Advertisement

Id helix–loop–helix proteins inhibit nucleoprotein complex formation by the TCF ETS‐domain transcription factors

Paula R. Yates, Graham T. Atherton, Richard W. Deed, John D. Norton, Andrew D. Sharrocks

Author Affiliations

  1. Paula R. Yates1,
  2. Graham T. Atherton2,
  3. Richard W. Deed2,
  4. John D. Norton2 and
  5. Andrew D. Sharrocks*,1
  1. 1 Department of Biochemistry and Genetics, The Medical School, University of Newcastle upon Tyne, Newcastle upon Tyne, NE2 4HH, UK
  2. 2 Cancer Research Campaign Department of Gene Regulation, Paterson Institute for Cancer Research, Christie Hospital NHS Trust, Wilmslow Road, Manchester, M20 9BX, UK
  1. *Corresponding author. E-mail: a.d.sharrocks{at}ncl.ac.uk

Abstract

The Id subfamily of helix–loop–helix (HLH) proteins plays a fundamental role in the regulation of cellular proliferation and differentiation. Id proteins are thought to inhibit differentiation mainly through interaction with other HLH proteins and by blocking their DNA‐binding activity. Members of the ternary complex factor (TCF) subfamily of ETS‐domain proteins have key functions in regulating immediate‐early gene expression in response to mitogenic stimulation. TCFs form DNA‐bound complexes with the serum response factor (SRF) and are direct targets of MAP kinase (MAPK) signal transduction cascades. In this study we demonstrate functional interactions between Id proteins and TCFs. Ids bind to the ETS DNA‐binding domain and disrupt the formation of DNA‐bound complexes between TCFs and SRF on the c‐fos serum response element (SRE). Inhibition occurs by disrupting protein–DNA interactions with the TCF component of this complex. In vivo, the Id proteins cause down‐regulation of the transcriptional activity mediated by the TCFs and thereby block MAPK signalling to SREs. Therefore, our results demonstrate a novel facet of Id function in the coordination of mitogenic signalling and cell cycle entry.

Introduction

The ternary complex factors (TCFs) comprise a subfamily of ETS‐domain transcription factors (reviewed in Treisman, 1994). These proteins form ternary complexes with the serum response factor (SRF) and serum response elements (SREs) found in immediate‐early genes such as c‐fos (reviewed in Treisman, 1994; Cahill et al., 1996). Three mammalian TCFs (Elk–1, SAP‐1 and SAP‐2/ERP/Net) have been identified (Rao et al., 1989; Dalton and Treisman, 1992; Giovane et al., 1994; Lopez et al., 1994; Price et al., 1995). Each protein shares four major regions of primary amino acid sequence similarity. The ETS DNA‐binding domain is located at the N‐terminus of the protein and together with a second conserved region, the B‐box, is sufficient for ternary complex formation with SRF and the c‐fos SRE (Dalton and Treisman, 1992; Janknecht and Nordheim, 1992; Marais et al., 1993; Giovane et al., 1994; Janknecht et al., 1994; Price et al., 1995). The C‐terminal domain (C‐box) is a target for phosphorylation by MAP kinases (MAPKs) and acts as an autonomous transcriptional activation domain in a heterologous context (reviewed in Treisman, 1994). The fourth conserved region (D‐domain) is located immediately N‐terminal from the C‐box and acts as a MAPK‐targeting domain (Yang et al., 1998a,b). Ternary complex formation by TCFs requires both protein–DNA interactions mediated by the ETS‐domain (Dalton and Treisman, 1992; Janknecht and Nordheim, 1992) and protein–protein interactions with SRF mediated by the B‐box (Shore and Sharrocks, 1994). In the case of Elk–1 and SAP‐1, ternary complex formation is stimulated by MAPK phosphorylation of the TCFs (Gille et al., 1992, 1995a,b; Shore et al., 1996).

The ternary TCF–SRF–SRE complex plays a pivotal role in the activation of immediate‐early genes such as c‐fos in response to a series of mitogenic and stress stimuli (reviewed in Cahill et al., 1996). The rapid induction of c‐fos and other immediate‐early genes is followed by a rapid shut‐off back to basal levels (reviewed in Herschman, 1991). Genes encoding members of the Id family of helix–loop–helix (HLH) proteins also represent immediate‐early genes which are markedly and rapidly induced by serum and other mitogenic stimuli (Christy et al., 1991; Deed et al., 1993; Barone et al., 1994; Hara et al., 1994; reviewed in Norton et al., 1998). The prototype, Id1, was originally identified as an inhibitor of DNA binding by basic helix–loop–helix (bHLH) proteins (Benezra et al., 1990). Subsequently, three further genes were identified that encode the related proteins Id2, (Sun et al., 1991; Biggs et al., 1992), Id3 (Christy et al., 1991; Deed et al., 1993) and Id4 (Riechman et al., 1994). The Id proteins contain functional HLH dimerization motifs but lack the DNA‐binding basic region found in the bHLH proteins. Id‐mediated inhibition of DNA binding by bHLH proteins is thought to occur by sequestration of these proteins into inactive dimers which cannot bind DNA due to the absence of a basic region in the Id proteins (Benezra et al., 1990; Sun et al., 1991; reviewed in Norton et al., 1998).

In addition to their association with bHLH transcription factors, Id proteins have been shown to interact with the non‐HLH proteins, retinoblastoma protein (pRB) (Iavarone et al., 1994) and MIDA1 (Shoji et al., 1995). In the case of pRB, binding of Id2 to the pocket domain results in the inhibition of RB function in inducing cell cycle arrest. Several lines of evidence therefore show that Id proteins function to promote cell cycle entry, to stimulate proliferation and to block cellular differentiation (reviewed in Norton et al., 1998).

Interactions between several ETS‐domain and bHLH transcription factors have recently been demonstrated or inferred. Ets‐1 and USF‐1 interact via the ETS‐domain of Ets‐1 to cause cooperative activation of the HIV‐1 enhancer (Sieweke et al., 1998). Similarly, cooperative interactions between Ets‐1 and the bHLH proteins E47 and TFE3 lead to enhanced transcription from the immunoglobulin μ heavy‐chain gene enhancer (Dang et al., 1998). The TCF SAP‐2/Net/ERP has also been shown to interact with bHLH proteins through a HLH‐like domain located immediately C‐terminal to the B‐box (Maira et al., 1996). This domain also functions intramolecularly to inhibit DNA binding by SAP‐2.

In this study, we investigated interactions between the Id HLH proteins and the TCF subfamily of ETS‐domain transcription factors. We demonstrate direct interactions between Ids and the ETS DNA‐binding domain of the TCFs. Binding of Ids disrupts ternary complex formation with SRF at the c‐fos SRE in vitro. Furthermore, overexpression of Ids attenuates MEK‐induced SRE‐activation in vivo. Together, our results implicate Ids as novel negative regulators of TCF function and provide further insights into the regulation of proliferation and differentiation by these two classes of transcription factors.

Results

The TCF ETS‐domain transcription factors interact with the HLH protein Id2

Physical interactions between several ETS‐domain and bHLH transcription factors were demonstrated recently (Maira et al., 1996; Dang et al., 1998; Sieweke et al., 1998). As negative transcriptional regulators and, due to their up‐regulation following serum stimulation (reviewed in Norton et al., 1998), the Id subfamily of HLH proteins represents candidate interacting proteins that might down‐regulate TCF activity. Initially, therefore, we tested whether the TCFs can interact with Id family members.

In vitro glutathione S‐transferase (GST) pull‐down experiments were carried out using full‐length TCFs and a GST–Id2 fusion protein (Figure 1B). All three TCFs interact with Id2, although the efficiency of interaction with Elk–1 is lower than with SAP‐1 and SAP‐2 (Figure 1B, lanes 7–9). This suggests that the Ids interact with a conserved region of the TCFs. In order to map the Id‐interaction domain on Elk–1, a series of C‐terminally truncated proteins was tested for their ability to interact with GST–Id2 (Figure 1C). All of the Elk–1‐truncated proteins bound to GST–Id2, indicating that the N‐terminal ETS DNA‐binding domain contained in Elk–1‐93 was sufficient to mediate this interaction. This domain is highly conserved among the TCFs, exhibiting 72% identity among all three proteins. Further C‐terminal truncations within the ETS‐domain resulted in proteins that were unable to bind to Id2 (data not shown), indicating that the integrity of the ETS‐domain is essential for the interaction of Elk–1 with Id2.

Figure 1.

The TCFs interact with Id2 through the ETS‐domain. (A) Schematic illustration of the domain structure of the proteins used. Numbers indicate the positions of the N‐ and C‐terminal residues with respect to the full‐length proteins. (B) GST pull‐down analysis of full‐length TCFs with GST (lanes 4–6) and a GST–Id2 fusion protein (lanes 7–9). Twenty per cent of the input proteins are shown in lanes 1–3. (C) Mapping the Id2 interaction domain on Elk–1. GST pull‐down analysis of full‐length Elk–1 with GST (lane 1) and full‐length or truncated Elk–1 derivatives with GST–Id2 (lanes 2–5). (D) Elk–1 and Id2 interact in vivo. (i) Schematic illustrating the domain structure of full‐length Id2 (amino acids 1–134) containing the conserved HLH domain (represented as black boxes) and Flag‐epitope‐tagged full‐length Elk–1 (amino acids 1–428). (ii) Co‐immunoprecipitation of Id2 and Elk–1 from Cos‐7 cells. Cells were either mock transfected with 1 μg pCMV5 and 1 μg pcDNA3 (lane 1), 1 μg of pcDNAId2 and 1 μg pCMV5 (lane 2), or 1 μg of pcDNA3Id2 and 1 μg of pCMVElk–1 (lane 3). A marker of bacterially purified Elk–1 protein (20 ng) is shown in lane M.

In order to examine whether the Id proteins interact with the TCFs in vivo, Cos‐7 cells were transfected with expression vectors encoding Id2 and Flag‐epitope‐tagged Elk–1. Complexes were immunoprecipitated using anti‐Id2 antibodies followed by detection of co‐precipitating Elk–1 by Western blotting using an anti‐Flag antibody (Figure 1D). When both Id2 and Elk–1 were co‐expressed, Elk–1 was co‐precipitated with Id2 (Figure 1D, lane 3), indicating that the two proteins interact in vivo.

Id1, Id2 and Id3 interact with the TCFs

The ability of different Id proteins to interact with the TCFs was subsequently tested. GST‐fusion proteins containing the N‐terminal ends of Elk–1 and SAP‐1 (including the ETS‐domain; Figure 2A) were tested for their ability to bind to in vitro translated Id1, Id2 and Id3 (Figure 2B and C). GST–MyoD was used as a positive control as MyoD has been shown previously to bind to the Id proteins (Benezra et al., 1990; Sun et al., 1991). All three Id proteins bind to GST–Elk–1‐205, although their relative binding efficiency varies in the order Id2>Id3≫Id1 (Figure 2B). Similarly, all three Id proteins bind to GST–SAP‐1‐298 with a similar relative binding efficiency profile (compare Figure 2C with 2B). All three Id proteins are therefore able to bind to Elk–1 and SAP‐1 in vitro, but of these, Id2 binds with the highest efficiency.

Figure 2.

Interaction of TCFs with Ids in vitro. (A) Diagrammatic illustration of Elk–1‐205, Sap‐1‐298 and MyoD fused to GST. The amino acids of the proteins included in the constructs are shown. GST–Elk–1‐205 and GST–SAP‐1‐298 lack the D‐domain and C‐terminal transcriptional activation domain. The location of the ETS‐domain and B‐box of the TCFs is shown and the helices within the HLH motif in MyoD are indicated by black boxes. Bacterially expressed GST‐fusion proteins were purified and equal molar concentrations used in all assays. (B) Pull‐down assays of 35S‐labelled Id1, Id2 and Id3 by immobilized GST, GST–MyoD or GST–Elk–1‐205. (C) Pull‐down assay of 35S‐labelled Id1, Id2 and Id3 by immobilized GST, GST–MyoD or GST–SAP‐1‐298. Equal molar amounts of 35S‐labelled Id protein were used in each reaction and 20% of the input protein is shown.

Mapping the TCF interaction determinants on the Id proteins

Id1, Id2 and Id3 all interact with the TCFs, albeit at a lower level with Id1. The Id proteins exhibit sequence similarity not only in their HLH domains but also in several other motifs located outside this region (reviewed in Norton et al., 1998). The N‐terminal region has been shown recently to play a role in the promotion of apoptosis induced by Id2 (Florio et al., 1998). In order to determine which of these domains are required for interaction with the TCFs, we used a GST pull‐down assay, with GST fusions to various parts of Id2 and either in vitro translated SAP1‐156 (including the ETS‐domain, Figure 1A) or E47 (Figure 3B and C).

Figure 3.

Interaction with the TCFs requires the Id HLH domain. (A) Schematic illustration of the domain structure of the GST–Id2 proteins used in this study. The black boxes represent the two helices of the HLH domain. Numbers indicate the position of the N‐ and C‐terminal residues with respect to the full‐length Id2 protein. (B and C) Mapping the SAP‐1‐ and E47‐interaction domains on Id2. GST pulldown analysis of 35S‐labelled SAP‐1 (1–156) (B) and 35S‐labelled E47 (C) with either full‐length GST–Id2 (lane 2) or truncated GST–Id2 derivatives (lanes 3–8). Equal molar amounts of each GST‐fusion protein were used in each reaction and 20% of each input protein is shown (lane 1).

SAP‐1 interacts strongly with the proteins possessing a HLH motif, wild‐type Id2, Id2‐ΔN and Id2‐ΔC (Figure 3B, lanes 2, 5 and 6). Interaction is also seen with the Id2 HLH domain alone, albeit at a reduced level [Figure 3B, lane 7, ∼15% wild‐type (WT) protein]. However, very little interaction is seen with the double proline mutant Id2‐HLH(pro) (Figure 3B, lane 8, <2% WT protein). Similarly, SAP‐1 does not interact with either the N‐ or C‐terminal regions of Id2 (Figure 3B, lanes 3 and 4). A known Id‐binding protein, E47, was used as a positive control in these experiments and its binding to the Id2 deletions closely resembles that of SAP‐1 (compare Figure 3B with 3C). Similar results were obtained using MyoD in place of E47 (data not shown). These results therefore indicate that the region of Id2 which interacts with SAP–1 is also required for binding to bHLH proteins such as E47 and MyoD.

Together, our data indicate that the HLH motif of the Id proteins is sufficient for binding to the TCFs, but maximal interaction requires the presence of sequences in either the N‐ or C‐terminal regions.

Ids inhibit ternary complex formation by TCFs and SRF at the c‐fos SRE

Ids bind to, and subsequently inhibit the DNA‐binding activity of, bHLH proteins (Benezra et al., 1990; Sun et al., 1991; reviewed in Norton et al., 1998). Similarly, as Id proteins bind to the TCFs, they may also affect the DNA‐binding activity of TCFs. The ability of Id proteins to disrupt the formation of ternary TCF–SRF–SRE complexes was therefore tested.

Ternary complex formation by both Elk–1 (Figure 4A, lanes 1 and 2) and SAP‐1 (Figure 4A, lanes 3 and 4) was inhibited by Id2. In contrast, the formation of binary SRF–SRE complexes was unaffected (Figure 4A and B, lanes 1–4, and Figure 4C, lanes 1–3). The relative ability of Id1, Id2 and Id3 to abrogate ternary complex formation was investigated (Figure 4B). Each of the Id proteins reduced ternary complex formation by full‐length SAP‐1. Id2 reduced complex formation to the greatest degree, followed by Id3 and then Id1. These results correlate with the strength of interaction detected using in vitro pull‐down assays (Figure 2C).

Figure 4.

Ids disrupt ternary complex formation by Elk–1 and SAP‐1. Gel retardation analysis of complex formation by TCFs in the presence of Id2. (A) Binary (2°) and ternary (3°) complex formation by full‐length Elk–1 (lanes 1–2), SAP‐1 (lanes 3–4), coreSRF and the c‐fos SRE in the absence (lanes 1 and 3) or the presence (lanes 2 and 4) of in vitro translated Id2 protein. Id2 was added prior to addition of the SRE to the binding reaction. (B) Ternary complex formation by full‐length SAP‐1 and the c‐fos SRE in the absence of Id (lane 1) or the presence of equal molar amounts of in vitro translated Id1, Id2 and Id3 proteins (lanes 2–4). (C) Binary (2°) and ternary (3°) complex formation in the absence of TCFs (lanes 1–3) in presence of Elk1‐168 (lanes 4–6) or SAP1‐156 (lanes 7–9), coreSRF and the c‐fos SRE. Increasing amounts (0 μl, lanes 1, 4 and 7; 0.4 μl, lanes 2, 5 and 8; 1.6 μl, lanes 3, 6 and 9) of in vitro translated Id2 protein were added prior to the addition of SRE to the binding reaction. Three‐fold (Elk1‐168) and 6‐fold (SAP1‐156) reduction in ternary complex formation is observed at the highest concentrations of Id2 used. (D) As in (C), except that SRF was omitted from the reaction and the E74‐binding site was used.

The C‐terminally truncated Elk–1 and SAP‐1 proteins, Elk1‐168 and SAP1‐156 (Figure 1A), containing the ETS‐domain and B‐box regions, are sufficient for ternary complex formation (Dalton and Treisman, 1992; Janknecht and Nordheim, 1992) and interaction with Id proteins (Figure 1C). The Id proteins also inhibit ternary complex formation by Elk1‐168 and SAP1‐156 in a dose‐dependent manner (Figure 4C, lanes 4–9) with 50% inhibition occurring at approximately a 1:1 stoichiometry (data not shown). The inhibitory activity of the Id proteins therefore acts on one of the two conserved N‐terminal domains in the TCFs.

TCFs can also bind to high‐affinity ets motifs in the absence of SRF (Janknecht et al., 1994; Shore and Sharrocks, 1995). However, in contrast to their effect on ternary complex formation, the Ids do not affect the ability of Elk1‐168 and SAP1‐156 to bind to the high‐affinity E74 site (Figure 4D).

Collectively, these data indicate that Id proteins inhibit ternary complex formation by interacting with the N‐terminal part of the TCF protein.

Ids actively displace TCFs by inhibiting ETS‐domain–DNA interactions

Ids can inhibit the formation of ternary TCF–SRF–SRE complexes (Figure 4). The ability of Ids to displace TCFs from preformed ternary complexes was investigated. Following ternary complex formation by SAP1‐156, Id2 was added and samples tested (Figure 5A). Rapid decay of the ternary complex was observed, indicating that Ids can actively disrupt ternary complexes (Figure 5A, lanes 1–6).

Figure 5.

Id2 actively disrupts complex formation by TCFs by acting on the ETS DNA‐binding domain. (A) Gel‐retardation analysis of binary (2°) and ternary (3°) complex formation by SAP1‐156, coreSRF and the c‐fos SRE in the presence of Id2. Ternary complex formation was allowed to proceed for 30 min prior to the addition of Id2. Samples were withdrawn at the indicated times following Id2 addition. (B) Sequences of the ets‐binding motifs within the E74, CECI and SRE sites. Bases identified as important for TCF binding from a c‐fos SRE in the presence of Id2. Ternary complex formation was Bases which differ from the optimal E74 motif are highlighted. (C) Gel‐retardation analysis of binary (2°) complex formation by Elk1‐93 and SAP1‐92 on the CECI site. Increasing amounts (0 μl, lanes 1 and 4; 0.4 μl, lanes 2 and 5; 1.6 μl, lanes 3 and 6) of Id2 were added prior to adding the CECI site to the reaction.

Both protein–DNA and protein–protein interactions are essential for ternary complex formation by TCFs (Dalton and Treisman, 1992; Janknecht and Nordheim, 1992; Shore and Sharrocks, 1994). Ids might therefore act to disrupt either of these two types of intermolecular interactions. Binding of TCFs to the high‐affinity E74 motif was unaffected by Ids (Figure 4D). However, the ets motif within the c‐fos SRE represents a lower affinity TCF‐binding motif (Shore and Sharrocks, 1995). We therefore tested whether Id2 could disrupt the binding of TCFs to lower affinity ets motifs such as in the CECI site (Ling et al., 1998). The binding of the ETS‐domains from Elk–1 and SAP‐1 to this site were both inhibited in a dose‐dependent manner by Id2 (Figure 5C, lanes 1–6). Consistent with the observation that Ids interact with the ETS‐domain (Figure 1C), Id2 inhibits DNA binding by this isolated TCF domain (Figure 4C).

Together these results indicate that Ids can actively displace preformed ternary complexes and that this inhibition occurs, at least in part, by affecting DNA binding by the ETS‐domain on suboptimal ets motifs.

c‐fos and Id1–3 expression show a temporal overlap following serum induction

c‐fos and Id1–3 represent immediate‐early genes (Christy et al., 1991; Deed et al., 1993; Barone et al., 1994; Hara et al., 1994; reviewed in Herschman, 1991; Murphy et al., 1997) that are activated following stimulation with serum and other mitogens. In order to determine the relative temporal expression profiles of these genes in NIH 3T3 cells following serum stimulation, we analysed their expression using Northern blotting (Figure 6). The expression of c‐fos peaked 30 min after serum stimulation, was maintained at a high level for 60 min and then gradually returned to basal levels after 2.5 h. Id1, Id2 and Id3 were induced between 30 and 60 min after serum stimulation and their expression was maintained for at least 4 h, consistent with previous studies (Barone et al., 1994; Hara et al., 1994). In contrast, Id4 was not detectably expressed in this cell type (data not shown).

Figure 6.

c‐fos and Id1–3 expression show a temporal overlap following serum induction. NIH 3T3 cells were serum starved for 24 h, then stimulated with 15% fetal calf serum (FCS) and harvested at the times indicated. Northern blots are shown after probing for each Id gene and for c‐fos. RNA loadings were normalized by reference to the 28S rRNA, visible on an ethidium bromide‐stained gel (shown in the bottom panel).

Thus, significant overlap in the expression profile of the Ids and c‐fos is observed, with the expression of the Ids coinciding with the shut off of the c‐fos promoter.

Ids inhibit c‐fos promoter activation via the SRE

The TCFs play a pivotal role in c‐fos induction following growth factor and mitogenic stimulation, and the Id proteins interact and inhibit ternary complex formation by the TCFs. The temporal overlap in Id and c‐fos expression suggests that the Id proteins might participate in the down‐regulation of c‐fos and other immediate‐early genes. To test this hypothesis, we examined the effect of overexpression of Id2 on the activity of a c‐fos promoter‐driven reporter construct (Figure 7A). A constitutively active MEK protein was co‐transfected to stimulate the c‐fos reporter via the extracellular signal‐related kinase (ERK) pathway (Yang et al., 1998a). Co‐transfection of increasing quantities of Id2 caused a dose‐dependent reduction in the activity of the reporter in response to MEK induction. At the highest level of transfected Id2, the reporter was inhibited to below basal levels (Figure 7A). ERK‐mediated induction of the c‐fos promoter is thought to take place mainly via the TCF component of the ternary TCF–SRF–SRE complex (reviewed in Treisman, 1994). If Id proteins function by interacting with Elk–1 and disrupting this ternary complex, then overexpression of Elk–1 should rescue the activity of the Id‐inhibited c‐fos promoter. Co‐transfection of Id2 reduced MEK‐inducible activation of the c‐fos promoter. However, co‐transfection of increasing quantities of Elk–1 caused a dose‐dependent rescue of c‐fos reporter gene activity (Figure 7B), indicating that inhibition of the c‐fos promoter by the Ids possibly involves interaction with the TCFs bound at the SRE.

Figure 7.

Id2 attenuates ternary‐complex‐mediated reporter activity at the c‐fos SRE. The layout of the reporters is represented as diagrammatic inserts. (A) Id2 down‐regulates a c‐fos promoter‐regulated CAT reporter activated by the MEK/ERK pathway. NIH 3T3 cells were co‐transfected with 1 μg of −711fosCAT reporter vector, 0.5 μg pCMVMEK‐1(DN) either in the absence or in the presence of 0.5, 1.0, 3.0 and 6.0 μg of pcDNAId2, respectively. (B) Elk–1 can rescue the down‐regulatory effect of Id2. NIH 3T3 cells were co‐transfected with 1 μg of −711fosCAT reporter vector, 0.5 μg pCMVMEK‐1(CA) and 4.0 μg of pcDNAId2 either in the absence or in presence of 0.5 or 2.5 μg of pCMVElk–1. (C) The SRE is sufficient to mediate the Id down‐regulation of the fos promoter. NIH 3T3 cells were co‐transfected with 1 μg of SREx2‐luc reporter vector, 0.5 μg pCMVMEK‐1(DN) either in the absence or presence of 0.5, 1.0, 3.0 and 6.0 μg of pcDNAId2, respectively. Data are presented relative to the reporter plasmid alone (taken as 1). All values and standard errors were calculated from averages of triplicate samples and are representative of two or three independent experiments.

In order to determine whether the Id proteins cause c‐fos promoter inhibition through the SRE, the ability of Id2 to inhibit the activity of an SRE‐driven reporter construct was analysed. Co‐transfection of MEK enhances the activity of the SRE‐luc reporter. However, co‐transfection of increasing quantities of Id2, causes a dose‐dependent reduction in the activity of the reporter in response to MEK induction. As observed on the intact c‐fos promoter, at the highest level of transfected Id2, the SRE‐driven reporter was inhibited to below basal levels (Figure 7C).

Together, these results demonstrate that the Id proteins can act to inhibit MEK‐inducible c‐fos promoter activation. This inhibition occurs via the TCF component of the ternary complex formed on the c‐fos SRE.

Discussion

The TCF subfamily of ETS‐domain transcription factors plays a pivotal role in regulating the expression of immediate‐early genes such as c‐fos and egr‐1 in response to a variety of extracellular signals including various mitogens and growth factors (reviewed in Cahill et al., 1996). These signals are transduced to the TCFs by the ERK MAPK pathway (reviewed in Treisman, 1994). Genes encoding the Id proteins are also activated in an immediate‐early fashion following mitogenic stimulation. Here, we demonstrate that following serum induction, the expression of Id1–3 in fibroblasts is activated shortly after c‐fos and is maintained as c‐fos expression is subsequently down‐regulated, suggesting a possible role for the Ids in this down‐regulation process. Indeed, overexpression of Id2 causes a reduction in the activity of a c‐fos promoter‐driven reporter construct. This inhibition occurs via the SRE and can be rescued by Elk–1 overexpression, indicating that the ternary TCF–SRF–SRE complex is the target for Id‐mediated inhibition of c‐fos promoter activity. Consistent with this hypothesis, we show that the Id proteins bind to and inhibit DNA binding by the TCF component of this ternary complex. Collectively, our results reveal a novel facet of Id protein function in the coordination of early responses to mitogenic signalling.

Mechanism of Id‐mediated inhibition of DNA binding by TCFs

Interactions between the Id proteins and the TCFs are mediated by the ETS‐domain of the TCF component and the HLH motif in the Ids. In the Ids, additional N‐ and C‐terminal residues apparently enhance these interactions, most probably by stabilizing the structure of the HLH motif. This interpretation is consistent with the recent observation that regions from outside the HLH motif are required for the stability of the HLH motif in the E‐class of bHLH proteins (Goldfarb et al., 1998). Id2 interacts preferentially with and inhibits DNA binding by SAP‐1 and SAP‐2, which is consistent with the greater similarity shared between these two proteins when compared with Elk–1. Interactions between Id proteins and ETS‐domain proteins from other subfamilies, including PEA3 and Fli–1, can also be detected (P.R.Yates and A.D.Sharrocks, unpublished data), which is consistent with the role of the highly conserved ETS‐domain in this interaction. However, in the case of the non‐TCF ETS‐domain proteins, the functional consequences of these interactions are currently unclear. Interestingly, the HLH motif in the Id proteins exhibits sequence similarity with the NID domain in SAP‐2 which acts in cis to inhibit DNA binding by SAP‐2 (Maira et al., 1996). This raises the possibility that the Id proteins inhibit DNA binding in an analogous manner to the NID. Inhibition of DNA binding by Id proteins takes place in a site‐dependent manner. The binding of TCFs to low‐affinity ets motifs (such as in the c‐fos SRE), either alone or in ternary complexes with SRF, is inhibited by Ids but binding to high‐affinity ets motifs (such as the E74 site) is unaffected. Ids can inhibit DNA binding by TCFs when added before or after the formation of TCF–DNA complexes, suggesting that they are able to disrupt preformed ternary complexes. This also suggests that the mechanism of inhibition of DNA binding possibly involves an allosteric effect rather than simple blocking of the DNA‐binding surface. Indeed, the observation that Ids are incapable of blocking the binding of the TCFs to the high‐affinity E74 site supports this hypothesis. A model has been proposed previously in which residues in SAP‐1 can act as allosteric effectors which alter DNA contacts made by other residues and hence DNA‐binding specificity, but do not affect binding to high‐affinity sites (Shore et al., 1996; Mo et al., 1998). The binding of Id proteins may promote similar allosteric changes which only affect binding to lower affinity sites.

Id1, Id2 and Id3 can all interact with and inhibit DNA binding by the TCFs. However, Id1 binds and inhibits with a lower efficiency, suggesting some specificity in their interaction. Id proteins have previously been shown to inhibit DNA binding by bHLH proteins (Benezra et al., 1990; Sun et al., 1991; reviewed in Norton et al., 1998) by forming inactive heterodimers. However, little sequence similarity exists between the ETS‐domain and bHLH proteins, indicating that the mechanisms of inhibiting DNA binding possibly differ. Id proteins do not, however, act as pleiotropic inhibitors of DNA binding as they do not affect DNA binding by the MADS‐box proteins SRF (Figure 4) and MEF2A (E.C.Roberts and A.D.Sharrocks, unpublished data).

The role of Ids in regulating immediate‐early gene expression

Following mitogenic stimulation, the Id genes are expressed in a classic immediate‐early manner (Christy et al., 1991; Deed et al., 1993; Barone et al., 1994; Hara et al., 1994). However, their expression is delayed in comparison with other immediate‐early genes such as c‐fos (Figure 6). It has previously been demonstrated that the product of another immediate‐early gene (egr‐1) acts to up‐regulate the expression of Id1 (Tournay and Benezra, 1996) and probably also Id3 (Christy et al., 1991), suggesting the existence of a cascade of genes which are sequentially activated following serum stimulation. In this study, we demonstrate that the Id proteins can inhibit the c‐fos promoter, suggesting the existence of a negative feedback loop which acts to down‐regulate genes such as c‐fos and egr‐1. Consistent with this hypothesis is the observation that the Ids elicit their function by affecting the activity of the TCFs which have been implicated in regulating both of these genes (reviewed in Treisman, 1994; Watson et al., 1997). Activation of the TCFs is mediated by phosphorylation of their transcriptional activation domains by MAPKs and leads to enhancement of both their DNA binding and transcriptional activation activities (reviewed in Treisman, 1994). However, the occupancy of the ets motif within the c‐fos SRE does not change following growth factor stimulation (Herrera et al., 1989). An exchange of active and inactive TCF molecules has been suggested to occur with the phosphorylated form displacing the non‐phosphorylated form (Gille et al., 1992). Alternatively, the TCFs could be phosphorylated while bound to the promoter (reviewed in Treisman, 1994). Recently, it has been demonstrated that the major constituent of ternary TCF–SRF–SRE complexes in several cell types is Net‐b, a truncated version of the murine SAP‐2 homologue Net (Giovane et al., 1997). Net‐b lacks the transcriptional activation domain and acts as a dominant‐negative inhibitor of SRE function. Thus, one role of the Id proteins in this context would be to promote dissociation of the phosphorylated TCFs and allow their replacement with either inactive or dominant‐negative forms of the TCFs, thereby providing an elegant mechanism for down‐regulating and resetting the promoter.

In summary, we have demonstrated direct functional interactions between the Id HLH proteins and members of the TCF subfamily of ETS‐domain transcription factors. These observations increase the repertoire of proteins known to be inhibited by Id proteins. In addition, the discovery that TCFs interact with Id proteins contributes to our understanding of the intricate mechanisms by which these transcription factors are regulated. Together, our results contribute to the understanding of the early events that follow mitogenic signalling and progression through the G1 phase of the cell cycle and how they are controlled in an integrated manner.

Materials and methods

Plasmid constructs

The following plasmids were used to express C‐terminal hexa histidine‐tagged proteins in Escherichia coli. pAS413 encodes full‐length Elk–1 (amino acids 1–428) (Ling et al., 1997). pAS275 encodes full‐length SAP‐1 (amino acids 1–431) (Shore et al., 1996).

The following plasmids were used to express GST‐fusion proteins in E.coli. pAS58 (encoding coreSRF amino acids 132–222), pGEX–MyoD [encoding GST–MyoD; full‐length MyoD (amino acids 1–318) fused to GST; kindly provided by H.Weintraub] and pGNElk (encoding GST–Elk–1‐205; Elk–1 amino acids 1–205 fused to GST) (Gille et al., 1995a) have been described previously. pGEX–Id2 encodes GST–Id2 (Id2 amino acids 1–134 fused to GST) (Deed et al., 1997). GST–Id2 deletion constructs, pAS910, pAS911, pAS912, pAS913, pAS914, encoding amino acids 1–36, 76–134, 36–134, 1–76, 36–76 of Id2 fused to GST, respectively, were constructed by PCR amplification from pcDNA3Id2 using oligonucleotide primer pairs ADS630–ADS612, ADS614–ADS631, ADS606–ADS631, ADS630–ADS613, ADS606–ADS613, respectively, and ligated as BamHI–XhoI fragments into the same sites of pGEX‐KG (Guan and Dixon, 1991). pAS915 (GST–Id2 deletion construct containing HLH domain amino acids 36–76 with the point mutations S44P and L68P) was constructed by PCR amplification from pCDNAId2H1H2br using the oligonucleotide pair ADS606–ADS613, followed by ligation as a BamHI–XhoI fragment into the same sites of pGEX‐KG. pAS535 (encoding GST–SAP‐1‐298; SAP‐1 amino acids 1–298 fused to GST) was constructed by ligating an NcoI–SacI fragment into the same sites of pGEX‐KG.

The following plasmids were used in mammalian cell transfections. pF711CAT contains human c‐fos promoter −711 to +42 ahead of the CAT gene (Treisman, 1985). pSRE‐luc contains two copies of the c‐fos serum response element (nucleotides −357 to 275) upstream from a minimal tk promoter and the luciferase gene (Seth et al., 1992). The expression vector for constituitively active MEK‐1(ΔN S218E–S222D) (pCMV–MEK–DN) was provided by N.Ahn (Mansour et al., 1994). pAS383 encodes full‐length Flag‐epitope‐tagged Elk–1 driven by a CMV promoter (Yang et al., 1998a).

The following plasmids were used for in vitro transcription/translation and/or transfection purposes. pcDNA3Id1 (encoding full‐length Id1; amino acids 1–148), pcDNA3Id2 (encoding full‐length Id2; amino acids 1–134), pcDNA3Id3 (encoding full‐length Id3; amino acids 1–119) (Norton and Atherton, 1998), pAS728 (encoding full‐length Elk–1; amino acids 1–428) (Ling et al., 1997), pAS96 (encoding Elk–1‐168; amino acids 1–168) and pAS72 (encoding Elk–1‐93; amino acids 1–93) (Shore and Sharrocks, 1994), pAS136 (encoding SAP‐1‐92; amino acids 1–92), pAS168 (encoding SAP‐1‐156; amino acids 1–156) (Shore and Sharrocks, 1995), pT7.SAP‐1 (encoding full‐length SAP‐1; amino acids 1–431) (Dalton and Treisman, 1992) and pT7.SAP‐2 (encoding full‐length SAP–2; amino acids 1–407) (Price et al., 1995) have been described previously. pAS700 (encoding Elk–1‐306; amino acids 1–306) was constructed by ligating the NcoI–BglII fragment from pAS413 into the NcoI–BamHI‐cleaved pAS37 (Sharrocks et al., 1993a). A full‐length cDNA encoding E47 (Sun et al., 1991) was cloned into pcDNA3 for in vitro transcription/translation.

Protein expression

The synthesis of proteins by in vitro transcription and translation was carried out with the TNT‐coupled reticulocyte lysate system (Promega, WI) according to the manufacturer's recommendations. Newly synthesized 35S‐labelled proteins were analysed by SDS–PAGE followed by visualization and quantification by phosphoimager analysis (Fuji BAS‐1500 phosphoimager and TINA 2.08e software). GST fusion proteins (Shore and Sharrocks, 1994) and hexahistidine‐tagged proteins (Yang et al., 1998a) were purified as described previously. The purity of proteins was determined by SDS–PAGE and the concentration by Coomassie Blue protein assay reagent (Pierce, IL).

In vitro protein–protein interaction assays

Interactions between GST–Elk–1‐205 and GST–SAP‐1‐298 and in vitro translated Ids and between GST–Id2 and truncated derivatives of GST–Id2 and in vitro translated full‐length and truncated derivatives of Elk–1 and SAP‐1 were investigated using pull‐down assays as described previously (Shore and Sharrocks, 1994).

Gel‐retardation assays

Gel‐retardation assays were performed with 32P‐labelled probes as described previously (Sharrocks et al., 1993b). The binding sites include the c‐fos SRE, the E74 site (Shore and Sharrocks, 1994) and the CECI site (Ling et al., 1998). The CECI site is an artificial DNA‐binding site which contains a low‐affinity SRF binding site and an ets motif corresponding to the S30 site identified from a DNA‐binding site selection study (Shore and Sharrocks, 1995). DNA–protein complexes were formed at room temperature for 15 min using bacterially expressed coreSRF and bacterially expressed or in vitro‐translated TCF derivatives. In vitro translated Id proteins were either added at the same time as the TCFs or after the end of this incubation period. When Ids were added following pre‐incubations, reactions were allowed to proceed at room temperature for a further 15 min. Generally, reactions were normalized for reticulocyte lysate content. Protein–DNA complexes were analysed on non‐denaturing 5% polyacrylamide gels cast in 0.5× Tris–borate–EDTA and visualized by autoradiography and phosphorimaging.

Cell culture transfection and reporter gene assays

NIH 3T3 and Cos‐7 cells were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS) (Gibco‐BRL, MD) and 25 mM glucose. Triplicate transfection experiments were carried out in six‐well plates using Superfect reagent according to the manufacturer's recommendations (Qiagen, Hilden, Germany). For reporter gene assays, c‐fos‐promoter‐driven or SRE‐driven reporters were co‐transfected alongside vectors encoding constituitively active MEK‐1, Id2 and Elk–1. DNA concentrations were normalized with pcDNA3 vector.

Transfected cells were incubated in Optimem (Gibco‐BRL, MD) for 24 h after transfection and lysed in 200 μl lysis buffer [100 mM K2PO4, pH 7.8, 0.2% Triton X‐100, 0.5 mM dithiothreitol (DTT)]. Cell debris was removed by centrifugation at 14 000 g for 15 min at 4°C and the supernatant assayed for CAT activity using a TLC assay as described previously (Ling et al., 1997). Luciferase assays were carried out as described previously (Yang et al., 1998). Data were quantified by phosphoimager analysis and the data presented graphically using Microsoft Excel software. Transfection efficiency was monitored by measuring the β‐galactosidase activity from co‐transfected pCH110 plasmid (0.5 μg) (Pharmacia KB Biotechnology Inc., Uppsala, Sweden). β‐galactosidase activities were determined by the use of a chemiluminescent substrate in the Galacto‐light Plus kit (tropix, MA).

Immunoprecipitation

The antibody matrix was prepared by covalently coupling an Id2‐specific rabbit polyclonal antibody (Santa Cruz Biotechnology Inc., CA) to protein A beads (Gersten and Marchlonis, 1978). Cos‐7 cell extracts containing overexpressed Id2 and Flag‐tagged Elk–1 proteins were prepared from two 35‐mm dishes in 400 μl of Triton lysis buffer (20 mM Tris pH 7.4, 137 mM sodium chloride, 25 mM β‐glycerophosphate, 50 mM sodium fluoride, 2 mM EDTA, 10% glycerol, 1% Triton X‐100, 2 mM sodium pyrophosphate, 10 mM MgCl2) containing protease inhibitors (at final concentrations: leupeptin 2 μg/ml, pepstatin A 1 μg/ml, PMSF 100 μg/ml and aprotinin 2 μg/ml). Antibody matrix (25 μl) was incubated with the cell extract with rotation for 4 h at 4°C. Complexes were washed three times with Triton lysis buffer (TLB) and subjected to SDS–PAGE followed by Western blot analysis. Immunoprecipitated Elk–1 was detected by immunoblot analysis with a mouse monoclonal anti‐M2 Flag antibody (Kodak, NY). Immune complexes were detected using horseradish‐peroxidase‐conjugated secondary antibody followed by enhanced chemiluminescence (Amersham, Amersham, UK).

Northern hybridization

Total cellular RNA was prepared by using an RNAzol extraction kit (Cinna‐Biotech, USA) and 10 μg samples were electrophoresed on 1.2% agarose–2.2 M formaldehyde gels. After staining with ethidium bromide, gels were blotted onto GeneScreen Plus nylon mebranes (NEN‐Dupont, MA) and hybridized with random‐primed 32P‐labelled probes essentially as described previously (Deed et al., 1993).

Figure generation

Figures were generated electronically from scanned images of autoradiographic immages by using Picture Publisher (Micrografix) and Powerpoint 7.0 (Microsoft) software. Results are representative of the original autoradiographic images.

Acknowledgements

We would like to thank Margaret Bell for excellent technical assistance and Katherine Stewart for secretarial assistance. We would also like to thank Bob Liddell for DNA sequencing, and Alan Whitmarsh, Roger Davis, Harold Weintraub, Richard Treisman and Peter Shaw for reagents. We are grateful to Janet Quinn for comments on the manuscript and members of our laboratory for helpful discussions. This work was supported by the Wellcome Trust and the UK Cancer Research Campaign [CRC]. A.D.Sharrocks is a Research Fellow of the Lister Institute of Preventative Medicine.

References