derived from a T cell acute lymphoblastic leukemia patient and has been shown to express Tal1 (44,45). Overexpression of E47 stimulated the expression of the E box reporter by 50-fold (Fig. 4A). Moreover, cotransfection of Tal1 with E47 was able to diminish the stimulation significantly, whereas cotransfection of E-T/2 with E47 retained the activation of the E box reporter gene. Thus, as in HeLa cells, it appears that Tal1 can inhibit E47 activity in T cells.
Experiments have demonstrated that endogenous complexes of Tal1/E47 can be immunoprecipitated from Jurkat cells (35). However it is not known if these complexes are transcriptionally active. The level of luciferase activity measured when the FIG. 2. Panel A, activation of pEbox-LUC by Tal1-E47 fusion proteins. HeLa cells were transfected with 0.5 g of each construct and 5 g of the pEbox-LUC reporter construct plus 1 g of pCMV-LacZ as an internal control. Panel B, effect of Tal1-E47 fusion proteins on E47 activation of pEbox-LUC. HeLa cells were cotransfected with 0.15 g of each construct and 0.15 g of E47 plus 5 g and 1 g of pEbox-LUC and pCMV-LacZ, respectively. In both panels A and B, cells were harvested 48 h posttransfection and assayed for luciferase and ␤-galactosidase activities. Each luciferase activity was corrected for ␤-galactosidase activity and presented as a percentage of the corrected activity in E47-transfected cells. Each value represents the mean Ϯ S.D. from at least three independent transfections. In certain cases, the error bars are too small to be visible. E box luciferase reporter construct is transfected alone into Jurkat cells can be taken as a measure of endogenous E protein activity. Furthermore, if this activity is truly the result of E proteins, then cotransfection of Id1 should lower this activity. Endogenous E protein activities in Jurkat cells appeared to be very low in that cotransfection with Id1 could not reduce the expression of the reporter further. 3 In addition, when the E-T/2 construct was transfected into Jurkat cells without E47, it resulted in a 13.5-fold activation of reporter expression (Fig.  4B). In contrast, Tal1 did not increase reporter expression at all. Because the E-T/2 fusion protein cannot form homodimers to activate transcription, the transcriptional activation by E-T/2 proteins is probably caused by the formation of heterodimers between E-T/2 and endogenous E47 or E47-like proteins. Also, because E-T/2 contains the same bHLH domain as Tal1, it may be able to compete with endogenous Tal1 to bind to E47 and activate the expression of the E box reporter. Together, these data suggest that endogenous Tal1/E47 complexes, although readily detectable, are transcriptionally inactive, thereby perhaps reducing the expression of genes FIG. 3. Panel A, effects of Tal1 and E-T/2 on E47-mediated activation of p21-LUC. 2 g each of E47/pcDNA3 and p21-LUC was cotransfected into HeLa cells with or without the indicated amounts of either pTal1 or pE-T/2. 1 g of pCMV-LacZ was cotransfected as an internal control. Cells were harvested 48 h posttransfection and assayed for luciferase and ␤-galactosidase activities. Each luciferase activity was corrected for ␤-galactosidase activity and presented as a percentage of the corrected activity in E47transfected cells. E47-activated p21-LUC expression was 20-fold. Each value represents the mean Ϯ S.D. from at least three independent transfections. In certain cases, the error bars are too small to be visible. Panel B, effects of Tal1 and E-T/2 on E47-mediated induction of CIP1 expression. 10 g of E47/pcDNA3 was transfected into 293T cells with or without the indicated amounts of pTal1 and pE-T/2. Total RNA samples isolated 48 h posttransfection were analyzed using Northern blots with radioactive probes for human CIP1 and glyceraldehyde-3-phosphate dehydrogenase (GAPDH). The hybridization signals were quantitated using a PhosphorImager. The level of CIP1 mRNA in each transfected cell is normalized against that of glyceraldehyde-3phosphate dehydrogenase and shown in the bar graph. The amount of CIP1 mRNA in E47-transfected cells presented in the bar graph is the average of the two samples shown in the gels. Panel C, effects of Tal1 and E-T/2 on E47-mediated expression of the CIP1 protein. 293T cells were transfected with 5 g of E47/pcDNA3 with or without the indicated amounts of either pTal1 or pE-T/2. Cell lysates were prepared 48 h after transfection and loaded onto 12% SDS-polyacrylamide gel electrophoresis. The CIP1 protein was detected by Western blotting. controlled by E47 and eventually leading to leukemogenesis in T cells.

DISCUSSION
The Tal1 protein has traditionally been grouped as a class II bHLH protein because of its tissue-specific expression and its dependence on class I bHLH proteins to bind to DNA. However, Tal1 differs functionally from other class II proteins, such as MyoD and Beta2, whose heterodimers with E47 transactivate appropriate reporter genes much more efficiently than E47 homodimers (46,47). When Tal1 is coexpressed with E47, the transactivating activity is reduced dramatically compared with E47 homodimers. We have attributed this inhibitory effect primarily to the lack of a transactivation domain in Tal1 which is compatible with the transactivating function of E47. To extrapolate from our data, it appears that E47 requires two sets of activation domains from both partners to activate transcription. Heterodimers between E47 and proteins with the NH 2 terminus of Tal1, or its NH 2 -terminal deletion mutant, or the NH 2 terminus of E47 containing deleted activation domains all failed to stimulate the expression of the reporter gene. The fact that heterodimers between E47 and MyoD or Beta2 can activate transcription would suggest that MyoD and Beta2 contain activation domains either compatible with the activation domain of E47 or capable of activating transcription on their own. Tal1, on the other hand, lacks such a domain. Unlike the Id proteins that sequester the E proteins into complexes unable to bind DNA (42,43), Tal1/E47 heterodimers can interact with DNA, but they are not able to activate transcription. Therefore, this ability of Tal1 to occupy control elements normally bound by E47 homodimers could render Tal1 a more powerful inhibitor than the Id1 proteins.
It is interesting that the NH 2 terminus of Tal1 displays the properties of an activation domain when it is fused to the bHLH domain of E47 or the Gal4 DNA binding domain such that homodimers of the fusion proteins can form. This putative activation domain appears inactive in the Tal1/E47 heterodimers, suggesting that it too requires two copies for transcriptional activation. Because Tal1 is known not to form homodimers, it is difficult to imagine that two copies of this putative activation domain could ever exist in a complex to activate transcription. Nevertheless, it remains a formal possibility that when Tal1 forms multiprotein complexes in certain circumstances, the activation domain of Tal1 can be functional.
Can Tal1 mediate transcriptional activation? Heterodimers between E47 and Tal1 have been shown to form ternary complexes with the Lim-only proteins, LMO1 and LMO2, through the bHLH domain of Tal1 (48). Coexpression of Tal1 and LMO1 or LMO2 in T cells appears to enhance the ability of Tal1 to activate transcription of a reporter gene driven by multiple copies of the E box (49). However, it is not clear whether the level of this activation is comparable to the expression activated by E47 homodimers. By using an artificial promoter sequence as a probe, in which an E box is placed next to a GATA site, Tal1 is found to form a higher order complex that includes Tal1, E47, GATA-1, LMO2, and the Lim domain binding protein 1. Transcriptional activity directed by this artificial promoter can be stimulated by overexpressing all of the components in the complex (16). These data would suggest that Tal1 may potentially be involved in transcriptional activation in certain specific settings where all of these factors are present, perhaps during hematopoiesis. However, the physiological target genes that can be activated by this multimeric complex have yet to be found.
Despite the potential of Tal1 in multifactor complexes to activate transcription of certain genes, it is distinctly possible that in its ectopic setting of T cells, Tal1 causes oncogenesis through its ability to act as an inhibitor of E47-mediated cellular processes. E47 is known to suppress cell growth in NIH3T3 fibroblasts upon overexpression (24), presumably through its ability to activate the expression of growth-suppressive genes such as CIP1. In T lymphocytes, E47 apparently acts as a tumor suppressor because mutations of the E2A gene that codes for E12 and E47 lead to the development of T cell lymphoma at extremely high frequencies in E2A-deficient mice (50). Transgenic mice overexpressing Tal1 under the promoter of lymphoid cell kinase gene also develop T cell lymphoma but at a lower incidence and later age (17,18). Importantly, the ability of Tal1 to cause tumors is enhanced by coexpression of casein kinase II␣ (17), which can inhibit the activity of E47 homodimers (51), thus supporting the hypothesis that Tal1 acts as an inhibitor of E47. Interestingly, transgenic mice overexpressing LMO1 or LMO2 also develop T cell malignancies (52). Tal1, although unable to cause tumors when expressed under the control of the SIL or CD2 promoter, increases the incidence and rate of tumor formation of the LMO1 and LMO2 transgenic mice, respectively (53,54). However, it remains to be determined whether this effect of Tal1 is caused by the inhibition of E47 or because of the transcriptional activation involving both Tal1 and LMO proteins.
In summary, previous evidence and our current results have FIG. 4. Panel A, effects of Tal1 and E-T/2 on E47-mediated activation of pEbox-LUC in Jurkat T cells. 20 g of E47/pcDNA3 and 10 g of pEbox-LUC were cotransfected into Jurkat cells with or without 20 g of either pTal1 or pE-T/2. Panel B, the E-T/2 fusion protein can activate pEbox-LUC on its own in Jurkat T cells. 20 g of pEbox-LUC was cotransfected into Jurkat cells with either E47/pcDNA3, pTal1, or pE-T/2. In both panels A and B, 10 g of pCMV-LacZ was cotransfected as an internal control. Cells were harvested 48 h posttransfection and assayed for luciferase and ␤-galactosidase activities. The luciferase values were corrected for ␤-galactosidase activity and are given as fold of activation over reporter alone. The numbers represent the mean Ϯ S.D. of at least three independent transfections. In certain cases, the error bars are too small to be visible. led us to propose that one of the mechanisms by which Tal1 causes T cell lymphoma is by inhibiting the transcriptional activation by E47. Two other proteins whose bHLH domains are highly homologous to Tal1, Tal2 and Lyl1, are also involved in T cell leukemia (7,8). These proteins also bind to E47 and abrogate its transactivation properties, 3 thus suggesting a common mechanism for all three proteins in leukemogenesis. What would be the genes normally controlled by E47 and inhibited by Tal1? It is unlikely that the down-regulation of the CIP1 gene, although it may occur, is the sole reason for leukemogenesis because CIP1-deficient mice do not develop T cell lymphoma (55). Additional genes in T cells controlled by E47 remain to be identified in the future. These genes may be crucial for the homeostasis of the T cell compartment by regulating cell cycle progression or apoptosis.