The ETS Domain Transcription Factor Elk-1 Contains a Novel Class of Repression Domain

Abstract

The ETS domain transcription factor Elk-1 serves as an integration point for different mitogen-activated protein (MAP) kinase pathways. Phosphorylation of Elk-1 by MAP kinases triggers its activation. However, while the activation process is well understood, its downregulation-inactivation is less well characterized. The ETS DNA-binding domain plays a role in the downregulation of Elk-dependent promoter activity following mitogenic activation by recruiting the mSin3A-HDAC complex. Here we have identified a novel evolutionarily conserved repression domain in Elk-1, termed the R motif, which serves to reduce the basal transcriptional activity of Elk-1 and dampen its response to mitogenic signals. This domain is highly potent and portable and can repress transcription in trans. The R motif is related to the CRD1 repression domain in p300 and can functionally replace this domain and confer p21^waf1/cip1 inducibility on p300. However, the R motif acts in a context-dependent manner and is not p21^waf1/cip1 responsive in Elk-1. Thus, the Elk-1 R motif and the p300 CRD1 motif represent a new class of repression domains that are regulated in a context-dependent manner.

The ternary complex factors (TCFs) represent a subfamily of ETS-domain transcription factors (reviewed in references 26, 31, 32, and 36). They are thought to act primarily through forming nucleoprotein complexes with serum response factor (SRF) and composite binding sites that contain CArG (for SRF) and ets (for the TCFs) binding motifs. These proteins have been implicated in transducing mitogenic and stress signals to immediate-early genes, such as c-fos, to elicit a transient and rapid gene activation response (reviewed in reference 3). This transient response necessitates a tightly controlled series of activation and repression events to ensure a correctly timed spike of induction. Much focus has been centered on how the activation process takes place, although less is known about how the low level basal state is maintained and reestablished following gene activation.

Each of the TCFs contains four domains of high sequence and functional similarity. The N-terminal part of these proteins contains the ETS DNA-binding domain. The B-box region that is responsible for interactions with SRF is located downstream from this (9, 27). The C-terminal end of the protein contains a transcriptional activation domain (C domain) (13, 20) which is preceded by a short MAP kinase docking motif (D domain) (5, 12, 35, 36). Following binding of the MAP kinases to this motif, the TCFs are phosphorylated in their transcriptional activation domains at multiple sites, which leads to potentiation of their transcriptional activation potential (7, 8, 14, 20, 22). In addition to these conserved domains, the TCFs also contain regions of lower similarity that might define the individual functions of the proteins. Indeed, in the case of SAP-2, two repression domains have been identified. The NID is located immediately downstream from the B box (18) whereas the CID is located in the central portion of the protein (4). While the molecular function of the NID is unknown, the CID acts by recruiting the CtBP corepressor complex. Neither repression domain is conserved in Elk-1.

Recently, we demonstrated that the Elk-1 ETS domain participates in the downregulation of its activity. The ETS domain has been shown to be the target of Id proteins that downregulate promoter-bound TCF complexes (40). In addition, the ETS domain plays a pivotal role in recruiting the Sin3A/HDAC corepressor complex that represses Elk-1-regulated promoters following mitogenic stimulation (39). However, little is known about how Elk-1 is kept inactive and how the c-fos promoter is shutdown prior to receiving activating stimuli. This is particularly important as Elk-1 or other TCFs are thought to continuously occupy target promoters, such as c-fos (10), and are thought to always be part of a complex with CBP (15).

In this study, we have dissected the C-terminal region of Elk-1 and identified a novel evolutionarily conserved repression domain, the R motif. This domain is conserved in many other important transcription factors, including p300. The domain is highly portable and is sufficient to mediate transcriptional repression and can be regulated in heterologous contexts. In Elk-1, the main function of this domain appears to be to dampen down the activity of its transcriptional activation domain and thus elicit a further tier of regulation on this transcription factor.

MATERIALS AND METHODS

Plasmid constructs.

The following plasmids were constructed and used for mammalian cell transfections. pG5-E1B-Luc (25), pG5-TK-Luc (pAS1467) (39), and pG5-E4-Luc (kindly provided by K. H. Klempnauer) (16) contain five GAL4 DNA binding sites cloned upstream of a minimal E1B, TK, and E4 promoter element, respectively, and the firefly luciferase gene. The L8G5E1a-Luc and LexA-VP16 constructs were provided by C. Lemercier (17). The Fos-Luc reporter containing the c-fos promoter (−711 to −3) fused to the luciferase gene (pAS822) was kindly provided by P. Shaw. p(5xSRE)-Luc (Invitrogen) and p(PEA3-AdML)-Luc (2; kindly provided by John Hassell) reporters have been described previously. The GAL4-E1B CAT reporter plasmid and RSV p21^waf1/cip1 expression plasmid have been described previously (29). pCMV5-MEK-1 (ΔN S218E-S222D) encodes constitutively active MEK-1 (19), pcDNA3-F-MKK6(E) (6), pcDNA3F-p38β₂ (6), pAS571 (pCMV-GAL) (35), pAS572 [pCMV-GAL-Elk(205-428)] (35), pAS900 [pCMV-GAL-Elk(310-428)] (36), pAS1351 [pCMV-GAL-Elk(330-428)] (36), pAS883 [pCMV-GAL-MEF2A(266-416)] (36), pAS1556 [pCMV-GAL-Elk(1-93)] (39), pAS1559 [pCMV-GAL-Elk(1-206)] (39), pAS1561 [pCMV-GAL-Elk(1-428)] (39), pCMV-GAL-CBP(1098-2414) (29), and pCMV-GAL-p300(192-1044) (29) were described previously.

pAS1550 [pCMV-GAL-LIN-1(147-180)], pAS1551 [pCMV-GAL-Elk(168-428)], pAS1552 [pCMV-GAL-Elk(93-428)], pAS1553 [pCMV-GAL-Elk(230-428)], pAS1554 [pCMV-GAL-Elk(236-428)], pAS1555 [pCMV-GAL-Elk(244-428)], pAS1556 [pCMV-GAL-Elk(252-428)], pAS1557 [pCMV-GAL-Elk(260-428)], pAS1558 [pCMV-GAL-Elk(290-428)], pAS1559 [pCMV-GAL-Elk (205-230)], pAS1560 [pCMV-GAL-Elk(205-244)], pAS1561 [pCMV-GAL-Elk(205-260)], pAS1562 [pCMV-GAL-Elk(205-347)], pAS1563 [pCMV-GAL-Elk(205-375)], pAS1564 [pCMV-GAL-Elk(205-399)], pAS1565 [pCMV-GAL-Elk(168-205)], pAS1566 [pCMV-GAL-Elk(230-260)], pAS1567 [pCMV-GAL-Elk(244-260)], pAS1568 [pCMV-GAL-Elk(230-244)], and pCMV-GAL-Elk(230-260) containing the mutations E247A (pAS1569), V248A (pAS1570), K249A (pAS1571), V250A (pAS1572), E251A (pAS1573), K254A (pAS1574), E255A (pAS1575), and E256A (pSA1576) were constructed by ligating the SalI/XbaI PCR fragments into the same sites of pAS571.

pAS1597 [pCMV-GAL-p300(985-1004)-CBP(1098-2414)], pAS1598 [pCMV-GAL-Elk(205-260)-CBP(1098-2414)] and pAS1599 [pCMV-GAL-CBP(1019-1082)-CBP(1098-2414)] were constructed by ligating the NotI/XbaI PCR fragments into same sites of pCMV-GAL-CBP(1098-2414). pAS1577 [pCMV-GAL-Elk(1-428)Δ(205-260)], pAS1578 [pCMV-GAL-Elk(94-428)Δ(205-260)], pAS1579 [pCMV-GAL-Elk(260-428,S383A/S389A)], pAS1580 [pCMV-GAL-Elk(230-260)-MEF2A(266-416)], and pCMV-GAL-Elk(230-428) containing mutations L234A (pAS1581), V236A (pAS1582), E237A (pAS1583), L240A (pAS1584), R242A (pAS1585), L244A (pAS1586), E247A (pAS1587), V248A (pAS1588), K249A (pAS1589), V250A (pAS1590), E251A (pAS1591), K254A (pAS1592), E255A (pAS1593), E256A (pAS1594), K249V250E251/AAA (pAS1595), and S383A/S389A (pAS1596) were constructed by ligating the SalI/XbaI PCR PCR fragments into the same sites of pAS571. The mutations were introduced by a two-step PCR protocol using a mutagenic primer and two flanking primers as described previously (28).

pAS1827 [encoding Flag-tagged full-length Elk-1(K249A)] and pAS1828 (encoding Flag-tagged Elk-1ΔR-full-length Elk-1 with amino acids 205 to 260 deleted) were constructed by PCR amplification followed by ligation of StuI/XbaI-cleaved products into the same sites in pAS728. pAS1829 [encoding CMV-driven full-length Elk-1(K249A)] and pAS1830 (encoding CMV-driven Elk-1ΔR) were constructed by ligating HindIII/XbaI fragments from pAS1827 and pAS1828, respectively, into the same sites in pCMV5. pAS383 (encoding Flag-tagged wild-type Elk-1 has been described previously; 35).

pGal4 p300(192-1004) was created by PCR using Pwo polymerase (Roche). PCR products were cleaved with NotI and BglII and ligated into the same sites in pVR1012 Gal4 (29). The antisense primer was designed without a termination codon and also created an additional XbaI site (5′ to the BglII site). The XbaI and BglII sites were then used to insert double-stranded oligonucleotides encoding CRD1(amino acids 1018 to 1026) and Elk-1 R motif (amino acids 246 to 259) at the carboxy terminus of p300 to create pGal4-p300(192-1004)+ CRD1 and pGal4-p300(192-1004) plus R motif.

All PCR-derived constructs were verified by automated dideoxy sequencing.

Tissue culture, cell transfection, and reporter gene assays.

293 cells were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS) (Gibco BRL). Transfection experiments with 293 cells were carried out using Polyfect transfection reagent (Qiagen) in 33-mm-diameter dishes. Mixtures of 2 μg of plasmid DNA, 10 μl of polyfect reagent, and 100 μl of opti-MEM (Gibco BRL) were incubated for 12 min at room temperature. The complexes were subsequently incubated with cells for 1.5 h. Cells were then washed with serum-free DMEM and harvested 18 h later. For luciferase reporter gene assays, 0.5 μg of reporter plasmid and 0.25 μg of pCH110 were cotransfected with 0.05 μg of GAL4-fusion expression plasmids with or without 0.1 μg of LexA-VP16 vector. Cell extracts were prepared and luciferase and β-galactosidase assays were carried out as described previously (35). Cells were serum starved for 10 h and subsequently treated with 50 nM epidermal growth factor (EGF) (Sigma) or 330 nM of Trichostatin A (TSA) for further 12 h before harvesting.

U2OS cells were transfected using calcium phosphate as previously described (33). In each transfection, 5 μg of Gal4E1B-CAT reporter plasmid, 50 ng of pGal4-p300(192-1004) plus CRD1 or plus Elk-R, 0.5 ng of Gal4-Elk constructs, and 4 μg of pRSV-p21 or RSV-ADH control were used. In all experiments, cells were harvested 36 h after transfection and chloramphenicol acetyltransferase (CAT) activity was assayed with 100 to 200 μg of protein from whole-cell extracts.

Western blot analysis.

Aliquots of 30 μl of total lysate prepared for reporter gene assays were subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) followed by Western blot analysis. Anti-GAL4(DBD) antibody (sc-577x; Santa Cruz Biotechnology) and horseradish peroxidase (HRP)-conjugated secondary antibodies (Transduction Laboratory) were used in the immunoblot analysis, followed by SuperSignal west dura extended duration substrate (Pierce), and visualized on a BIO-RAD Fluor-S MultiImager and Quantity One software (BioRed). Data from Western blottings are computer-generated images (Quantity One; BIO-RAD).

RESULTS

Mapping of repression domain in C terminus of Elk-1.

In order to identify potential novel transcriptional regulatory domains in Elk-1, we tested the ability of a series of truncated GAL-Elk-1 fusion proteins to regulate the activity of a GAL4 site-driven luciferase reporter gene. Full-length Elk-1 shows repressive activity on a GAL4-driven thymidine kinase (TK) promoter when fused to the GAL4 DNA binding domain. Deletion of the ETS DNA-binding domain of Elk-1 alleviated this repressive activity (Fig. 1B; 39). However, little activation of the reporter above basal levels was observed, even after deletion to amino acid 205 (Fig. 1B). Indeed, in comparison to GAL4 alone, which consistently caused a two- to threefold induction of reporter activity, reduced reporter activation was observed in the presence of GAL-Elk(205-428). This suggested the presence of a repression domain embedded in the C terminus of Elk-1 that acts to suppress the activity of the C-terminal transactivation domain (C domain). To map the location of this repression domain, a series of C-terminal (Fig. 1C) and N-terminal (Fig. 1D) deletions were constructed. Upon deletion into the activation domain (positions 205 to 375), even more repressive activity was observed. This repressive activity was retained and not alleviated until deletion to amino acid 230, indicating that the C-terminal extent of the repression domain corresponded to amino acid 260 (Fig. 1C). Repressive activity was maintained upon N-terminal deletion until amino acid 252 (252 to 428), when partial relief was observed. Upon further deletion to amino acid 260, repressive activity was lost and activation was observed (Fig. 1D). Importantly, the loss of repression cannot be attributed to reductions in expression levels [e.g., compare GAL-Elk(230-428) and GAL-Elk(330-428); Fig. 1D, inset]. Thus, the N-terminal end of the repression domain maps to amino acid 244.

FIG. 1.

Mapping C-terminal Elk-1 repression domain. (A) Schematic of GAL-driven TK promoter-reporter construct. (B to E) Reporter gene analysis of activities of indicated GAL-Elk fusion proteins (relative to the basal level of the reporter construct; taken as 1) in 293 cells. Control, no GAL4 fusion protein was transfected. The known Elk-1 domains (ETS [B to D]) are in white boxes. The grey box represents the repression motif (R motif) that we have identified. The insets to each part are Western blots probed with Gal4 antibodies to show the relative expression levels of each of the fusion proteins. Arrows indicate the band corresponding to the GAL4 fusion protein with the correct size. Luciferase assays are representative of at least two independent experiments (standard errors are shown; n = 2).

Based on the deletion studies, we expressed amino acids 230 to 260 as a GAL fusion protein. This motif is sufficient to repress the reporter activity. In contrast, an equivalently sized piece of Elk-1(168-205) was unable to repress transcription (Fig. 1E). As GAL-Elk(230-260) is expressed to higher levels than GAL4 (Fig. 1E, inset), the lack of activity is not due to reduced expression. Indeed, where tested, the expression of the GAL4 DNA binding domain in isolation was either undetectable or less than any of the GAL fusions investigated in this study (data not shown). Thus, amino acids 230 to 260 in Elk-1 constitute an autonomously acting repression domain (hereafter referred to as the R motif) that is sufficient to repress transcription. In Elk-1, this domain suppresses the basal activity of its activation domain.

R motif represents potent repression domain.

The TK promoter-driven reporter gene is a low-activity promoter that is only moderately enhanced (two- to threefold) in the presence of the GAL4 DNA binding domain. To establish whether the Elk-1 R motif represents a strong repression domain, we analyzed its activity on a composite reporter element consisting of an E1A promoter preceded by LexA and GAL4 binding sites (Fig. 2A). In the presence of the highly active Lex-VP16 fusion protein, this reporter is activated to high levels (Fig. 2B; 17). A series of GAL4-Elk-1 fusion proteins were then added in trans to investigate whether they contained any repressive activity (Fig. 2B). In the context of the 205-428 construct, the analysis of N- and C-terminally deleted proteins showed a virtually identical profile as the TK promoter (data not shown). Also, as observed with the TK promoter, the region encompassing amino acids 230 to 260 is a potent repression domain on the LexA-GAL4-Luc reporter. GAL-Elk(230-260) is expressed at similar or lower levels than the other constructs, further underscoring the potency of this motif (Fig. 2B). Deletions within this domain result in a loss of this repressive activity. Significantly, the region 244 to 260 is inactive as a repression domain, although deletion analysis suggests that this region contains repressive activity (Fig. 1). Furthermore, amino acids 244 to 347 exhibit repressive activity on this reporter (data not shown), suggesting that while amino acids 244 to 260 are required for transcriptional repression, residues located either N or C terminal to this are required to constitute a functional repression domain (see Discussion).

FIG. 2.

Elk-1 R motif is potent repressor of transcription. (A) Schematic of LexA-GAL-driven E1A promoter-reporter construct. LexA-VP16 represents an activator whereas GAL-Elk is a repressor (or activator) in this context. (B to D) Reporter gene analysis of the activities of the indicated GAL-Elk fusion proteins (relative to the basal level of the reporter construct; taken as 1) in 293 cells. Control, no GAL4 fusion protein was transfected. The presence of LexA-VP16 is indicated. The insets to each part are Western blots probed with Gal4 antibodies to show the relative expression levels of each of the fusion proteins. Arrows indicate the band corresponding to the GAL4 fusion protein with the correct size. Luciferase assays are representative of at least two independent experiments (standard errors are shown; n = 2).

While amino acids 230 to 260 constitute an autonomously functioning repression domain, it is important to verify that this motif is important in the context of the full-length protein. While full-length Elk-1(1-428) is a potent repressor in the presence of Lex-VP16 on the LexA-GAL4-Luc reporter, deletion of the R motif (amino acids 230 to 260) reduced repression, with a ∼3-fold increase in reporter activity (Fig. 2C). However, full relief of repression is not observed in the presence of the Elk-1 ETS domain which is itself known to act as a repressive motif (39). We therefore examined the role of the R motif in the absence of the ETS domain (Fig. 2D). Amino acids 94 to 428 did not affect the activity of the reporter. However, upon deletion of the R motif, synergistic activation of transcription was observed in the presence of Lex-VP16. Thus, deletion of the R motif in the context of full-length Elk-1 or Elk-1(94-428) reduces its repressive activity and allows the activity of the C-terminal transcriptional activation domain to be manifested. While deletion of the R motif leads to a ∼3-fold enhancement of reporter activity, deletion of the ETS domain results in a 40-fold enhancement of reporter activity (Fig. 2C and D). However, the simultaneous deletion of the R motif and the ETS domain leads to ∼150-fold increases in transcriptional activation, indicating that the R motif and ETS domain act synergistically to repress Elk-1 activity.

These results therefore demonstrate that the R motif of Elk-1 represents a potent transcriptional repression domain that can act in trans or in cis to reduce the potency of different transcriptional activation domains.

R motif functions through enhancing deacetylation.

It is becoming apparent that one of the major mechanisms by which transcriptional repression is effected is through the recruitment of histone deacetylase complexes and reduction of the acetylation status of promoters and/or associated regulatory proteins (reviewed in reference 21). Indeed, the Elk-1 ETS domain represses transcription, at least in part, by the recruitment of the mSin3A-HDAC complex (39). To establish whether the Elk-1 R motif might repress transcription through a similar mechanism, we tested the effect of the deacetylase inhibitor trichostatin A (TSA) on repression mediated by various GAL-Elk fusion proteins (Fig. 3B and C). Regions 244 to 428 and 205 to 428, which exhibited strong repressive activity (Fig. 1D), both showed high sensitivity to TSA (Fig. 3B). Furthermore, the isolated R motif retained TSA sensitivity that is similar to that observed with the Elk-1 ETS domain (Fig. 3C; 39). Together, these data therefore indicate that this motif acts, at least in part, through increasing deacetylase action at the GAL-driven TK promoter. One way in which the R motif might function is through recruitment of histone deacetylase-containing complexes, but we were unable to show association with either class I (HDAC-1) or class II (HDAC-4) deacetylases (39; data not shown; see Discussion).

FIG. 3.

R motif functions by increasing histone deacetylase activity in promoter- and context-specific manner. (A) Schematic of the GAL-driven TK promoter-reporter construct. (B and C) Reporter gene analysis of the activities of the indicated GAL-Elk fusion proteins on the GAL-driven TK promoter-reporter construct in 293 cells. Control, no GAL4 fusion protein was transfected. The activities of each construct were tested in the presence and absence of TSA and presented as relative luciferase activities and n-fold derepression (relative to the basal level of the reporter construct; taken as 1). (D) Reporter gene analysis of the activities of the indicated GAL-Elk fusion proteins in 293 cells on the GAL-driven promoter-reporter constructs shown above each graph (TK, E1A, and E4). The reporter gene activities in the presence of the GAL fusion proteins are shown on the left of each graph, and the n-fold derepressions observed in the presence of TSA are shown on the right of each graph (both relative to the basal level of the reporter construct; taken as 1). A Western blot probed with Gal4 antibodies to show the relative expression levels of each of the fusion proteins is shown on the right. Luciferase assays are representative of at least two independent experiments (standard errors are shown; n = 2).

The Elk-1 ETS domain exhibited promoter selectivity towards repression via histone deacetylase-dependent mechanisms. We therefore compared the repressive effects of the R motif on three different GAL-driven reporters (TK, E1B, and E4) (Fig. 3D). The pattern of repression on each promoter by regions containing the ETS domain and the R motif was virtually identical. Similarly, region 230-428 showed repressive activity on all reporter constructs and upon removal of the R motif (amino acids 260 to 428), this repression was replaced by activation. However, the reliance on histone deacteylases for this repressive activity differed on each promoter. Elk-1(1-206) showed TSA sensitivity on the TK promoter but not on the E1B and E4 promoters (Fig. 3D; 39). Similarly, Elk-1(205-260) showed TSA sensitivity on the TK promoter but not the E1B promoter but differed in that TSA sensitivity was observed on the E4 promoter. In the context of the C-terminal activation domain, TSA sensitivity was observed on each promoter, but this was lost upon deletion of the R motif (Fig. 3D). Thus, on its own, the R motif shows a promoter-specific dependence on histone deacetylase activity for its action, but this specificity is lost in the presence of the C-terminal activation domain. This different behavior presumably reflects the presence or absence of other factors in the basal promoters that can recruit histone acetylases, which is bypassed in the presence of the Elk-1 TAD that can recruit CBP (13; see Discussion).

Collectively, these data demonstrate that the R motif of Elk-1 represses transcription, at least in part, by enhancing the activity of histone deacetylases at promoters.

Identification of critical residues in R motif.

Deletion analysis indicated that residues that constitute the R motif (amino acids 230 to 260) form an autonomous repression domain and that residues 244 to 260 appear important for this function (Fig. 1 and 2). In order to identify residues that are critical for this function, we carried out alanine-scanning mutagenesis on the R motif (Fig. 4). In particular, we targeted charged or hydrophobic residues in this region. Initially, we tested GAL-Elk fusion proteins in the context of Elk(230-428) on the composite LexA-GAL4-Luc reporter (Fig. 4A and B). None of the point-mutated proteins tested between amino acids 230 and 244 exhibited reduced repressive activity (Fig. 4B). However, two residues located between amino acids 244 and 260 (K249 and E251) are important for the repressive activity (Fig. 4A). These two residues are also important for repression when tested in the context of the isolated R motif (Fig. 4C). We also investigated whether these residues were critical for the histone deacetylase-dependent action of the R motif (Fig. 4D). While repression by wild-type GAL-Elk(230-428) and the point mutant E247A showed high TSA dependency, this was reduced to near basal levels in the K249A mutant, a triple mutant of K249, V250, and E251, and a mutant with a complete R-motif deletion [GAL-Elk(260-428)]. Thus, residues that are critical for the repressive effect of the R motif are also important for enhancing the activities of histone deacetylases.

FIG. 4.

Identification of important residues for repressive function of R motif. (A to C) Reporter gene analysis of the activities of the indicated GAL-Elk(230-428) (A and B) or GAL-Elk(230-260) (C) fusion proteins on the LexA-GAL-driven E1A promoter-reporter construct (relative to the basal level of the reporter construct; taken as 1) in 293 cells. The presence of LexA-VP16 is indicated. The insets to each part are Western blots probed with Gal4 antibodies to show the relative expression levels of each of the fusion proteins. (D) Reporter gene analysis of the activities of the indicated GAL-Elk fusion proteins on the GAL-driven TK promoter-reporter construct in 293 cells. The activities of each construct were tested in the presence and absence of TSA. The reporter gene activities in the presence of the GAL fusion proteins are shown on the upper part of graph, and the n-fold derepression observed in the presence of TSA is shown on the lower part (both relative to the basal level of the reporter construct; taken as 1). Luciferase assays are representative of at least two independent experiments (standard errors are shown; n = 2).

R motif represses Elk-1 activity on Elk-1-responsive reporters.

. The R motif acts to repress transcription in the context of GAL4 fusion proteins and GAL4 site-driven reporter plasmids (Fig. 1 to 4). To study the role of the R motif in regulating the activity of Elk-1 in the absence of the GAL4 DNA binding domain, we disrupted the R motif in the context of full-length Elk-1 and tested the activities of these mutant proteins on Elk-1-responsive reporter constructs.

First, we deleted the R motif from Elk-1 and compared the ability of the resulting protein (Elk-1ΔR) with its wild-type counterpart to activate a luciferase reporter gene driven by the intact c-fos promoter (nucleotides −711 to −3) (Fig. 5A). Wild-type Elk-1 does not enhance the reporter over basal levels, suggesting that the reporter is already saturated by endogenous TCFs. However, Elk-1ΔR causes an enhancement in reporter activity over basal levels, demonstrating that deletion of the R motif makes this protein become more active.

FIG. 5.

Deletion of R motif enhances Elk-1 activity on Elk-1-responsive reporters. (A, C, and D) Reporter gene analysis of the activities of the indicated Elk-1 derivatives on the c-fos promoter- (A), SRE- (C), or PEA3- (D) driven luciferase reporter constructs (relative to the basal level of the reporter construct; taken as 1) in 293 cells. (B) Western blot probed with Flag antibodies to show the expression of Elk-1 derivatives.

To further demonstrate the role of the R motif in the context of full-length Elk-1, we investigated two additional reporter genes containing either multiple SREs (in which Elk-1 cooperates in a complex with SRF) or PEA3 binding sites (where Elk-1 binds autonomously). Here we used full-length Elk-1 that contained a single point mutation in the R motif, K249A, that leads to partial relief of repression (Fig. 4). In comparison to wild-type Elk-1, Elk-1(K249A) exhibited enhanced transactivation properties on both the SRE-Luc (Fig. 5C) and PEA3-Luc (Fig. 5D) reporters. The expression levels of wild-type Elk-1 and its two mutant counterparts are very similar, indicating that differing levels do not explain the differences in transcriptional activation properties that we observe (Fig. 5B).

Thus, in the context of full-length Elk-1, the R motif plays an important role in reducing its transactivation properties on Elk-1-responsive reporters.

Elk-1 R motif is portable repression domain.

The sequence of the C-terminal end of the Elk-1 R motif was used to search for similar motifs in other proteins (34). Several similar motifs were identified in proteins, including p300/CBP (29), for which repressive activity has been ascribed (Fig. 6A). In particular, the two residues that are critical for the repressive activity of the R motif are conserved in p300. We therefore asked whether these motifs are interchangeable.

FIG. 6.

The Elk-1 R motif is part of a novel class of repression motifs and can function in heterologous contexts. (A) Alignment of C-terminal end of R motif of Elk-1 with CRD1 motif from p300 and putative motifs found in other transcription factors. The arrows represent the key amino acids involved in repression in Elk-1 (Fig. 4). Residues conserved with Elk-1 are shaded in grey. Acidic residues found in the C-terminal part of this motif are shaded in black. (B to E) Reporter gene analysis of the activities of the indicated GAL-CBP (B), GAL-p300/GAL-Elk (C and D), and GAL-LIN-1 (E) fusion proteins. Assays were carried out in 293 cells (B and E) or U2OS cells (C and D). The addition of p21 or LexA-VP16 is indicated next to each column. Data are shown relative to the basal level of the reporter in conjunction with GAL4 (B and E) (taken as 1) or as absolute values (C and D). Schematics of the reporter constructs (B and E) are given above each graph, and GAL fusion proteins are shown adjacent or above (C and D) the graphs. The CRD1 motif from p300 is shown as a black box, the R motif from LIN-1 is shown as a white box, and the R motif from Elk-1 is shown as a grey box. Luciferase and CAT assays are representative of at least two independent experiments (standard errors are shown; n = 2 or 3). Western blots probed with Gal4 antibodies to show the expression of fusion proteins are shown in B and E). Arrows indicate the band corresponding to the GAL4 fusion protein with the correct size.

Firstly, we fused the R motif to the transactivation domain of CBP in place of the repression domain in CBP (CRD1). Deletion of the CRD1 region (amino acids 1019 to 1082) results in enhancement of the transactivation capacity of CBP (Fig. 6A; 29). However, the insertion of the homologous domain from p300 or the R motif from Elk-1 returns the activity of CBP back to its basal level (Fig. 6B). Western blotting demonstrates that this loss of transactivation activity in the CBP fusions is not due to reduced expression (Fig. 6B, bottom). The CRD1 motif is known to permit induction of the activity of the p300 transactivation domain by the Cdk inhibitor p21^waf1/cip1 in U2OS cells (Fig. 6C; 29); therefore, we tested whether the R motif of Elk-1 could mediate a similar function. In its native context, the R motif was unable to permit p21^waf1/cip1 inducibility of Elk-1 [Fig. 6C, Elk(205-428)]. However, when fused to p300, the Elk-1 R motif inhibited basal activity of the activation domain and permitted p21^waf1/cip1 inducibility (Fig. 6D).

LIN-1 is the most closely related protein to Elk-1 found in C. elegans (1). We therefore tested whether a putative repression motif related to the Elk-1 R motif (Fig. 6A) exhibited repressive activity. When fused to the GAL DNA-binding domain, this motif was capable of repressing the activity of the powerful VP16 activation domain in trans (Fig. 6E). Thus, the R motif appears to be an evolutionarily conserved feature of Elk-1 at both the sequence and functional levels.

Collectively, these data demonstrate that the Elk-1 R motif is an evolutionarily conserved domain that can act in cis in heterologous contexts to suppress the basal activity of different activation domains. However, the local context of the activation domain can dictate how it responds to regulatory proteins, such as p21^waf1/cip1.

R motif affects enhancement of activation domain activity by MAP kinase cascades.

The activity of the Elk-1 transcriptional activation domain (TAD) is greatly enhanced following activation of the MAP kinase cascades (7, 20). To establish whether the R motif still functions to repress transcription and reduce the potency of the Elk-1 transactivation domain following MAP kinase-mediated phosphorylation, we compared the inducibility of the Elk-1 TAD in the presence and absence of the R motif (Fig. 7A). In the presence of the Erk pathway activator, MEK-1, induction of the Elk-1 transcriptional activation domain was observed irrespective of the presence of the R motif. However, the induced level in the presence of the R motif [GAL-Elk(230-428)] was the same as the basal level in the absence of this motif [GAL-Elk(260-428)], demonstrating the importance of this motif in inhibiting the basal activity of the transactivation domain.

FIG. 7.

The R motif regulates the level of EGF/MAP kinase-mediated activation of Elk-1. (A to E) Reporter gene analysis of the activities of the indicated GAL-Elk (A, D, and E) or GAL-MEF2A (B and C) fusion proteins on the LexA-GAL-driven E1A (A) and GAL-driven E1B (B-E) promoter-reporter constructs in 293 cells. Control, no GAL4 fusion protein was transfected. (A) The presence of LexA-VP16 and constitutively active MEK-1 (A) or cotransfected MKK6/p38β₂ (C) are indicated. (D) Stimulation by EGF is indicated. The activities of each construct relative to basal reporter levels (A to D) and n-fold inducibility relative to EGF-stimulation of wild-type GAL-Elk(230-428) (taken as 1) (E) are shown. Luciferase assays are representative of at least two independent experiments (standard errors are shown; n = 2). Western blots probed with Gal4 antibodies to show the expression of fusion proteins are shown (B and D). (F) Model for how R motif represses activity of Elk-1 TAD. SRF (dark grey ellipse) and Elk-1 (domains shown by light grey ellipses) are shown complexed on a SRE.

We also tested whether the Elk-1 R motif could inhibit the TAD of a different MAP kinase target, MEF2A. Fusion of the R motif to the MEF2A TAD leads to reduced activity of this domain (Fig. 7B). The suppression of the MEF2A transactivation domain is even more apparent when assayed in the context of the LexA-GAL4-E1A-luc reporter (data not shown). We investigated the inducibility of the R motif-MEF2A chimera (Fig. 7C). Activation of the p38 MAP kinase pathway led to enhanced activity of this chimera, both in the presence and absence of the R motif. Again, MAP kinase activation in the presence of the R motif [GAL-Elk(230-260)-MEF2A(266-416)] was similar to the basal level of activity in the absence of this motif [GAL-MEF2A(266-416)].

We also tested the ability of the R motif to affect the activity of Elk-1 in response to growth factor (EGF) stimulation of the Erk pathway. As observed with MEK stimulation, the presence of the R motif in GAL-Elk(230-428) reduced the level of activation elicited by EGF stimulation [compare to GAL-Elk(260-428); Fig. 7D]. This stimulation was lost in the absence of the critical Ser383/389 phosphoacceptor motifs in Elk-1, demonstrating that EGF stimulation is not able to elicit any derepression via the R motif in the absence of MAP kinase-mediated Elk-1 phosphorylation. These Ser383/Ser389 mutant versions of GAL-Elk fusions were expressed to similar levels as their wild-type counterparts (Fig. 7D, bottom). In addition to reducing the overall magnitude of induction by EGF, the actual n-fold inducibility was reduced in the presence of the R motif [compare GAL-Elk(230-428) and GAL-Elk(260-428); Fig. 7E). Furthermore, mutations that reduced the repressive capacity of the R motif (K249A) enhanced the response to EGF, while those that did not affect its activity also did not affect EGF responsiveness (E247A) (Fig. 7E).

Collectively, these results therefore demonstrate that the R motif acts to inhibit both the basal and activated levels of the Elk-1 TAD. Thus, the major function of this motif appears to be to keep Elk-1 in an inactive state in the absence of MAP kinase-mediated phosphorylation.

DISCUSSION

The ETS domain transcription factor Elk-1 is commonly thought of as an activator protein (reviewed in references 31 and 37). However, it was recently shown that Elk-1 can also repress transcription via its N-terminal ETS DNA-binding domain and recruitment of the Sin3A/HDAC complex (39). Here, we demonstrate that Elk-1 contains a second repression domain, the R motif, which is located in the C-terminal part of the molecule. SAP-1 and SAP-2 are two closely related proteins with similar conserved domains involved in DNA binding (ETS), SRF binding (B box), MAP kinase binding (D domain), and transcriptional activation (C domain). However, the R motif is not conserved in these other family members, suggesting that it provides unique properties for Elk-1. Evolutionary conservation of this motif is, however, suggested by its presence in the Caenorhabditis elegans ETS domain protein LIN-1. This protein shows several other conserved features, including a high level of conservation in the DNA-binding domain (1) and also conservation in the MAP kinase docking and regulatory phosphorylation sites (12). Furthermore, genetic experiments have indicated a potential repressive role for LIN-1 (30). SAP-2 contains two different repression domains, the NID and CID (4, 18). The repressive properties of SAP-1 have not been investigated, but it exhibits strong homology to the SAP-2 NID. In both cases, the ETS domain represents an additional potential repression domain. Thus, individual members of this subfamily of proteins have recruited different repression domains to modulate their transcriptional activities in different ways, thereby contributing to their unique regulation and functions.

The Elk-1 R motif is a powerful repression motif that can act in trans and in cis to reduce the activity of different activation domains. In the context of Elk-1, the R motif appears to be a weaker repression domain than the ETS domain (Fig. 2). However, deletion of both domains leads to synergistic transcriptional activation (Fig. 2), suggesting that they act cooperatively to block different activities rather than acting individually where additive effects might be expected. Thus, both motifs play important roles in Elk-1. In addition to the Elk-1 TAD, the R motif can dampen the activity of the activation domains found in VP16 (in trans; Fig. 2) and CBP, p300, and MEF2A (in cis; Fig. 6). At least part of this inhibitory activity is due to increasing the activity of histone deacetylases at target promoters (Fig. 3). However, we have been unable to detect interactions with class I or II HDACs (HDAC1 and -4) (data not shown), suggesting that these putative interactions are either weak or, alternatively, that the R motif might act to inhibit histone acetylases. In the latter scenario, the net result at the promoter would be enhanced activity of histone deacteylases and hence transcriptional repression. The observation that CBP appears to be constitutively recruited by Elk-1 is consistent with this hypothesis. Furthermore, the dependence of repression by the R motif on HDAC activity exhibits promoter specificity (Fig. 3). While fused to the Elk-1 activation domain, R motif-mediated repression is HDAC independent on the E1B basal promoter. Similar results were observed for repression mediated by the Elk-1 ETS domain on this promoter (39). When assayed as an isolated domain, further differences were seen and the R motif acted in an HDAC-independent manner on both the E1B and E4 promoters. Thus, other core promoter-binding factors must dictate the importance and ability of the R motif to regulate transcription via HDAC-dependent mechanisms. The CRD1 motif in p300 also represses in a TSA-sensitive manner, but again, no direct interactions with HDACs could be detected (data not shown). This further underlines the functional homology between the R motif and CRD1 (see further discussion below).

The minimum autonomously acting repression motif in Elk-1 encompassed 31 amino acids (230 to 260), although alanine-scanning studies indicated that the critical residues (K249 and E251) reside in the C-terminal part of this motif (Figs. 2 and 4). Consistent with this is the observation that Elk-1(244-428) exhibits repressive activity. However, this also suggests that N- and C-terminal residues are required for correct folding or presentation of this motif as residues 244 to 260 are insufficient to mediate repressive activity (Fig. 2). While K249 and E251 are clearly the most important amino acids in the R motif, other amino acids must also play key roles, as only partial derepression is seen upon their mutation (Fig. 4). Further multiple-residue mutagenesis experiments are required to identify such amino acids. Database searching (34) enabled us to identify several proteins that exhibit repressive activities and contain related motifs (Fig. 6A). An alignment of these motifs shows strict conservation of the two critical residues (KXE motif), with additional conservation around it. An additional feature is the presence of an acidic patch downstream from the KXE motif. Previous studies have demonstrated that acidic peptides tend to act as activation domains whereas basic peptides act as repression domains (23, 24). However, this novel class of repression motifs we have identified clearly do not fit this paradigm due to their net negative charge (−5 in the Elk-1 R motif). The R motif also shows similarity to the repressive SC motif that acts to inhibit transcriptional synergy at complex promoter elements (11). The SC motifs contain a central KXE motif but, unlike the R motif, downstream acidic residues are not always present, and functionally, SC motifs cannot repress transcription in trans when not tethered to a transactivation domain. Functional conservation of the R motif is also suggested by experiments in which chimeric Elk-1-p300 proteins were analyzed, in which the CRD1 repression domain in p300 (29) was replaced with the Elk-1 R motif (Fig. 6). Significantly, the Elk-1 R motif also conferred p21^waf1/cip1 inducibility on the p300 transcriptional activation domain, demonstrating that it can mediate specific regulatory activities in a context-dependent manner. Furthermore, the region of the CRD1 motif compared to the Elk-1 R motif in these assays is much shorter than the original domain analyzed (29) and underlines the importance of the region conserved between these two proteins (Fig. 6A).

In the context of Elk-1, the R motif is not p21^waf1/cip1 responsive (Fig. 6). Instead, this repression domain appears to act to dampen the activity of the C-terminal activation domain (Fig. 7F). It is currently unclear whether this acts intramolecularly to inhibit the activity of the TAD directly and/or alternatively by recruiting repressive activities that act directly on the promoter. In the absence of the R motif, the Elk-1 activation domain shows a high basal level activity that is similar to the activity of the activation domain in the context of the intact C terminus once activated by MAP kinase-mediated phosphorylation (Fig. 7). This reduction in basal TAD activity is not due to changes in Ser383/389 phosphorylation, as similar basal levels are observed in the presence or absence of the R motif (data not shown). It is critically important that the activation domain is inactive in the absence of mitogenic signaling to stop promiscuous immediate-early gene induction. Thus, the R motif may have evolved to control the activity of an intrinsically active TAD. However, by analogy with the CRD1 domain in p300, it appears likely that specific signals or pathways may target the R motif in Elk-1 to elicit derepression and potentially permit synergistic activation in combination with the MAP kinase pathways. Future work will be aimed at identifying such pathways.