A repetitive element containing a critical tyrosine residue is required for transcriptional activation by the EWS/ATF1 oncogene

Abstract

Chromosomal fusion of the N-terminal region of the Ewings Sarcoma Oncogene (EWS-activation-domain, EAD) to the DNA-binding domains of a variety of cellular transcription factors produce oncogenic proteins (EWS-fusion proteins (EFPs)) that cause distinct malignancies. In EFPs, the EAD acts as a potent transcriptional activation domain and this ability is repressed in the context of normal, non-tumorigenic, EWS. Trans-activation by the EAD is therefore a specific characteristic of EFPs and it is thought that EFPs induce tumorigenesis via improper transcriptional activation of cellular genes. Functional elements required for transcriptional activation are dispersed throughout the EAD, as are thirty-one copies of a Degenerate Hexapeptide Repeat (DHR, consensus SYGQQS). This suggests that the EAD contains a highly reiterated functional element related to DHRs. Here we show that in the context of EWS/ATF1, the EFP that causes malignant melanoma of soft parts, trans-cooperation by small regions of the EAD (30 residues) results in potent transcriptional activation dependent on the conserved tyrosine residues present in DHRs. These findings provide the first evidence for a role of DHRs in EAD-mediated trans-activation and demonstrate that the EAD represents a novel tyrosine-dependent transcriptional activation domain.

Introduction

Chromosomal translocations involving fusion of the N-terminal region of the Ewings Sarcoma Oncogene (EWS) to a variety of cellular transcription factors, produce dominant oncogenes that cause distinct sarcomas (May and Denny, 1997; Rauscher, 1997; Ron, 1997). For all of the above malignancies, the EWS-fusion proteins (EFPs) function as potent transcriptional activators dependent on the EWS N-terminal region (EWS-activation-domain (EAD)) and a C-terminal DNA-binding domain contributed by the fusion partner. The spectrum of malignancies associated with EFPs are thought to arise via EFP-induced transcriptional deregulation, with the tumor type specified by the EWS fusion partner and cell type.

EWS/ATF1 (the EFP which causes malignant melanoma of soft parts, MMSP (Zucman et al., 1993) is formed by fusion of the EAD to the C-terminal region of ATF1 (Figure 1). ATF1 itself activates transcription upon phosphorylation by protein kinase A (Lee and Masson, 1993; Ribeiro et al., 1994) while in contrast, EWS/ATF1 functions as a potent constitutive activator of several PKA-inducible promoters (Brown et al., 1995; Fujimura et al., 1996; Pan et al., 1998) dependent on both the EAD (Brown et al., 1995) and the DNA-binding (bZIP) domain of ATF1 (Fujimura et al., 1996; Pan et al., 1998). Regions of ATF1 other than the bZIP domain, including the glutamine rich activation domain, are not required for strong activation by EWS/ATF1 (Brown et al., 1995; Pan et al., 1998), indicating that the EAD contributes most of the activity. Studies of EWS/ATF1, other EFPs (May et al., 1993; Ohno et al., 1993; Prasad et al., 1994; Sanchez-Garcia and Rabbitts, 1994) and Gal4/EAD fusions (Lessnick et al., 1995; Pan et al., 1998; Kim et al., 1998) have provided some insights into the mechanism of EAD action. First, it has been established that the EAD can act autonomously as an activation domain (Lessnick et al., 1995; Pan et al., 1998; Kim et al., 1998). Second, multiple dispersed elements are required for full EAD activity (Pan et al., 1998; Lessnick et al., 1995; Kim et al., 1998). Third, binding of the EAD to the RNA polymerase II subunit hsRPB7 (Petermann et al., 1998; Bertolotti et al., 1998) and the correlation between hsRPB7 binding and trans-activation (Li and Lee, 2000), suggested an important role for hsRPB7. Such a role has recently been demonstrated directly using a yeast assay for EAD-mediated trans-activation (Zhou and Lee, 2001).

Figure 1
figure1

Structure and function of EWS/ATF1. EWS/ATF1 is a naturally occurring oncogenic EWS-Fusion-Protein (EFP) that is a causative agent of malignant melanoma of soft parts (Zucman et al., 1993). EWS/ATF1 contains the C-terminal region of ATF1 (residues 66–271) and the N-terminal region of EWS (the EWS-Activation Domain, EAD). The bZIP domain of ATF1 mediates dimerization and DNA-binding. EWS/ATF1 lacks the PKA phosphoacceptor site of ATF1 and is a potent constitutive activator of ATF-dependent promoters, dependent on the EAD. The N-terminal 82 residues of the EAD binds to the RNA Pol II subunit hsRPB7, residues 228–264 (ZFM1) bind to the transcriptional repressor/splicing protein ZFM1/SF1 (Zhang et al., 1998) and residues 258–280 (IQ) contain an IQ domain involved in calmodulin binding and phosphorylation by protein kinase C (Deloulme et al., 1997). The EAD contains thirty-one Degenerate Hexapeptide Repeats (DHRs (purple boxes)); consensus sequence SYGQQS that constitute 70% of the EAD. The sequences of all DHRs present in the EAD are summarized (box marked DHRs (EAD); numbers represent the per cent occurrence for each amino acid). The sequence of the EAD region (residues 8–40) present in Z33 is shown and contains four DHRs (DHR1–4) and spacer residues. DHR residues (purple) with conserved tyrosine (red, position 2) and glutamine (red, position 4) and spacer residues (grey) are shown. Below the amino acid sequence present in Z33, DHRs 1–4 are represented by distinct ovals to represent their different primary sequences. For DHRs 1–4, amino acid substitutions in each position relative to consensus are summarized (box marked DHRs (Z33))

Structure/function relationships for the EAD itself are poorly characterized. Similar to other activation domains (Triezenberg, 1995) the EAD has little predicted secondary structure in solution but is rich in certain amino acids including proline, glutamine and serine/threonine. Another similarity is the presence of a repetitive primary structure (Delattre et al., 1992) which, for the EAD, consists of thirty-one copies of a Degenerate Hexapeptide Repeat (DHR, consensus SYGQQS, Figure 1). DHRs also confer some distinctive features to the EAD. First, the absolutely conserved tyrosine in position 2 of the DHR (Delattre et al., 1992, see Figure 1) suggests an important function for this residue that is not shared by other activation domains. Second, the EAD has a high content of neutral polar residues and a very low content of hydrophobic residues. This latter characteristic is in contrast to both acidic (Regier et al., 1993; Blair et al., 1994) and glutamine rich activators (Gill et al., 1994; Tanaka and Herr, 1994) in which bulky hydrophobic residues are critical for function.

A salient feature of the EAD is that it is almost entirely composed (70%) of DHRs (Figure 1) suggesting that DHRs must play a crucial role in trans-activation. However, synthetic activators containing consensus DHRs fused to ATF1 have little transcriptional activity on a reporter containing a single ATF1 binding site (Pan et al., 1998). In contrast, the most active region of the EAD (residues 1–86, (Pan et al., 1998)) contains more degenerate DHRs (Figure 1). This suggests that deviations away from consensus may increase the trans-activation potential of DHRs. To date, DHRs have not been systematically studied because analysis of such highly repetitive elements requires a sufficiently sensitive assay for small regions of the EAD (approximately 40 residues or 120 nucleotides). Here we show that promoters containing multiple activator binding sites can support potent trans-activation by a protein (Z33) that contains only 33 residues of the EAD fused to ATF1. Mutational analysis of Z33 using the above assay demonstrates that DHRs and the conserved tyrosine residues therein, play a crucial role in trans-activation by EWS/ATF1. In addition, our results show that the EAD represents a novel, tyrosine-dependent, transcriptional activation domain.

Results

An assay for small regions of the EAD

The goal of our studies was to examine the involvement of DHRs in EAD-mediated trans-activation. This task requires analysis of multiple simultaneous mutations and can only be achieved via total synthesis of mutated oligonucleotides and their subsequent functional analysis. This in turn requires establishment of a transcription assay for small regions of the EAD (40 residues) containing only a limited number of DHRs. For two main reasons, we started our analysis by testing the extreme N-terminal region of the EAD (Figure 1) fused to ATF1. First, in our previous studies of EWS/ATF1, this region has higher activity than other regions of the EAD (Pan et al., 1998). Second, hsRPB7 is known to bind to residues 1–82 of the EAD (Petermann et al., 1998) and hsRPB7 is required for EAD-mediated trans-activation (Zhou and Lee, 2001).

Initially, we tested the feasibility of using a previously described transient assay (Brown et al., 1995) for EWS/ATF1 or mutants thereof. Briefly, an expression vector for the activator protein and a CAT reporter linked to a truncated somatostatin promoter (Δ(−71)SomCAT, which contains a single ATF binding site) are transiently introduced into cells lacking EWS/ATF1 (JEG3 cells) and transcriptional activity is monitored by CAT assay. In this assay, the ATF1 portion of EWS/ATF1 (Δ325, Figure 2a) is essentially inactive (Figure 2b) and addition of the N-terminal 86 residues of the EAD (Δ87C) gives readily detectable trans-activation (80-fold greater than Δ325) as previously shown (Pan et al., 1998). Further deletion analysis indicated that the determinants of trans-activation for Δ87C cannot be mapped to a particular region but are dispersed. For example, the N-terminal 42 residues (Δ42, Figure 2b) exhibited only modest activity (10-fold greater than Δ325). Significantly however, duplication of residues 1–42 (Δ42D) gave activity comparable with that of Δ87C, consistent with the idea that Δ87C contains a repeated functional element that cooperates synergistically in cis.

Figure 2
figure2

An assay for trans-activation by small regions of the EAD. (a) Structure of activators. Δ325 contains ATF1 residues 66–271, corresponding to the region of ATF1 present in EWS/ATF1. Δ87C, Δ42 and Δ42D each contain Δ325 fused to different residues of the EAD as indicated. ZΔE is equivalent to Δ325 with the Zta bZIP domain in place of the ATF1 bZIP domain. Z57 contains ZΔE fused to EAD residues 1–57. (b) Trans-activation assays using a single site reporter. JEG3 cells were transfected with 5 μg pΔ(−71)SomCAT as reporter and 5 μg of activator plasmids expressing Δ325, Δ87C, Δ42 and Δ42D as indicated at the top. CAT assays and Western blots of epitope tagged proteins were performed at 40 h post-transfection. An autoradiogram of a representative CAT assay is shown to the left (c=chloramphenicol and ac=acetylated chloramphenicol) and the corresponding Western blot to the right. The position of molecular weight standards (Biorad pre-stained low size range) are indicated to the right. (c) Trans-activation assays using a multiple site reporter. Arrangement of the figure and experimental procedure is the same as in (b). Experiments were performed using 5 μg pZ1E4TCAT (Z1), pZ3E4TCAT (Z3) or pZ7E4TCAT (Z7) as reporters and 5 μg of activator plasmids expressing ZΔE or Z57

The low absolute activity (Δ42 has only 0.6% of the activity of the intact EAD) and minimal trans-activation capacity of Δ42 were not amenable to further analysis. In light of the cis-cooperation exhibited by Δ42D (Figure 2b) and many other regions of the EAD (data not shown) we reasoned that small regions of the EAD might be able to cooperate in trans. We tested this idea by using a series of reporters containing different numbers of activator binding sites (Carey et al., 1992). We also exploited a heterologous bZIP domain, from the Zta protein (Carey et al., 1992) as previously used to study ATF1 (Ribeiro et al., 1994) and EWS/ATF1 (Pan et al., 1998). This ‘bZIP swap’ eliminates the potential influence of endogenous dimerization partners of ATF1 (Ribeiro et al., 1994) and is useful for DNA-binding studies of mutant activators due to lack of endogenous proteins with Zta DNA-binding specificity (see Figure 3). Initially, a protein called Z57 (Figure 2a) containing the N-terminal 57 residues of the EAD, the region of ATF1 present in EWS/ATF1 (minus the ATF1 bZIP domain) and including the Zta bZIP domain, was tested for the ability to activate Zta reporters containing 1, 3 and 7 Zta binding sites (pZ1E4TCAT, pZ3E4TCAT and pZ7E4TCAT, respectively). Z57 failed to detectably activate pZ1E4TCAT (Figure 2c) but activation was readily detected using pZ3E4TCAT (18-fold activation) and was very potent using pZ7E4TCAT (285-fold activation). The control protein lacking the EAD (ZΔE) exhibited minimal activity using pZ7E4TCAT as reporter (Figure 2c). Thus, multiple Zta binding sites increased trans-activation by Z57 in a highly synergistic manner, dependent on the EAD. Such synergistic activity has been reported for the Zta protein using the same reporters (Carey et al., 1992) and is due to a direct effect on trans-activation and not DNA-binding (Carey et al., 1992) indicating that synergistic activation by Z57 most likely reflects a direct increase in trans-activation. As a final step in assay design we established that a protein called Z33 (which is identical to Z57 except that it contains only residues 8–40 of the EAD) strongly activated pZ7E4TCAT (156-fold activation, Figure 3). The region of the EAD present in Z33 is small enough to allow efficient construction of desired mutants, as described, and thus allowed the mutational analysis described below.

Figure 3
figure3

Mutational analysis of Z33. (a) Quantitation of trans-activation. JEG3 cells were transfected with 5 μg of Z7E4TCAT as reporter and 0.15 μg plasmid expressing activator protein. Trans-activation is expressed as a per cent of Z33 with the s.e.m. indicated as an error bar. Absence of an error bar indicates that the data represent the average from two experiments only. (b) Expression of mutant proteins. Expression levels of mutant proteins (except Y1–4, F12, F34) as determined by Western blot. (c) DNA-binding assays. Extracts from transfected cells were incubated with 1 ng of 32P-labeled probe containing a Zta binding site and DNA-protein complexes resolved on non-denaturing polyacrylamide gels. Test proteins are indicated at the bottom, positions of free probe and DNA-protein complex to the side and the presence of excess competitor oligonucleotide (100 ng) at the top. The unmutated competitor oligonucleotide (wt) contains a Zta binding site with the TTGCTAA core motif (Ribeiro et al., 1994) and for the mutant competitor oligonucleotide (m) the core motif is mutated to GACACAC

Role of tyrosine residues in trans-activation by Z33

Systematic analysis of individual residues within DHRs (in Z33) is complicated because of the degeneracy and because alanine substitutions are naturally present in positions 1, 3, 5 and 6 (Figure 1). In contrast, the Tyr residue at position 2 is absolutely conserved and the Gln residue in position 4 is strongly conserved. We therefore tested the effect of altering these positions to Ala (Figure 3). Simultaneously changing Tyr to Ala in all four DHRs (YA) reduced trans-activation to almost background levels (activity of YA is 1.1% of Z33, Figure 3a) while changing the Gln to Ala has a relatively modest effect on trans-activation (28% of Z33). Thus we conclude that for Z33, the Tyr residue in position 2 of the DHRs is critical for trans-activation while the Gln in position 4 is much less important.

To further examine the role of Tyr residues we analysed a series of additional mutants. Changing any two out of four Tyr residues (Figure 3a. Y12 (3.2% of Z33), Y23 (3.8%), Y13 (4.5%), Y24 (1.9%) and Y14 (6.7%); for nomenclature Y12 has Tyr to Ala changes in DHR1 and DHR2 and so forth) also greatly reduced trans-activation (average of 4% of Z33) and changing one Tyr alone (Y1, 21% of Z33; Y2, 27%; Y3, 24%; and Y4, 30%) had a significant but much smaller effect in each case (Figure 3a, average of 25% of Z33). All of the mutant proteins were expressed at similar levels (Figure 3b and data not shown). To gain a more precise insight into the structural requirement for tyrosine residues we altered Tyr to Phe. Simultaneously changing all four Tyr residues to Phe (YF) had a significant effect on trans-activation (9.9% of Z33) but had much less effect than the corresponding changes to Ala (1.1% of Z33). Similarly, changing only two Tyr residues to Phe had a relatively modest effect (F12, 31% of Z33; F34, 33% of Z33) compared with the corresponding effect of changes to Ala (average 4% of Z33). Thus the aromatic ring of the conserved tyrosine residues contributes significantly to trans-activation by Z33.

To confirm that the effects of altering Tyr residues were direct effects on trans-activation and were not due to effects on DNA-binding, we performed gel mobility shift assays (Figure 3c). Extracts from cells transfected with ZΔE and Z33 were incubated with a 32P-labeled probe containing a Zta binding site. A single DNA-protein complex of different mobility was observed for the two extracts (Figure 3c) and no complex was detected using extracts from untransfected cells (data not shown). DNA-protein complex formation was competed by an excess of unlabeled oligonucleotide containing a Zta binding site (wt) but not by an oligonucleotide containing a mutated Zta binding site (m). Dependence on exogenous protein, differential mobility and the observed DNA-binding specificity demonstrated that the DNA-protein complexes formed are due ZΔE and Z33 proteins. We tested selected mutants (YA, Y13, Y24, Y14 and YF) and found that all proteins tested have DNA-binding activity comparable with Z33.

We conclude from the above results that all of the conserved tyrosine residues present in the DHRs of Z33 contribute to trans-activation. Since the effect of altering each Tyr residue is similar (either singly or in pairs), this provides strong evidence that the individual Tyr residues are functioning in the same manner, as part of a repetitive element, the DHR. In addition, the Tyr residues strongly synergise with each other. For example, mutation of any two out of four Tyr residues is sufficient to reduce Z33 activity by approximately 25-fold instead of the twofold reduction predicted for an additive effect.

Role of DHRs in trans-activation by Z33

Given the effect of Tyr mutations described above and the fact that DHRs constitute 70% of Z33, it seemed possible that DHRs are the only functional elements within Z33. In addition, if Z33 contains a functional element not related to DHRs, then simultaneous alteration of the spacer residues (which are dispersed and together constitute the remaining 30% of Z33) would most likely disrupt such an element. We therefore tested the effect of different spacer mutations (S1–S4) on trans-activation (Figure 4). The amino acid changes are shown in Figure 4 and the rationale for the choice of residues was as follows. For S1, the spacers were substituted by those (of unrelated sequence) between the DHRs present in EAD residues 43–75. For S2 and S3, the spacer is changed to AQ and AQQ respectively. These sequences were chosen because AQQ corresponds to the functional spacer between DHR1 and DHR2. For S4, the spacer was changed to SGG which forms a flexible linker (Netzer and Hartl, 1997). All of the above mutants (S1–S4) exhibit only modestly reduced trans-activation compared with Z33 (S1, 62% of Z33; S2, 43% of Z33; S3, 42% of Z33; S4, 38% of Z33). Thus the above results indicate that specific spacer residues between DHRs are not critical for trans-activation by Z33. We also tested the effect of deleting the spacers entirely (ΔS) and found that this also had a relatively small effect on trans-activation (ΔS had 29% of the activity of Z33). This latter result indicates that DHRs alone have significant activity and that there is a weak requirement, if any, for a spacer between DHRs.

Figure 4
figure4

Role of DHRs in trans-activation by Z33. EAD residues 8–40 present in Z33 are shown at the top and are fused to ATF1 and Zta (not shown). The structure of different mutants is diagrammed below, with DHRs (numbered ovals or labeled with C to represent the consensus sequence SYGQQS) and spacer residues (open boxes marked S for WT or with the specific amino acid residues indicated for mutants). Experiments were performed as described in Figure 2 (with 5 μg of Z7E4TCAT reporter) except that 0.15 μg of activator plasmids (0.3 μg for S1) were used for transfection. Quantitation of trans-activation (expressed as per cent of Z33 with the s.e.m. indicated as an error bar) is shown graphically to the right. At the bottom, a Western blot shows expression levels for each activator, following transfection of cells with 5 μg of activator plasmids

As a final test of the involvement of DHRs in trans-activation by Z33 we asked whether different DHRs could functionally substitute for each other (Figure 4). Replacing the DHRs present in Z33 with consensus DHRs (CR) produced relatively modest levels of trans-activation (CR has 9.2% of the activity of Z33) indicating that consensus DHRs are not highly effective substitutes for the DHRs present in Z33. This result is consistent with the previous observation that synthetic activators containing only consensus DHRs are inactive on a single site reporter (Pan et al., 1998). Next, we asked whether a single DHR from Z33 can substitute for other DHRs present in Z33 (4R4 contains four copies of DHR4) or whether scrambling the order of DHRs in Z33 (SR2) maintains activity. Both 4R4 and SR2 had moderately reduced activity compared with Z33 (4R4, 28% of Z33; SR2 25.5% of Z33) indicating that DHRs can functionally substitute for each other quite effectively. With respect to the somewhat lower activity of 4R4 and SR2 compared with Z33, we suggest that due to the highly synergistic action of DHRs (as implied by the effect of Tyr mutations (Figure 3)) small alterations might result in relatively large changes in activity. Thus the small loss of activity observed for 4R4 and SR2 might readily be accounted for by juxtaposition of unfavorable sequences that either strongly impair a single DHR or slightly impair multiple DHRs. In addition it is important to emphasize that both 4R4 and SR2 are still approximately 40-fold more active than the control protein ZΔE and are therefore potent activators. Taken together, the above results show that DHRs represent a functional unit that is critical for trans-activation by Z33 and suggest that amino acid substitutions away from the DHR consensus (such as those present in Z33) significantly increase trans-activation.

Discussion

Structure of the EAD

Using EWS/ATF1 as a model to study transcriptional activation by EFPs, we herein provide evidence that DHRs play a major role in EAD-mediated trans-activation. This conclusion is based on several key findings as follows: (1) the absolutely conserved tyrosine residues present in DHRs play a critical role in trans-activation by Z33; (2) all of the conserved tyrosine residues present in Z33, either alone or when tested in pairs, make a similar contribution to trans-activation; (3) specific spacer sequences within Z33 play no obvious role in trans-activation; (4) DHRs can functionally substitute for each other, indicating that DHRs represent a functional unit.

With respect to the precise role of tyrosine side chains in trans-activation by the EAD, potential effects of Tyr phosphorylation remain to be examined. However, we suggest that Tyr phosphorylation is unlikely to be involved in trans-activation. First, although EWS can associate with several tyrosine kinases (Guinamard et al., 1997; Felsch et al., 1999; Kim et al., 1999) and the EAD contains multiple potential tyrosine kinase interaction sites (Mayer et al., 1992; Songyang et al., 1993; 1995), we have been unable to detect Tyr phosphorylation within the N-terminal 86 residues of the EAD (KAW Lee, 2001, unpublished results) which corresponds to the most active region of the EAD (Pan et al., 1998). Second, the effect of altering Tyr to Phe in two out of four DHRs in Z33 is not dramatic (only threefold) and is considerably less than the corresponding Tyr to Ala changes (25-fold). Thus our data are more consistent with the possibility that the aromatic ring of the conserved Tyr in DHRs plays a direct role in trans-activation.

Besides the conserved tyrosine, the degree of degeneracy within DHRs does not allow a clear definition of DHRs and our study does not address the potential effect of serine and threonine phosphorylations (although again we have been unable to detect such phosphorylations within Z33, KAW Lee, 2001, unpublished results). However our results do enable some additional insights. Consensus (SYGQQS) DHRs have relatively low activity and amino acid substitutions (such as those present in Z33) increase activity. This is consistent with the previous observation that synthetic activators containing only consensus DHRs are inactive when assayed on a single site reporter (Pan et al., 1998) and clearly shows that, with the exception of Tyr, enrichment of particular amino acids (notably serine, glycine and glutamine) is not the primary determinant of EAD activity. The major effect of amino acid substitutions present in Z33 is to decrease the number of polar residues relative to the consensus DHR, suggesting that a slightly less hydrophilic surface is favorable for trans-activation. Overall, our results suggest that the EAD is composed of a number of critical tyrosine residues surrounded by polar/uncharged residues that favor hydrogen bonding interactions and/or flexibility.

Mechanism of EAD action

The results presented here and our previous findings (Pan et al., 1998) suggest that DHRs may be the sole determinant of EAD activity. First, we have shown that DHRs are sufficient for trans-activation by a small region of the EAD (present in Z33). Second, the other known elements of the EAD including the ZFM1 binding site (residues 228–264) and the IQ domain (residues 258–280) have a minor effect on trans-activation when deleted from the EAD (Pan et al., 1998) and in any event, contain multiple DHRs. Third, duplication of EAD residues 1–86 creates a protein with sixteen DHRs which has high absolute activity (only threefold lower than the intact EAD) when assayed on a single site reporter (Pan et al., 1998). Considering all of the above, we favor a model in which multiple DHRs (each with variable activity) strongly cooperate, in cis, to create an efficient activation domain.

Several considerations suggest that our findings for EWS/ATF1 are most likely to be applicable to EFPs in general. We have previously shown that except for the ATF1 bZIP domain (which is required to recruit the EAD to the promoter), other regions of ATF1 do not play a major role in trans-activation by EWS/ATF1 (Brown et al., 1995; Pan et al., 1998). In addition, the EAD is a very potent activation domain when fused to the DNA-binding domain of Gal4 (Kim et al., 1998; KAW Lee, 2001 unpublished results) and can therefore function autonomously as an activation domain. Finally, several EAD mutants produce similar relative levels of trans-activation in the context of either EWS/ATF1 or Gal4 (Li and Lee, 2001, unpublished results). We note, however, that in contrast to our studies, the extreme N-terminal region of the EAD appears to be less important for overall EAD activity in certain situations (Lessnick et al., 1995), suggesting that the activity of particular regions of the EAD and hence, particular DHRs, may be influenced by either cell type or EFP context. This possibility remains to be thoroughly examined.

The presence of small highly reiterated elements that function synergistically is a common feature of activation domains (Seipel et al., 1992; Blair et al., 1994; Tanaka and Herr, 1994). Synergistic effects appear to involve three basic mechanisms, namely, cooperative DNA-binding, stimulation of different steps in transcription complex assembly and multiple protein/protein contacts that increase transcription complex stability (reviewed by Carey, 1998). Two factors indicate that cooperative DNA-binding does not account for our results. First, synergistic activity has been reported for the Zta protein itself using the reporters that we have used (Carey et al., 1992) and is due to a direct effect on trans-activation. Second, the ability of small regions of the EAD to synergise, in cis, on a single site reporter (Figure 2c) is unlikely to involve effects on DNA binding. The EAD is known to directly contact a limited number of transcriptional components including multiple TAFs (Bertolotti et al., 1998), and two subunits of RNA polymerase II, hsRPB7 (Petermann et al., 1998; Bertolotti et al., 1998) and hsRPB5 (Bertolotti et al., 1998). Synergistic activation by cis-linked DHRs within the EAD might therefore reflect a highly efficient increase in the local concentration of DHRs in the vicinity of critical target proteins such as hsRPB7. To date, hsRPB7 has only been demonstrated to interact with the N-terminal 82 residues of the EAD (Petermann et al., 1998) but it is of interest to determine whether hsRPB7 can bind throughout the EAD, as might be predicted.

Relationship of the EAD to normal activators

Superficially, the EAD has much in common with several other transcriptional-activation domains (Triezenberg, 1995). However, the EAD lacks functionally important hydrophobic side chains characteristic of both acidic (Regier et al., 1993; Blair et al., 1994) and glutamine rich (Gill et al., 1994; Tanaka and Herr, 1994) domains and the DHR also lacks the proline rich feature important for activation by CTF1 (Kim and Roeder, 1994). In addition, Tyr is less effective than Phe in position 442 of the VP16 acidic activation domain (Regier et al., 1993) while for the EAD, Tyr is more effective than Phe. These observations demonstrate that the EAD represents a novel tyrosine-dependent transcriptional activation domain. The EAD shows a strong resemblance to the activation domain of the stage-specific activator protein (SSAP) (Benuck et al., 1999) which is rich in glutamine, glycine and also tyrosine and has only a low content of acidic residues. For SSAP, a genetic screen identified several acidic gain of function mutations but no effects on tyrosine were reported (Benuck et al., 1999). Given the similarities between the EAD and SSAP it would be of interest to determine the role of tyrosine residues in trans-activation by SSAP and to establish whether SSAP and the EAD activate transcription by a related mechanism.

Future prospects

The work of several groups has revealed that EFPs are involved in tumor maintenance (reviewed by Kovar et al., 1999) raising the possibility that EAD inhibitors will have therapeutic potential. The absolute tumor specificity of EFPs together with the clear functional distinctions between EFPs and EWSs further suggests that EFPs are attractive therapeutic targets. Our results identify a small peptide motif, the DHR, that may represent an effective target for drug design/screening. Moreover, since DHRs are present in multiple copies within the EAD it may be that small molecule inhibitors that target DHRs would be particularly effective. Such a scenario has been proposed in the case of other transcriptional activators (Nyanguile et al., 1997; Pollock and Gilman, 1997). Further understanding of the mechanism of EAD action will depend on multiple approaches involving characterization of the key molecular contacts between the EAD and transcriptional components and the development of in vitro systems that can be biochemically manipulated. Overall, the panel of mutants described here and their functional characterization in vivo, provide useful reagents and a firm basis for future molecular studies of the EAD.

Materials and methods

Plasmids

pΔ(−71)SomCAT (Montminy et al., 1986), pΔ87C, pΔ207C, pΔ325 (Pan et al., 1998), pSVEZΔA (Li and Lee, 2000) and pZ1E4TCAT, pZ3E4TCAT and pZ7E4TCAT (Carey et al., 1992) are as previously described. pΔ42 was obtained by partial digestion of pΔ87C (Pan et al., 1998) with BglII and NdeI and insertion of an oligonucleotide encoding EAD residues 34–42 to leave a unique BglII site at position 42. pΔ42D was obtained by inserting a BglII ended PCR product into pE42 digested with BglII. pPBWT was obtained by inserting PstI/BglII-ended oligonucleotides encoding EAD residues 12–57 into pΔ207C digested with BglII and PstI. pZ57 was obtained by digestion of pSVEZΔA (containing the Zta bZIP domain) with HindIII and NdeI and insertion of a HindIII/NdeI partial fragment from pPBWT. pZΔE was obtained by digestion of pSVEZΔA with HindIII and NdeI and insertion of a HindIII/NdeI fragment from pΔ325. All other mutants were constructed as follows. Four oligonucleotides (each of 65 nucleotides) were synthesized such that when ligated together (via a 5 base pair cohesive overhang) produce the coding sequence of the EAD (from residues 8–42 or the desired mutations), an efficient translation initiation site (GAGAAAATGGCG) and HindIII/BglII overhangs. Oligonucleotides were inserted in one step into pSVEZΔA digested with HindIII and BglII and recombinants sequenced (using an ABI automated sequencer) to identify clones without extraneous mutations.

Transfections, CAT assays and Western blotting

Transfections by calcium phosphate co-precipitation, CAT assays and Western blotting (Pan et al., 1998) using antibody KT3 (MacArthur and Walter, 1984) were carried out as previously described except that cell extracts used for CAT assays were also used directly for Western blots. For experiments using pΔ(−71)SomCAT as reporter, cells were transfected with 5 μg of activator plasmid. For experiments using pZ7E4TCAT as reporter, activation by Z33 is in the linear range at or below 0.3 μg and unless stated otherwise, 0.15 μg of activator plasmid was used for transfections. For quantitation, per cent conversion of unacetylated to acetylated 14C-chloramphenicol under linear assay conditions (for both transactivation and CAT assays) was determined by excision of spots from the TLC plate and quantitation using a liquid scintillation counter. Most experiments were repeated multiple times to allow calculation of standard error values (s.e.m.). In cases without error bars the experiment was performed only twice.

Gel mobility shift assays

Assays were performed as previously described (Krajewski and Lee, 1994) except for the following. Cell extracts in CAT assay lysis buffer (in 250 mM Tris-Cl pH 7.8, 1 mM EDTA) were used directly for DNA-binding assays. Fifteen μl reactions contained 1 μl (adjusted to equal A280) of CAT extract which equals 1% of the extract prepared from 107 cells. For competition experiments, all components including competitor oligonucleotide but excluding cell extract were mixed, followed by addition of cell extract.

References

  1. Benuck ML, Li Z, Childs G . 1999 J. Biol. Chem. 274: 25419–25425

  2. Bertolotti A, Melot T, Acker J, Vigneron M, Delattre O, Tora L . 1998 Mol. Cell. Biol. 18: 1489–1497

  3. Blair WS, Bogerd HP, Madore SJ, Cullen BR . 1994 Mol. Cell. Biol. 14: 7226–7234

  4. Brown AD, Lopez-Terrada D, Denny CT, Lee KAW . 1995 Oncogene 10: 1749–1756

  5. Carey M . 1998 Cell 92: 5–8

  6. Carey M, Kolman J, Katz DA, Gradoville L, Barberis L, Miller G . 1992 J. Virol. 66: 4803–4813

  7. Delattre O, Zucman J, Plougastel B, Desmaze C, Melot T, Peter M, Kovar H, Joubert I, de Jong P, Rouleau G, Aurias A, Thomas G . 1992 Nature 359: 162–165

  8. Deloulme JC, Prichard L, Delattre O, Storm DR . 1997 J. Biol. Chem. 272: 27369–27377

  9. Felsch JS, Lane WS, Peralta EG . 1999 Curr. Biol. 9: 485–488

  10. Fujimura Y, Ohno T, Siddique H, Lee L, Rao VN, Reddy ESP . 1996 Oncogene 12: 159–167

  11. Gill G, Pascal E, Tseng ZH, Tjian R . 1994 Proc. Natl. Acad. Sci. 91: 192–196

  12. Guinamard R, Fougereau M, Seckinger P . 1997 Scand. J. Immunol. 45: 587–595

  13. Kim TK, Roeder RG . 1994 Nucleic Acids Res. 22: 251

  14. Kim J, Lee KAW, Pelletier J . 1998 Oncogene 16: 1021–1030

  15. Kim J, Lee JM, Branton PE, Pelletier J . 1999 Proc. Natl. Acad. Sci., USA 96: 14300–14305

  16. Kovar H, Aryee D, Zoubek A . 1999 Curr. Op. Oncol. 11: 275–284

  17. Krajewski W, Lee KAW . 1994 Mol. Cell. Biol. 14: 7204–7210

  18. Lee KAW, Masson N . 1993 Biochem. Biophys. Acta. 1174: 221–233

  19. Lessnick SL, Braun BS, Denny CT, May WA . 1995 Oncogene 10: 423–431

  20. Li KKC, Lee KAW . 2000 J. Biol. Chem. 275: 23053–23058

  21. MacArthur H, Walter G . 1984 J. Virol. 52: 483–491

  22. May WA, Lessnick SL, Braun BS, Klemsz M, Lewis BC, Lunsford LB, Hromas R, Denny CT . 1993 Mol. Cell. Biol. 13: 7393–7398

  23. May WA, Denny CT . 1997 Curr. Top. Micro. Immun. 220: 143–150

  24. Mayer BJ, Jackson PK, Van ER, Baltimore D . 1992 Mol. Cell. Biol. 12: 609–618

  25. Montminy MR, Sevarino KA, Wagner JA, Mandel G, Goodman RH . 1986 Proc. Natl. Acad. Sci., USA 83: 6682–6686

  26. Netzer WJ, Hartl FU . 1997 Nature 388: 343–349

  27. Nyanguile O, Uesugi M, Austin DJ, Verdine GL . 1997 Proc. Natl. Acad. Sci., USA 94: 13402–13406

  28. Ohno T, Rao VN, Reddy ESP . 1993 Cancer Res. 53: 5859–5863

  29. Pan S, Ming KY, Dunn TA, Li KKC, Lee KAW . 1998 Oncogene 16: 1625–1631

  30. Petermann R, Mossier BM, Aryee DNT, Khazak V, Golemis EA, Kovar H . 1998 Oncogene 17: 603–610

  31. Pollock R, Gilman M . 1997 Proc. Natl. Acad. Sci., USA 94: 13388–13389

  32. Prasad DDK, Ouchida M, Lee L, Rao VN, Reddy ESP . 1994 Oncogene 9: 3717–3729

  33. Rauscher III FJ . 1997 Curr. Top. Micro. Immun. 220: 151–162

  34. Regier JL, Shen F, Triezenberg SJ . 1993 Proc. Natl. Acad. Sci., USA 90: 883–887

  35. Ribeiro A, Brown AD, Lee KAW . 1994 J. Biol. Chem. 269: 31124–31128

  36. Ron D . 1997 Curr. Top. Micro. Immun. 220: 131–142

  37. Sanchez-Garcia I, Rabbitts TH . 1994 Proc. Natl. Acad. Sci., USA 91: 7869–7873

  38. Seipel K, Georgiev O, Schaffner W . 1992 EMBO J. 11: 4961–4968

  39. Songyang Z, Shoelson SE, Chaudhuri M, Gish G, Pawson T, Haser WG, King F, Roberts T, Ratnofsky S, Lechleider RJ, Neel BG, Birge RB, Fajardom JE, Chou MM, Hanafusa H, Schaffhausen B, Cantley LC . 1993 Cell 72: 767–778

  40. Songyang Z, Carraway III KL, Eck MJ, Harrison SC, Feldman RA, Mohammadi M, Schlessinger J, Hubbard SR, Smith DP, Eng C, Lorenzo MJ, Ponder BAJ, Mayer BJ, Cantley LC . 1995 Nature 373: 536–539

  41. Tanaka M, Herr W . 1994 Mol. Cell. Biol. 14: 6056–6067

  42. Triezenberg SJ . 1995 Curr. Opin. Genet. Dev. 5: 190–196

  43. Zhang D, Paley AJ, Childs G . 1998 J. Biol. Chem. 273: 18086–18091

  44. Zhou H, Lee KAW . 2001 Oncogene 20: 1519–1524

  45. Zucman J, Delattre O, Desmaze C, Epstein AL, Stenman G, Speleman F, Fletchers CDM, Aurias A, Thomas G . 1993 Nature Genetics 4: 341–345

Download references

Acknowledgements

We thank Dr Zhang Mingjie for many discussions during the course of this work, Dr Chris Rock for helpful comments on the manuscript and Kim KC Li for excellent technical assistance. This work was supported by a Hong Kong Government Research Grants Council grant (award HKUST 6106/98M) to KAW Lee.

Author information

Correspondence to Kevin A W Lee.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Feng, L., Lee, K. A repetitive element containing a critical tyrosine residue is required for transcriptional activation by the EWS/ATF1 oncogene. Oncogene 20, 4161–4168 (2001) doi:10.1038/sj.onc.1204522

Download citation

Keywords

  • EWS/ATF1 oncogene
  • transcription
  • activation domain
  • tyrosine

Further reading