Báo cáo khoa học: Domains of ERRcthat mediate homodimerization and interaction with factors stimulating DNA binding potx

Fusion proteins of the germ cell nuclear factor GCNF/NR6A1 with ERRc showed that the characteristic feature to be stimulated by additional factors can be transferred to a heterologous pr

Domains of ERRc that mediate homodimerization and interaction with factors stimulating DNA binding

Moritz Hentschke, Ute Su¨sens and Uwe Borgmeyer

Zentrum fu¨r Molekulare Neurobiologie Hamburg (ZMNH), Universita¨t Hamburg, Germany

The estrogen receptor-related receptor c (ERRc/ERR3/

NR3B3) is an orphan member of the nuclear receptor

superfamily closely related to the estrogen receptors To

explore the DNA binding characteristics, the protein–DNA

interaction was studied in electrophoretic mobility shift

assays (EMSAs) In vitro translated ERRc binds as a

homodimer to direct repeats (DR) without spacing of the

nuclear receptor half-site 5¢-AGGTCA-3¢ (DR-0), to

exten-ded half-sites, and to the inverted estrogen response element

Using ERRc deletion constructs, binding was found to be

dependent on the presence of sequences in the ligand binding

domain (LBD) A far-Western analysis revealed that ERRc

forms dimers even in the absence of DNA Two elements,

located in the hinge region and in the LBD, respectively, are necessary for DNA-independent dimerization DNA bind-ing of bacterial expressed ERRc requires additional factors present in the serum and in cellular extracts Fusion proteins

of the germ cell nuclear factor (GCNF/NR6A1) with ERRc showed that the characteristic feature to be stimulated by additional factors can be transferred to a heterologous protein The stimulating activity was further characterized and its target sequence narrowed down to a small element in the hinge region

Keywords: orphan nuclear receptor; transcription factor; estrogen receptor-related; DNA binding; dimerization

The nuclear receptors (NR) comprise a family of

transcrip-tional regulators involved in a wide variety of biological

processes, such as embryonic development, differentiation,

and homeostasis This family includes ligand-dependent

transcription factors for steroid hormones, estrogens,

thy-roid hormones, retinoids, vitamin D, and other

hydropho-bic compounds [1] In addition, several members are orphan

receptors for which ligands have yet to be identified [2,3]

Nuclear receptors exhibit a modular structure with

func-tionally separable domains (A/B, C, Dand EF) [4] The

most highly conserved region of these proteins is the

DNA-binding domain (DBD, C-domain), which contains two

zinc-binding modules that fold to form a single structural

domain [5] They confer binding to a core recognition motif,

or a NR half-site, resembling the sequence 5¢-AGGTCA-3¢

Most receptors bind as homodimers or heterodimers

to palindromes or to direct-repeated sequences of the

AGGTCA motif [6] However, a subset of orphan receptors

bind an extended NR half-site with the core sequence

5¢-TCAAGGTCA-3¢as monomers The C-terminal

exten-sion (CTE) of the DBD contributes to the specific

interac-tion by base specific contacts in the minor grove of the

DNA The C-terminal domain (EF) has an intrinsic ligand-binding function, a ligand-dependent transactivation func-tion (AF-2), and a dimerizafunc-tion interface The variable, N-terminal domain (A/B) is important in transcriptional regulation of some nuclear receptors, and a short variable domain (D) with a nuclear localization motif is thought to

be the hinge between C and EF

Based on the evolution of the conserved DBD and of the ligand-binding domain (LBD), the superfamily has been divided into six subfamilies and 26 groups of receptors [7] Subfamily 3 comprises three groups, the estrogen receptors ERa and ERb [8,9], the estrogen receptor-related receptors (ERRs) and one receptor each for the three steroid hormone classes: glucocorticoids, mineralocorticoids, progestin, and androgen [10] ERRa and ERRb were initially isolated because of their homo-logy to ERa [11] Although structurally related, no natural ligand is known for the ERRs Both receptors bind to extended NR half-sites and to classical estrogen receptor response elements (EREs), inverted repeats of the

NR half-site separated by three base pairs [12–14] Both types of sequence element function as response elements

of ERa as well, suggesting a functional relationship between these receptors [15] Putative common target genes of ERs and ERRs, such as lactoferrin, aromatase and osteopontin [15–19], and common coactivators [14] further strengthen the view of a functional interference of these receptors Although monomeric binding of ERRa has been suggested [12,20], homodimer binding was demonstrated by cotranslation of ERRa and truncated ERRa, generating an intermediate band in electrophoretic mobility shift assay (EMSA) [13,15,19]

Transfection studies revealed ERR-dependent activation

of promoters with EREs or extended half-sites Activation

of the reporter genes occurred in the absence of any exogen-ous added ligand Interestingly, studies by Vannacker et al

Correspondence to U Borgmeyer, ZMNH, Universita¨tsklinikum

Eppendorf, Martinistrasse 52, 20246 Hamburg, Germany.

Fax: + 49 40 42803 5101, Tel.: + 49 40 42803 6622,

E-mail: uwe.borgmeyer@zmnh.uni-hamburg.de

Abbreviations: CTE, C-terminal extension; DBD, DNA binding

domain; DR, direct repeat; ERR, estrogen receptor-related receptor;

EMSA, electrophoretic mobility shift assay; ERE, estrogen response

element; GCNF, germ cell nuclear factor; GST, glutathione

S-transferase; LBD, ligand-binding domain, NR, nuclear receptor.

Note: a web site is available from http://www.zmnh.uni-hamburg.de

(Received 25 April 2002, revised 26 June 2002,

accepted 10 July 2002)

show the requirement of a serum factor for transcriptional

activation [13]

By several means, a novel nuclear receptor was isolated

from human and mouse cDNA libraries [21–24] Because

sequence comparisons reveal high homology to ERRa and

ERRb, the receptor was given the systematic name NR3B3,

and the trivial names ERRc and ERR3 ERRc is much

more closely related to ERRb than to ERRa However, the

DBDs of all ERRs are more than 90% conserved In the

adult mouse, ERRc is highly expressed in heart, brain,

kidney and skeletal muscle [25] We have previously

described its spatial pattern of expression during embryonic

development and in the mature mouse brain [26] In the

adult brain, high transcript levels were observed in the

isocortex, the olfactory system, cranial nerve nuclei, and

major parts of the coordination centers, a pattern that is

established in the embryo During development expression

is prominent in the nervous system [27] The gene is

preferentially transcribed in already differentiating areas of

the nervous system establishing many features of the adult

expression pattern This expression pattern suggests

func-tions of the receptor not shared with its two close

homologues Different isoforms have been described in

mouse and human, differing in the length of their

N-terminal domains [24,25,28] Binding to an extended

NR half-site has been performed with in vitro translated

ERRc2 [28] The authors conclude that ERRc2 binds as a

monomer to extended half-sites Hong et al (1999)

dem-onstrated ERRc-dependent activation of reporter genes

controlled by estrogen response elements in the absence of

any added ligand An AF-2 activation domain bound by the

coactivator GRIP1 primarily mediates the transcriptional

activation [24] Recent studies demonstrated binding and

antagonistic function of the synthetic estrogen receptor

modulators 4-hydroxytamoxifen to ERRc [29,30] The

crystal structure of the human ERRc LBDbound to the

SRC-coactivator peptide has been resolved In the crystal,

the LBDadopts a transcriptionally active conformation

suggesting that putative steroidal ligands would function as

antagonist [31]

Here, we describe the binding characteristics of mERRc2

The receptor binds to DR-0, extended half-sites, and to

classical EREs Interestingly, efficient binding depends on

additional factors present in the serum and in cellular

extracts We present a sequence in the hinge region as the

target site of these activities ERRc binds as dimer to

DNA Dimerization depends on sequence elements,

pre-sent in the DBD, in the hinge region and in the LBD The

C-terminal dimerization motifs function independent of

DNA

M A T E R I A L S A N D M E T H O D S

Plasmid constructs

Full-length ERRc2 cDNA was amplified by PCR with Pfu

polymerase (Stratagene) from a mouse embryonic day 15

brain cDNA The forward primer, c2-start (5¢-AAAG

CTTGCCGCCACCATGGATTCGGTAGAACTTTGC

CT-3¢), includes HindIII and NcoI restriction sites, a Kozak

consensus site [32], the translational start codon of ERRc2

(underlined) and additional 20 nucleotides of the coding

sequence The reverse primer, c2-stop (5¢-GGAT

CCTCAGACCTTGGCCTCCAGCATTTC-3¢), includes

a BamHI restriction site, the translational stop codon (underlined) and 21 nucleotides complementary to the coding sequence The product was cloned into the SrfI site

of Script vector (Stratagene) to generate pCMV-ERRc2 The correct integration was verified by sequencing The SalI linearized plasmid pCMV-ERRc2 served as a template to generate epitope-tagged and truncated con-structs of ERRc All products were cloned into the pGEM-T Easy vector for sequence verification To generate

in vitro translation plasmids, the inserts were isolated and cloned into pSPUTK vector (Stratagene) through either NcoI and SalI, or NcoI and BamHI sites Inserts of clones with internal NcoI or BamHI sites were isolated by partial digestion For the N-terminal truncation, DN-ERRc, c2-stop and the forward primer DN (5¢-ACCATGGTAG ATCCCCAGACCAAGTGTGAA-3¢) were used in the amplification It includes an NcoI restriction site, a new translational start codon (underlined) and a 21-nucleotide sequence coding for amino acids 111–117 of ERRc2 (all numbers according to GenBank accession number AF117254) For the C-terminal truncations the start primer c2-VSVG-start (5¢-ACCATGGAGTACACCGACATCG AGATGAACAGGCTGGGCAAGGATTCGGTAGAA CTTTGCCTGCCT-3¢ that includes a translational start codon (underlined), a sequence coding for an epitope of the vesicular stomatitis virus glycoprotein (VSV-G) and the reverse primers: D10 5¢-AGTCGACTCAAAGTTTGT GCATGGGCACTTTGCC-3¢ (ERRc-448), D50 5¢-AGT CGACTCACATGTGCTGGCCAGCCTCGTAATC-3¢ (ERRc-408), D82 5¢-AGTCGACTCAATTAGCAAGAG

(ERRc-331), D173 5¢-AGTCGACTCAATGTTTTGCCCATCCA ATGAT-3¢ (ERRc-285), D240 5¢-AGTCGACTCAGTTC TCAGCATCTATTCTGCGCTT-3¢ (ERRc-218), were used in the amplification, respectively The reverse primers, named according to the extent of the resulting protein truncation, contain SalI restriction sites, translational stop codons (underlined) and 21–24 nucleotides complementary

to the ERRc coding sequence The position of the C-terminal amino acid of proteins derived from the respective products is given in parentheses Fusion proteins GE-1, GE-2, and GE-3 of N-terminal parts of murine germ cell nuclear factor (mGCNF) and C-terminal parts of ERRc were generated by in vitro translation The respective NcoI/SalI- and SalI/BamHI-fragments were generated by PCR and cloned in a double ligation reaction in the pSPUTK vector, digested with NcoI and BamHI

The following oligonucleotides were used:

GCNF-start 5¢-ACCATGGAGCGGGACGAACGGCC ACCTAGC-3¢, c2-stop, G2r 5¢-AGTCGACTTCTTCT TCTGATATCTGGACTGG-3¢(GCNF 1–167), E2f 5¢-AGTCGACAGAATAGATGCTGAGAACAGCCCA-3¢ (ERRc 213–458), G3r 5¢-AGTCGACCAGACTGTAG GACTGAGGGTCCAG-3¢(GCNF 1–271), and E3f 5¢-AGTCGACCATTTGTTGGTGGCTGAACCAGAG-3¢ (ERRc 240–458) The SalI restriction sites are underlined and the respective amino acids encoded by the amplified fragment are given in parentheses

For GE-1, NcoI/AflII- and AflII/BamHI-fragments were generated and cloned into pSPUTK The oligonucleotides GCNF-start, c2-stop, G1r 5¢-ACTTAAGCATGCCCA

TCTGGAGACACTTGAG-3¢ (GCNF 1–140), and E1f

5¢-ACTTAAGGAAGGGGTCCGTCTTGACAGAGTG-3¢

(ERRc 196–458) were used for the amplification The AflII

restriction sites are underlined A schematic view of the

constructs is given in Fig 4A

Generation of antibodies

The peptide AcNH2-YDDCSSTIVEDPQTK-CONH2

rep-resenting amino acids 101–115 of ERRc2 was synthesized

and cross-linked via the C-terminal lysine to keyhole limpet

hemocyanin Eurogentec performed all procedures,

inclu-ding the immunization of rabbits The serum of the second

boost was used

Bacterial expression of ERRc

The NcoI–HindIII insert of pCMV-ERRc2 containing the

whole coding sequence was cloned into pGEX-KG

expres-sion vector (Amersham Biosciences) The resulting plasmid

pGEX-KG-ERRc2 coding for a fusion protein of

glutathi-one S-transferase (GST) and ERRc was transformed into

Escherichia coliBL21 Cells were grown in 500 mL Lennox

L broth base containing 200 lgÆmL)1ampicillin to an D600

of 0.8–1.0 Subsequently, cells were induced under constant

shaking with 1 mMisopropyl thio-b-D-galctoside for 3 h at

37C The cells were harvested and resuspended in 10 mL

ice cold phosphate-buffered saline (NaCl/Pi), lysed by

sonication and centrifuged at 4C with 20 000 g for

15 min The GST fusion protein was purified from the

supernatant using glutathione–Sepharose 4B beads

accord-ing to the manufacturer’s instructions (Amersham

Biosciences)

Electrophoretic mobility shift assays (EMSAs)

Single-stranded oligonucleotides were purchased

(Meta-bion) and annealed in 10 mM Tris/HCl, pH 7.5, 60 mM

NaCl and stored at)20 C Double-stranded

oligonucleo-tides had 5¢ overhangs of four nucleooligonucleo-tides on both strands

For EMSAs, double-stranded oligonucleotides were labeled

using Klenow polymerase (Roche) with [a-32P]dATP

(Amersham Biosciences) and unincorporated nucleotides

were removed by gel filtration on Sephadex G25 spin

columns (Roche) Labeled oligonucleotides were stored at

4C in 10 mM Tris/HCl, pH 7.5, 1 mM EDTA, 60 mM

NaCl

In vitro translation was performed using the

SP6-polymerase TNT Reticulocyte Lysate System (Promega)

according to the manufacturer’s instructions and stored at

)70 C Binding reactions were performed in a total volume

of 12 lL consisting of 20 mMHepes pH 7.4, 80 mMNaCl,

20 mMKCl, 2 mMdithiothreitol, 1 lg Cot-1 DNA and, if

not stated otherwise, 1 lL of reticulocyte lysate or cellular

extract Complete Protease Inhibitor was added according

to the manufacturer specifications (Roche) Binding

reac-tions were incubated for 30 min followed by the addition of

2 lL of the labeled oligonucleotides and incubated further

for 30 min at room temperature For the supershift and for

the analysis of the serum activity, 2 lL of serum diluted in

NaCl/Pi was added before loading and incubated for an

additional 30 min Complexes were resolved by

nondena-turing PAGE in 0.5· Tris/borate/EDTA (45 m Tris base,

45 mMboric acid, 1 mMEDTA) at 4C at 20 VÆcm)1for

4 h The gels were dried, analyzed with the Fujix BAS 2000 bioimaging system by theTINATMsoftware (Raytest) and exposed to BioMax MR film (Kodak)

Oligonucleotides used were as follows: SIS 5¢-ctaca gaAGGTCAAGGTCAaatgaag-3¢; LFRE 5¢-gttgcaCCT TCAAGGTCAtctgaac-3¢; D R-0 5¢-agcttcAGGTCAAGG TCAgagagct-3¢; D R-0 A 5¢-agcttcACCTCAAGGTCAga gagct-3¢; ERE 5¢-gttcAGGTCActgTGACCTgacctg-3¢ Sequences corresponding to half-sites are capitalized The sequence of one strand is shown after the fill-in reaction Serum treatment

The serum was stored at)20 C Aliquots were incubated for 20 min at 22C, 65 C, 70 C, 75 C, 80 C, and 95 C, respectively Samples were centrifuged for 10 min at

13 000 g and the supernatant was used in EMSA

Treatment with 4 volumes of organic solvents was for

20 min at room temperature Samples were centrifuged for

10 min at 13 000 g The supernatant of the precipitation with ethanol, methanol, isopropanol, and acetone was dried

in a speed-vac concentrator and suspended in 0.5 volumes NaCl/Pi The precipitates were dried at room temperature and resuspended in 1 volume of NaCl/Pi The organic phase

of the extraction with ethanol and with chloroform were dried and resuspended in 0.5 volumes of NaCl/Pi Charcoal treatment was overnight

Cell lysates Cells were grown to approximately 80% confluence on

92 mm tissue culture dishes, washed twice with NaCl/Pi, and harvested in 1.5 mL NaCl/Piby gently scraping with a rubber policeman Cells were centrifuged with 300 g and the pellet was resuspended in lysis buffer (3 lLÆmg)1, 20 mM Tris/HCl, 100 mMNaCl) Cells were lysed by freeze–thaw, centrifuged (16 000 g) and the supernatant was stored at )80 C

Far-Western based protein–protein interaction For the far-Western overlay binding assay 3 lL of reticulocyte lysate programmed to synthesize the indicated proteins was subjected to SDS/PAGE using 10% acryl-amide and transferred by semidry electroblotting to poly(vinylidene difluoride) (PVDF) membranes Further incubations were carried out on an orbital shaker The proteins were partially renatured by first incubating the membrane in 6M guanidine/HCl, which was stepwise diluted in buffer A (25 mMHepes, pH 7.5, 25 mMNaCl,

5 mM MgCl2, 1 mM dithiothreitol) to 0.187M After renaturation, the membrane was incubated at room temperature for at least 2 h in buffer A with 0.05% NP40 and 5% milk powder The membranes were then overlaid overnight at 4C with ERRc, synthesized by

in vitro translation in the presence of [35S]methionine (>1000 CiÆmmol)1; Amersham Biosciences) and diluted

1 : 400 in buffer B (20 mM Hepes, pH 7.5, 75 mM KCl, 0.1 mM EDTA, 2.5 mM MgCl2, 1 mM dithiothreitol, 1% milk powder, 0.05% NP40) The membranes were then washed three times in buffer B, each wash lasting at least

10 min Signals were detected with a Fujix BAS 2000

bioimaging analyzer and autoradiographed with Kodak

BioMax MR film

R E S U L T S

Increased DNA binding ofin vitro generated ERRc

in the presence of serum

In order to analyze the DNA interaction of ERRc, the

full-length cDNA coding for ERRc2 was cloned into an in vitro

translation vector and a rabbit antiserum was generated

The antiserum, aERR, was directed against the peptide

AcNH2-YDDCSSTIVEDPQTK-CONH2, encoded by the

exon that also codes for the amino acids of the first

zinc-finger Western blot analysis revealed that the antiserum

recognizes in vitro expressed ERRc (not shown) The

DNA-binding specificity was determined by incubation of

in vitro translated ERRc and incubated with the GCNF

response element SIS [33], and with the ERRa response

element LFRE [16], both sharing the core sequence

5¢-TCAAGGTCA-3¢, followed by an electrophoretic

mobility shift analysis (EMSA) (Fig 1A,B) A weak

complex was observed on both elements in the absence

of the antiserum Although the intensity of this complex varied slightly in the presence of serum, the most remark-able difference is a tremendous increase of two new protein–DNA complexes in the presence of serum Appar-ently, these novel bands are ERRc–DNA complexes bound by one and two antibodies, respectively The experiment offers three major conclusions Firstly, two elements, the DR-0 element of the human bPDGF promoter (SIS) and an extended half-site of the lactoferrin promoter (LFRE) are bound by ERRc These elements have previously been shown to be binding sites for GCNF, and for both, ERRa and GCNF, respectively [16,33,34] Secondly, the antiserum recognizes the native protein when

it is bound to DNA Thirdly, the DNA binding activity is promoted by the presence of aERR To distinguish between the effect of specific ERRc-antibodies and an undefined function of the serum, binding was performed in the presence or absence of the preimmune serum Again,

an increase of binding was observed; however, as expected, this was mainly due to an increase of the faint complex present in the absence of serum (Fig 1C) In the presence

Fig 1 Binding of ERRc is modulated by the presence of serum EMSA of in vitro translated ERRc with SIS, a D R-0 element and with LFRE an extended half-site (A, B) Supershift of ERRc-SIS (A) and ERRc-LFRE (B) complexes by increasing amounts of antiserum a-ERR Constant amounts of the binding element and of in vitro generated ERRc were subjected to electrophoresis in the absence (lanes 1), and in the presence of 0.004 lL (lanes 2), 0.008 lL (lanes 3), 0.016 lL (lanes 4), 0.03 lL (lanes 5), 0.06 lL (lanes 6), 0.13 lL (lanes 7), 0.25 lL (lanes 8), 0.5 lL (lanes 9), 1 lL (lanes 10), and 2 lL (lanes 11) of a-ERR (C) Increasing amounts of the preimmune serum (PIS) result in an increase of the ERRc-SIS complex The binding reaction was subjected to electrophoresis in the absence (lane 1), and in the presence of ERRc (lanes 2–12) with increasing amounts of PIS [0.004 lL (lane 3) to 2 lL (lane 12)] (D) Binding was performed in the absence (lanes 1–7), and in the presence of 1 lL PIS (lanes 8–12) with increasing amounts of ERRc (0.06 lL in lanes 1 and 7; 0.13 lL in lanes 2 and 8; 0,25 lL in lanes 3 and 9; 0.5 lL in lanes 4 and 10; 1 lL in lanes 5 and 11; 2 lL in lanes 6 and 12) The ERRc–DNA complexes are marked by an arrow, the ERRc–DNA complexes bound

by a-ERR are indicated by arrowheads SIS and LFRE indicate free DNA.

of constant amounts of serum, less in vitro translated

ERRc was necessary for DNA binding (Fig 1D) Taken

together, these experiments reveal that although a specific

protein–DNA complex of in vitro generated ERRc is

formed in the absence of serum, lower

ERRc-concentra-tions are needed in the presence of serum

Increased binding activity in the presence of serum

is heat sensitive

Having identified serum as a stimulating factor, we next

thought to elucidate the nature of this activity To initiate

the characterization of the stimulating serum effect, its

sensitivity against heat was tested Rabbit serum was treated

for 20 min at various temperatures, centrifuged, and the

supernatant was analyzed by EMSA (Fig 2A) The

stimu-lating effect, still present at a temperature of 75C, was

absent after incubation at 80C Precipitation of the

proteins was not observed at 70C, some precipitation

occurred at 75C, and massive precipitation was found at

higher temperatures Hence, the stimulating factor in the

serum is either heat-sensitive, e.g a protein, or associated

with the precipitate

Characterization of the stimulating activity

For further characterization of the stimulating factor, the

serum was subjected to various treatments Whereas a size

exclusion assay with a Bio-Gel P30 spin column of an exclusion limit of about 40 000 Da demonstrated that the activity was in the fraction of the large molecules, a microdialysis with a nitrocellulose membrane with a pore size of 0.025 lm did not diminish the effect (not shown) After precipitation with ethanol, methanol, isopropanol or acetone, the activity was detected in the precipitate (Fig 2B) Because the effect might be due to a small molecule tightly associated with a protein, the serum was subjected to several extraction methods Extraction using ethanol, chloroform,

or ether could not separate the activity from the hydrophilic phase In addition, the activation factor did not quantita-tively interact with charcoal (Fig 2B) Hence, the factor in the serum may be a protein, e.g serum albumin, stabilizing a conformation with a higher DNA affinity, or a small molecule tightly associated with a protein

To distinguish an indirect mechanism mediated by constituents of the reticulocyte lysate from a direct effect

on ERRc, binding of bacterial expressed GST–ERRc fusion protein was investigated No binding of the affinity purified fusion protein was detected in the absence of serum Again, addition of serum greatly enhanced binding

to DR-0, thereby excluding indirect mechanisms (Fig 3A) In addition, expression in E coli allowed analyzing of a possible direct effect of the reticulocyte lysate on DNA binding Indeed, the addition of lysate, programmed to synthesize the unrelated protein luciferase, stimulated binding of ERRc (Fig 3A) Furthermore,

Fig 2 Characterization of the activating func-tion of the serum (A) Binding of ERRc to SIS was analyzed in the presence (lanes 1–10) and in the absence of rabbit serum (lane 11) Prior to binding, the serum was subjected to increasing temperatures as indicated (B) Binding of ERRc to SIS in the absence (lane 1) and in the presence of bovine serum (lane 2–18) Serum was not treated (lanes 2, 18), or precipitated with ethanol (lanes 3, 4), meth-anol (lanes 5, 6), isopropmeth-anol (lanes 7, 8) or aceton (lanes 9, 10), as indicated The preci-pitates (lanes 3, 5, 7, 9) and the respective supernatants (lanes 4, 6, 8, 10) were tested After organic extraction, with ethylacetate (lanes 11, 12), chloroform (lanes 13, 14), and diethylether (lanes 15, 16), the hydrophilic (lanes 11, 13, 15) and the organic phase (lanes 12, 14, 16) were analyzed In lane 17 the binding reaction was supplemented with charcoal-treated serum The ERRc–DNA complexes are marked by arrows, f indicates free DNA.

bovine serum albumin and highly purified human serum

albumin, both activated DNA binding of the bacterial

expressed protein (Fig 3B) However, ovalbumin does

not enhance DNA binding, suggesting that the effect is

not a pure function of the protein concentration (data not

shown)

The detection of factors stimulating the DNA binding

activity in reticulocyte lysate suggests that cellular

constit-uents may have a stimulating activity To address this issue,

we tested whole cell extract derived from CV-1 cells, NIH/

3T3 cells and P19 cells, respectively (Fig 3C) All extracts

stimulated the binding activity GST–ERRc fusion proteins

suggesting a physiological function of the enhancement of

DNA binding

A sequence element in the hinge region is essential

for the stimulating effect

As demonstrated above, limiting factors greatly enhanced

the formation of ERRc–DNA complexes As a

conse-quence, it should be possible to map elements in the receptor

as targets of these factors

To this end, the truncated protein ERRc-218 coding for

the first 218 amino acids, and two fusion construct of the

N-terminal part of GCNF with the C-terminal part of

ERRc, GE-2 and GE-3 (Fig 4A), were tested GE-2 covers

amino acids 1–167 of GCNF and 213–458 of ERRc,

whereas GE-3 covers amino acids 1–271 of GCNF and

240–458 of ERRc A SalI restriction site at the fusion codes

for two additional amino acids, valine and aspartic acid In

both fusion proteins DNA binding is mediated by GCNF

The truncated in vitro translated protein ERRc-218, lacking

amino acids forming the LBDand the C-terminal part of

the hinge region, still binds to DNA, and the addition of

serum results in increased binding (Fig 4B) Consequently,

the LBDis not necessary for the activating function of the

serum Hence, an allosteric conformational switch by

binding of a steroid ligand bound to a carrier in the serum

is very unlikely At least some of the target sequences must

be located either in the A/B domain, the DBD, or the hinge region As expected, binding of GCNF is not increased by serum addition The same is true for GE-3 in which most of the LBDof GCNF is replaced by that of ERRc, further demonstrating that the LBDis not involved in the activation We conclude that the LBDis neither essential for the activation, nor does its fusion to a homologous protein result in a transfer of the activity However, the binding of GE-2, in which the C-terminal part of the hinge and the LBDof GCNF are replaced by the corresponding domains of ERRc, is greatly stimulated by serum (Fig 4B) Accordingly, the ERRc-hinge region confers the activation ERRc-218 and GE-2, both affected by the addition of serum, have a sequence overlap of six amino acids These results suggest a central role of the common sequence,

NH2-RIDAEN-COOH, in the stimulating effect Three of these amino acids are charged, further implying that the stimulating effect is not induced by lipophilic ligand receptor interaction A comparison with the homologous receptors ERRa, ERRb, ERa, and ERb and a data base search in the nonredundant protein data base revealed that the

RIDAEN element is unique for ERRc

ERRc binds as a homodimer to DNA Dimerization is essential for the function of most nuclear receptors Previously, ERRc was reported to bind as a monomer to DNA [28] However, a recent report assumes that ERRc binds also as a dimer to DNA [35] For ERRa and ERRb, monomeric and dimeric binding has been demonstrated [3] The repeat nature of the binding site, and the fact that ERRc–DNA complexes have a mobility very similar to a GCNF homodimer and to a PPARc/RXRc heterodimer (not shown), suggest that ERRc binds to DNA preferentially as a dimer

To address the dimerization properties of ERRc in solution, we constructed the mutant DN-ERRc in which the entire N-terminal domain of ERRc is deleted (Fig 4A) This mutant still binds to DR-0 and forms protein–DNA

Fig 3 Binding of a bacterial expressed GST–

ERRc fusion protein depends on factors present

in serum and in cellular extracts The purified

GST–ERRc fusion protein was tested for

binding to the SIS element Only the upper

half of the autoradiograph is shown (A)

Binding without additional factors, and in the

presence of fetal bovine serum (FCS) and

reticulocyte lysate (RL), as indicated (B)

Binding in the presence of bovine serum

albumin (BSA) and human serum albumin

(HSA), respectively (C) Whole cell extracts of

the kidney derived cell line CV-1, of NIH/3T3

fibroblasts, and of the embryonal carcinoma

cell line P19, were incubated with the SIS

ele-ment in the absence and in the presence of the

GST–ERRc fusion protein, as indicated.

complexes with a mobility higher than that of the wild-type

receptor (Fig 5, compare lanes 1 and 2) The mixing of

ERRc with DN-ERRc results in the formation of

DNA-bound ERRc/DN-ERRc heterodimers, which migrate with

a mobility intermediate between those of the homodimeric

ERRc and DN-ERRc complexes (Fig 5) Dimeriziation is detected on DR-0, and also in weaker complexes formed on the extended half-site DR-0 A, and on ERE, an inverted repeat with a spacing of three base pairs, the classical estrogen response element

DNA binding of C-terminal deletion mutants Dimerization motifs are commonly found in the DBDs including the CTE and in the C-terminus of nuclear receptors [36–38] To identify sequence elements in the LBDthat contribute to DNA binding, a series of C-terminal truncated ERRc polypeptides comprising the first 218–448 amino acids of the 458 amino acid full-length protein were generated (Fig 4A) An SDS/PAGE analysis of the proteins generated by in vitro translation in the presence

of [35S]methionine demonstrated their synthesis in similar amounts (not shown) Binding to DR-0 was tested in comparison to the full-length protein, to DN-ERRc, to GE-2, and to GE-3 ERRc-448, lacking the C-terminal nine amino acids, the sequence harboring the H12 a helical region still binds to DNA [31] (Fig 6) Although the protein migrates faster during denaturing gel electrophoresis, the protein–DNA complex has a slightly reduced mobility when compared to the full-length ERRc This may be either due

to a conformational change or to differences in the surface charge distribution of the truncated receptor Further truncation of additional 41 amino acids in mutant ERRc-408 gives rise to a much weaker complex indicating

a reduced DNA affinity that may be the result of an impaired folding or a reduced dimerization function Again, the complexes migrate slightly slower when compared to

Fig 5 ERRc binds as a homodimer to DNA Binding of the full-length

ERRc (lanes 1, 4, 7), the N-terminal truncated protein DN-ERRc

(lanes 2, 5, 8) and a mixture of both proteins (lanes 3, 6, 9) were

subjected to an EMSA with the indicated D NA elements (SIS: a D R-0

element of the bPDGF promoter; DR-0A: an extended half-site; ERE,

an estrogen response element of the vitellogenin promoter) The

position of the ERRc–DNA complexes (double arrow) of the

DN-ERRc–DNA complexes (arrow), and of the heterodimer

(arrowhead) are indicated.

Fig 4 Localization of the ERRc domain involved in the enhanced DNA binding (A) Schematic view of truncated ERRc and fusion proteins of GCNF and ERRc used in this study The position of the N-terminal A/B-domain, the DBD (C-domain), the hinge region (D- D omain), and the LBD(EF domain) are indicated For the truncated protein, the first and last amino acid is indicated with respect to the full-length protein In the chimeras GE-1, GE-2, and GE-3 the numbering refers to the amino acids of GCNF and ERRc, respectively (B) Binding of the truncated protein ERRc-218, of the fusion proteins GE-2 and GE-3, and of GCNF to SIS in the presence and in the absence of a rabbit serum (RS), as indicated The positions of the complexes of SIS with ERRc-218 (double arrow), GE-2 (open arrowhead), GE-3 (filled arrowhead), and GCNF (arrow) are indicated.

full-length proteins All three proteins form an additional

weak and faster migrating complex, apparently a monomer

The intensity of this band is not affected by the truncations,

indicating that reduced binding of the dimer is due to

inefficient dimerization The truncated ERRc-408 lacks the

a helices 10–12 Helices 9 and 10 have been implicated in

dimerization of various nuclear receptors A crystal

struc-ture of the RXRa LBDrevealed a dimer interface formed

mainly by helix 10 and, to a lesser extent, helix 9 and the

loop between helix 7 and helix 8 [39] A weak dimer is

formed by ERRc-376, a truncated protein lacking helix 9

Further truncation of helices 6–8 in ERRc-331, and helices

4–8 in ERRc-285 results in much smaller, weak complexes

In contrast, the smallest truncated protein, ERRc-218,

lacking the whole LBDand part of the hinge region shows a

robust complex (Fig 6) This protein consists of the DBD

and includes 25 amino acids of the Ddomain and therefore

the CTE Several conclusions can be drawn from the

binding analysis According to the conserved a helical

sandwich structure, as determined for ERa [40,41] and

more recently for ERRc [31], a dimerization function can be

assigned to a region containing a helices 10 and 11 The

increase of binding by the additional deletion of the a helices 1–3 and of the C-terminal part of the hinge region (compare ERRc-218 and ERRc-285) suggests that these elements offer some steric hindrance for dimerization or DNA binding An additional dimerization function can be assigned to the N-terminal 218 amino acids In analogy to other nuclear receptors, this function is proposed to be located in the DBD including the CTE [36,37] Taken together, these results imply homophilic interaction of ERRc on various NR response elements mediated by at least two dimerization modules

Dimerization function of ERRc Two nuclear receptor dimerization interfaces have been defined, one within the DBD and one within the LBD A two step-model for dimeric binding of RXR heterodimers has been proposed First, heterodimers would be formed through their dimerization interfaces contained in the LBD, and in a second step the DBDs would be able to bind with high affinity to DNA [42] In order to analyze to what extent dimerization of the truncated proteins is impaired in the absence of DNA, C-terminal deletion proteins ERRc-448 to ERRc-218 were separated by SDS/PAGE and subjected to

a far-Western analysis, a method based on direct protein-interaction Only ERRc-448 was identified as binding partner of the full-length protein labeled by incorporation

of [35S]methionine (Fig 7A) Further truncation of 41 amino acids abolishes the homophilic interaction This result is in agreement with the DNA-binding analysis: highly reduced binding of the truncated proteins is most likely the result of the deletion of a dimerization function, which can be located to the a helical region 10–11 On the other hand, the smallest deletion mutant tested, ERRc-218, binds to DNA but does not dimerize with the full-length protein under far-Western conditions (Fig 7A) The dimer-ization of this mutant may be dependent on the presence of the DNA-response element As in solution, the full-length protein binds to DN-ERRc, the N-terminal truncated protein The chimeric protein GE-1 (GCNF1-140/ ERRc196–458) containing the 263 C-terminal amino acids

of ERRc is efficiently bound by labeled ERRc (Fig 7B) This interaction indicates that the dimerization motifs in the C-terminus function independently of the motifs in the DBD Decreasing amounts of ERRc-specific residues in GE-2 and GE-3 are accompanied by reduced and abolished interaction, respectively Hence, additional amino acids in the Dand helix 1 region are important for dimerization For DNA-independent dimerization, both elements, one located between 219 and 239 and the second between amino acids

409 and 448 are necessary

D I S C U S S I O N

In this study, we show that ERRc binds to a DR-0 element, but also to extended half-sites ERRs have a conserved DBD Therefore, it is not surprising that they all bind

to elements with the extended half-site element TCAAGGTCA In addition, a weak complex was detected

on ERE, an inverted response element There are conflicting results in the literature as to whether ERRs bind as monomers or dimers Our results show that ERRc binds preferentially as a dimer to all of these elements This has

Fig 6 Binding of truncated and chimeric receptors to a DR-0 element.

The full-length protein (lane 1), C-terminal deletions (lanes 2–7), the

N-terminal deletion DN-ERRc (lane 8), and fusion proteins GE-2 and

GE-3 (lanes 9 and 10) were tested in EMSA Equal amounts of primed

reticulocyte lysate and labeled SIS-binding site were used in each lane.

The positions of the DNA complexes with ERRc (filled arrow),

ERRc-218 (open arrow), DN-ERRc (double arrow), GE-2 (open

arrowhead) and GE-3 (closed arrowhead)are indicated The position

ERRc monomers bound to DNA in lanes 1–3 is indicated by the

bracket.

been demonstrated by mixing of an N-terminal truncated

protein with the full-length protein An intermediate band in

an EMSA is confirmation of dimerization Additionally,

protein–DNA complexes of the orphan receptor GCNF

that binds to DR-0 as a dimer show a very similar migration

[43] An important future question is the identification of

functional binding sites and the analysis of a possible

cross-talk of receptors with a similar binding site specificity To

further characterize functional domains of the protein,

binding of C-terminal truncated proteins to SIS, a DR-0

element of the bPDGF promoter was analyzed by EMSA

Surprisingly, binding of some of the truncated protein gave

rise to a slower migrating complex This phenomenon has

also been observed for truncated GCNF bound to DR-0

[44] Because the analysis was performed under

nondena-turing conditions, a reasonable explanation is a less compact

structure of the truncated protein, or differences in the

surface charge distribution of the truncated receptor A

faster migrating weak complex that appears to be a

monomer shows a similar behavior (Fig 6, lanes 1–3)

However, in contrast to the dimer, the intensity of this band

is not affected by the deletion, suggesting a reduced

dimerization function The truncated protein ERRc-218,

which contains the DBD including its C-terminal extension

binds to DNA, suggesting that ERRc has a DNA–

dependent dimerization interface The weak complexes

formed by 408, 376, 331, and

ERRc-285 further strengthens the assumption that these trunca-tions have a distorted DNA-independent dimerization

As an independent approach we subjected various deletion mutants and fusion proteins to a direct analysis

of protein–protein interaction by far-Western blots The interaction of the mutated proteins with the radioactive full-length ERRc supported the results of the EMSA It is important to note that deletion of the N-terminal domain does not influence the dimerization properties of the receptor However, the C-terminal LBDis important for homophile interactions The deletion ERRc-408 lacking the helix 10/11 does not dimerize The crystal structures of the LBDs of hRXRa, hRARc, hTRa, and hERa show that this dimerization is mediated mainly by helices 9 and 10 [39,40,45–47] A recent analysis of the ERRc LBDshows that it adopts a canonical three-layered a helical sandwich structure and superimposes well with the hER LBD[31]

In addition, the analysis allowed the study of the interaction of a fusion protein with a heterologous DBD Although ERRc does not bind to GCNF, the fusion protein GE-1, composed of the N-terminal GCNF portion with the DBD and the C-terminal portion of ERRc is bound by full-length ERRc Therefore, the dimerization function in the C-terminus works independently of the dimerization func-tion in the DBD In addifunc-tion, the analysis shows that both, the C-terminal (Fig 7A, compare lanes 2 and 3), and the N-terminal truncation of the C-terminus (Fig 7B, compare

Fig 7 A Far-Western analysis deciphers the DNA binding-independent dimerization function of ERRc (A) C-terminal deletion mutants of ERRc were separated by SDS/PAGE, blotted to a membrane filter, and probed for interaction with35S-radiolabeled ERRc (lanes 2–7) The probe, separated on the same gel is shown in lane 1 The arrow indicates the position of ERRc-448 (B) The full-length protein (lane 1), the proteins GE-1, GE-2, GE-3 (lanes 2–4), DN-ERRc (lane 5), and GCNF as a negative control (lane 6) were separated by SDS/PAGE and subjected to a Far-Western analysis as described in A The arrow indicates the position of ERRc (C) Schematic representation of ERRc The position of the dimerization motifs is indicated by black bars, the numbers refer to the amino acids important for dimerization.

lanes 3 and 4) abolish dimerization Therefore, at least two

dimerization interfaces in the C-terminus exist, one located

between amino acid 213 and 239, and the second between

amino acids 409 and 448 The C-terminal interface includes

helix 10, whose function in dimerization is well established

for several receptors For the further N-terminal located

interface, the presence of amino acids in the hinge region up

to helix 1 in the LBDis important: the CTE is not essential

Interestingly, Tetel et al reported that the minimal fragment

mediating progesterone receptor homodimerization was the

hinge-LBDconstruct [48] In addition, GST pull down

experiments reveal the importance of theD-Domain of the

thyroid hormone receptor for homodimerization and

hete-rodimerization with RXR However, in the same

experi-mental design, the EF domain of the RXR formed

heterodimers with the thyroid hormone receptor [49] The

His-tagged ERRc LBDforms dimers in solution [31]

The discrepancy could be due to the fact that in our study

the binding partner is immobilized, the GE-3 starts 11

amino acids further to the C-terminus, or that the GCNF

fusion affects dimerization On the other hand it is possible

that the His-tag influences protein interaction [50]

The serum effect is very surprising because in vivo, ERRc

should never be in a direct contact with the serum However,

it is possible that a serum factor enters the cell

Co-transfec-tion with ERRs and a reporter gene also suggest a funcCo-transfec-tion

of serum in transcription activation [13] Because we

achieved activation of binding by purified serum albumin,

it appears more likely that the endogenous activators differ

from the serum factor Preliminary results in our laboratory

(M Hentschke, unpublished observations) show that at

least two active fractions can be separated by ion exchange

chromatography and by gel filtration chromatography of

crude P19 cell extracts The identification of the active

components in these fractions will be an important

prere-quisite to analyze the mechanism underlying the

phenom-enon A specific effect should be dependent on the presence

of sequence elements present in ERRc but not in GCNF

Therefore, we have focussed on the target protein, ERRc

Indeed, the C-terminal deletions reveal that even the binding

of the smallest protein analyzed is activated by additional

factors However, neither binding of GCNF, nor of the

chimera GE-3 is influenced by additional factors However,

binding of GE-2 with 27 additional amino acids is clearly

stimulated by additional factors Taken together these

experiments reveal that amino acids 1–218, and amino acids

213–458 fused to GCNF can mediate this increase in DNA

binding Although, the importance for efficient binding of

additional factors has been shown for additional nuclear

receptors, to our knowledge this is the first example where a

short sequence with a central function in mediating this

effect has been identified for ERRs

The question arises as to whether there are other

receptors whose binding depends on additional proteins

Indeed, there are several reports about cellular extracts,

necessary for efficient binding of steroid hormone receptors

[51,52] The function of the high-mobility group box

proteins, HMG-1 and HMG-2, members of the nonhistone

chromatin proteins, has been analyzed in more detail They

are recruited to DNA by steroid hormone receptors and

although very abundant, subsequently led to an increase in

transcriptional activity in transient transfection assays

[53–56], but have no effect on binding of several nonsteroid

hormone receptors [54] HMG-1/-2 appear to act by facilitating receptor interaction with target DNA sites [56] The HMG box contacts the DNA in the minor groove introducing a strong bend [57] Therefore, the HMG box proteins have been proposed to substitute for the lack of a minor groove-interacting surface in the DBD of the steroid hormone receptors [54,56] However, they do not result in the supershift of the retarded bands that would be expected

if HMGs were present in the complex A deletion analysis of the androgen receptor indicated that that HMG-1 needs at least part of the CTE and of the hinge region for the stimulation of receptor DNA binding [58] Whether the observed effect on DNA binding of ERRc can be mediated

by HMG box proteins is presently unknown HMG-1 is a very conserved and abundant protein, which interacts with many apparently unrelated proteins [59] The recent iden-tification of SRY, a nuclear HMG box-containing protein

as an interaction partner of the androgen receptor suggests that additional differential expressed HMG box proteins may be identified as interaction partner of nuclear receptors [60] The analysis of the ERRc LBDstructure revealed that the ligand free conformation is the transcriptionally active form suggesting that alternative mechanisms may be important to regulate the activity of this true orphan [31]

A systematic approach will be necessary to identify the most efficient interaction partners and to understand how these additional proteins succeed to increase DNA binding of ERRc and therefore modulate the activity of this orphan receptor

A C K N O W L E D G E M E N T S

We thank Prof Schaller for the support of this work This project was supported by a fellowship to M H through the Graduiertenkolleg 255 and is part of his doctoral thesis Special thanks go to Drs Irm Hermans-Borgmeyer and Sabine Hoffmeister-Ullerich for the fruitful discussion throughout the project, to Simon Hempel for help with the figures and to Cornelia Meyer, Mirja Bernhardt and Anja Nitzsche for assistance during their practical training.

R E F E R E N C E S

1 Mangelsdorf, D.J., Thummel, C., Beato, M., Herrlich, P., Schu¨tz, G., Umesono, K., Blumberg, B., Kastner, P., Mark, M., Chambon, P & Evans, R.M (1995) The nuclear receptor super-family: the second decade Cell 83, 835–839.

2 Enmark, E & Gustafsson, J.A (1996) Orphan nuclear receptors – the first eight years Mol Endocrinol 10, 1293–1307.

3 Gigue`re, V (1999) Orphan nuclear receptors: from gene to func-tion Endocrinol Rev 20, 689–725.

4 Green, S & Chambon, P (1988) Nuclear receptors enhance our understanding of transcription regulation Trends Genet 4, 309– 314.

5 Renaud, J.P & Moras, D (2000) Structural studies on nuclear receptors Cell Mol Life Sci 57, 1748–1769.

6 Khorasanizadeh, S & Rastinejad, F (2001) Nuclear–receptor interactions on DNA-response elements Trends Biochem Sci 26, 384–390.

7 Laudet, V (1997) Evolution of the nuclear receptor superfamily: early diversification from an ancestral orphan receptor J Mol Endocrinol 19, 207–226.

8 Green, S., Walter, P., Kumar, V., Krust, A., Bornert, J.M., Argos,

P & Chambon, P (1986) Human oestrogen receptor cDNA: sequence, expression and homology to v-erb-A Nature 320, 134– 139.