| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Nuclear Receptor Heterodimer
Center for Biotechnology (F.F.W., K.R.S., E.T., D.F.)
Department of Biosciences Karolinska Institute NOVUM, S-141 57
Huddinge, Sweden
Department of Medical Nutrition
(J.-A.G.) Karolinska Institute F-60 NOVUM,
S-141 86 Huddinge, Sweden
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
complex. We show that in vitro the
steroid receptor coactivator SRC-1 can be recruited by RXR
upon addition of its ligand, and to OR1 upon addition of
22(R)-OH-cholesterol, demonstrating that the latter can act
as a direct ligand to OR1. Additionally, heterodimerization is
sufficient to recruit SRC-1 to OR1/RXR, indicating SRC-1 as a
molecular mediator of dimerization-induced activation. In transfection
experiments, coexpression of a nuclear receptor-interacting fragment of
SRC-1 abolishes constitutive activation by OR1/RXR, which can be
restored by overexpression of full-length SRC-1. This constitutes
evidence for an in vivo role of SRC-1 in
dimerization-induced activation by OR1/RXR.
Additionally, we show that the nuclear receptor-interacting protein
RIP140 binds in vitro to OR1 and RXR with requirements
distinct from those of SRC-1, and that binding of the two cofactors is
competitive. Taken together, our results suggest a complex modulation
of differentially induced transactivation by OR1/RXR coregulatory
molecules.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
c, tau-4) (3, 6, 7, 8). While some family members bind to DNA as either monomers or homodimers, a large number of them have been shown to recognize their specific DNA response elements as heterodimers with one particular receptor variant, the 9-cis retinoic acid (9cRA) or retinoid X receptor (RXR) (9). Accordingly, the corresponding heterodimer complexes are potentially activated by two types of ligands, i.e. via both dimerization partners. However, the LBD also plays a major role in receptor dimerization, and the two functions are structurally not separable since they comprise overlapping regions. Ligand binding has been shown to induce a significant conformational change in receptor LBDs (10, 11, 12), which among other things is thought to reposition H12 in a different intramolecular context to form part of a complete AF-2 surface as a prerequisite for transactivation (4, 13, 14). It is thus not surprising that ligand binding and dimerization can affect each other in various ways. For example, ligand-binding can favor or disfavor homo- vs. heterodimerization, and heterodimerization can prevent or, on the contrary, be necessary for ligand binding by one of the partners (15, 16, 17, 18).
The nuclear receptor OR1 (UR, RIP15, NER, LXRß) (19, 20, 21) and
its paralog RLD-1 (LXR) (22, 23) are heterodimerization partners of
RXR and have recently been shown to be activated by various oxysterols,
which are cholesterol metabolites and precursors in the biosynthesis of
steroid hormones and bile acids (24, 25). The identification of this
novel class of nuclear receptor activators suggested a role of the
OR1/RLD-1 nuclear receptor subgroup in cholesterol homeostasis and/or
the regulation of downstream biosynthetic processes, which was recently
elegantly demonstrated by targeted gene inactivation of the LXR gene
in mice (26, 27). Since in the complex with OR1 or LXR the partner
protein RXR remains highly permissive for activation by its ligand 9cRA
(21, 23), a vitamin A metabolite, the heterodimer is a potential target
for two very different classes of signaling molecules. Additionally, we
have recently demonstrated that although neither OR1-/RLD-1-LBD nor
RXR-LBD transactivate by themselves, the heterodimer can confer
constitutive transactivation (28); to date, all other described RXR
heterodimers, with the exception of MB67/CAR (29, 30) and perhaps Nurr1
(15), appear to be ligand dependent for transactivation.
OR1/RXR-mediated constitutive activation might be of more general
interest also for other nuclear receptor-mediated signaling pathways
because OR1/RXR binds a motif (DR4) that is recognized by several other
nuclear receptor dimers, e.g. an RXR heterodimer with the
thyroid hormone receptor (TR) or a COUP-TF homodimer, and could thus
modulate or interfere with transcriptional regulation by these
receptors independent of ligand. Several lines of evidence indicate
that the observed constitutive transactivation is based on
dimerization-induced activation of OR1 upon interaction with RXR,
which constitutes a novel mechanism of nuclear receptor activation and
makes the OR1/RXR heterodimer the first nuclear receptor complex
described to be activated in three different ways (28). Given that the
relative strength of constitutive vs. ligand-dependent
transcriptional activity of the heterodimer has been observed to vary
greatly under different conditions, constitutive complex activation
might play an important regulatory role not only by conferring basal
transcriptional activation but also by tuning the ligand sensitivity of
the heterodimer.
In recent years, a number of coregulatory proteins that mediate transcriptional regulation exerted by nuclear receptors have been described (see Refs. 31, 32 for reviews). Although there are some reports of direct interaction of family members with central components of the transcriptional machinery (e.g. Refs. 33, 34, 35), recruitment of intermediary factors seems to be of more general relevance. For example, several factors have been described that, upon their direct or indirect recruitment by the DNA-bound nuclear receptors, alter the acetylation status of neighboring histones (36, 37, 38, 39, 40, 41) and thereby the accessibility of the promoter region to further protein binding, revealing one mechanism by which nuclear receptor cofactors regulate transactivation.
In this work, we have extended our previous work on the nuclear
receptor OR1/RXR heterodimer to study the effect of differential
activation of the complex on interaction with some known coregulatory
proteins, with a focus on the nuclear receptor coactivator SRC-1 (42)
and the coregulator RIP140 (43). Molecules of the SRC-1 coactivator
class are recruited to nuclear receptors upon ligand binding and have
recently been demonstrated to possess histone acetyltransferase
activity (41) and to be able to recruit further histone
acetyltransferases (44). We show that SRC-1 interacts with the
OR1/RXR heterodimer upon activation by any of the three modes, but
not constitutively with either receptor. Thus, our experiments
demonstrate 1) that 22(R)-OH-cholesterol
[22(R)-HC] can act as a true ligand for OR1, and 2)
that, just like ligand activation, dimerization-induced activation may
act through SRC-1 as a molecular mediator. In transient transfection
studies, it is shown that SRC-1 can indeed mediate the constitutive
transcriptional activation by the OR1/RXR complex. For the nuclear
receptor-interacting protein RIP140, a distinct function in mammalian
transcriptional regulation has so far not been identified, but a
regulatory role by competition with SRC-1 has been proposed (45). We
present evidence that, in contrast to SRC-1, RIP140 displays
constitutive and only moderately ligand-enhanced interaction with
either partner of the OR1/RXR heterodimer in vitro, and
that binding to the heterodimer cannot occur simultaneously with SRC-1.
We discuss the possible roles of the accessory proteins in
transcriptional regulation by the OR1/RXR heterodimer and the
implications for differential activation by this multiply activatable
nuclear receptor complex.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Heterodimer Interacts with SRC-1 in
Vitro has been shown to be
transcriptionally activated in three ways, i.e. by
activators of each partner as well as by dimerization itself, we wished
to determine how these modes of activation correlated with recruitment
of a member of the best characterized nuclear receptor accessory
protein group (41, 42, 46, 47, 48, 49, 50, 51, 52), the coactivator SRC-1 (42). To address
this question, we tested the ability of the DNA-bound OR1/RXR
complex to interact with a fragment (SRC-iad, Fig. 1
) comprising a previously described
nuclear receptor interaction domain of SRC-1 (46, 52) in an
electrophoretic mobility shift assay (EMSA). In
vitro-translated receptors were bound to radiolabeled
oligonucleotides containing a DR4 response element in the absence or
presence of purified glutathione-S-transferase
(GST)-tagged SRC-iad produced in Escherichia coli (Fig. 2A). In contrast to a TR/RXR
heterodimer, the OR1/RXR heterodimer was able to recruit the
GST-fused SRC-1 fragment, but not the GST portion alone (data not
shown), in the absence of added ligand. This recruitment was not
affected by a point mutation in the AF-2 motif of RXR, but was
completely abolished by the corresponding mutation in H12 of OR1.
|
|
), a clear enhancement of
SRC interaction by addition of 9cRA could be observed (lanes 1–6
vs. 7–12). Strikingly, this ligand effect was lost when the
RXR H12 mutant was employed (lanes 25–36). Somewhat unexpectedly,
we were not able to observe any enhancement of SRC-iad interaction with
the OR1/RXR heterodimer by addition of its activator and proposed
ligand 22(R)-HC (lanes 13–24). However, a ligand effect on
SRC recruitment by OR1 could be demonstrated on a GAL4-OR1 dimer (Fig. 2C), which is not preactivated by dimerization. The enhancement of OR1
interaction with SRC-iad upon addition of 22(R)-HC indicates
that this metabolite is indeed an OR1-ligand.
Interaction of SRC-1 with the DNA-Bound OR1/RXR Heterodimer Is
Ligand Independent
Recruitment of SRC-1 by the OR1/RXR heterodimer in the absence
of added ligand was in line with the described dimerization-induced
activation of OR1 by interaction with RXR. However, since the
presence of nuclear receptor ligands in the in vitro
translation system cannot be excluded, interaction of the OR1/RXR
heterodimer with SRC-1 was further investigated using
His6-tagged receptor proteins overexpressed in and purified
from E. coli. As shown in Fig. 2D, SRC-iad interacted with
the DR4-bound OR1/RXR-
N heterodimer in the absence of added
ligand. Since this system is highly unlikely to contain any
steroid-like components, this result confirms that the OR1/RXR
heterodimer can recruit SRC-1 in a ligand-independent fashion. The
interaction could be further enhanced by the RXR ligand, 9cRA, but,
again, no enhancement by addition of 22(R)-HC was observed
on the dimerization-activated complex.
Recruitment of SRC-1 in Vitro Is Triggered by
Dimerization of OR1 with RXR
To be able to further dissect the effect of dimerization on
recruitment of SRC-1, we employed a GST-fusion protein pull-down assay.
First, we tested the interaction of SRC-1 with the GST-fused LBDs of
both receptors in the absence or presence of their respective
activators. As illustrated in Fig. 3A, in vitro translated and radiolabeled SRC-1 (aa 1–1061)
shows no significant constitutive interaction with either of the
unliganded LBDs and no interaction at all with the GST-fusion moiety.
In contrast, interaction is strongly enhanced upon addition of 9cRA or
22(R)-HC with the respective receptor LBD, indicating that
SRC-1 can be recruited to both heterodimer partners by ligand
activation. Thus, this result confirms that 22(R)-HC can act
as a direct ligand to OR1 and independently of RXR
heterodimerization. While mutation of the AF-2 motif of OR1 abolished
ligand enhancement of SRC-1 interaction, the corresponding RXR
mutant still retained some 9cRA responsiveness, indicating that this
mutant is leaky toward both ligand binding and activation.
|
). (Note that SRC-1 recruitment by RXRm + 9cRA is
lost due to the additional washing step involved in this procedure.) In
this setup, it can be shown that SRC-1 is bound far more efficiently by
OR1-LBD dimerized with RXR-LBD than by the individual LBDs (lane 11
vs. lanes 1 and 5). Great care was taken to use equivalent
amounts of GST-fused proteins as bait, and the effect was observed in
several independent experiments and even when more free than
heterodimerized OR1 was offered. The experiment thus demonstrates that
coactivator recruitment can indeed be triggered by the third mode of
activation, heterodimerization. As an important control, this effect is
not observed for the heterodimer of TR and RXR (lane 19). In
conclusion, SRC-1 displays the characteristics predicted for a
molecular mediator of dimerization-induced activation of the OR1/RXR
complex.
In this extended pull-down assay, both 9cRA and 22(R)-HC
enhance recruitment of SRC-1 by the heterodimer (lanes 11–14),
demonstrating that both ligands can further activate the complex. In
agreement with our previous observations (28), mutation of the
RXR-AF-2 does not affect the dimerization-activated state (lane 15),
while it is sensitive to mutation of the AF-2 motif of OR1 (lane 16).
This result is not due to impaired heterodimerization since neither
mutation had a visible effect on heterodimer-LBD recruitment as judged
by Coomassie staining (data not shown).
RIP140 Interacts with OR1 and RXR in Vitro
We next tested the activation dependency of OR1/RXR interaction
with another coregulatory protein, the human RIP140 (43, 53).
Significant constitutive interaction with a 728-amino acid
carboxy-terminal portion of hRIP140 (RIP140-N, see Fig. 1) is found
for the unliganded receptor LBDs in the pull-down assay (Fig. 4A, lanes 3 and 7). This can be enhanced
only moderately by addition of the respective ligands (lanes 4 and 8).
As for SRC-1, interaction of RIP140-N with the OR1/RXR
heterodimer (lane 9) is stronger than for the individual LBDs. However,
the observed increase is weaker, and, given the background of
constitutive interaction, the increased interaction might be explained
by cooperativity of binding alone or by the recruitment of one molecule
of RIP140-N per heterodimer partner and might therefore not be
activation triggered. Mutation of the AF-2s of either receptor has no
clear effect on RIP140-N interaction in the combinations tested
(lanes 10 and 11). Finally, the notion that the presentation of two
LBDs in itself leads to increased enhanced binding of RIP140-N is
supported by the fact that it also occurs upon TR dimerization with
RXR (lanes 13–15), corresponding to a nonactivated receptor
state.
|
),
which had been expressed in and purified from E. coli, with
the DNA-bound OR1/RXR heterodimer (Fig. 4B). In the absence of added
ligand, a slight supershift upon addition of RIP140-iad is observed
(lanes 1–4; cf. also lanes 8 and 9). In comparison, in the
presence of 9cRA a similar supershift is observed already with lower
amounts of RIP140-iad and its migration further slowed in a
dose-dependent manner (lanes 5–7). While the reasons for this gradual
supershift remain to be analyzed further, a possible explanation is
that two molecules of RIP140-iad bind to the 9cRA-activated OR1/RXR
heterodimer. In the presence of limiting amounts of RIP140-iad,
fluctuation between ternary and quaternary complexes might cause
smearing.
The experiment was extended to examine the effects of H12 mutations as
well (lanes 8–13). Mutation of the AF-2 motif of RXR leads to a
loss of the additional supershift, while mutation of H12 of OR1
abolishes interaction with RIP140-iad, establishing that both H12
receptor domains play a role in interaction with RIP140. It should be
noted that the fragment of RIP140 used here was smaller than the one
used in the pull-down assay, containing only two instead of four
nuclear receptor-interaction motifs (46) (cf. Fig 1A), which could
explain the observed difference in sensitivity for mutation of H12 of
RXR.
Binding of RIP140 Interferes with SRC-1 Recruitment by
OR1/RXR
Since both SRC-1 and RIP140 can bind to the 9cRA-activated
OR1/RXR heterodimer, we asked whether the interaction can take place
simultaneously, or if binding of one factor prevents interaction with
the other. First, we performed an SRC-1 pulldown with the dimerized,
liganded, or unliganded OR1/RXR LBDs in the absence or presence of
excess purified His10-RIP140-iad (Fig. 5A). In all three activation states,
RIP140-iad clearly reduced heterodimer interaction with SRC-1,
indicating that binding of RIP140 vs. SRC-1 is competitive.
Second, in an EMSA (Fig. 5B), His10-RIP140-iad delayed the
recruitment of GST-SRC-iad in the 9cRA-activated state (lanes 10–14
vs. 15–19) and possibly weakly in the absence of
added ligand (lanes 7–9 vs. 2–4). However, it should be
kept in mind that the strong 9cRA sensitivity of RIP140-iad observed
here is most likely due to the fact that this is only a short fragment
of RIP140. Slower complexes that might contain both RIP140-iad and
SRC-iad were not observed. Taken together, the results suggest that
interaction of OR1/RXR with RIP140 and SRC-1 is mutually
exclusive.
|
observed in in vitro experiments
suggests a role of SRC-1 as a mediator of transcriptional activation by
the complex of these receptors. In transient transfection experiments
in CV1 cells employing OR1 and RXR and a 3xDR4-TK-luciferase
reporter, cotransfection of a plasmid encoding SRC-1 (aa 1–1061) did
not yield enhanced constitutive transactivation (data not shown).
To test the idea that this could be due to the presence of saturating
endogenous amounts of proteins of the SRC-1 class, we used a truncated
version of SRC-1 comprising only a nuclear receptor interaction domain
(aa 216–394), which can be expected to act as a dominant negative
interactor. Indeed, cotransfection of a construct expressing this
fragment of SRC-1 efficiently interfered with activation by OR1/RXR
(Fig. 6A). The rather high basal activity
of the 3xDR4-TK-luciferase reporter remained largely unaffected,
indicating that the observed effect was due to interaction of the
nuclear receptor interaction domain of SRC-1 with OR1/RXR and not
other factors acting on the promoter employed. Conversely, the
transcriptional enhancement by OR1/RXR that was affected by
cotransfection of the dominant negative SRC-1 fragment could be
restored by additional coexpression of excess amounts of SRC-1 (Fig. 6B). Thus, SRC-1 cannot only interact with the OR1/RXR heterodimer
but can also functionally mediate dimerization-induced activation by
this nuclear receptor heterodimer in intact cells.
|
Heterodimer), cotransfection of the corresponding
OR1 mutant with RXR and a 3xDR4-TK-LUC reporter not only completely
abolished dimerization-induced activation and
22(R)-HC-enhanced activation but also affected
9cRA-dependent transactivation. Taken together, and taking into account
our previous data obtained with GAL4-chimeric receptor variants (28),
the results show that mutation of H12 of OR1 abolishes ligand
inducibility of OR1 as well as dimerization-induced activation of the
heterodimeric complex and, additionally, affects the 9cRA
responsiveness of RXR.
|
abolishes the 9cRA-responsiveness of the OR1/RXR
complex but leaves dimerization-induced activation unaffected (28).
This seemed to be in conflict with reports that a complex of the OR1
paralog LXR with an H12 point-mutated variant of RXR (L455A,
E456A, here referred to as RXR-le) retained 9cRA inducibility, which
was interpreted to indicate that the 9cRA-response of the heterodimer
is mediated by the activation function of LXR (54). We therefore
decided to analyze the different mutant variants in comparison. In a
transient cotransfection assay (Fig. 7B), both RXR-ml and RXR-le
displayed residual 9cRA responsiveness on a 3xDR1-TK-LUC reporter,
while the deletion variant RXR was silent. When tested as
GAL4-fusion variants of RXR (Fig. 7C), the point mutants displayed a
clear 9cRA-response (see also Ref. 54) and thus proved to be leaky with
regard to 9cRA-activation. The observed adverse effects of these point
mutations on transactivation from the DR1 element (Ref. 54 and Fig. 7B)
might be due to a reduction in homodimerization efficiency or other
functions of RXR specifically required for transactivation from a
DR1.
Judging from the activity obtained with the GAL4-chimera, mutant RXR-ml
is somewhat more stringent than mutant RXR-le and was therefore used
for all other experiments described here. In the pull-down assay,
mutation of the AF-2 motif of RXR reduced, but did not abolish,
9cRA-dependent recruitment of SRC-1, again indicating that the mutation
is leaky, and had no effect on its dimerization-induced recruitment.
However, 9cRA-enhanced interaction of the OR1/RXRml heterodimeric
complex with the SRC-iad or RIP140-iad fragments in EMSA could no
longer be observed.
Taken together, the results are in line with our previous conclusions
that 9cRA induction, but not dimerization-induced activation, is
dependent on an intact H12 in RXR, while dimerization-induced
activation is eliminated by mutations in H12 of OR1.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
and proposed the dimerization-induced occurrence of a
conformational change in the LBD, similar to that otherwise obtained
upon ligand binding, as the underlying mechanism. Since ligand-induced
transactivation by nuclear receptors has recently been shown to be
mediated by direct interaction with certain coactivator proteins, we
have in the present work addressed the idea that dimerization-induced
activation of OR1 upon interaction with RXR might lead to the
recruitment of such factors as well. Here, we demonstrate that in
vitro interaction with the nuclear receptor coactivator SRC-1 is
indeed triggered by dimerization of OR1 with RXR, with roles for the
AF-2 motifs of OR1 and RXR as predicted by our previous studies
(28). The identification of the SRC-1 coactivator as a
dimerization-induced interactor of OR1/RXR provides a molecular
basis for the concept of dimerization-induced activation and, despite
contrary reports (55), strongly supports the notion that
dimerization-induced activation is indeed a ligand-independent
event.
As reported before for nuclear receptors (e.g. Refs. 42, 56, 57, 58), we
show that SRC-1 is recruited to glutathione-Sepharose-bound GST fusions
of both RXR- and OR1-LBD in an activator-dependent fashion. The
experiment provides strong evidence for direct interaction of the
activator employed, 22(R)-HC, with the OR1-LBD, indicating
that it is not only an activator but indeed a ligand to OR1 and that
its binding to OR1 is independent of RXR. As a central result for
this study, the pull-down experiments demonstrated that while SRC-1 is
unable to interact efficiently with nonactivated OR1- or RXR-LBD, it
is readily recruited by the heterodimerized LBDs. The fact that this is
not observed for a heterodimer of RXR- and TR-LBDs rules out the
possibility that this is a general, merely synergistic, effect of weak
binding to two heterodimerized nuclear receptor LBDs. In agreement with
previous observations from transient transfection studies that
dimerization-activated OR1/RXR-heterodimer can be superactivated by
either ligand (Ref. 28 and our unpublished results), both 9cRA
and 22(R)-HC additionally enhanced recruitment of SRC-1 by
dimerized OR1/RXR.
Similar results showing interaction of the DNA-bound OR1/RXR
heterodimer with SRC-1 in the absence of added ligand were obtained in
EMSA studies, although 22(R)-HC-enhanced recruitment could
only be demonstrated when a homodimeric GAL4-OR1 fusion variant was
employed. OR1 expressed in and purified from E. coli was
additionally employed to confirm the conclusion from the pull-down
assay that, in contrast to a TR/RXR heterodimer, a heterodimer of
purified OR1 and RXR is activated with respect to interaction with
SRC-1 in a system devoid of potential sources of OR1 or RXR ligands.
It should be noted that the EMSA experiments involved the central
nuclear receptor interaction domain of SRC-1 only (cf. Fig. 1A).
In both pull-down and EMSA assays, the roles of the carboxy-terminal
AF-2 motifs of the partner receptors for SRC-1-interaction triggered by
the three modes of activation were in close agreement with our
observations from previous transfection studies (28). Ligand-induced
SRC-1-interaction was affected by mutations in H12 of the respective
liganded receptor, while dimerization-induced interaction was dependent
on an intact H12 of OR1 but not of RXR.
In transient transfection studies, the occurrence of
dimerization-induced activation by OR1/RXR from a DR4-containing
reporter in the absence of cotransfected SRC-1 or other nuclear
receptor coactivators indicated that such coactivators are
constitutively expressed in the cells. It was therefore not surprising
that cotransfection of SRC-1 did not yield any further transcriptional
activation. Instead, we employed a dominant negative variant of SRC-1
(SRC-iad, containing an nuclear receptor-interaction domain but no
transcriptional activation domain), which efficiently interfered with
OR1/RXR-mediated transcriptional activation, suggesting an
interaction of SRC-iad with OR1/RXR that blocks interaction of the
heterodimer with endogenous nuclear receptor coactivators.
Subsequently, by titrating in functional SRC-1 to compete with dominant
negative SRC-iad for OR1/RXR interaction, transactivation by the
complex was regained. Thus, SRC-1 can act as a molecular mediator of
dimerization-induced transcriptional activation by the OR1/RXR
heterodimer.
The precise role of a second nuclear receptor-interacting protein
employed in this work, RIP140, in transcriptional regulation by nuclear
receptors remains to be resolved. While there are several reports that
it can act as a nuclear receptor coactivator (53, 59, 60) and that
nuclear receptor interaction is dependent on ligand and an intact AF-2
motif (43, 56, 61), other studies in mammalian cell systems (43) and in
yeast (45, 62) failed to demonstrate any transcriptional enhancement
effect. Moreover, negative effects of RIP140 on transcriptional
activation have also been reported (43, 62, 63). A model has recently
been put forward according to which RIP140 indirectly regulates nuclear
receptor AF-2 activity by competition for coactivators such as SRC-1
(45). In view of the fact that RIP140 seems to share a number of
interaction characteristics with SRC-1, including a common interaction
motif (46), we were interested in its ability to interact with the
different activated states of the OR1/RXR heterodimer.
In our studies, RIP140 displayed OR1/RXR interaction characteristics
that were distinct from those of SRC-1. First, the pull-down assay
revealed that RIP140 (aa 431-1158) is able to interact with OR1,
RXR, and TR in their nonactivated states and that ligand
activation plays only a minor role for RIP140 recruitment under these
conditions. Second, in the pull-down assay, interaction was unaffected
by mutation of the AF-2 motif of either OR1 or RXR, supporting the
notion that RIP140 interaction with these nuclear receptors is not
strictly activation dependent and involves other nuclear receptor
structural features apart from this motif. However, the EMSA, in which
a shorter portion of RIP140 (RIP140-iad) had to be employed, due to
technical limitations of protein expression in E.
coli, illustrated a contribution of receptor activation to
RIP140 recruitment. A constitutive supershift of the DR4-bound
OR1/RXR heterodimer upon addition of RIP140-iad could be shown to be
dependent on the AF-2 motif of OR1 but not RXR. Moreover, addition
of 9cRA resulted in a complex of further decreased mobility, which was
RXR-AF-2 dependent. The latter observation also indicates that the
heterodimer can be activated in at least two steps with respect to
RIP140 interaction and raises the interesting possibility that two
molecules of RIP140 might be recruited by a fully activated receptor.
It should be pointed out, however, that wild-type RIP140 contains at
least two nuclear receptor interaction domains (53), only one of which
was studied here (cf. Fig 1A), and which in a natural situation might
bind simultaneously to one or two receptor LBDs.
Our in vitro studies indicate that binding of fragments of
RIP140 and SRC-1 to an OR1/RXR heterodimer is mutually exclusive and
that the two factors therefore compete for binding, which is in line
with the model proposed by Treuter et al. (45) based on
studies on RXR/PPAR (peroxisome proliferator-activated
receptor)-heterodimers. As discussed above, the competing factors SRC-1
and RIP140 display rather different sensitivity toward dimerization and
ligand activation, and it therefore seems likely that the
OR1/RXR-heterodimer will differentially recruit these two (and
other) coregulators depending on its specific activation state.
The OR1/RLD-1 heterodimer with RXR is the only nuclear receptor
complex known to date that can be activated in three different ways, by
two classes of ligands from different metabolic pathways and in a
ligand-independent fashion. Thus, its net transcriptional potential
will, on the one hand, depend on the occurrence of dimerization-induced
activation (the exact prerequisites of which remain unresolved) and the
availability of ligand. On the other hand, transcriptional activation
and ligand sensitivity will be governed by the relative availability of
coregulatory factors supporting or inhibiting transcriptional
activation by the OR1/RXR heterodimer in its respective activation
state. For example, the nuclear receptor coactivator SRC-1 interacts
with the OR1/RXR-complex in all three activation states, and more
strongly when activated by both dimerization and ligand. However, in
the presence of RIP140, which appears to be a much weaker coactivator
if a coactivator at all (43, 45), the two factors will compete for
binding to the dimerization-activated heterodimer. Upon ligand
addition, binding of RIP140 appears to be enhanced to a much lesser
extent than that of SRC-1, leading to an alteration in the balance of
binding of the two factors in favor of SRC-1. Thus, in the presence of
sufficient amounts of RIP140, the relative effect of ligand-
vs. dimerization-activation would be enhanced. Similarly,
different corepressors might require different activation signals for
displacement and thereby suppress coactivator recruitment by one
pathway until they are released upon activation by another. In fact,
the nuclear receptor corepressors, N-CoR (64) and SMRT (65), can be
recruited by the OR1-LBD in vitro equally efficiently as by
TR-LBD (our unpublished results). The relative expression levels of
both receptors and coregulators are therefore likely to regulate
signaling via OR1 and RXR in a highly complex fashion.
Indeed, such mechanisms might account for the fact that in transient
transfection studies in which LXR mediates 9cRA-induction of
transcriptional activation due to low endogenous amounts of RXR,
constitutive activation is only observed upon RXR cotransfection
(23, 24, 54), which results in a higher number of
LXR/RXR-heterodimers for which coregulators would have to
compete. In the rat, OR1 is a ubiquitously expressed nuclear receptor
(20, 21), while expression of its paralog LXR is more restricted
(22, 23), but little is known so far about the role and regulation of
expression levels of these two nuclear receptors. Different paralogous
variants of RXR show different tissue-specific expression patterns
(66), but since RXR serves as a heterodimerization partner for a large
number of nuclear receptors, its relative availability cannot be
concluded from expression levels alone.
As a side aspect of this study, we have obtained data relevant to
previous reports on allosteric effects in OR1/RXR
heterodimerization. In the case of OR1/RXR, dimerization leads to
allosteric activation of one partner (28). A related variant of
allosteric control, the "phantom ligand effect" (67), has been
described in which ligand binding by one heterodimer partner (RXR)
leads to transcriptional activation mediated by the AF-2 motif of the
other (RAR). Similarly, it was proposed for a heterodimer of the OR1
paralog LXR with RXR that the 9cRA-response is dependent not on
H12 of RXR itself, but of the partner LXR (55). This conflicted
with our own conclusions (28) that the AF-2 motif of RXR is required
for retinoid signaling in RXR heterodimers with a rat LXR-homolog
(RLD-1) or OR1. In the present study, we have shown in transient
transfections that the RXR-AF-2 point mutant used in the study on
LXR (55) is leaky with respect to 9cRA-dependent transcriptional
activation by RXR-homodimers and is thus not a stringent tool with
which to exclude a role of RXR-AF-2 in heterodimers with LXR/OR1.
A second argument for the attribution of 9cRA-induced activation to the
AF-2 of LXR/OR1 is the observation (54) that its mutation abolishes
9cRA-induced transcriptional activation by the RXR-heterodimer (see
also Ref. 28 and this study). In accordance with this, we have shown
here that OR1-H12 mutation also abolishes 9cRA-induction of binding of
an SRC-1 fragment to DR4-bound OR1/RXR in an EMSA. However, the
adverse effect of LXR-/OR1-H12 mutation on the 9cRA-responsiveness
of RXR is reminiscent of the abolition of RXR ligand
permissiveness upon heterodimerization with the retinoic acid receptor
(RAR) (Refs. 15, 16 ; see also Ref. 17). This provides the
alternative explanation that the OR1-AF-2 does not itself mediate
9cRA-induction of the heterodimer, but that an OR1-H12 mutant
adventitiously resembles RAR in that it imposes a conformation upon
RXR in which RXR is nonpermissive for activation by its retinoid
ligand. In summary, the question of whether LXR/OR1 is activated
upon 9cRA-binding to RXR remains unresolved. However, in view of the
fact that the dimerization-activated OR1/RXR-complex can be
superactivated by either ligand (Ref. 28 and our unpublished
results) and that activation by both ligands is more than additive
(24), it seems unlikely that the OR1-H12 should mediate activation by
all three signals.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
(68)
were expressed from a slightly modified vector pGEM-3Z (Promega Corp., Madison, WI) containing the complete open reading frame
preceded by an improved Kozak region. Receptor expression in CV-1 cells
was from vector pCMV5. GAL4-OR1, GAL4-OR, and GAL4-RXR have been
described previously (28). All H12-mutants (see Fig. 1B) were obtained
by PCR, and sequences of resultant clones were confirmed by sequencing.
pmT-TR was a gift of Stefan Nilsson (KaroBio, Huddinge,
Sweden).
pCMV5-SRC-1 was derived from a gift of B.W. O’Malley (i.e.
as described in Ref. 42 but in vector pCMV5). pCMV5-SRC-iad comprised a
fragment thereof encoding amino acids 216–394, fused in-frame to an
initiator codon and ha-epitope as in pBK-CMV-ha. hRIP140-N was
encoded by a DNA fragment isolated in the laboratory in a yeast
two-hybrid screen using PPAR-LBD as a bait against a human liver
cDNA library (45) and expressed from pBK-CMV-ha, which contains the
sequence
GATCCGCCGCCACCATGGATTACCCATACGACGTCCCAGACTACGCTCAGATCTCCGAATT-CGC
GC inserted into BamHI/XhoI, providing an
initiator codon and a favorable surrounding sequence as well as
encoding a hemagglutinin epitope (ha, underlined).
pBK-CMV-ha-RIP140-N comprised amino acids 431-1158 of human RIP140
(43).
UASx4-TK-LUC has been described previously (19). 3xDR4-TK-LUC and 3xDR1-TK-LUC contained three copies of the sequences GACCGTTCTGGGTCACGAAAGGTCA- AGCGC and GACCGTTCTAAAGGTCAAAGGTCAAGTGGC, respectively, in vector Rsr-TK-LUC. Rsr-TK-LUC is a derivative of pGL3-Basic (Promega Corp.) containing an RsrII site before HindIII in the polylinker, and a HindIII/BglII fragment of the TK-promoter of UASx4-TK-LUC cloned into HindIII/NcoI.
Plasmids for Protein Overexpression in E. coli
GST-fusion proteins were expressed from pGEX-4T (Pharmacia Biotech, Piscataway, NJ), His6-fusion proteins
(except full-length OR1, RXR-N; see below) were expressed from
modified pET-15b, and the His10-RIP140-iad fusion were
expressed from pET-19b (both Novagen, Madison, WI). pET-RXR-lbd
comprised amino acids 203–467 of rat RXR (see Fig. 1B for the
mutant). The fragment used for pGEX-SRC-iad was as described above.
pET-RIP140-iad comprised amino acids 747-1158 of human RIP140.
His6-tagged full-length OR1 was expressed from modified
pET-3a (Novagen) that encoded the sequence MHHHHHHIEGR preceding OR1.
His6-tagged RXR-N was expressed accordingly, but lacked
codons 11–111 of RXR.
All vector constructs were made by PCR-based amplification of the respective fragments including introduction of convenient restriction sites, and subsequent assembly; the PCR-derived sequence parts were then either replaced by template material or confirmed by sequencing.
Transfection and Preparation of Cell Extracts
CV-1 cells were split into plates of 3.6 mm diameter to achieve
approximate confluency at the time of harvest. DMEM was used containing
10% FCS, 100 U/ml penicillin, and 100 µg/ml streptomycin at 37 C in
5% CO2. After 24 h, cells were transfected with DOTAP
(Boehringer Mannheim, Indianapolis, IN) according to the
instructions provided by the supplier; transfection mixes contained 200
ng of corresponding constructs in expression vector pCMV-5 (if not
indicated otherwise) and 1 µg of the respective TK-LUC reporter.
Twelve hours later, the medium was exchanged for fresh medium with or
without inducers. 9cRA was used at 0.1 µM, T3
was used at 1 µM final concentration, and corresponding
controls received an equal amount of solvent. Cells were harvested at
36 h after transfection; results are given as means of three
independent experiments.
EMSA
The oligonucleotide probe
(GACCGTTCTGGGTCACGAAAGGTCAAGCGC/GTCGCGCTTGACCTTTCGTGACCCAGAACG
for OR1 or TR with RXR;
AGCTTACTAGTTCGGAGGACAGTCCTCCGTCTAGAGCT/CTAGACGGA-
GGACTGTCCTCCGAACTAGTA for GAL4-chimera) was labeled using DNA
polymerase Klenow fragment and [-32P]dCTP and purified
on Sephadex G-50 NICK columns (Pharmacia Biotech). Binding
mixes contained 1–4 µl of in vitro translated protein
(T7-TNT coupled transcription/translation system, Promega Corp.) as well as appropriate additives [final concentration 50
mM KCl, 5 mM MgCl2, 20
mM HEPES, pH 7.5, 200 µg/ml BSA, 4 mM
spermidine, 8% glycerol, 50 µg/ml poly(dI-dC)·poly(dI-dC), 75
µg/ml salmon sperm DNA; additionally 20 µM
ZnCl2 for UAS-binding, 1 mM EDTA, 1
mM dithiothreitol (DTT), and 0.01% Triton X-100 for
DR4-binding]. Candidate interacting proteins and ligands or solvent
were added where indicated (1/20 volume 9cRA in dimethylsulfoxide
(DMSO) to yield 50 µM, 1/20 volume 22(R)-HC to
yield 250 µM, 1/20 volume T3 in PBS to yield
50 µM). After 20 min preincubation on ice, approximately
10 fmol (20,000 cpm) labeled probe were added. Five percent
polyacrylamide gels were prepared and run in 0.25x TBE (22.5
mM Tris-borate, 0.5 mM EDTA).
Expression and Purification of GST- and His-Tagged
Proteins
E. coli BL21(DE3)LysS cells carrying the appropriate
expression vectors (except pET-3a-His6-OR1,
pET-3a-His6-RXR-N, see below) were grown at 30 C in
LB-medium containing 0.5% casamino acids, 0.5% glucose, 100 µg/ml
ampicillin, and 34 µg/ml chloramphenicol until they reached an
OD600 = 0.5, and were then induced with 1
mM isopropylß-D-thioglactopyranoside for 100
min. Cells were harvested by centrifugation and frozen on dry ice and
then thawed in 12 ml STE (10 mM Tris/HCl pH 8.8, 150
mM NaCl, 1 mM EDTA) with 0.2 mM
PMSF and 1 mM DTT freshly added. Lysozyme (0.1 mg/ml) was
added, and lysis was allowed to occur under rotation at 4 C for 20 min.
After addition of 2.1 ml 10% sarkosyl and 100 µl 0.5 M
EDTA, the lysate was homogenized by drawing it into a syringe using
needles with a diameter of 1.2 mm (twice) and 0.6 mm (twice) and pulse
sonicated three times for 20 sec at medium intensity and 40%
duty cycle. The homogenates were then centrifuged at 24,000 x
g for 20 min at 4 C, and the supernatants were frozen and
stored at -80 C. Note that for preparation of His-tagged proteins, DTT
and EDTA were omitted and the final concentration of sarkosyl was
reduced to 1%. His-tagged proteins were purified on TALON affinity
resin (CLONTECH Laboratories, Inc., Palo Alto, CA), and
GST-tagged proteins were purified on glutathione-Sepharose 4B
(Pharmacia Biotech) in batch procedure according to the
instructions provided by the suppliers. Purity and stability of the
purified proteins were determined by SDS-PAGE analysis and subsequent
Coomassie blue staining. Protein concentrations were determined by
Bradford-assay (Bio-Rad Laboratories, Inc., Richmond,
CA).
For His6-OR1 and His6-RXR-N, the transformed
bacterial cells were grown in M9 minimal medium containing 1% casamino
acids, 0.002% thiamin B1, 300 µg/ml ampicillin, and 72 µg/ml
chloramphenicol until they reached an approximate
OD600 = 0.6, and then induced with 1 mM
isopropylß-D-thioglactopyranoside and harvested after
2–3 h. Cell pellets were resuspended in buffer A (20 mM
Tris/HCl, pH 8.0, 100 mM NaCl, 1 mM PMSF) and
broken in a French Press. Cell debris was sedimented by 15 min
centrifugation at 16.000 x g, and the supernatant was
applied on a TALON affinity column (CLONTECH Laboratories, Inc.) run on a Biological Chromatography System (Bio-Rad Laboratories, Inc.). The column was washed with 5–10
mM imidazole, and the protein was eluted in a linear
imidazole gradient (0–500 mM imidazole in buffer A).
Fractions were analyzed on a protein gel, and pure fractions were
combined, ammonium precipitated, and dialyzed against 20 mM
Tris/HCl, pH 8.0, 100 mM KCl, 50% glycerol, 1
mM PMSF before storage at -80 C.
In Vitro Protein-Protein Interaction Assay (GST
Pull-Down)
Sepharose beads were pretreated according to the instructions of
the manufacturer. GST-fusion proteins were bound to
glutathione-Sepharose for 2 or more hours at 4 C under rotation (100
µl extract, 10–20 µl beads per sample) and subsequently washed
three times in pull-down buffer (PDB: 50 mM
KPi, pH 7.4, 100 mM NaCl, 1 mM
MgCl2, 10% glycerol, 0.1% Tween 20). For studies on
heterodimeric receptor LBDs, 25 µg purified His6-tagged
RXR-LBD were added to the loaded beads and allowed to bind for 1 h
or longer at 4 C under rotation. The supernatant was then replaced by
200 µl PDB + 1.5% BSA (fraction V, Sigma Chemical Co.,
St. Louis, MO). For the competition experiment, 50 µg of purified
His10-tagged RIP140-iad were allowed to prebind to the
dimers for 45 min at this stage. One to 3 µl of in vitro
translates (TNT-coupled transcription/translation system, Promega Corp.) of the constructs to be analyzed were then added per
sample and incubated for about 2 h as before. At the same time,
ligand or corresponding solvent was added where indicated (1/100
volume; 9cRA in DMSO to yield 10 µM, 22(R)-HC
in DMSO to yield 50 µM, or 10 µM
T3 in PBS to yield 10 µM). Finally, samples
were washed three times in PDB, and the beads were collected by
centrifugation, boiled for 5 min in a total volume of 60 µl 1x
SDS-PAGE buffer and one fourth were subjected to SDS-PAGE. Equal
protein loads were confirmed by Coomassie staining before
autoradiography.
|
ACKNOWLEDGMENTS |
|---|
. |
FOOTNOTES |
|---|
This study was supported by Grant 13X-2819 from the Swedish Medical Research Council .
1 Current address: Eberhard-Karls-Universität Tübingen,
Institut für Zellbiologie, Abteilung Molekularbiologie, Auf der
Morgenstelle 15, 72076 Tübingen, Germany. 
2 Current address: University of Oslo, Department of Biochemistry,
P.O. Box 1041 Blindern, N-0317 Oslo, Norway.
Received for publication June 22, 1998. Revision received March 12, 1999. Accepted for publication March 16, 1999.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
: binding and activation correlate with antidiabetic
actions in db/db mice. Endocrinology 137:4189–4195[Abstract]
ligand-binding domain
bound to all-trans retinoic acid. Nature 378:681–689[CrossRef][Medline]
. Nature 383:728–731[CrossRef][Medline]
. Cell 93:693–704[CrossRef][Medline]
and orphan nuclear receptor OR1 reveals
dimerization-induced activation as a novel mechanism of nuclear
receptor activation. Mol Cell Biol 17:3977–3986[Abstract]
