The structure of corepressor Dax-1 bound to its
target nuclear receptor LRH-1
Elena P. Sablina, April Woodsa, Irina N. Krylovab, Peter Hwanga, Holly A. Ingrahamb, and Robert J. Flettericka,1
Departments ofaBiochemistry and Biophysics andbMolecular and Cellular Pharmacology, University of California, San Francisco, CA 94143
Communicated by John D. Baxter, Methodist Hospital Research Institute, Houston, TX, October 8, 2008 (received for review July 1, 2008)
The Dax-1 protein is an enigmatic nuclear receptor that lacks an
expected DNA binding domain, yet functions as a potent corepres-
sor of nuclear receptors. Here we report the structure of Dax-1
bound to one of its targets, liver receptor homolog 1 (LRH-1).
Unexpectedly, Dax-1 binds to LRH-1 using a new module, a repres-
sor helix built from a family conserved sequence motif, PCFXXLP.
Mutations in this repressor helix that are linked with human
endocrine disorders dissociate the complex and attenuate Dax-1
function. The structure of the Dax-1:LRH-1 complex provides the
molecular mechanism for the function of Dax-1 as a potent tran-
Dax-1 ? LRH-1 ? nuclear receptor ? regulation ? structure
the X chromosome gene 1) (1) is well known for its role in human
pathophysiology. Duplication of the DAX-1 gene causes pheno-
typic sex reversal in XY individuals (2), and mutations in DAX-1
are responsible for adrenal hypoplasia congenita, an inherited
disorder of adrenal gland development (3). During embryogen-
esis, Dax-1 functions to direct cell differentiation in testes and
adrenal tissues (1). In adult physiology, Dax-1 acts as a global
repressor of many nuclear receptors, including SF-1, Nur77,
ERR?, ER, AR, PR, and LRH-1 (4–13). Dax-1 also is indis-
pensable to maintaining the pluripotent state of embryonic stem
cells (14, 15).
There is a little information on either the structure or regu-
latory mechanisms of Dax-1. Dax-1 belongs to a unique family
of nuclear receptors (NR0B1) that lack the essential DNA
binding domain. Instead, the human Dax-1 N terminus consists
of three sequence repeats that include the LXXL/ML motif
(‘‘LXXL/ML boxes’’ 1–3) (16). This unique N-terminal exten-
sion is thought to play a role in subcellular distribution and
nuclear localization of Dax-1 (17). No homologues for the
N-terminal region of Dax-1 are known, but its C-terminal
domain is a clear homologue of the nuclear receptor ligand-
binding domain (LBD) (1). To date, no hormone for Dax-1 has
remains under debate (4, 5, 7–9, 12, 18–20). The elucidation of
Dax-1 mechanisms has been frustrated by a lack of high-
resolution structural information. Here we report the first
structure of Dax-1 bound to its physiological target, nuclear
receptor liver receptor homolog 1 (LRH-1; NR5A2).
LRH-1 was first discovered in the liver and intestine, where it
regulates genes controlling bile acid synthesis and cholesterol
homeostasis (21–24). Recently, LRH-1 was found in human
steroidogenic tissues and was shown to activate transcription of
genes encoding steroidogenic enzymes (25). In particular, reg-
ulation of the CYP19A gene encoding aromatase, which converts
androgens to estrogens, gives LRH-1 a pivotal role in estrogen
signaling (25–28). Similar to Dax-1, LRH-1 is indispensable to
maintaining the pluripotent state of embryonic stem cells (29).
Unlike other nuclear receptors that function as homodimers
or heterodimers, LRH-1 binds DNA with high affinity as a
monomer (30, 31). In contrast to hormone-controlled nuclear
receptors, physiological ligands for LRH-1 have not yet been
he orphan nuclear receptor DAX-1/Dax-1 (dosage-sensitive
sex-reversal adrenal hypoplasia congenital critical region on
identified, consistent with the fact that NR5A receptors activate
reporter gene transcription in the absence of exogenously added
ligands (32). Structural studies of LRH-1 (33–37) have revealed
its LBD in the active conformation and suggested phosphatidy-
linositols as potential candidate hormones for this receptor (34);
however, whether these ligands stabilize the LBD or function as
regulating hormones remains to be determined. The hinge
region preceding the LRH-1 LBD provides additional sites for
receptor regulation through posttranslational modification (38).
Recent studies have found that the two targets of our work,
LRH-1 and Dax-1, are coexpressed in the ovary, where they
show that Dax-1 is a key physiological regulator of LRH-1
transcriptional activity and LRH-1-mediated steroidogenesis.
The present work provides the first structural and functional
analysis of a regulatory Dax-1:LRH-1 assembly and suggests a
mechanism for Dax-1 function as a potent transcriptional
Preparation and Characterization of the (Dax-1)2:LRH-1 Heterotrimer.
To evaluate whether Dax-1 can bind to LRH-1 in vitro, we
performed the standard GST pull-down assay using bacterially
expressed and purified GST-LRH-1 LBD fusion protein and in
vitro transcribed and translated35S-labeled full-length Dax-1.
The results of this experiment show that Dax-1 interacted with
LRH-1 LBD in the absence of any added hormones or coregu-
latory proteins. Moreover, under the same conditions, the
observed Dax-1-LRH-1 binding exceeded the analogous inter-
actions with nuclear receptor SF-1, another functional target of
Dax-1 (supporting information (SI) Fig. S1).
Because multiple regions of Dax-1 have been reported to bind
nuclear receptors (4, 5, 7–9, 12), we assessed the binding of five
different fragments of Dax-1 to LRH-1 LBD: its N-terminal
region (aa 1–208), the LBD (aa 205–472), the LBD with
preceding LXXL/ML repeats (aa 138–472 and 70–472), and
full-length Dax-1 (aa 1–472). Of these fragments, only the
putative Dax-1 LBD produced a stable complex with LRH-1
(Fig. S2). Further biochemical analyses of the purified Dax-
1:LRH-1 complex showed that the assembly is a heterotrimer
with a Dax-1:LRH-1 ratio of 2:1 (Fig. S3 A and B). Consistent
with these data, analytical ultracentrifugation revealed the pres-
ence of a single protein species with a molecular mass of ?90
kDa, which agrees with the calculated molecular mass of the
We characterized the binding affinity of Dax-1 LBD for
Author contributions: E.P.S., H.A.I., and R.J.F. designed research; E.P.S., A.W., and I.N.K.
performed research; E.P.S., I.N.K., and P.H. contributed new reagents/analytic tools; E.P.S.,
I.N.K., P.H., H.A.I., and R.J.F. analyzed data; and E.P.S., H.A.I., and R.J.F. wrote the paper.
The authors declare no conflict of interest.
Data deposition: The atomic coordinates and structure factors have been deposited in the
Protein Data Bank, www.pdb.org (PDB ID code 3F5C).
1To whom correspondence should be addressed. E-mail: firstname.lastname@example.org.
This article contains supporting information online at www.pnas.org/cgi/content/full/
© 2008 by The National Academy of Sciences of the USA
November 25, 2008 ?
vol. 105 ?
LRH-1. Direct binding experiments using surface plasmon res-
onance showed that these proteins interact with high affinity
(Kd? 0.9 ?/- 0.1 ?M; Fig. S3C), comparable to the reported
affinities of other nuclear receptors for their functional regula-
tors (34, 43–45). Notably, the binding affinity of Dax-1 LBD for
LRH-1 exceeds the affinity of Dax-1 peptides containing the
N-terminal LXXL/ML motifs for LRH-1 (34) by more than
30-fold. This comparative analysis suggests that the observed
interactions between Dax-1 and LRH-1 LBDs are functionally
The purified Dax:LRH-1 complex was crystallized, and its
three-dimensional structure was solved using the molecular
replacement method. Consistent with the biochemical data, the
asymmetric unit of the crystalline complex revealed one (Dax-
1)2:LRH-1 heterotrimer (Fig. 1). The structure of the complex
was refined to 3.0 Å with Rfree/R values of 26.1/23.4 (Fig. S4) and
included amino acid residues 318–559 for LRH-1 LBD and
250–315, 353–472 and 250–312, 353–472 for the first and second
molecules of Dax-1, respectively. The N-terminal region (aa
205–249) of both Dax-1 molecules was not visible on the electron
density maps of the complex and not included in the structure.
Details of the structure determination and refinement are
summarized in Table S1.
Dax-1 Dimerization and Its Determinants. The trimeric structure of
the complex suggested that Dax-1 may form a homodimer. To
probe the existence of a Dax-1 dimer, we coexpressed three pairs
of Dax-1 deletion mutants tagged with either His6or S tags (Fig.
2A). Analysis of purified His6-tagged proteins revealed that the
Dax-1 fragment 205–472 used in our structural analysis is a
dimer in solution, as determined by its association with the
S-tagged Dax-1 (Fig. 2B). This finding was confirmed by size
exclusion chromatography revealing the presence of a single
protein species with molecular mass of ?60 kDa in a purified
Dax-1 205–472 protein sample (Fig. S5A). This value agrees with
the calculated molecular mass for the Dax-1 dimer. In contrast,
a shorter fragment of Dax-1 corresponding to the visible Dax-1
LBD (aa 245–472) is a monomer, as determined by Western blot
and size exclusion chromatography analyses (Figs. 2 and S5B).
We conclude that 40 residues (aa 205–245) N-terminal to the
visible Dax-1 LBD are sufficient to form and maintain a stable
Dax-1 LBD in the (Dax-1)2:LRH-1 heterotrimer is a direct
consequence of the dimeric state of Dax-1 (Fig. S5C).
Specific Structural Features of the Dax-1 LBD. The dimerization
domain of Dax-1 was not visualized on electron density maps of
the (Dax-1)2:LRH-1 trimer, likely because of its flexible nature
(Fig. S5C). The residues constituting the dimerization domain of
Dax-1 precede helix H3 and correspond to helices H1 and H2 in
other nuclear receptors. The absence of H1 and H2 in Dax-1 is
consistent with secondary structure predictions suggesting that
the N-terminal region of Dax-1 LBD lacks the expected signa-
tures for these helices. Notably, the N- and C-terminal bound-
aries of the visible structure of Dax-1 (aa 250–472) match the
minimal transcriptional silencing domain of Dax-1 defined in
mutational and cellular studies (18). A second disordered region
in Dax-1 LBD (aa 314–352) corresponds to the loop connecting
helices H5 and H7. The length of this loop varies among Dax-1
LBDs from different species, and large deletions in this region
do not alter either the subcellular localization or the corepressor
function of Dax-1 (46, 47). With the exception of these two
regions, the rest of the Dax-1 LBD folds into a well-defined,
Remarkably, the ligand-binding pocket, which usually forms
the hydrophobic core of nuclear receptor LBDs, is absent in
Dax-1. Three structural elements of Dax-1—helices H11, H5,
and the short ?-strand following H5—reroute in a short cut to
eliminate the pocket (Fig. 3 A and B). The resulting volume of
the ligand-binding cavity is about 80 Å3(compared with 830 Å3
in mLRH-1) (33), which is insufficient to accommodate even a
small ligand (Fig. 3C). Mass spectroscopy analysis of the purified
(Dax-1)2:LRH-1 complex confirms the absence of any ligand
bound to the Dax-1 LBD.
is shown in yellow; the two Dax-1 molecules are shown in blue. The structural
elements involved in binding of the first Dax-1 to LRH-1 are shown in rose (RH
site) and dark red (AF-2 region). The structural elements at the second Dax-
1-LRH-1 interface are shown in light and dark green and are indicated.
The three-dimensional structure of the Dax-1:LRH-1 complex. LRH-1
repeats (indicated by arrows); the 1/2 repeat is the N-terminal to the visible Dax-1 LBD (blue). Pairs of coexpressed Dax-1 fragments 245–472, 205–472, and
138–472 with either His6or S tags are shown as black bars. (B) Western blot analysis showing formation of Dax-1 homodimers. Coexpressed fragments were
purified using Ni-NTA matrix and analyzed by Western blot using the S-tagged antibody. Lanes 2, 4, and 6 correspond to the soluble bacterial extracts; lanes 1,
3, and 5 correspond to the purified Dax-1 proteins.
Structural determinants of Dax-1 dimerization. (A) Schematic representation of Dax-1. The N-terminal domain of Dax-1 (gray) has 3 and 1/2 structural
Sablin et al.
November 25, 2008 ?
vol. 105 ?
no. 47 ?
Functional Conformations of Dax-1 and LRH-1 LBDs. Similar to the
structure of mLRH-1 alone (33), the mLRH-1 LBD in the trimer
is without hormone and in an active conformation, as defined by
the position of helix H12 (Fig. 1). Furthermore, superimposition
of the complexed LRH-1 with free LRH-1 (33) shows nearly
identical LBD structures (with a rmsd for 242 C? atoms of 0.9
Å), demonstrating that LRH-1 does not assume a different
conformation when inactivated by Dax-1. The trimer also can
form with mLRH-1 mutant A268W, a functional receptor with
an occluded ligand-binding cavity (33), indicating that Dax-1
does not require an LRH-1 hormone for binding.
Both Dax-1 molecules in the complex are found in an identical
conformation (with an rmsd for 183 C? of 0.7 Å), which differs
from that of the LRH-1 LBD. Each is characterized by docking
of helix H12 into its own coactivator-binding groove (Figs. 1 and
3A). In other nuclear receptors, this groove is known to bind the
nuclear receptor coactivator boxes containing the LXXLL motif
(43, 48, 49). In Dax-1, the conserved M464, M467, and L468
from H12 (MXXML; Fig. 3A) mimic the Leu residues from the
LXXLL motif. The docked position of H12 in Dax-1 is consistent
with its function as a repressor that does not need to bind
Architecture of the (Dax-1)2:LRH-1 Heterotrimer. The two bound
molecules of Dax-1 are curiously positioned in the complex (Fig.
1). The first Dax-1 LBD is docked into the coactivator groove of
LRH-1, forming a large binding interface of ?1350 Å3. Surpris-
ingly, the core structural element used by Dax-1 for binding to
LRH-1 is a compact loop connecting helices H3 and H4. This
loop forms a short helix, termed a ‘‘repression helix’’ (RH) (Fig.
4A and B), bordered by two conserved Pro residues, P275 and
P281, which introduce two turns into the Dax-1 polypeptide
chain. The helical conformation of the RH site is very stable and
is supported by multiple intramolecular contacts (Fig. 4B). This
property is rather uncommon for accessible surface loops, which
often are flexible. Another unusual property of the RH is the
presence of a cluster of solvent-exposed hydrophobic residues
(C276, I279, and L280; Fig. 4B), which suggests that this site
might have evolved for intermolecular protein interactions.
Superimposition of the (Dax-1)2:LRH-1 complex with the
structures of hLRH-1 bound to the GRIP-1 peptide (34, 37)
shows that the position and conformation of the RH in the
complex are similar to those of the docked regulatory peptide.
However, the RH site of mDax-1 does not have the LXXLL
motif present in nuclear receptor coactivators. Instead, the RH
site includes a sequence motif, 275-PCFXXLP-281, conserved
among all members of the NR0B1 subfamily. Thus, the exposed
hydrophobic residues C276, I279, and L280 from the Dax-1
repressor helix (Fig. 4B) substitute for the three Leu residues
from the LXXLL motif of the coactivators. The N-terminal
region of Dax-1 helix H9 with preceding loop L8–9 (Fig. 4A)
extends this interface with LRH-1 to ?1350 Å2.
LBD, just outside the ligand-binding pocket of LRH-1 (Fig. 1).
Curiously, for its interaction with LRH-1, the second Dax-1 uses
the same site, consisting of the RH and the N-terminal part of
helix H9 (Fig. 4C). The second Dax-1-LRH-1 interface includes
the N- and C-terminal parts of LRH-1 helices H7 and H11,
whereas helix H12 of Dax-1 is in an ‘‘inactive’’ conformation and is docked in the coactivator groove; the residues participating in the docking interactions are
filling the putative ligand-binding cavity of Dax-1 are shown as stick models.
Structural comparison of LRH-1 (yellow) and Dax-1 (blue) LBDs. (A) The AF-2 sites of LRH-1 and Dax-1. Helix H12 of LRH-1 is in an ‘‘active’’ conformation,
elements involved in binding. (B) Architecture of the RH site. The repressor helix is bordered by two conserved Pro residues, P275 and P281 (shown as spheres).
The helical conformation of the RH site is supported by multiple intramolecular interactions. Residues participating in binding with LRH-1 are shown in bold.
(C) A magnified view of the secondary Dax-1-LRH-1 interface indicating the residues participating in binding interactions.
Binding interactions between Dax-1 (blue) and LRH-1 (yellow). (A) A magnified view of the primary Dax-1-LRH-1 interface indicating the structural
www.pnas.org?cgi?doi?10.1073?pnas.0808936105Sablin et al.
respectively, and is less extensive (?700 Å2). Interestingly, the
second Dax-1 LBD places the side chain of semiconserved Q397
in loop L8–9 at a distance of direct contact with residues at the
mouth of the LRH-1 pocket. In LRH-1 from all species except
rodents, this is the coordination site of the phosphate group of
the bound phospholipid (Fig. S6) (34). Thus, a bound LRH-1
ligand might influence the second Dax-1 in vivo. We observe a
weak binding interface (?200 Å2; Fig. 1) between the visible
structures of Dax-1 LBDs in the heterotrimer, formed between
residues positioned on the exposed faces of helix H9 of the two
Dax-1 LBDs. This observation is consistent with the dimeric
state of Dax-1 (Fig. 2).
Mutational Analysis of the Dax-1-LRH-1 Interfaces. To date, more
than 30 human Dax-1 missense mutations have been associated
with adrenal disease (3). All mutations affect evolutionarily
conserved residues positioned within the visible structure of
Dax-1, suggesting that this structure corresponds to a critical
functional domain. Most mutations affect buried Dax-1 residues
and are predicted to destabilize the structural core of the LBD
(50); however, one described mutation in hDax-1 affects a
solvent-exposed hydrophobic residue (L278P in hDax-1 (3);
L280P in mDax-1). Strikingly, L280 in mDax-1 is one of the
conserved residues involved in the binding of Dax-1 to LRH-1
(Fig. 4B). Substitution of L278 for Pro in hDax-1 likely would
distort the helical conformation of the RH site, thereby weak-
ening the binding interactions of Dax-1 and LRH-1. Consistent
with this prediction, equivalent mutation L280P in mDax-1
resulted in diminished repression of LRH-1 by Dax-1 (Fig. 5A).
Mutating the adjacent hydrophobic I279 to negatively charged
Asp eliminated the binding of Dax-1 to LRH-1 (Fig. S7) and
produced a variant that no longer overpowered the coactivator
(I279D in Fig. 5B). Consistent with these results, double-mutant
I279A/L280A (Fig. 5C) attenuated the ability of Dax-1 to act as
a potent corepressor of LRH-1 in cellular reporter assays. Taken
together with structural data, these findings confirm the func-
tional importance of the discovered RH and its conserved motif
PCFXXLP for regulatory interactions of Dax-1 with LRH-1.
Complementary to the described mutations in Dax-1 RH site,
LRH-1 mutations R380E and E553R in the primary AF-2 site
eliminated Dax-1 binding to the receptor (Fig. S7). In contrast
(Y538D, V541D, and Y538D/V541D; Fig. 4C) exhibited no
effect on Dax-1 binding (Fig. S7). These results suggest that
binding of the second Dax-1 to LRH-1 is mediated by the first
Dax-1 docked into the AF-2 groove, likely through direct
Dax-1-Dax-1 interactions that we see in part in the crystal
structure. These data also are consistent with the hypothesis that
Dax-1 binds to LRH-1 as a homodimer, with determining
interactions at the primary AF-2 binding site. Although neither
of the secondary site LRH-1 mutants produced noticeable
these mutants did attenuate the inhibitory function of Dax-1 in
cell-based transcription assays (Fig. 5C). An effect of these
mutants was apparent when they were paired with the RH site
mutant I279A/L280A, suggesting that the presence of the second
Dax-1 in the complex contributes to the repressor efficiency.
Mechanism of Transcriptional Inhibition by Dax-1. The data de-
scribed in this work demonstrate that the Dax-1 LBD interacts
directly with the LBD of LRH-1. Our finding that Dax-1 binds
in the coactivator groove of LRH-1 is consistent with previous
data showing that repression by Dax-1 could be through com-
petitive inhibition (9). A novel twist that our analysis introduces
LBD and contains the subfamily conserved sequence motif
PCFXXLP (Fig. 4B). The observed novel mode of interaction
does not contradict the previous assignment for the N-terminal
repeats as the region responsible for nuclear localization of
Dax-1 (17). Furthermore, the observed docking of Dax-1 LBD
into the coactivator groove of LRH-1 is consistent with the
recently demonstrated mode of repression of nuclear receptors
Nur77 and ERR? by Dax-1, for which Dax-1 LBD was shown to
be sufficient (7, 8).
Our work suggests two reasons why Dax-1 functions as an
efficient transcriptional inhibitor that can overpower coactiva-
tors. First, the binding interactions of Dax-1 at the AF-2 site are
more extensive than those of a coactivator with a single LXXLL
motif. The repression site of Dax-1 is expanded by residues
presented by the well-ordered helix H9. Second, the RH site of
Dax-1 is well structured and stabilized through extensive in-
tramolecular interactions. The stability of the RH site contrasts
with the dynamic nature of LXXLL motifs of nuclear receptor
coactivators that are followed by flexible loops (43), which
require prestructuring for binding to nuclear receptors. We
conclude that the increased interaction surface and the preor-
dered state of the RH site underlie the efficiency of Dax-1 as a
Dimerization of Dax-1 and Possible Functional Role of the Dax-1
Dimer. Dax-1 binds to primary and secondary sites on LRH-1.
What are these sites’ contributions to the trimer assembly? As
LRH-1 by Dax-1 in HepG2 cells. Luciferase activity was measured after cotrans-
fection of vectors encoding the Aro-Luc reporter, mLRH-1, and either wild-
type or mutant mDax-1. Mutation L280P in the RH site of mDax-1 (L278P in
Dax-1 in the presence of coactivator SRC-1. LRH-1 and Dax-1 variants were
cotransfected into HepG2 cells with or without SRC-1, as indicated by ‘‘?.’’
Dax-1 is a potent repressor of LRH-1 that overpowers activation by SRC-1;
(C) Diminished repression of Dax-1 by parallel mutations at the Dax-1 RH site
and at the secondary binding site of LRH-1. Either wild-type or I279A/L280A
Dax-1 RH mutant was cotransfected into HepG2 cells with either wild-type or
mutant Y538D LRH-1.
Regulatory interactions between Dax-1 and LRH-1. (A) Repression of
Sablin et al.
November 25, 2008 ?
vol. 105 ?
no. 47 ?
discussed earlier, LRH-1 mutations in the primary AF-2 site
block formation of the complex. In contrast, LRH-1 mutants
with altered secondary sites behave similarly to wild-type pro-
tein. In the context of the structure, our data indicate that Dax-1
binds LRH-1 as a dimer with determining interactions at the
primary AF-2 site.
What is the structure of the Dax-1 dimer? Although our group
and others (51) have observed formation of Dax-1 dimers, the
(Dax-1)2:LRH-1 structure does not reveal an obvious configu-
ration for the Dax-1 homodimer. A previously proposed model
suggests that Dax-1 molecules might dimerize through docking
of the N-terminal LXXLL motifs into the coactivator grooves of
their respective partners (51). Our results argue against this
model, because the structure of Dax-1 reveals occupation of the
AF-2 site by helix H12. Our findings also demonstrate that Dax-1
can dimerize in the absence of the LXXLL motifs, and that the
presence of 40 residues preceding helix H3 is sufficient for Dax-1
dimerization. The dimerization domain of Dax-1 is not seen in
the (Dax-1)2:LRH-1 trimer, likely because of its flexible linkage
to the visible Dax-1 LBDs.
In this work, we show Dax-1 to be a dimeric repressor with one
repression site present on each of the two LBDs. Its binding
partner LRH-1 has only one AF-2 site. Why is the Dax-1 dimer
needed to bind to the LRH-1 monomer? What could be the role
of the second Dax-1 in the regulatory complex? One possibility
is that the second Dax-1 LBD might enhance the repressive
power, providing more contacts to LRH-1 and thus ensuring
repression and domination of Dax-1 over coactivators. Consis-
tent with this idea, binding experiments showed that the dimeric
Dax-1 binds LRH-1 with higher affinity (Kd? 0.9 ? 0.1 ?M; Fig.
S3C) compared with the truncated monomeric Dax-1 LBD (aa
245–472; Kd? 4.8 ? 0.2 ?M; Fig. S8). In addition, the flexible
association between the two Dax-1 molecules might facilitate the
binding of different nuclear receptors with variable molecular
surfaces. The weakly ordered contact relating the two Dax-1
LBDs in the trimer may be a functional feature that allows the
pair to adjust and perform a ‘‘claw-like’’ clasping, to accommo-
date different nuclear receptor LBDs. We also note that the
placement of the second Dax-1 at the entrance to the LRH-1
ligand-binding pocket makes the second Dax-1 LBD a candidate
to serve as a ‘‘sensor’’ of the receptor’s ligand state. An
alternative possibility is that the second Dax-1 LBD might simply
provide an additional binding interface and link the regulatory
assembly to other components of the transcriptional machinery.
The dimerization of Dax-1 in vivo might depend on its concen-
tration in cells, be controlled by specific co-regulatory proteins,
or be triggered by specific signaling events; therefore, both
monomeric and dimeric forms of Dax-1 might exist and function
in different cellular contexts. Clearly, gaining further insight into
the dynamics of Dax-1 dimerization and identifying relevant
partners of Dax-1 will help translate this unusual structure into
In summary, we describe the first three-dimensional structure
of Dax-1 in association with one of its physiological targets, the
nuclear receptor LRH-1. The RH in the Dax-1 LBD, which
includes a subfamily conserved sequence motif, 275-PCFXXLP-
281, mediates this association. Our findings demonstrate that
Dax-1 functions as a ligand-independent nuclear receptor and
explain its function as a competitive transcriptional corepressor.
The three-dimensional structure of Dax-1 also rationalizes the
naturally occurring missense mutations in the human Dax-1
leading to congenital X-linked adrenal hypoplasia. This work
should aid in the design of further experiments aimed at
clarifying the function of Dax-1 as a potent transcriptional
Materials and Methods
LBDs were expressed in Escherichia coli and immobilized on glutathione-
agarose beads (Sigma).35S-radiolabelled full-length mDax-1 was prepared in
vitro in a reticulocyte lysate using the TNT T7 quick-coupled transcription-
translation system (Promega) according to the manufacturer’s protocol. The
immobilized GST proteins with specifically bound mDax-1 were eluted, re-
solved by SDS-PAGE, and visualized by autoradiography.
Preparation, Crystallization, and Crystallographic Analysis of the Dax-1:LRH-1
Complex. A DNA fragment encoding mLRH-1 LBD (aa 313–560) with His6tags
and a cleavage site for tobacco etch virus (TEV) protease was cloned into a
1–208, 205–472, 138–472, 70–472, and 1–472 were cloned into pETDuet-1
plasmid (Novagen). For coexpression of Dax-1 with LRH-1, BL21(DE3)Star cells
(Invitrogen) were cotransformed with the corresponding plasmids. Coexpres-
sion was induced with 0.1 mM isopropyl ?-D-1-thiogalactopyranoside (IPTG),
and cells were grown for 12 h at 15 °C. Cell lysates were loaded onto Ni-NTA
matrix, and specifically bound proteins were eluted with 200 mM imidazole.
These experiments showed that only the Dax-1 LBD forms a stable complex
was purified using size-exclusion chromatography (Superdex 75, Amersham
Pharmacia). The Dax-1:LRH-1 complex was crystallized by the vapor diffusion
method, using a reservoir solution containing 0.1 M Hepes (pH 7.5), 20% PEG
8000, and 1 mM Deoxy-BigChap. Before data collection, crystals were flash-
frozen in liquid nitrogen. X-ray diffraction data were obtained at ?180 °C to
3.0 Å at the Advanced Light Source beamline 8.3.1 (? ? 1.1 Å) using a single
crystal. Data were integrated and scaled using DENZO and SCALEPACK. The
crystal was of the space group P43with cell dimensions of a ? b ? 103.4 Å and
c ? 117.5 Å, and it contained one (Dax-1)2:LRH-1 heterotrimer in the asym-
metric unit. The structure of the complex was determined by the molecular
replacement method using the CNS algorithms with a starting model derived
from the atomic coordinates for mLRH-1 (PDB ID 1PK5). Electron-density maps
based on coefficients 2Fo? Fcwere calculated from the phases of the initial
model. Subsequent rounds of model building and refinement were performed
the structure was checked using simulated annealing composite omit maps.
Preparation of His6- and S-Tagged Dax-1 Fragments for Western Blot Analysis.
Three DNA fragments encoding mDax-1 residues 245–472, 205–472, and
pETDuet-1 and pACYCDuet-1 vectors, respectively (Novagen). For coexpres-
sion of the His6- and S-tagged fragments, BL21(DE3)Star cells (Invitrogen)
were cotransformed with the corresponding plasmids. Coexpression was in-
duced with 0.1 mM IPTG, and cells were grown for 12 h at 15 °C. Cell lysates
with 200 mM imidazole. The proteins were resolved by SDS-PAGE and ana-
lyzed by Western blot, using mouse anti-S-tag monoclonal antibody and goat
anti-mouse IgG alkaline phosphatase conjugate (Novagen), and then devel-
oped with colorimetric detection reagents.
Surface Plasmon Resonance (SPR) Assay. The His6-tagged mLRH-1 LBD and
Dax-1 fragments 205–472 and 245–472 were expressed and purified using
affinity and size-exclusion chromatography, and the purified LRH-1 LBD was
immobilized to an Ni2?-tri-NTA chip (52). For SPR assays using anti-GST chips,
a DNA fragment encoding mLRH-1 LBD was cloned into a pGEX-4T-1 vector
(Amersham Pharmacia). The GST-LRH-1 protein was purified using glutathi-
one sepharose beads and size-exclusion chromatography and immobilized to
to a CM5 chip. All measurements were performed using a Biacore T100
over immobilized LRH-1 and reference surfaces. The equilibrium dissociation
constants (Kd) were determined using nonlinear regression analysis for the
association and dissociation phases of the sensorgrams.
Plasmids, Site-Directed PCR Mutagenesis, and Transfection. The Aro-Luc re-
porter vector pGL2 was purchased from Promega; pcDNA3, pcDNA3-
FLAGmDax-1, and pSG5-SRC-1, pcineo-HAmLRH-1 plasmids were described
earlier. Mutagenesis of all LRH-1 and Dax-1 constructs was performed using
the QuikChange site-directed mutagenesis kit (Stratagene) according to the
manufacturer’s protocol. Before the transcriptional assays, expression of all
mutant proteins was tested in Cos7 cells by Western blot analysis of nuclear
and cytoplasmic extracts using the corresponding antibodies to epitope tags.
www.pnas.org?cgi?doi?10.1073?pnas.0808936105Sablin et al.