Multiple mutations contribute to repression by the v-Erb A oncoprotein

Abstract

The v-Erb A oncoprotein of avian erythroblastosis virus is derived from c-Erb A, a hormone-activated transcription factor. Notably, v-Erb A has sustained multiple mutations relative to c-Erb A and functions as a constitutive transcriptional repressor. We report here an analysis of the contributions of these different mutations to v-Erb A function. Our experiments demonstrate that two amino-acid differences between v-Erb A and c-Erb A, located in the ‘I-box,’ alter the dimerization properties of the viral protein, resulting in more stable homodimer formation, increased corepressor binding, and increased target gene repression. An additional amino-acid difference between v- and c-Erb A, located in helix 3 of the hormone binding domain, renders corepressor binding by the viral protein more resistant to release by thyroid hormone. Finally, we report that a C-terminal truncation in v-Erb A not only inhibits exchange of corepressor and coactivator, as previously noted, but also permits v-Erb A to recruit both SMRT and N-CoR corepressors, whereas c-Erb A is selective for N-CoR. The latter two mutations in v-Erb A also impair its ability to suppress c-Jun function in response to T3 hormone. We propose that the acquisition of oncogenic potential by the v-Erb A protein was a multistep process involving a series of mutations that alter the transcriptional repressive properties of the viral protein through multiple mechanisms.

Introduction

The avian erythroblastosis virus (AEV) induces both fibrosarcomas and erythroblastosis in susceptible hosts (Graf and Beug, 1983; Privalsky, 1992; Beug et al., 1996; Samarut, 1996; Rietveld et al., 2001). AEV possesses two oncogenes, v-erb A and v-erb B. V-erb B is essential for oncogenic transformation of both fibroblasts and erythroid cell progenitors, whereas v-erb A cooperates with v-erb B to enhance the neoplastic phenotype (Graf and Beug, 1983; Privalsky, 1992; Beug et al., 1996; Samarut, 1996; Rietveld et al., 2001). For example, erythroid cells transformed by the v-erb B gene alone spontaneously differentiate into more mature reticulocytes and erythrocytes; coexpression of v-erb A blocks this terminal differentiation, keeping the erythroid cells in a more immature, more highly proliferative state (Graf and Beug, 1983; Vennstrom et al., 1987; Gandrillon et al., 1989; Privalsky, 1992; Beug et al., 1996; Samarut, 1996; Rietveld et al., 2001). V-erb A also can extend the proliferative capacity of v-erb B-transformed fibroblasts (Gandrillon et al., 1987; Desbois et al., 1991).

V-erb B is a virally transduced and mutated version of the cellular gene for epidermal growth factor (EGF) receptor, and encodes a constitutively active tyrosine-kinase (Ullrich et al., 1984). In contrast, v-erb A is a transduced and mutated version of the host cell gene (c-erb A) for thyroid hormone receptor-α (TR-α), a member of the nuclear receptor family of hormone-regulated transcription factors (Sap et al., 1986; Weinberger et al., 1986). The c-Erb A protein functions in normal cells by binding to specific DNA sequences (denoted response elements) and regulating the expression of adjacent target genes in response to T3 hormone (Mangelsdorf et al., 1995; Ribeiro et al., 1998; Zhang and Lazar, 2000; Wondisford, 2003). In the absence of T3 hormone, c-Erb A typically recruits corepressors and represses transcription; conversely, the binding of T3 hormone induces a conformational change in the C-terminal helix 12 of c-Erb A, resulting in release of corepressor, the recruitment of coactivators, and the activation of target gene transcription (Glass and Rosenfeld, 2000; Privalsky, 2004). Corepressors and coactivators function, in turn, by modifying the chromatin template and by interacting with the general transcriptional machinery so as to inhibit or enhance transcription, respectively (Glass and Rosenfeld, 2000; Privalsky, 2004).

The v-Erb A protein has sustained a series of alterations relative to c-Erb A, including an N-terminal fusion of viral ‘gag’ sequences, a C-terminal deletion, and 13 internal amino-acid changes (Figure 1) (Sap et al., 1986; Weinberger et al., 1986). The C-terminal deletion in v-Erb A removes helix 12 of c-Erb A, resulting in constitutive corepressor binding by v-Erb A and a constitutive transcriptional repression phenotype (Munoz et al., 1988; Zenke et al., 1988; Damm et al., 1989; Sap et al., 1989). As a consequence, v-Erb A can function as a dominant-negative inhibitor of c-Erb A function, and v-Erb A has been proposed to operate in the erythroleukemic cell by interfering with thyroid hormone signaling (Damm et al., 1989; Evans, 1989; Sap et al., 1989; Schroeder et al., 1992b; Samarut, 1996; Bauer et al., 1998; Stunnenberg et al., 1999; Thormeyer and Baniahmad, 1999; Urnov et al., 2000; Rietveld et al., 2001). However, simple dominant-negative mutants in human thyroid hormone receptors are associated with an endocrine disorder, resistance to thyroid hormone (RTH) syndrome, rather than neoplasia (Usala, 1991; Refetoff, 1993; DeGroot, 1996; Chaterjee, 1997). We therefore examined if the additional alterations sustained by v-Erb A relative to c-Erb A contribute to the oncogenic properties of the viral protein. We report here that amino-acid changes in a dimerization interface (the I-box) and in helix 3 of the hormone binding domain further enhance the transcriptional repression properties of v-Erb A by stabilizing homodimer formation and by interfering with T3 binding. We also report that the loss of helix 12 from v-Erb A not only stabilizes corepressor binding, but also alters the corepressor specificity of the v-Erb A protein, allowing it to efficiently bind both N-CoR and SMRT members of the corepressor family. The helix 3 mutation and the C-terminal truncation in v-Erb A also work together to mitigate c-Jun inhibition by T3. The implications of these observations for v-Erb A-mediated oncogenesis are discussed.

Figure 1
figure1

Schematic of c-Erb A, gag-v-Erb A, and chimeras. The structures of the wild-type c-Erb A protein (chicken thyroid hormone receptor α-1; CCCC), gag-v-Erb A (gVVVV), and the chimeric proteins used in this study are shown. The DNA binding domain (DBD), hormone binding domain (HBD), I-box, helices 1–3 (H1–3), and helix 12 regions are indicated for c-Erb A. Retroviral gag sequences, N- and C-terminal deletions, and amino-acid substitutions (asterisks) within the v-Erb A sequence are also shown and are numbered based on the c-Erb A sequence. The locations of the sites used for constructing the various chimeras are indicated by vertical dotted lines; chimera proteins are named based on the origin of swapped fragments (right and bottom of the panel)

Results

Substitutions in the v-Erb A I-box enhance homodimerization and impair RXR heterodimerization

C-Erb A can bind to DNA response elements as a protein monomer, homodimer, or heterodimer with other members of the nuclear receptor family, such as the retinoid X receptors (RXRs) (Lazar et al., 1991; Forman et al., 1992; Glass, 1994). It has been reported that gag-v-Erb A is impaired in its ability to form heterodimers with RXR compared to c-Erb A (Forman et al., 1989; Selmi and Samuels, 1991; Yen et al., 1994; Wahlstrom et al., 1996; Bauer et al., 1997; Shen and Subauste, 2000; Yoh and Privalsky, 2001; Zubkova and Subauste, 2003). To investigate this phenomenon in more detail, we compared DNA binding by the viral and cellular proteins using an electrophoretic mobility shift assay (EMSA) and a radiolabeled DNA probe (a divergent repeat-6 response element, DIV-6, derived from the chicken F2 lysozyme promoter; Baniahmad et al., 1992). We confirmed that both gag-v-Erb A and c-Erb A formed homodimers on this element, but that gag-v-Erb A was more resistant to heterodimer formation with RXR than was c-Erb A (Figure 2); compare c-Erb A (denoted CCCC in our nomenclature; Figure 1) to gag-v-Erb A (denoted gVVVV). We next mapped the sequences responsible. A gag-c-Erb A protein fusion (gCCCC) displayed similar dimerization properties as did native c-Erb A (CCCC); reciprocally, a v-Erb A protein lacking gag sequences (VVVV) mimicked gag-v-Erb A (gVVVV) in this assay (Figure 2). These results indicated that the impaired RXR heterodimerization properties of gag-v-Erb A are not due to the gag sequences. We next analyzed a series of c-/v-Erb A proteins representing reciprocal exchanges of sequence within the Erb A coding region itself (Figure 1). Analysis of these v-/c-Erb A chimeras localized the relevant sequences to a region bearing four amino-acid substitutions between c- and v-Erb A (Figure 3, compare the efficient RXR heterodimerization properties of the VVCV construct with the inefficient RXR heterodimerization properties of the CCVC construct). Notably, this region includes the ‘I-box’ previously implicated as a nuclear receptor dimerization interface (Forman and Samuels, 1990; Au-Fliegner et al., 1993; Glass, 1996; Perlmann et al., 1996; Reginato et al., 1996; Ribeiro et al., 2001).

Figure 2
figure2

Resistance of v-Erb A to heterodimerization with RXR. (a) The relative ability of a v-Erb A construct lacking gag sequences (VVVV) to form RXR heterodimers compared to c-Erb A (CCCC) was determined. C-Erb A (approximately 40 ng) or v-Erb A (approximately 20 ng) was synthesized using a baculovirus/Sf9 cell expression system and was incubated with a radiolabeled DIV-6 DNA probe together with increasing amounts of RXRγ (from 0 to approximately 22 ng). A representative gel electrophoretogram is shown. Arrows indicate the relevant protein/DNA complexes. Free probe was present in excess in all EMSA experiments; to save space, the position of the free DNA probe was cropped off the panel. (b) Quantification of the RXR heterodimerization properties of v-Erb A versus c-Erb A. The ability of increasing amounts of RXR protein to convert Erb A homodimers to RXR/Erb A heterodimers was tested using EMSA and a radiolabeled DIV-6 DNA probe. Different Erb A protein constructs either containing or lacking gag sequences (Figure 1) were synthesized by in vitro translation, mixed with increasing quantities of RXR (as indicated below the panel, 11 ng/μl), and the formation of Erb A homodimer/DNA complexes versus RXR/Erb A heterodimer/DNA complexes was quantified. All experiments were performed in the absence of T3

Figure 3
figure3

Impaired RXR heterodimerization properties of v-Erb A map to two amino-acid substitutions, P363S and T370A, in the I-box. An EMSA protocol as in Figure 2 was employed to determine the ability of c-Erb A, v-Erb A, and the various v-/c-Erb A chimeras (Figure 1) to bind to a radiolabeled DIV-6 DNA element as homodimers or as heterodimers with RXR. Erb A protein concentration levels were adjusted to yield equal quantities of homodimers in the absence of RXR. Increasing concentrations of RXR (indicated on left) were added and the amount of radiolabeled DIV-6 probe migrating as an Erb A homodimer or as an RXR heterodimer was quantified. All experiments were performed in the absence of T3. The mean and standard deviation of three separate experiments are shown

To identify the specific amino-acid substitutions responsible, we exchanged each of the four I-box amino acids individually, or in combination, in the v- and c-Erb A sequences. Reciprocal substitutions into codons 324 and/or 378 had little or no effect on the dimerization properties of their parental proteins (data not shown). However, introduction of the v-Erb A amino acids at codon 363 or 370 into c-Erb A (i.e. CCCC(P363S) or CCCC(T370A)) reduced the ability of c-Erb A to form heterodimers with RXR (Figure 3). Conversely, introduction of the reciprocal c-Erb A-like substitutions into v-Erb A (i.e. VVVV(S363P) or VVVV(A370T)) enhanced the ability of v-Erb A to heterodimerize with RXR (Figure 3). A double mutant of c-Erb A bearing both v-Erb A-like substitutions (CCCC(P363S,T370A)) displayed the weak heterodimerization properties characteristic of v-Erb A, whereas the reciprocal double mutant of v-Erb A (VVVV(S363P,A370T)) displayed the stronger heterodimerization properties characteristic of c-Erb A (Figure 3). We conclude that two amino-acid substitutions, at codons 363 and 370, are jointly responsible for the impaired RXR heterodimerization properties of the viral oncoprotein.

The c- and v-Erb A protein concentrations in Figure 3 were adjusted to yield equal levels of homodimers in the absence of RXR. Notably, when assayed at equimolar protein concentrations, gag-v-Erb A displayed an elevated ability to form homodimers relative to c-Erb A (Figure 4). These enhanced homodimerization properties of gag-v-Erb A were most evident on the DIV-6 element, although they could also be detected on a direct-repeat (DR-4) element (Figure 4 and data not shown). Assays using our chimeric constructs demonstrated that the same two I-box substitutions responsible for decreased RXR heterodimerization by v-Erb A were also responsible for its enhanced homodimerization properties when assayed alone (Figure 4, compare the reciprocal substitutions in VVVV and in CCCC with the wild-type proteins). We conclude that two amino-acid changes in a receptor dimerization interface in v-Erb A simultaneously enhance homodimerization, and reduce RXR heterodimerization by the viral oncoprotein.

Figure 4
figure4

Enhanced homodimer properties of v-Erb A map to two amino-acid substitutions, P363S and T370A, within the I-box. An EMSA protocol similar to that in Figure 2 was employed to determine the ability of c-Erb A, v-Erb A, and the various v-/c-Erb A chimeras (Figure 1) to bind as homodimers to a radiolabeled DIV-6 DNA element. All Erb A constructs were prepared at identical molarities (13 pmol/μl) and were tested over an identical range of protein concentrations (left of the panel). The amount of Erb A homodimer/DNA complex formed at each protein concentration was quantified; the mean and range of two experiments are shown

The I-box substitutions also inhibit the ability of v-Erb A to form heterodimers with RAR and to bind to DNA as a monomer

C-Erb A forms heterodimers not only with RXRs but also with retinoic acid receptors (RARs) (Glass et al., 1989; Yen et al., 1992; Lee and Privalsky, 2005). RAR/c-Erb A heterodimers display unique transcriptional regulatory properties and form with efficiencies comparable to RXR/c-Erb A heterodimers on appropriate DR-4 and DR-5 elements (Lee and Privalsky, 2005). We therefore next examined the ability of v-Erb A to heterodimerize with RAR relative to that of c-Erb A. V-Erb A shared the ability of c-Erb A to form heterodimers with RARs on a DR-4 element, but was severely impaired in its ability to form RAR heterodimers on a DR-5 element (Figure 5a). Neither v- nor c-Erb A efficiently formed RAR heterodimers on a DIV-6 or INV-0 DNA element (Lee and Privalsky, 2005 and data not shown). Surveying our chimeric c-/v-Erb A constructs demonstrated that the restricted RAR heterodimerization properties of the viral oncoprotein mapped to the same two codons, 363 and 370, that were responsible for its impaired RXR heterodimerization properties, with both amino-acid substitutions contributing to this phenotype (Figure 5a).

Figure 5
figure5

Contribution of v-Erb A I-box substitutions to changes in v-Erb A heterodimerization with RAR, monomeric DNA binding, and resistance of homodimers to disruption by T3. (a) The ability of c-Erb A, v-Erb A, or v-/c-Erb A chimeras (Figure 1) to form heterodimers with RAR was tested using an EMSA protocol and radiolabeled DR-4 (top) or DR-5 (bottom) DNA probes. The Erb A construct of interest was mixed with a fixed concentration of RAR and the amount of RAR/Erb A heterodimer complex formed, relative to Erb A homodimer formed in the absence of RAR, was quantified. The mean and standard deviation of three experiments are shown. (b) The ability of c-Erb A chimera proteins to bind to DNA as monomers was tested using the EMSA protocol with the DIV-6 probe in the absence of T3, as in Figure 2a. The appropriate region of a representative electrophoretogram is presented (top of the panel). The amount of monomer complexes was quantified; the mean and standard deviation of three experiments are shown (bottom of the panel). (c) The stability of Erb A homodimers to disruption by T3 was tested using the EMSA protocol and v-Erb A, c-Erb A, or v-/c-Erb A chimeras (Figure 1). The Erb A proteins were isolated from transfected COS-1 cells and the amount of Erb A homodimer/DNA complex formed by each protein on a radiolabeled DIV-6 probe was determined at different T3 concentrations. A representative electrophoretogram is presented (left of the panel). The data were also quantified for these and additional chimeras (right of the panel); homodimer formation in the absence of T3 was defined as 100%. The mean and range of two independent experiments are shown

A third mode of DNA recognition by c-Erb A is the ability to bind to suitable response elements as a protein monomer (Lazar et al., 1991; Forman et al., 1992; Darling et al., 1993; Glass, 1994; Katz and Koenig, 1994; Subauste and Koenig, 1995, 1998; Wahlstrom et al., 1996; Zubkova and Subauste, 2002, 2003). These c-Erb A monomers can be observed on DNA elements containing a single half-site, or as part of a mixed population of monomers, homodimers, and heterodimers on response elements containing two or more half-sites. We observed that the ability of v-Erb A to form monomers on DNA was much weaker than that of c-Erb A (Figure 5b). Analysis of our chimeric constructs again implicated the I-box sequences in this phenomenon and demonstrated that substitution of the two I-box codons in c-Erb A with the corresponding v-Erb A sequences (P363S/T370A) reduced or abolished monomeric DNA binding (Figure 5b). However, and in contrast to our studies on heterodimer formation, a c-Erb A mutant bearing only the v-Erb A alanine substitution at codon 370 was strongly impaired in monomer binding, whereas the v-Erb A serine substitution at codon 363 had little or no effect on monomer binding (Figure 5b). Intriguingly, v-Erb A constructs bearing the reciprocal mutation(s) in v-Erb A did not regain monomeric binding (data not shown), indicating that although the amino acid at position 370 was necessary, it was not sufficient for efficient monomer binding to DNA.

We conclude that, relative to c-Erb A, the gag-v-Erb A oncoprotein (a) more efficiently forms homodimers at a given protein concentration, (b) is more resistant to formation of RXR heterodimers on a variety of response elements, (c) is unable to form RAR heterodimers on a subset of response elements, and (d) is impaired in the ability to bind to DNA as a receptor monomer. All of these properties map to one or both of two amino-acid substitutions (P363S and T370A) acquired by the viral oncoprotein within the I-box dimerization interface.

The v-Erb A I-box sequence also contributes to stabilizing receptor homodimer complexes against disruption by T3

Although c-Erb A homodimers assemble with high affinity on many DNA elements in the absence of hormone, these c-Erb A homodimers typically dissociate in the presence of T3, whereas RXR/c-Erb A heterodimers are stable to T3 (Lazar et al., 1991; Forman et al., 1992; Ribeiro et al., 1992; Yen et al., 1992, 1994; Miyamoto et al., 1993; Piedrafita et al., 1995; Yoh and Privalsky, 2001). This T3-driven dissociation of homodimers may, in fact, function to drive an exchange of the c-Erb A/c-Erb A homodimers implicated in transcriptional repression for the RXR/c-Erb A heterodimers involved in transcriptional activation. Consistent with these prior studies, c-Erb A homodimers (CCCC) bound to a DIV-6 element were dissociated by addition of T3 (Figure 5c). V-Erb A homodimers (VVVV) were resistant to dissociation by T3 under the same conditions (Figure 5c), as expected from the known inability of v-Erb A to bind to T3 hormone. Notably, several of our v-/c-Erb A chimeras that did bind T3 were nonetheless resistant to T3 in this assay. In particular, the CCVC and CVVC chimeras retained substantial homodimer formation at even the highest T3 concentrations tested (Figure 5c), despite the ability of these chimeras to bind to T3 and the presence of an intact c-Erb A-derived helix 12 (see below). All four amino-acid substitutions in the I-box of v-Erb A contributed to this hormone resistance (data not shown). These results indicate that the I-box amino-acid substitutions acquired by v-Erb A favor a homodimeric mode of DNA binding by multiple mechanisms, including rendering the v-Erb A homodimer more resistant to dissociation by T3.

V-Erb A displays an enhanced ability to bind to, and an altered preference for, corepressors compared to c-Erb A

C-erb A homodimers recruit corepressor much more efficiently than do c-Erb A/RXR heterodimers (Cohen et al., 1998, 2000; Yoh and Privalsky, 2001; Zubkova and Subauste, 2004). We inquired if this same preference of corepressor for homodimers was also true for v-Erb A. We employed an EMSA supershift assay in which interaction of v- or c-Erb A with corepressor was detected as a supershift of the receptor/DNA complex to a slower electrophoretic mobility (Figure 6a, and quantified in Figure 6b). As anticipated, both v-Erb A and c-Erb A homodimers were strongly supershifted by an N-CoR construct, whereas RXR heterodimers of either c-Erb A or v-Erb A were inefficiently supershifted by N-CoR (Figure 6).

Figure 6
figure6

Preferential interaction of corepressors with Erb A homodimers versus RXR/Erb A heterodimers. EMSA was employed to determine the ability of Erb A homodimers and RXR/Erb A heterodimers to interact with corepressors in the absence of hormone. C-Erb A protein (CCCC) or v-Erb A (VVVV) protein (20 ng; isolated from recombinant baculovirus/Sf9 cells) were mixed with the N-CoR or SMRT constructs indicated (200 ng/μl) in the absence or presence of RXRγ (5.5 ng). The proteins were incubated with a radiolabeled DIV-6 DNA probe, and the resulting protein/DNA complexes were resolved by native gel electrophoresis. (a) A representative electrophoretogram is presented; the positions of homodimers, heterodimers, and complexes supershifted by the N-CoR or SMRT corepressor are indicated. (b) The amount of Erb A homodimer or RXR/Erb A heterodimer DNA complex was quantified, as was the amount of the corresponding corepressor-bound (supershifted) Erb A or RXR/Erb A DNA complexes

SMRT corepressor is closely related to N-CoR but differs in its affinity for different nuclear receptors (Privalsky, 2004). Although alternative mRNA splicing can produce corepressors with differing receptor affinities, c-Erb A generally interacts more strongly with N-CoR than with SMRT (Zamir et al., 1997; Cohen et al., 1998, 2000, 2001; Webb et al., 2000; Makowski et al., 2003; Goodson et al., 2005). To examine this question for v-Erb A, we tested SMRT constructs in place of N-CoR in our EMSA supershifts. Both c- and v-Erb A homodimers were supershifted by the SMRT construct, whereas the corresponding RXR heterodimers were resistant to interaction with SMRT. Interestingly, although c-Erb A displayed a stronger interaction with N-CoR than with SMRT, v-Erb A displayed a near-equal ability to interact with either N-CoR or SMRT (Figure 6, compare the SMRT and N-CoR supershifts, filled bars, for VVVV+RXR in the top right panels with the corresponding corepressor supershifts for CCCC+RXR in the bottom right panels). Neither v-Erb A nor c-Erb A bound significantly to negative controls, such as a nonrecombinant GST construct (data not shown).

We used our chimeras to map the sequences responsible for the acquisition of enhanced SMRT binding by v-Erb A. Converting the I-box domain of c-Erb A to that of v-Erb A (CCVC) or reciprocally converting the I-box domain of v-Erb A to that of c-Erb A (VVCV) did not alter the relative preference of the parental constructs for N-CoR versus SMRT (Figure 7). Similarly, exchanging larger regions of the ligand binding domains of v- and c-Erb A (e.g. CVCC versus VCVV, or CVVC versus VCCV) also failed to exchange their preference for N-CoR versus SMRT (Figure 7). However, reintroducing the C-terminal helix 12 sequences of c-Erb A into v-Erb A (VVVC) converted the N-CoR=SMRT binding properties characteristic of v-Erb A into the N-CoR>SMRT binding properties of c-Erb A; the reciprocal elimination of helix 12 from c-Erb A produced a chimera (CCCV) that displayed the N-CoR=SMRT binding properties of the native v-Erb A oncoprotein (Figure 7). We conclude that not only does v-Erb A exhibit an enhanced ability to bind corepressors in the presence of RXR compared to c-Erb A, but also that the loss of helix 12 alters the v-Erb A corepressor specificity, permitting both SMRT and N-CoR to be recruited at near-equal efficiencies by the viral oncoprotein.

Figure 7
figure7

Mapping the enhanced ability of v-Erb A to interact with SMRT to the deletion of helix 12. The ability of c-Erb A, v-Erb A, and our v-/c-Erb A chimeras (Figure 1) to interact with the SMRT and N-CoR corepressors was determined using the EMSA protocol in Figures 5c and 6. Fixed amounts of each Erb A protein, isolated from transfected COS-1 cells, were mixed with increasing amounts of an N-CoR or SMRT construct. The amount of homodimer supershifted by the corepressor was quantified for each Erb A construct. The maximum N-CoR supershift was defined as 100. The mean and standard deviation of three experiments are shown

The amino-acid substitutions in v-Erb A contribute combinatorially to the constitutive repression and dominant-negative properties of this oncoprotein

We next assayed the ability of our wild-type and chimeric Erb A constructs to regulate expression of a DIV-6-thymidine kinase promoter (TK)-luciferase reporter in transfected CV-1 cells (Figure 8). Both the ability of the Erb A chimera to regulate the reporter alone (left panels) and to act as dominant-negative inhibitor when cointroduced with a wild-type c-Erb A (right panels) were assayed. CV-1 cells lack endogenous c-Erb A, and T3 had little or no effect on expression of the reporter in the absence of an ectopically expressed receptor (Figure 8, dashed line). Introduction of c-Erb A repressed the DIV-6-TK-luciferase reporter in the absence of T3, but activated it in the presence of T3 (Figure 8, compare to the basal levels seen with the empty pSG5 vector). As expected, introduction of v-Erb A into these cells repressed the DIV-6-TK-luciferase reporter in both the presence and the absence of T3 and also strongly interfered with T3-mediated reporter gene activation in a dominant-negative fashion when cointroduced with c-Erb A (Figure 8).

Figure 8
figure8

Combinatorial effects of v-Erb A mutations on reporter gene repression and dominant-negative properties. Transient transfections of CV-1 cells were performed to determine the ability of each Erb A construct to repress or activate a DIV-6-TK-luciferase reporter gene when introduced alone (left panels), or to act as a dominant negative when cointroduced with c-Erb A (right panels). The transfected cells were incubated at the concentrations of T3 indicated below the panels and the luciferase activity, relative to an internal β-galactosidase control, was determined. The average and standard deviation of three separate experiments are provided. An empty pSG5 vector was used in the upper left panels as a minus Erb A control. The results for c-Erb A (solid line) and for v-Erb A (dotted line) are replicated in every panel to permit easier comparison with the properties of the different chimeric constructs

The loss of the C-terminal c-Erb A helix 12 domain from v-Erb A has been proposed to be the major structural basis behind the loss of transcriptional activation by the viral oncoprotein (Munoz et al., 1988; Damm et al., 1989; Sap et al., 1989; Zenke et al., 1990; Saatcioglu et al., 1997). Consistent with this proposal, a c-Erb A protein bearing the v-Erb A C-terminal truncation (CCCV) retained the ability to repress reporter gene expression, but was unable to derepress or activate target genes in response to T3 and functioned as a dominant-negative inhibitor when coexpressed with wild-type c-Erb A over a range of T3 concentrations (Figure 8). However, the CCCV chimera was not as potent a dominant-negative inhibitor as was v-Erb A itself (e.g. VVVV) (Figure 8, right panels). Conversely, restoring the C-terminal helix 12 domain of c-Erb A to the v-Erb A protein (VVVC) generated a chimeric protein that derepressed at high T3 concentrations, yet failed to mediate significant reporter gene activation in response to hormone when introduced alone, and retained substantial dominant-negative activity when cointroduced with c-Erb A (Figure 8). Taken together, these results indicate that additional mutations beyond the loss of helix 12 also contribute to the dominant-negative properties of v-Erb A.

We therefore next examined the contributions of the I-box domain. Interestingly a c-Erb A construct (CCVC) containing the v-Erb A I-box activated the DIV-6-TK-luciferase reporter somewhat more strongly than did wild-type c-Erb A (compare CCVC to CCCC, either assayed alone or when cointroduced with c-Erb A) (Figure 8). Reciprocal replacement of the v-Erb A I-box with that of c-Erb A (VVCV) reduced the dominant-negative properties of v-Erb A when cointroduced with c-Erb A, but little or no observable effect on v-Erb A function when assayed alone (Figure 8). We suggest that the I-box sequence in v-Erb A enhances the ability of either v-Erb A or c-Erb A to bind to the DIV-6 element as a homodimer. For c-Erb A, this enhanced homodimer binding results in stronger activation in the presence of T3; for v-Erb A, this stabilization of homodimer binding enhances the ability of v-Erb A to function as a dominant-negative inhibitor of wild-type c-Erb A function.

Swapping of the helix 1–3 sequences between v- and c-Erb A revealed a distinct phenomenon. Replacement of the c-Erb A helix 1–3 region with that of v-Erb A (CVCC) significantly reduced reporter gene activation in response to a range of T3 concentrations, and generated a moderate dominant-negative activity when coexpressed together with wild-type c-Erb A (Figure 8). A combined c-Erb A chimera containing both the helix 1–3 and I-box substitutions of v-Erb A (CVVC) displayed a more severe impairment in transcriptional activation, and a more potent dominant-negative activity, than did the helix 1–3 substitutions alone (Figure 8). The reciprocal replacement of the v-Erb A helix 1–3 region with that of c-Erb A had little effect alone, but further reduced the dominant-negative properties of v-Erb A when cointroduced together with the I-box mutation (Figure 8, compare VCVV and VCCV to VVVV). Therefore, the helix 1–3 amino-acid sequences found in v-Erb A, but not in c-Erb A, further enhance the repressive and dominant-negative properties of the viral oncoprotein.

An amino-acid substitution in helix 3 contributes to the dominant-negative properties of v-Erb A by inhibiting T3 binding and corepressor release

To better understand the molecular basis behind these effects of helices 1–3 on the dominant-negative properties of v-Erb A, we examined the effect of these mutations on the binding and release of corepressor in response to T3, employing our EMSA supershift assay (Figure 9a). The c-Erb A homodimer bound N-CoR in the absence of T3, and released from this corepressor in response to relatively low concentrations of T3 (Figure 9a). A v-Erb A possessing the helix 1–3 sequences of c-Erb A (CVCC) also bound N-CoR in the absence of T3, but required significantly more T3 to release corepressor than did wild-type c-Erb A (Figure 9a). Comparable effects were observed assaying the effects of these substitutions on the T3-mediated release of SMRT corepressor (Figure 9a). Thus, the v-Erb A helix 1–3 substitutions appear to contribute to the repressive actions of the viral oncoprotein by interfering with corepressor release in response to hormone.

Figure 9
figure9

Mapping of the impaired corepressor release and T3 binding properties of v-Erb A to a K231N substitution in helix 3. (a) The release of corepressors from the Erb A constructs by T3 was tested employing an EMSA supershift protocol as in Figure 7. The amount of each Erb A homodimer/DIV-6 DNA complex supershifted by each corepressor construct (either N-CoR (top) or SMRT (bottom); 200 ng each) was determined over a range of T3 concentrations (indicated below the panel). The mean and range of two independent experiments are shown. (b) The ability of c-Erb A or the v-/c-Erb A chimeras to assume a protease-resistant conformation in response to T3 was assayed using an elastase protocol. 35S-labeled Erb A proteins were incubated at the T3 concentrations indicated below the panel and then exposed to elastase and the formation of a protease-resistant protein core was determined by SDS–PAGE and phosphorimager quantitation (input=100%). The mean and standard deviation of three experiments are shown

Given this requirement for higher than normal levels of T3 to release corepressor by constructs bearing the v-Erb A helix 1–3 sequence, we investigated if these substitutions decreased the affinity of the receptor for hormone. Binding of hormone by nuclear receptors causes a conformational change in their hormone binding domain, resulting in a compaction of the protein chain about the hormone ligand and producing a protease-resistant polypeptide core (Leng et al., 1993; Keidel et al., 1994). This hormone-generated protease resistance, when determined over a range of hormone concentrations, is a useful measure of hormone binding avidity (Yoh and Privalsky, 2002; Farboud and Privalsky, 2004). Notably, significantly more T3 was required to confer protease resistance on a c-Erb A construct bearing the v-Erb A helix 1–3 sequence than required for wild-type c-Erb A (Figure 9b, compare CCCC with CVCC and VVVC).

There are four amino-acid differences in helices 1–3 between v- and c-Erb A. Two of these, P189L and P201L, are in an ‘Ω-loop’ that is thought to fold over to help seal the T3 binding cavity of c-Erb A on binding hormone agonist (Wagner et al., 1995). Interestingly, changing the c-Erb A prolines at these positions to leucines (CCCC(P189L) or CCCC(P201L)) or the double mutation (CCCC(P189L,P201L)) had relatively little effect on T3 binding (Figure 9b). Neither did a lysine to arginine substitution at codon 184 (CCCC(K184R)) (Figure 9b). In contrast, a c-Erb A bearing the v-Erb A lysine to asparagine substitution at the end of helix 3 (CCCC(K231N)) required considerably higher levels of T3 for protease protection and for corepressor release than wild-type c-Erb A (Figure 9a and b). We conclude that the lysine in helix 3 is required for high-affinity T3 binding by the c-Erb A protein, and that its substitution by asparagine in v-Erb A renders corepressor release dependent on higher than normal levels of T3 hormone.

Mutations in v-Erb A also counteract the antiproliferative effects of T3 on c-Jun function

In addition to regulating target genes through direct DNA binding, wild-type c-Erb A can also regulate transcription indirectly through combinatorial protein–protein interactions with other transcription factors, such as c-Jun. C-Erb A enhances the transcriptional activity of c-Jun in the absence of T3, and suppresses it in the presence of T3 (Desbois et al., 1991; Sharif and Privalsky, 1992). V-Erb A retains the ability to enhance c-Jun function, but has lost the ability to suppress in response to hormone, and this constitutive c-Jun enhancement contributes to the pro-proliferative actions of v-Erb A in fibroblasts (Desbois et al., 1991; Sharif and Privalsky, 1992). To better define this loss of T3 inhibition of c-Jun by v-Erb A, we tested our panel of v-/c-Erb A chimeras for the ability to regulate c-Jun function using a c-Jun-responsive luciferase reporter gene. As noted previously, wild-type c-Erb A enhanced expression of the c-Jun-responsive reporter in the absence of T3, and this was reversed at intermediate T3 concentrations (Figure 10, ‘CCCC’). V-Erb A induced a comparable enhancement of c-Jun function that was not significantly inhibited by T3 (Figure 10, ‘VVVV’). This failure of v-Erb A to suppress c-Jun activity in response to hormone was the consequence of both the loss of helix 12 (Figure 10, compare the actions of VVVV to VVVC and CCCC to CCCV), and the helix 3 domain substitution (compare the actions of VVVV to VCVV or VCCV, and of CCCC to CVCC or CVVC). The substitutions in the I-box dimerization region, in contrast, appeared to have little effect on this phenotype (Figure 10, compare VVVV to VVCV and CCCC to CCVC). We suggest that the loss of hormone binding affinity associated with these mutations in v-Erb A contribute not only to its constitutive repression phenotype on c-Erb A-responsive genes, but also to its constitutive activation phenotype on c-Jun-responsive genes.

Figure 10
figure10

Ability of v-Erb A, c-Erb A, and chimeras to regulate c-Jun function. The various v-, c-, and chimeric Erb A constructs detailed in Figure 1 were tested for the ability to alter c-Jun function in transient transfections of CV-1 cells. The same overall protocol of Figure 8 was employed, except a c-Jun-responsive luciferase reporter (containing AP-1 binding sites from the collagenase promoter) was substituted for the DIV-6-TK-luciferase reporter, and 10 ng of a pRSV c-Jun expression vector was included in all assays

Discussion

The conversion of v-Erb A into a constitutive repressor reflects the acquisition of multiple genetic changes

V-Erb A functions as a constitutive repressor that can bind to TR target genes and interfere with T3-mediated transcriptional activation in a dominant-negative fashion (Evans, 1989; Beug et al., 1996; Samarut, 1996; Bauer et al., 1998; Thormeyer and Baniahmad, 1999; Urnov et al., 2000; Rietveld et al., 2001). Significantly, dominant-negative c-Erb A mutants do not appear to cause erythroleukemia in humans, but instead manifest as an endocrine disorder, RTH syndrome (Usala, 1991; Refetoff, 1993; DeGroot, 1996; Chaterjee, 1997). One notable distinction between the oncogenic v-Erb A and the non-leukemogenic RTH syndrome alleles is the multiplicity of the genetic lesions involved. In RTH syndrome, each mutant c-Erb A allele typically contains only single amino-acid difference from wild type (Usala, 1991; Refetoff, 1993; DeGroot, 1996; Chaterjee, 1997). The v-Erb A allele, in contrast, has sustained multiple genetic lesions relative to c-Erb A, including an N-terminal fusion of gag sequences, a C-terminal deletion, and 13 internal amino-acid substitutions (Sap et al., 1986). In this manuscript, we have further explored the contributions of these multiple v-Erb A mutations to its dominant-negative, repressive phenotype. We report that these changes in v-Erb A have altered its dimerization properties, its ability to recruit different corepressors, and its ability to bind and respond to T3 hormone. These changes appear to work together to favor the functions of v-Erb A as a transcriptional repressor and as a dominant-negative inhibitor of c-Erb A activity.

Changes in the I-box sequence of v-Erb A stabilize homodimer formation through multiple mechanisms, including inhibition of monomer binding, of heterodimerization with RXR, and of disruption by T3

V-Erb A has been reported to bind to DNA efficiently as a protein homodimer, whereas c-Erb A preferentially binds to DNA as a heterodimer with RXR (Forman et al., 1989; Selmi and Samuels, 1991; Au-Fliegner et al., 1993; Yen et al., 1994; Piedrafita et al., 1995; Wahlstrom et al., 1996; Bauer et al., 1997; Shen and Subauste, 2000; Yoh and Privalsky, 2001; Zubkova and Subauste, 2003). We report here that there are several separable components to this phenomenon. (a) When assayed alone, v-Erb A displays an approximately twofold greater ability to form homodimers on a DIV-6 DNA response element than does c-Erb A. (b) Although homodimers of c-Erb A are dissociated by T3 hormone, the corresponding v-Erb A homodimers are refractile to disruption by T3. (c) V-Erb A forms heterodimers with RXRs relatively inefficiently, and v-Erb A homodimers persist at RXR concentrations sufficient to convert most c-Erb A homodimers into heterodimers. (d) A final interesting feature is that v-Erb A is inefficient at binding to DNA as a protein monomer, whereas c-Erb A binds to many DNA elements as a mixture of receptor monomers and homodimers. Notably, these differences in v-Erb A all favor homodimer formation over alternative modes of DNA recognition. Homodimers of v- and c-Erb A recruit corepressors efficiently, whereas monomers and RXR heterodimers bind corepressor much more poorly (Cohen et al., 2000; Yoh and Privalsky, 2001; Zubkova and Subauste, 2004); these changes in the dimerization properties of v-Erb A therefore favor the ability of the oncoprotein to bind corepressor in the presence of RXRs and/or T3 relative to c-Erb A.

We mapped all four of these alterations in the dimerization properties of v-Erb A to the receptor I-box. This region represents an important site of receptor–receptor contact and mutations in the I-box are known to alter the dimerization properties of many nuclear receptors, including v- and c-Erb A (Forman and Samuels, 1990; Selmi and Samuels, 1991; Au-Fliegner et al., 1993; Glass, 1996; Perlmann et al., 1996; Reginato et al., 1996; Bauer et al., 1997; Shen and Subauste, 2000; Ribeiro et al., 2001). V-Erb A and (avian) c-Erb A differ in their sequence at four amino acids in the I-box region (Sap et al., 1986). We determined that two of these differences, at codons 363 and 370, were primarily responsible for the resistance to RXR heterodimerization by v-Erb A. This is consistent with a prior study comparing v-Erb A to rat c-Erb A (Zubkova and Subauste, 2003); note that a third substitution highlighted by this prior study, at codon 351 in our numbering system, is identical in v-Erb A and its avian c-Erb A progenitor, and is not relevant to our own analysis. Interestingly, introducing only one of our two substitutions into c-Erb A, at codon 370, was sufficient to disrupt monomer binding, suggesting that the protein surfaces involved in monomeric DNA binding overlap, but are distinguishable from those that allow heterodimerization with RXRs. Mapping the basis for the resistance of v-Erb A homodimers to T3 disruption was more complex, given that part of this phenomenon arises from poor T3 binding by v-Erb A (see below). Nonetheless, surveying a series of v- and c-Erb A chimeras that had comparable T3 binding properties demonstrated that homodimers bearing the v-Erb A I-box sequence are more resistant to T3 disruption than are homodimers bearing the c-Erb A I-box.

V-Erb A can heterodimerize with RARs on a restricted set of response elements

C-Erb A can efficiently form heterodimers not only with RXRs, but also with RARs; unlike RXR/c-Erb A heterodimers, these RAR/c-Erb A heterodimers can efficiently recruit corepressors and suppress gene expression (Lee and Privalsky, 2005). In the current study, we demonstrate that v-Erb A shares this ability of c-Erb A to heterodimerize with RARs, but only over a more limited set of response elements relative to c-Erb A. V-Erb A forms RAR heterodimers on DR-4 elements, but only very poorly on DR-5 elements; c-Erb A, in contrast, forms heterodimers with RAR to a significant extent on both elements. Interestingly, the alteration in RAR heterodimerization by v-Erb A mapped to the same I-box substitutions noted above, with the identity of the amino acids at codons 363 and 370 being of particular importance. Therefore, v- and c-Erb A probably use the same or similar interface for dimerization with RAR as for dimerization with RXR. This ability of v-Erb A to heterodimerize with RARs may contribute to its neoplastic functions, at least on certain response elements/target genes (Desbois et al., 1991; Schroeder et al., 1992b; Chen et al., 1993; Chen and Privalsky, 1993; Gandrillon et al., 1994).

Deletion of the C-terminal helix 12 alters the corepressor specificity of v-Erb A, as well as preventing the exchange of corepressor for coactivator in response to hormone

Binding of hormone agonist by most nuclear receptors reorients helix 12 so as to occlude a corepressor docking site and to form a novel, coactivator binding surface (Glass and Rosenfeld, 2000; Privalsky, 2004). The truncation of helix 12 in v-Erb A has been proposed to be the primary molecular defect behind the constitutive repression properties of the viral oncoprotein (Damm et al., 1989; Sap et al., 1989; Zenke et al., 1990). However, our own studies demonstrate that the C-terminal deletion alone does not account for all of the repressive properties of the viral oncoprotein. A c-Erb A protein bearing a C-terminal truncation is a less efficient dominant-negative inhibitor than is the v-Erb A protein, whereas restoring the C-terminus of c-Erb A to v-Erb A fails to restore fully transcriptional activation. These observations suggest that, although helix 12 deletion plays a significant role in aberrant function of v-Erb A, many of the additional mutations that have accumulated in the viral oncoprotein serve to further enhance its functions as a constitutive repressor. The I-box mutations, noted above, represent one of these additional determinants of dominant-negative function, as does the mutation in helix 3 discussed below.

Intriguingly, the loss of helix 12 not only contributes to the constitutive corepressor binding characteristics of v-Erb A, but also alters the corepressor specificity of the viral oncoprotein from that of c-Erb A. Whereas c-Erb A binds N-CoR more strongly than SMRT (Cohen et al., 1998, 2000, 2001; Webb et al., 2000; Makowski et al., 2003; Goodson et al., 2005; Zamir et al., 1997), we found that v-Erb A displayed a near-equal interaction with SMRT as with N-CoR. Restoring the c-Erb A helix 12 to v-Erb A restored the c-Erb A preference for N-CoR. These results suggest that helix 12 not only regulates access to the corepressor and coactivator docking surfaces, but can also play an active role in discriminating among the different types of corepressor recruited to this docking surface. It is possible that the alteration in corepressor selectivity of v-Erb A might favor target gene repression in cell types that express SMRT in preference to N-CoR.

A mutation in the helix 3 domain of v-Erb A impairs T3 binding

V-Erb A is significantly impaired in the ability to bind T3 hormone compared to c-Erb A; part of this loss in hormone affinity is due to truncation of helix 12, whereas another component has been mapped to amino-acid differences between v- and c-Erb A in helices 1–3 of the hormone binding domain (Munoz et al., 1988; Damm et al., 1989; Sap et al., 1989; Zenke et al., 1990). We mapped this second determinant responsible for the reduced T3 affinity of v-Erb A to a single amino-acid substitution, K231N, located at the end of helix 3. This result was unexpected. Examination of the c-Erb A structure has revealed an ‘Ω-loop’ between helices 2 and 3 that has been proposed to cap, and stabilize, T3 binding (Wagner et al., 1995). Two prolines that contribute to defining this Ω-loop in c-Erb A are leucines in v-Erb A (Wagner et al., 1995), and it appeared logical that these P to L substitutions might be the basis behind the reduced T3 affinity of the v-Erb A helix 1–3 chimeras. Nonetheless, the two P to L substitutions had little or no effect on T3 binding in our assays, whereas the helix 3K to N substitution did. Helix 3 is not known to contact the T3 ligand in c-Erb A, and we speculate that the K231N substitution must affect ligand binding by indirect effects on protein conformation. This mutation would be expected to favor repression under limiting T3 concentrations; consistent with this prediction, c-Erb A constructs bearing the K231N substitution retained corepressor in vitro and functioned as dominant-negatives in vivo at intermediate T3 concentrations.

V-Erb A is a product of the accumulation of multiple mutations that can cooperate to enhance the neoplastic phenotype

Overexpression of certain cellular proto-oncogenes, such as c-ras or c-myc, is sufficient to cause oncogenic transformation of cells in culture (e.g. Pincus et al., 1992; Soucek and Evan, 2002). However, full activation of the neoplastic potential of a proto-oncogene typically requires one or more mutations in the coding region, such that an aberrant protein product is produced. In many cases, the aberrant product is a hypermorph, mimicking the actions of the normal cell product but in an enhanced manner. The v-Erb A oncoprotein is, in contrast, an antimorph that interferes with the functions of its normal cellular counterpart, c-Erb A (Evans, 1989; Ghysdael and Beug, 1992; Yen et al., 1994; Beug et al., 1996; Samarut, 1996; Bauer et al., 1998; Thormeyer and Baniahmad, 1999; Urnov et al., 2000; Rietveld et al., 2001). V-Erb A bears multiple mutations relative to c-Erb A, including a C-terminal truncation and internal amino-acid substitutions. The C-terminal truncation clearly contributes to unmasking the oncogenic potential of v-Erb A by disrupting the helix 12 toggle that, in c-Erb A, couples hormone binding to corepressor release and coactivator acquisition; restoration of helix 12 in v-Erb A permits the v-Erb A-mediated block to erythroid transformation to be released in response to T3 (Damm et al., 1989; Glass et al., 1989; Sap et al., 1989; Pain et al., 1990; Forrest et al., 1990; Zenke et al., 1990; Saatcioglu et al., 1997; Privalsky, 2004). The significance of the additional, internal amino-acid substitutions that have accumulated in v-Erb A since its divergence from c-Erb A has been less clear. Are these mutations neutral and reflect only the known genetic instability of retroviruses and cancer cells, or do they contribute to v-Erb A function in the leukemic cell? Supporting a role for these additional mutations in v-Erb A biology, a c-Erb A lacking only helix 12 does not fully mimic the oncogenic phenotype induced by v-Erb A (e.g. Zenke et al., 1990; Disela et al., 1991).

Our own biochemical analyses indicate that many of these mutations serve to enhance transcriptional repression by v-Erb A. Some, such as the asparagine to lysine replacement in helix 3, impair T3 binding, and thereby stabilize corepressor recruitment at intermediate T3 levels. In fact, v-/c-Erb A chimeras possessing the v-Erb A lysine at this position confer increased proliferation and a greater resistance to T3-induced differentiation in erythroid cells than do chimeras that possess the c-Erb A Arg (Zenke et al., 1990; Schroeder et al., 1992a). This same mutation, operating together with the C-terminal helix 12 truncation, also conferred on v-Erb A a hormone-resistant enhancement of c-Jun function that has been previously implicated in augmenting fibroblast proliferation (Desbois et al., 1991; Sharif and Privalsky, 1992). Other v-Erb A mutations, such as the I-box mutations, favor homodimer formation, and therefore enhance the ability to recruit corepressors compared to receptor monomers or heterodimers with RXR. The biology of these I-box mutations appears to be somewhat complex. A c-Erb A-like substitution of one of these amino acids, at codon 363 in v-Erb A, can enhance erythroid transformation in certain assays (e.g. Damm et al., 1987); our data indicate that this substitution reduces homodimerization on certain DNA elements but preserves the lack of monomeric DNA binding seen for wild-type v-Erb A. These changes in v-Erb A dimerization may play a role in target gene recognition. V-Erb A homodimers preferentially bind to DNA response elements, such as DIV-6 elements, that differ from the DR-4 elements preferentially bound by RXR/c-Erb A heterodimers (Lazar et al., 1991; Baniahmad et al., 1992; Forman et al., 1992; Yen et al., 1994; Piedrafita et al., 1995; Wahlstrom et al., 1996; Harbers et al., 1998). We and others have suggested that changes in the DNA recognition domain of v-Erb A have altered its target gene specificity (e.g. Schroeder et al., 1992b; Chen and Privalsky, 1993; Chen et al., 1993; Gandrillon et al., 1994; Judelson and Privalsky, 1996). The changes in the dimerization properties of the viral oncoprotein described here may operate together with these changes in DNA sequence recognition, allowing v-Erb A to repress genes involved in leukemogenesis that are not normal targets of c-Erb A regulation. The I-box substitutions in v-Erb A may contribute differently to the regulation of different target genes and to the manifestation of certain transformation parameters but not others. We suggest that the multiple mutations in v-Erb A relative to c-Erb A have been selected during the evolution of this oncoprotein to enhance its dominant-negative properties and to adapt this protein from its normal function as a hormone receptor to its new role in leukemogenesis. We further suggest that the c-erb A alleles associated with RTH syndrome, which typically lack these additional adaptations, function as simple attenuators of c-Erb A function, and therefore give rise to endocrine malfunction rather than neoplasia.

Materials and methods

Molecular clones

The pSG5, and baculovirus expression vectors containing avian RXRγ, avian c-erb A (thyroid hormone receptor)-α, gag-v-erb A, and human RARα were previously described (Chen et al., 1993; Tzagarakis-Foster and Privalsky, 1998; Hauksdottir and Privalsky, 2001; Yoh and Privalsky, 2001). The pSG5-v-erb A vector was created by use of polymerase chain reaction (PCR) to delete all gag sequences and the first 12 codons of c-erb A from the pSG5-gag-v-erb A construct. The pSG5-gag-c-erb A vector was constructed reciprocally by fusing the gag sequences from pSG5-gag-v-erb A to codon 13 in c-erb A. Chimeras between chicken c-erb A and v-erb A were created by employing BstEII, BssHII, or SacI sites and exchanging equivalent restriction fragments from the corresponding pSG5 clones. Amino-acid substitutions were created by the QuikChange Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA, USA). Recombinant baculovirus clones of the c-erb A and v-erb A chimeras were generated by inserting EcoRI–EcoRI fragments from the corresponding pSG5 clones into the pFASTBAC1 vector using the Bac-to-Bac Baculovirus expression methodology (Invitrogen, Carlsbad, CA, USA). GST-corepressor constructs representing the S1 and S2 domains of SMRT (codons 2077–2517) or the N2 and N3 domains of N-CoR (codons 1681–2218) were generated as described previously (Lee and Privalsky, 2005).

The DIV-6-TK-luciferase reporter plasmid was constructed by excising the DR-4 element in the M-pTK-Luc vector (Hauksdottir and Privalsky, 2001) with XhoI and SalI, and replacing it with the chicken lysozyme F2 element (Baniahmad et al., 1992) (composed of two annealed oligonucleotides, IndexTerm5′-TCGAA TTATT GACCC CAGCT GACCT CAAGT TACG-3′ and IndexTerm5′-TCGAC GTAAC TTGAC CTCAG CTGGG GTCAA TAAT-3′ purchased from MWG Biotech, High Point, NC, USA). The collagenase AP-1 site-luciferase and pRSV-c-Jun constructs were previously described (Sharif and Privalsky, 1992; Starr et al., 1996).

Electrophoretic mobility shift assays

Nuclear receptors were prepared by expression in a recombinant baculovirus/Sf-9 cell system and isolated as nuclear extracts (Chen and Privalsky, 1993). GST-corepressor proteins for the supershift assays were produced in Escherichia coli strain BL21 transformed with the corresponding pGEX-KG vectors and purified by binding to and elution from a glutathione-agarose matrix (Sigma-Aldrich Inc., St Louis, MO, USA) (Guan and Dixon, 1991). The DIV-6 oligonucleotide used in the EMSAs, derived from the F2 cLys element, was IndexTerm5′-TCGAA TTATT GACCC CAGCT GACCT CAAGT TACG-3′.

EMSAs were performed by mixing 4 μl of appropriately diluted nuclear receptor preparation, 32P-radiolabeled oligonucleotide probe (20–30 ng of DNA, approximately 300 000 c.p.m.), and 10 μl of binding buffer (15 mM Tris, pH 7.6, 4.5% glycerol, 20 mg/ml BSA, 200 mM KCl, 3 mM MgCl2, 200 mg/ml poly(dI-dC), 2 mM dithiothreitol) in a total volume of 20 μl. The reaction mixture was incubated at 25°C for 25 min, and the resulting protein/DNA complexes were resolved by nondenaturing gel electrophoresis through 6% polyacrylamide gel (37.5 : 1 acrylamide–bisacrylamide) in 0.5 × Tris-borate-EDTA buffer. Radiolabeled protein/DNA complexes were visualized and quantified by phosphorimager analysis (STORM system; Molecular Dynamics Inc., Sunnyvale, CA, USA). For supershift experiments using corepressors and coactivator, the nuclear receptor preparations were incubated with the appropriate GST fusion protein for 5 min on ice. Hormones, if indicated, were added after the incubation with corepressors or coactivator.

Alternatively, EMSAs were also performed using nuclear receptors synthesized from pSG5 vectors by use of a Quick Coupled in vitro transcription/translation system (Yoh and Privalsky, 2001) or nuclear extracts from COS-1 cells transiently transfected with appropriate mammalian expression vectors. For the latter, COS-1 cells were maintained in Dulbecco's modified Eagle's medium (DMEM; Invitrogen, Carlsbad, CA, USA) supplemented with 10% heat-inactivated fetal bovine serum (HyClone, Logan, UT, USA) in a humidified atmosphere of 5% CO2 at 37°C. For transfections, cells were plated in 100 mm culture dish and transfections were performed using the Effectene protocol as recommended by the manufacturer (Qiagen, Valencia, CA, USA) using a total 5 μg of expression plasmid plus pUC18. Cells were harvested 48 h later by scrapping and were washed in cold phosphate-buffered saline (PBS). The cells were made to swell in hypotonic buffer (10 mM HEPES, pH 7.9, 10 mM KCl, 1 mM EDTA, 0.05% NP-40, 0.5 mM dithiothreitol, 1 × COMPLETE protease inhibitor cocktail) and were lysed using a Dounce homogenizer (15–20 strokes with a tight pestle). The nuclei were collected by centrifugation, resuspended in high-salt buffer (20 mM HEPES, pH 7.9, 420 mM KCl, 1 mM EDTA, 25% glycerol, 0.5 mM dithiothreitol, 1 × COMPLETE protease inhibitor cocktail) and were incubated on ice for 30 min. The nuclear extracts were clarified by centrifugation and were employed in EMSAs after appropriate dilution in dilution buffer (20 mM HEPES, pH 7.9, 300 mM KCl, 1 mM EDTA, 0.5 mM dithiothreitol, 25% glycerol).

Transient transfections

CV-1 cells were maintained as described for COS-1 cells. For transfections, 3.0 × 104 CV-1 cells were plated per well in a 24-well plate and were incubated for 24 h at 37°C. The cells were then washed with PBS and placed in DMEM supplemented with 10% hormone-depleted fetal bovine serum. Transfections were initiated employing the Effectene protocol described above, but using 100 ng of DIV-6-TK-luciferase reporter plasmid, 25 ng of pCH110 (as a β-galactosidase internal transfection control), the receptor expression vector as indicated, and sufficient pUC18 to bring the total DNA concentration to 250 ng (Yoh and Privalsky, 2001). After a 24 h incubation, the transfection medium was replaced with fresh, hormone-stripped medium, and either ethanol carrier or 3,3′,5-triiodo-L-thyronine (T3) was added. The cells were then incubated for an additional 24 h, washed with PBS, harvested, and lysed in 100 μl of Triton lysis buffer (0.2% Triton X-100, 91 mM K2HPO4, 9.2 mM KH2PO4). Luciferase and β-galactosidase activities were measured as previously reported (Yoh and Privalsky, 2001).

Protease resistance/hormone binding assay

35S-radiolabeled c- or v-Erb A proteins, wild-type or chimeric, were synthesized using the TnT in vitro transcription/translation method (Promega). The proteins were incubated with various concentrations of T3, exposed to 0.05 U of elastase (Sigma), and the recovery of a protease-resistant protein core was determined by SDS–PAGE and phosphorimager analysis as previously described for RARs (Farboud and Privalsky, 2004).

References

  1. Au-Fliegner M, Helmer E, Casanova J, Raaka BM and Samuels HH . (1993). Mol. Cell. Biol., 13, 5725–5737.

  2. Baniahmad A, Kohne AC and Renkawitz R . (1992). EMBO J., 11, 1015–1023.

  3. Bauer A, Mikulits W, Lagger G, Stengl G, Brosch G and Beug H . (1998). EMBO J., 17, 4291–4303.

  4. Bauer A, Ulrich E, Andersson M, Beug H and von Lindern M . (1997). Oncogene, 15, 701–715.

  5. Beug H, Bauer A, Dolznig H, von Lindern M, Lobmayer L, Mellitzer G, Steinlein P, Wessely O and Mullner E . (1996). Biochim. Biophys. Acta, 1288, M35–M47.

  6. Chaterjee VKK . (1997). Hormone Res., 48, 43–46.

  7. Chen H, Smit-McBride Z, Lewis S, Sharif M and Privalsky ML . (1993). Mol. Cell. Biol., 13, 2366–2376.

  8. Chen HW and Privalsky ML . (1993). Mol. Cell. Biol., 13, 5970–5980.

  9. Cohen RN, Brzostek S, Kim B, Chorev M, Wondisford FE and Hollenberg AN . (2001). Mol. Endocrinol., 15, 1049–1061.

  10. Cohen RN, Putney A, Wondisford FE and Hollenberg AN . (2000). Mol. Endocrinol., 14, 900–914.

  11. Cohen RN, Wondisford FE and Hollenberg AN . (1998). Mol. Endocrinol., 12, 1567–1581.

  12. Damm K, Beug H, Graf T and Vennstrom B . (1987). EMBO J., 6, 375–382.

  13. Damm K, Thompson CC and Evans RM . (1989). Nature, 339, 593–597.

  14. Darling DS, Carter RL, Yen PM, Welborn JM, Chin WW and Umeda PK . (1993). J. Biol. Chem., 268, 10221–10227.

  15. DeGroot LJ . (1996). Ann. Intern. Med., 125, 623.

  16. Desbois C, Pain B, Guilhot C, Benchaibi M, Ffrench M, Ghysdael J, Madjar JJ and Samarut J . (1991). Oncogene, 6, 2129–2135.

  17. Disela C, Glineur C, Bugge T, Sap J, Stengl G, Dodgson J, Stunnenberg H, Beug H and Zenke M . (1991). Genes Dev., 5, 2033–2047.

  18. Evans RM . (1989). Int. J. Cancer Suppl., 4, 26–28.

  19. Farboud B and Privalsky ML . (2004). Mol. Endocrinol., 18, 2839–2853.

  20. Forman BM, Casanova J, Raaka BM, Ghysdael J and Samuels HH . (1992). Mol. Endocrinol., 6, 429–442.

  21. Forman BM and Samuels HH . (1990). New Biol., 2, 587–594.

  22. Forman BM, Yang CR, Au M, Casanova J, Ghysdael J and Samuels HH . (1989). Mol. Endocrinol., 3, 1610–1626.

  23. Forrest D, Munoz A, Raynoschek C, Vennstrom B and Beug H . (1990). Oncogene, 5, 309–316.

  24. Gandrillon O, Ferrand N, Michaille JJ, Roze L, Zile MH and Samarut J . (1994). Oncogene, 9, 749–758.

  25. Gandrillon O, Jurdic P, Benchaibi M, Xiao JH, Ghysdael J and Samarut J . (1987). Cell, 49, 687–697.

  26. Gandrillon O, Jurdic P, Pain B, Desbois C, Madjar JJ, Moscovici MG, Moscovici C and Samarut J . (1989). Cell, 58, 115–121.

  27. Ghysdael J and Beug H . (1992). Cancer Surv., 14, 169–180.

  28. Glass CK . (1994). Endocr. Rev., 15, 391–407.

  29. Glass CK . (1996). J. Endocrinol., 150, 349–357.

  30. Glass CK, Lipkin SM, Devary OV and Rosenfeld MG . (1989). Cell, 59, 697–708.

  31. Glass CK and Rosenfeld MG . (2000). Genes Dev., 14, 121–141.

  32. Goodson ML, Jonas BA and Privalsky ML . (2005). J. Biol. Chem., 280, 7493–7503.

  33. Graf T and Beug H . (1983). Cell, 34, 7–9.

  34. Guan KL and Dixon JE . (1991). Anal. Biochem., 192, 262–267.

  35. Harbers M, Wahlstrom GM and Vennstrom B . (1998). J. Steroid Biochem. Mol. Biol., 67, 181–191.

  36. Hauksdottir H and Privalsky ML . (2001). Cell Growth Differ., 12, 85–98.

  37. Judelson C and Privalsky ML . (1996). J. Biol. Chem., 271, 10800–10805.

  38. Katz RW and Koenig RJ . (1994). J. Biol. Chem., 269, 18915–18920.

  39. Keidel S, LeMotte P and Apfel C . (1994). Mol. Cell. Biol., 14, 287–298.

  40. Lazar MA, Berrodin TJ and Harding HP . (1991). Mol. Cell. Biol., 11, 5005–5015.

  41. Lee S and Privalsky ML . (2005). Mol. Endocrinol., 19, 863–878.

  42. Leng X, Tsai SY, O'Malley BW and Tsai MJ . (1993). J. Steroid Biochem. Mol. Biol., 46, 643–661.

  43. Makowski A, Brzostek S, Cohen RN and Hollenberg AN . (2003). Mol. Endocrinol., 17, 273–286.

  44. Mangelsdorf DJ, Thummel C, Beato M, Herrlich P, Schütz G, Umesono K, Blumberg B, Kastner P, Mark M, Chambon P and Evans RM . (1995). Cell, 83, 835–839.

  45. Miyamoto T, Suzuki S and DeGroot LJ . (1993). Mol. Endocrinol., 7, 224–231.

  46. Munoz A, Zenke M, Gehring U, Sap J, Beug H and Vennstrom B . (1988). EMBO J., 7, 155–159.

  47. Pain B, Melet F, Jurdic P and Samarut J . (1990). New Biol., 2, 284–294.

  48. Perlmann T, Umesono K, Rangarajan PN, Forman BM and Evans RM . (1996). Mol. Endocrinol., 10, 958–966.

  49. Piedrafita FJ, Bendik I, Ortiz MA and Pfahl M . (1995). Mol. Endocrinol., 9, 563–578.

  50. Pincus MR, Chung D, Dykes DC, Brandt-Rauf P, Weinstein IB, Yamaizumi Z and Nishimura S . (1992). Ann. Clin. Lab. Sci., 22, 323–342.

  51. Privalsky ML . (1992). Biochim. Biophys. Acta, 1114, 51–62.

  52. Privalsky ML . (2004). Annu. Rev. Physiol., 66, 315–360.

  53. Refetoff S . (1993). Clin. Lab. Med., 13, 563–581.

  54. Reginato MJ, Zhang J and Lazar MA . (1996). J. Biol. Chem., 271, 28199–28205.

  55. Ribeiro RC, Apriletti JW, Wagner RL, West BL, Feng W, Huber R, Kushner PJ, Nilsson S, Scanlan T, Fletterick RJ, Schaufele F and Baxter JD . (1998). Recent Prog. Horm. Res., 53, 351–392; discussion 392–354.

  56. Ribeiro RC, Feng W, Wagner RL, Costa CH, Pereira AC, Apriletti JW, Fletterick RJ and Baxter JD . (2001). J. Biol. Chem., 276, 14987–14995.

  57. Ribeiro RC, Kushner PJ, Apriletti JW, West BL and Baxter JD . (1992). Mol. Endocrinol., 6, 1142–1152.

  58. Rietveld LE, Caldenhoven E and Stunnenberg HG . (2001). Oncogene, 20, 3100–3109.

  59. Saatcioglu F, Lopez G, West BL, Zandi E, Feng W, Lu H, Esmaili A, Apriletti JW, Kushner PJ, Baxter JD and Karin M . (1997). Mol. Cell. Biol., 17, 4687–4695.

  60. Samarut J . (1996). Curr. Top. Microbiol. Immunol., 212, 163–173.

  61. Sap J, Munoz A, Damm K, Goldberg Y, Ghysdael J, Leutz A, Beug H and Vennstrom B . (1986). Nature, 324, 635–640.

  62. Sap J, Munoz A, Schmitt J, Stunnenberg H and Vennstrom B . (1989). Nature, 340, 242–244.

  63. Schroeder C, Gibson L and Beug H . (1992a). Oncogene, 7, 203–216.

  64. Schroeder C, Gibson L, Zenke M and Beug H . (1992b). Oncogene, 7, 217–227.

  65. Selmi S and Samuels HH . (1991). J. Biol. Chem., 266, 11589–11593.

  66. Sharif M and Privalsky ML . (1992). Oncogene, 7, 953–960.

  67. Shen Q and Subauste JS . (2000). J. Biol. Chem., 275, 41018–41027.

  68. Soucek L and Evan G . (2002). Cancer Cell, 1, 406–408.

  69. Starr DB, Matsui W, Thomas JR and Yamamoto KR . (1996). Genes Dev., 10, 1271–1283.

  70. Stunnenberg HG, Garcia-Jimenez C and Betz JL . (1999). Biochim. Biophys. Acta, 1423, F15–F33.

  71. Subauste JS and Koenig RJ . (1995). J. Biol. Chem., 270, 7957–7962.

  72. Subauste JS and Koenig RJ . (1998). Mol. Endocrinol., 12, 1380–1392.

  73. Thormeyer D and Baniahmad A . (1999). Int. J. Mol. Med., 4, 351–358.

  74. Tzagarakis-Foster C and Privalsky ML . (1998). J. Biol. Chem., 273, 10926–10932.

  75. Ullrich A, Coussens L, Hayflick JS, Dull TJ, Gray A, Tam AW, Lee J, Yarden Y, Libermann TA, Schlessinger J, Downward J, Mayes ELV, Whittle N, Waterfield D and Seeburg PH . (1984). Nature, 309, 418–425.

  76. Urnov FD, Yee J, Sachs L, Collingwood TN, Bauer A, Beug H, Shi YB and Wolffe AP . (2000). EMBO J., 19, 4074–4090.

  77. Usala SJ . (1991). Thyroid, 1, 361–367.

  78. Vennstrom B, Beug R, Damm K, Engel D, Gehring U, Graf T, Munoz A, Sap J and Zenke M . (1987). Horm. Metab. Res. Suppl., 17, 14–19.

  79. Wagner RL, Apriletti JW, McGrath ME, West BL, Baxter JD and Fletterick RJ . (1995). Nature, 378, 690–697.

  80. Wahlstrom GM, Harbers M and Vennstrom B . (1996). Oncogene, 13, 843–852.

  81. Webb P, Anderson CM, Valentine C, Nguyen P, Marimuthu A, West BL, Baxter JD and Kushner PJ . (2000). Mol. Endocrinol., 14, 1976–1985.

  82. Weinberger C, Thompson CC, Ong ES, Lebo R, Gruol DJ and Evans RM . (1986). Nature, 324, 641–646.

  83. Wondisford FE . (2003). J. Investig. Med., 51, 215–220.

  84. Yen PM, Ikeda M, Brubaker JH, Forgione M, Sugawara A and Chin WW . (1994). J. Biol. Chem., 269, 903–909.

  85. Yen PM, Sugawara A and Chin WW . (1992). J. Biol. Chem., 267, 23248–23252.

  86. Yoh SM and Privalsky ML . (2001). J. Biol. Chem., 276, 16857–16867.

  87. Yoh SM and Privalsky ML . (2002). Methods Mol. Biol., 202, 129–152.

  88. Zamir I, Zhang J and Lazar MA . (1997). Genes Dev., 11, 835–846.

  89. Zenke M, Kahn P, Disela C, Vennstrom B, Leutz A, Keegan K, Hayman MJ, Choi HR, Yew N, Engel JD and Beug H . (1988). Cell, 52, 107–119.

  90. Zenke M, Munoz A, Sap J, Vennstrom B and Beug H . (1990). Cell, 61, 1035–1049.

  91. Zhang J and Lazar MA . (2000). Annu. Rev. Physiol., 62, 439–466.

  92. Zubkova I and Subauste JS . (2002). Biochem. Biophys. Res. Commun., 294, 35–41.

  93. Zubkova I and Subauste JS . (2003). Mol. Cell. Endocrinol., 199, 61–72.

  94. Zubkova I and Subauste JS . (2004). Mol. Biol. Rep., 31, 131–137.

Download references

Acknowledgements

We thank Liming Liu for superb technical assistance and KR Yamamoto for the generous gift of the AP-1-luciferase reporter. This work was supported by Public Health Service/National Institutes of Health award R37-CA53394.

Author information

Affiliations

Authors

Corresponding author

Correspondence to Martin L Privalsky.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Lee, S., Privalsky, M. Multiple mutations contribute to repression by the v-Erb A oncoprotein. Oncogene 24, 6737–6752 (2005). https://doi.org/10.1038/sj.onc.1208826

Download citation

Keywords

  • v-Erb A
  • c-Erb A
  • thyroid hormone receptor
  • T3R
  • dominant negative
  • c-Jun

Further reading

Search