T-cell proto-oncogene rhombotin-2 is a complex transcription regulator containing multiple activation and repression domains.

The LIM domain protein rhombotin-2 (RBTN-2/TTG-2/LMO2) is involved in many processes, including leukemogenesis and erythropoiesis. It is thought that the principle role of RBTN-2 in these processes is to regulate transcription. To examine the potential for RBTN-2 to modulate transcription, we constructed RBTN-2/GAL4 DNA-binding domain fusion proteins and measured their ability to activate transcription of a reporter gene construct. From these studies we identified a transcription activation domain within the NH2 terminus of RBTN-2. This activation domain was further localized within a proline-rich 19-amino acid region. A second activation domain of 11 amino acids was also identified. This domain was located within the COOH terminus of RBTN-2, and functioned in mammalian cells but not in yeast. Furthermore, the two LIM domains of RBTN-2 were shown to function as transcription repression domains. Each individual LIM domain acted as an independent transcription repression domain on a heterologous activation domain. However, in context of full-length RBTN-2, the LIM domains selectively repressed the NH2-terminal activation domain, but had no effect on the COOH-terminal domain. Overall, these results demonstrate that the T-cell oncogene RBTN-2 is a complex transcription factor possessing multiple transcription regulatory modules, including two activation domains and two repression domains.

Chromosomal translocations, juxtaposing two different regions of the genome, are the recurrent events in human cancers, particularly leukemias (1,2). Molecular analyses of the breakpoints in lymphocytic leukemias have revealed that genes specifying transcription factors are the most frequent target of the changes brought about by these translocations (2). Translocations cause two types of changes in the affected transcription factors. In one type, a chimeric fusion transcription factor is generated when the breakpoints occur in the coding regions of the affected genes. The resulting fusion transcription factor, possessing altered specificity, is produced in the cancer cells. The other type of change results in the increased, and usually ectopic, expression of a structurally intact transcription factor. This change most frequently occurs when translocation break-points occur in the regulatory region of a proto-oncogene, juxtaposing the proto-oncogene to either immunoglobulin or T-cell receptor genes in lymphocytic leukemias. Thus, either altered or ectopic expression of proto-oncogenes is thought to contribute toward the pathogenesis of human leukemias.
Translocations to the rhombotin-2 (RBTN-2/TTG-2/LMO2) locus represent the second type of change to a proto-oncogene. The RBTN-2 locus at chromosome 11p13 is the most frequent site of chromosomal translocation in childhood T-cell acute lymphoblastic leukemia (3)(4)(5). Translocations to this locus involve T-cell receptor genes, and almost invariably result in high expression of normal RBTN-2 (3,4). While RBTN-2 is expressed in virtually all tissues during fetal and adult life, it is not expressed in T-lymphocytes (4,6,7). Thus, the ectopic expression of RBTN-2 in thymocytes contributes to leukemogenesis as demonstrated in transgenic mice with enforced expression in T-cells (8 -10). A normal function for RBTN-2 in erythropoiesis has recently been found. Homozygous null mice lacking RBTN-2 die in utero around day 11 due to lack of red blood cells in the embryonic yolk sac (11). The mechanism of RBTN-2 protein in its leukemogenic and normal functions is implicated in transcriptional regulation because of its nuclear location in hematopoietic cells (7,11), and recent demonstration that RBTN-2 forms complexes with other transcription factors: namely tal-1, GATA-1, and GATA-2 (12)(13)(14).
Additional evidence that RBTN-2 functions in transcriptional regulation is derived from the structural similarity between RBTN-2 and known transcription factors. The RBTN-2 gene encodes a 158-amino acid protein, approximately twothirds of which is comprised of tandemly repeated cysteine-rich zinc-binding motifs called LIM domains (15,16). The LIM domain was originally identified in the transcription factors Lin-11 (17), Isl-1 (18), and Mec-3 (19). However, unlike these LIM-homeodomain transcription factors, RBTN-2 has no demonstrable DNA binding activity (15,16). Recently, the LIM domain has been shown to facilitate a variety of protein-protein interactions. In vitro studies have shown that the LIM domain can dimerize with other LIM domain (20,21) or other motifs such as helix-loop-helix domain (12,13). Thus, RBTN-2, lacking known transactivation domains and DNA-binding motifs, was thought to regulate transcription by binding other transcription factors and modulating their activity. However, recently, RBTN-2 was shown to possess transcription activity when fused to the DNA-binding domain of GAL4 (22) 1 . In this report, we confirm this transactivation activity and further show that RBTN-2 has complex transcriptional regulatory activities; it has two separate activation domains and two repression domains. Furthermore, we show that the LIM domains, in addition to their known function as protein-protein interaction domains, have transcription repression activity. These results support the notion that RBTN-2 may regulate the transcription of target genes in a complex manner that, when ectopically expressed in T-lymphocytes, may eventually lead to leukemias.
The other RBTN-2 constructs used in mammalian cells were generated by PCR. They include the NH 2 terminus region (RN, amino acids 1-29), the NH 2 terminus deletion mutant (RdN, amino acids 30 -158), the COOH terminus deletion mutant (RdC, amino acids 1-145), the LIM domain region (R2L12, amino acids 30 -147), the first LIM domain (R2L1, amino acids 30 -83), and the second LIM domain (R2L2, amino acids 94 -147). Of these PCR-generated mutants, RN, RdN, and RdC were cloned into the HindIII site of pM3 vector, while the LIM domain region mutants, R2L12, R2L1, and R2L2, were cloned into the EcoRI site of pM3 vector and the EcoRI site of the pM3/3CGln vector. The pM3/3CGln vector (a kind gift from Dr. Clare Sample, St. Jude Children's Research Hospital, Memphis, TN), encodes the GBD fused with the glutamine-rich activation domain (3CGln) from Epstein-Barr virus transcription factor, EBNA3C (26). The LIM domain mutants cloned into this vector were positioned before and in-frame with the 3CGln transactivation domain. Authenticity of the above constructs was confirmed by sequencing.
The CAT reporter gene vector pG5EC contains the gene encoding chloramphenicol acetyltransferase (CAT), which is under the control of a promoter with the adenovirus E1b TATA box and five consecutive GAL4-binding sequences (25). The pCMV-␤-gal plasmid contains the ␤-galactosidase gene under the control of the CMV promoter (Clontech, Palo Alto, CA).
Cell Lines and Transfections-The COS-1 cells and NIH 3T3 cells were grown in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin.
All transfections were performed using LipofectAMINE™ reagent as described by the supplier (Life Technologies, Inc.). The expression of fusion proteins in transfection assays was examined by Western blot using an antibody to the GAL4 DNA-binding domain (Santa Cruz Biotechnology, Santa Cruz, CA). GAL4 fusion proteins expressed in transfected cells were immunoprecipitated with the GAL4 DNA-binding domain antibody, and detected by Western blot using the same antibody.
Yeast Strain and Assays-The yeast strain Y190 (24) was provided by Dr. Stephen Elledge. It contains two reporter genes, HIS3 and lacZ, under the control of promoters containing the GAL4-binding sites (24,27). The yeast was grown on YPAD medium, or SD minimal medium supplemented as described (24).
The pAS2 constructs were transformed into Y190 as described (24). The activation of the reporter genes, HIS3 or lacZ, was judged by the ability of transformants: 1) to grow in the absence of histidine (50 mM 3-amino-1,2,4-triazole, a histidine synthesis inhibitor, was added to increase selection pressure) and 2) to turn blue on plates containing 5-bromo-4-chloro-3-indolyl-␤-D-galactoside (40 ng/ml), or by filter-lift assay as described by the supplier (Clontech).
CAT Assays-Two micrograms each of the following three plasmids (6 g total DNA), the CAT reporter construct pG5EC, GBD-RBTN-2 mutant fusions vectors (pM3), and pCMV-␤-gal, were co-transfected into COS-1 cell lines, using 24 l of LipofectAMINE™ reagent based on the protocol provided by the supplier (Life Technologies, Inc.). After 72 h, the cells were harvested and the amount of the cell lysate used in each CAT assay was normalized on the basis of the ␤-galactosidase activity of each extract. The CAT assays were performed as described (28). 1 and others (22) have found that RBTN-2, when fused to a DNA-binding domain, has transactivation activity in yeast. The full-length RBTN-2 coding sequence was cloned into the GBD fusion protein expression vector pAS2 and transformed into Saccharomyces cerevisiae strain Y190, which contains two reporter gene constructs: HIS3 and lacZ (24). As shown in Fig.  1, full-length RBTN-2, when fused to GBD, activated the reporter genes and the correspondent yeast transformants grew in the absence of histidine and turned blue in the filter-lift assay, indicating that RBTN-2 indeed has intrinsic transcription activation activity.

RBTN-2 Has Intrinsic Transcription Activity in Yeast-We
To identify the region(s) of RBTN-2 responsible for this transcription activity in yeast, we generated various deletion mutants of the RBTN-2 gene by the PCR (Fig. 1A). These deletion mutants were cloned into the pAS2 vector and transformed into the yeast strain Y190. As shown in Fig. 1 (panel A and B), deletion of the NH 2 -terminal 26 amino acids (RdN) abolished the activation activity of RBTN-2; while deletion of the COOH terminus (RdC) had no effect on this activity. Neither LIM domain alone nor LIM domains together (RL1, RL2 and RL12) showed any transcription activity. This result further supported the transactivation activity of RBTN-2 and indicated that the activation domain was located in the 26-amino acid sequence of the NH 2 terminus.

RBTN-2 Functions as a Transcriptional Regulator in Mammalian Cells-Since
RBTN-2 is a mammalian gene, we sought to establish whether RBTN-2 exhibits transcription activation in mammalian cells. Therefore, the full-length RBTN-2 coding sequence was cloned into the mammalian GBD fusion protein expression vector pM3 (25). This construct was co-transfected into COS cells with the CAT reporter gene construct pG5EC (25) which contains five tandemly repeated GAL4-binding sites in its promoter region. The transfected cells were incubated for 72 h for optimum CAT gene expression. After the incubation period, the cells were lysed and assayed for CAT activity. As shown in Fig. 2A, full-length RBTN-2, when fused to GBD, showed 3-fold more CAT activity (lane 2) as compared to control cells that were transfected with the empty pM3 (lane 1). The increase in transcriptional activity was low, but was consistently observed in six different experiments (range 3-6-fold over the background). Control experiments using a CAT reporter plasmid (E1b-CAT) lacking GAL4-binding sites showed no activity with any of the constructs tested (Fig. 2B). Thus, the transactivation activity of the fusion proteins was dependent on binding to the GAL4 target sites on the reporter plasmid. These results establish that, as in yeast, RBTN-2 has tran-scription activation in mammalian cells. These findings were confirmed when the NH 2 -terminal 29-amino acid region fused to GBD also showed 6-fold higher transcription activity ( Fig.  2A, lane 3) as compared to the control cells (lane 1). However, in contrast to yeast, we found that deletion of 29 amino acids from the NH 2 terminus region or RBTN-2 did not abolish the increase in the CAT activity ( Fig. 2A, lane 4), suggesting the presence of other transactivation domain(s). Overall, these results suggested a more complex transcription regulatory behavior of RBTN-2 in mammalian cells than that seen in yeast.
The NH 2 -terminal Activation Domain Is Located within a Proline-rich Region-To further localize the NH 2 terminus activation domain, a series of deletion mutants were made either by PCR or synthesis of oligonucleotides and then cloned into the pM3 vector (Fig. 3A). These constructs were transfected into COS cells along with the CAT reporter gene construct pG5EC. Transfected cells were incubated for 72 h and cell lysates were assayed for CAT activity. As shown in Fig. 3B, the first 10 amino acids of the NH 2 -terminal region (RN1-10) did not contribute to the transcription activation because this 10amino acid stretch did not show any significant activity over the background, whereas, the rest of the 19 amino acids (Fig.  3B, RN11-29) possessed full activity. Therefore, we further examined the transcription activity of this 19-amino acid region in detail. There are only 4 proline residues in full-length RBTN-2, and all these 4 prolines are present in this 19-amino acid stretch (Fig. 3A). When we deleted the sequence from 24 to 29 amino acids, which contains two consecutive prolines, the transcription activation was reduced by about 50% (Fig. 3B, RN11-23 and RN1-23), suggesting that proline residues play an important role in the transcription activation activity of the NH 2 terminus region. The sequence from residue 15 to 23 has been proposed, by Sanchez-Garcia and colleagues (22), to have homology with the acidic activation domain of VP16 and is included in this 19-amino acid stretch. However, the region containing these amino acids (Fig. 3B, RN14 -23) contributed only about 40% of the total activity, whereas a region lacking 4 amino acids of the "VP16 homology region" (RN11- 19) showed about 60% of the total activity. To ensure that the differences in activity were not due to varying amounts of fusion proteins, we examined the fusion protein levels by Western blot analysis (Fig. 3C). The results showed that very similar amounts of the fusion proteins were expressed in the transfected cells. Therefore, the differences in CAT activity reflect the differing abilities of the NH 2 -terminal deletion constructs to activate tran- scription. Based on this deletion analysis, we conclude that the complete activation domain of the NH 2 terminus resides in the 19-amino acid stretch from residue 11 to 29, and that the proline residues and VP-16 homology regions contribute significantly to the total transactivation activity of the NH 2 terminus region.
The COOH Terminus of RBTN-2 Has Transactivation Activity in Mammalian Cells-In contrast to yeast, deletion of the NH 2 terminus did not abolish the activity of RBTN-2 in mammalian cells (Fig. 2A, lane 4). These results show that additional activation domain(s) exist in regions other than the NH 2 terminus of RBTN-2. Since deletion of the COOH terminus greatly reduced the overall activity (Fig. 4, lane 4), we therefore tested the transcription activity of this region by cloning the COOH terminus of RBTN-2 into pM3 vector. COS cells transfected with the pM3 construct containing the COOH terminus region (11 amino acids) and the CAT reporter gene construct pG5EC showed 4 times more CAT activity than the control cells (Fig. 4, compare lanes 1 and 3). These results show that the COOH terminus region of RBTN-2 possesses transcription activation activity in mammalian cells. Surprisingly, the existence of the COOH terminus seems to be critical to the overall activation activity of RBTN-2 in mammalian cells, since its deletion (Fig. 4, lane 4) greatly reduced the overall transcription activation, even though the truncated RBTN-2 molecule contained the NH 2 -terminal activation domain.
LIM Domains of RBTN-2 Act as Transcriptional Repressors-We observed an interesting phenomenon in that the overall activation activity of RBTN-2 did not reflect the sum of the individual activation domains of the NH 2 terminus and the COOH terminus in mammalian cells. In fact, the activity of the NH 2 terminus domain alone is stronger than that of the fulllength RBTN-2 ( Fig. 2A, lanes 2 and 3), suggesting that the activity of this domain is somehow repressed in the context of the whole molecule. Besides the two activation domains at the two termini, the two LIM domains constitute virtually the rest of the RBTN-2 protein. Therefore, we examined the possibility that the LIM domains act as transcription repressor. We positioned the LIM domain region of RBTN-2 between the GBD and a heterologous activation domain. We used the glutaminerich region (3CGln) of the Epstein-Barr virus transcription factor EBNA3C which has been shown to possess transactivation activity (26). The transcription activity of the 3CGln (pM3/ 3CGln) is shown in lane 2 of  Fig. 2. A, schematic representation of the NH 2 -terminal region and its deletion constructs. Each construct is named based on the position of the first and the last amino acid residues. Each bar corresponds to the length of the amino acid sequence and the position of the construct relative to the NH 2 -terminal region. B, the relative CAT activity of the NH 2 -terminal deletion constructs. Quantification of CAT activity was performed using a PhosphorImager (Molecular Dynamics) and the associated software ImageQuant. The relative CAT activity of each construct is represented as a percentage of the CAT activity of the complete NH 2 terminus (RN1-29). C, equivalent amounts of GAL4 fusion proteins are expressed in transfection assays. Extracts from the transfected cells were immunoprecipitated with a GAL4 DNA-binding domain antibody, and fusion proteins were detected by Western blot using the same antibody. lineages (7,11); (ii) as an LIM-only protein, RBTN-2 interacts with the transcription factors Tal-1 (12, 13), GATA-1 and GATA-2 (14), and retinoblastoma-binding protein-2 1 ; (iii) RBTN-2 fused to GBD activates transcription (22). 1 In this study, we have confirmed that RBTN-2 functions as a transcription activator in both yeast and mammalian systems. In addition, we have localized the NH 2 terminus activation domain to a 19-amino acid region and identified a second activation domain at the COOH terminus of RBTN-2. The COOHterminal activation domain, unlike the NH 2 terminus activation domain, functions in mammalian cells but not in yeast. Furthermore, we have demonstrated that the LIM domains, currently recognized as protein-protein interaction domains, also function as repression domains. The presence of LIM domains result in the overall low transcription activation activity of full-length RBTN-2 compared to those of the individual activation domains at both termini.
Studies on the eukaryotic transcription factors indicate that, as opposed to DNA-binding domains, the transcription activation domains of these proteins are less well defined (29). However, based on primary amino acid sequences, several classes of activation domains have been loosely defined, e.g. acidic, proline-rich, glutamine-rich, serine/threonine-rich, and recently identified isoleucine-rich domains (30 -32). But some activation domains do not fit into any of these classes (33). In agreement with Sanchez-Garcia et al. (22), we have found that the NH 2 terminus region of RBTN-2 serves as an activation domain in yeast, as well as in mammalian cells. In addition, we have localized the activation domain to a 19-amino acid stretch which displays the full transcription activation activity of the NH 2 terminus region. It has been suggested that an 8-amino acid region (residues 15-23), with homology to the acidic activation domain of VP16 (34), may account for the transcription activity of RBTN-2 (22). However, we found that the "VP16 homologous sequence" in a deletion construct (RN14 -23, Fig.  3) gave only ϳ40% of the total activation activity of the NH 2 terminus domain. Thus, our data suggest that the VP16 homologous sequence contributes significantly but does not account for the total transcription activity of the NH 2 terminus activation domain of RBTN-2. This notion is further supported by our observation that an 8-amino acid stretch (RN11-19, Fig. 3) gave ϳ60% of the total activation activity even though it lacked 4 of the 8 amino acids constituting the VP16 homologous region.
We found that the aromatic amino acid proline seems to play an important role in the transcription activity of the NH 2 terminus activation domain. This conclusion is based on the following observations: (i) there are only 4 proline residues in RBTN-2, but all of them are present within this 19-amino acid activation domain; (ii) deletion of two consecutive prolines, Pro-24,25, from this region (RN1-23 and RN11-23, Fig. 3) resulted in an approximately 50% decrease in the transcription activity; (iii) the level of transcription activation exhibited by the NH 2 terminus activation domain of RBTN-2 is similar to that seen in other proline-rich domains of transcription factors, such as AP-2 (3-6-fold activity over the background) (35,36). These transcription factors usually act in a TATA-proximal position and have considerably lower activity compared to the transcription factors containing acidic-rich domain (35,36). However, further studies involving site-directed mutagenesis are required to determine whether or not proline residues play a critical role in the NH 2 -terminal activation domain of RBTN-2.
We also discovered a second activation domain in the COOH terminus region of RBTN-2 that was active in mammalian cells, but was silent in yeast. This activation domain, comprised of 11 amino acids, does not easily fit into any class of the activation domains, but it has four isoleucine residues. Thus, the COOH-terminal region may represent either a novel class or an isoleucine-rich activation domain (32). It is noteworthy that the COOH-terminal activation domain is required for the transactivation activity of the entire RBTN-2 protein. Removal of the COOH terminus alone results in silencing of the NH 2 terminus activation domain in mammalian cells (Fig. 4, lane 4).
The presence of two activation domains in RBTN-2 is not too surprising since it is not unusual for eukaryotic transcription factors to have multiple activation domains (37). For example, the HLH family member E2A contains at least two separate activation domains; one functions efficiently in a variety of mammalian cell lines, whereas the second functions preferentially in pancreatic beta cell lines (38). In some eukaryotic transcription factors, the situation is even more complex. c-Fos has three activation domains in its COOH terminus which show weak activity independently but act synergistically in combination (39,40). Recently, two additional activation domains and an inhibitory region has been identified in NH 2 terminus of c-Fos (39). Multiple, regulatory domains in the eukaryotic transcription activation factors, such as RBTN-2, may allow these proteins to respond to different signals and/or endow them with multiple functions.
The two activation domains of RBTN-2 behave quite differently in that the NH 2 -terminal one functions in both yeast and mammalian cells, whereas, the COOH-terminal one has activity only in mammalian cells. One possibility for this different behavior of activation domains is that it reflects the basic differences in the general transcriptional machinery between mammalian cells and yeast. Indeed, recent studies have shown that some activation domains, e.g. the class of glutamine-rich domains such as those of SP1, AP2, Oct-1, Oct-2 (29,31,41), do not stimulate transcription in yeast S. cerevisiae, even when human TATA-binding protein or human-yeast hybrid TATAbinding protein is supplied (31). Thus, the COOH-terminal activation domain of RBTN-2 may target transcription factor(s) in mammalian cells that are absent in yeast. The distinct behaviors of the two separate activation domains of RBTN2 imply that each domain may have different function. This view is consistent with the observations that RBTN-2 is a multifunctional protein. RBTN-2 is required for erythropoiesis (11), and is involved in T-cell leukemogenesis (8,10). In addition, RBTN-2 is implicated in the acute-phase response (7). In this context, it is noteworthy that RBTN-2 binds multiple protein partners, especially transcription factors through its LIM domains (12)(13)(14). 1 It is likely that each function requires different partner(s) that may interact with the distinct LIM domains of RBTN-2. Thus, the functions of RBTN-2 may depend upon the relative availability of the protein partners, that in turn may be influenced by the developmental stages, different tissue distribution, or cellular environment. In this context, it will be interesting to determine the relative activity of the two activation domains in different tissues.
We found that the overall transcription activity of RBTN-2 was lower than that of each individual activation domain, suggesting the presence of a repressor domain in RBTN-2. The most obvious candidates for the repressor effect are the LIM domains since, other than the NH 2 and COOH terminus activation domains, they virtually constitute the remaining portion of RBTN-2. Indeed, our results show that RBTN-2 LIM domains, either individually or in combination, suppress the transcription induced by a heterologous activation domain (the EBNA3C protein of the Epstein-Barr virus, Fig. 5). Since each LIM domain can inhibit transcription activation independently, we consider that RBTN-2 has two repressor domains. A similar "mix and match" of protein domain approach has led various investigators to identify repressor domains in different transcription factors such as WT1, eve, en, c-Evb-A, and v-Evba (37,42). These repressor domains are termed "portable" or "transferable" because they function in the context of heterologous activation and DNA-binding domains. Thus, RBTN-2 possesses two portable repressor domains. These repressor domains function in a position independent manner, i.e. repression was seen when LIM domains were positioned either amino-terminal (Fig. 5, lanes 3-5) or carboxyl-terminal (Fig. 4, lane  4) to activation domains. Despite this observation, the COOH terminus activation domain is not repressed by LIM domains (Fig. 2A, lane 4). Therefore, it is important to note that in context of the whole RBTN-2 protein, the LIM domains selectively repress the NH 2 terminus activation domain (Fig. 4, lane  4). The mechanism of this selective repression is not known.
The selective repression of the NH 2 -terminal domain provides a mechanism for differential activity of RBTN-2. It is possible that repression of the NH 2 -terminal domain is abrogated through binding LIM interacting proteins. Based on this hypothesis, LIM interacting proteins would be present in yeast, but not in COS cells, since the NH 2 -terminal domain is active in yeast despite the presence of the LIM domains (Fig. 1). It is likely that the NH 2 -terminal domain is also active in mammalian cells. The unmasking of the NH 2 -terminal domain activity in selective tissues, and/or during fetal development would be dependent on expression of LIM interacting proteins in these tissues. Tissue-specific unmasking of activation domains is known to occur, e.g. an activation domain of E2A functions only in pancreatic beta cells (38).
In summary, our study establishes that the primary cellular role of T-cell oncogene, RBTN-2, is to serve as a transcription regulator. The mapping of two distinct activation domains and two repression domains suggest a complex behavior of RBTN-2 on transcription regulation. In this context, RBTN-2 is similar to other complex, multifunctional transcription regulatory proteins such as p53 (43), c-Fos (39), SRF (44), and IRF (23), that possess multiple negative and positive regulatory domains. The presence of multiple transcription regulatory modules suggests that the overall activity of RBTN-2 depends on the interplay among these activation and repression domains, as well as on the differential binding of LIM domain to different target proteins during fetal development and in different tissues.