Review Nuclear retinoid receptors and the transcription of retinoid-target genes Julie Bastien, Ce ´cile Rochette-Egly * Institut de Ge ´ne ´tique et de Biologie Mole ´culaire et Cellulaire, CNRS/INSERM/ULP, UMR 7104, 1 rue Laurent Fries, BP 10142, Illkirch Cedex 67404, France Received 14 October 2003; accepted 2 December 2003 Received by A.J. van Wijnen Abstract The pleiotropic effects of retinoids are mediated by nuclear retinoid receptors (RARs and RXRs) which are ligand-activated transcription factors. In response to retinoid binding, RAR/RXR heterodimers undergo major conformational changes and orchestrate the transcription of specific gene networks, through binding to specific DNA response elements and recruiting cofactor complexes that act to modify local chromatin structure and/or engage the basal transcription machinery. Then the degradation of RARs and RXRs by the ubiquitin–proteasome controls the magnitude and the duration of the retinoid response. RARs and RXRs also integrate a variety of signaling pathways through phosphorylation events which cooperate with the ligand for the control of retinoid-target genes transcription. These different modes of regulation reveal unexpected levels of complexity in the dynamics of retinoid-dependent transcription. D 2004 Elsevier B.V. All rights reserved. Keywords: Retinoids; Nuclear receptors; Transcription; Degradation; Kinases; Phosphorylation; Ubiquitin– proteasome 1. Introduction Vitamin A and its active derivatives referred to as retinoids are non-steroid hormones which play a critical role in the development and homeostasis of virtually every vertebrate tissues through their regulatory effects on cell differentiation, proliferation and apoptosis (Ross et al., 2000; Altucci and Gronemeyer, 2001a). It has long been established that retinoids exert their action by regulating the expression of specific subsets of genes within target tissues. However, it is only during the last 15 years that the understanding for retinoids action rapidly increased, subse- quently to the cloning of nuclear retinoid receptors and the identification, within the promoters of retinoid-responsive genes, of elements exhibiting a high affinity for these receptors (for review, see (Chambon, 1996; Laudet and Gronemeyer, 2001), and references therein). Then these nuclear receptors have been shown to work as ligand- activated transcription activators in a spatiotemporal specific manner during embryonic development. During the last decade, the molecular rationale for retinoid receptors action has been facilitated by the identi- fication of the DNA- and ligand-binding domains (DBD and LBD, respectively) (Chambon, 1996), and by the determination of their crystal structure (Renaud and Moras, 2000). Moreover, a number of studies demonstrated that they have to contend with repressive chromatin structures in order to activate gene expression. Indeed, as most target genes are initially silent and packed in a dense chromatin structure, liganded retinoid receptors recruit a battery of intermediary proteins, including coactivators, chromatin remodellers and modifyers which act in a coordinated and/or combinatorial manner to decompact chromatin and direct RNA polymerase II (RNA Pol II) and the general transcription factors (GTFs) to the promoter (Dilworth and Chambon, 2001). Then an important question was what happens after the retinoid-activated receptors have bound their DNA reponse 0378-1119/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.gene.2003.12.005 Abbreviations: RAR, retinoic acid receptor; RXR, retinoid X receptor; RA, retinoic acid; DBD, DNA-binding domain; LBD, ligand-binding domain; RARE, retinoic acid response element; HAT, histone acetyltrans- ferase; HMT, histone methyltransferase; HDAC, histone deacetylase; CBP, CREB-binding protein; Cdk, cyclin-dependent kinase; TBP, TATA-binding protein; GTF, general transcription factor; MAPK, mitogen-activated protein kinase; P13K, phosphoinositide 3-kinase; PTEN, phosphatase and tensin homolog; PPAR, peroxisome proliferator activated receptor; LXR, liver X receptor; APL, acute promyelocytic leukemia. * Corresponding author. Tel.: +33-3-88-65-34-59; fax: +1-33-3-88-65- 32-01. E-mail address: [email protected] (C. Rochette-Egly). www.elsevier.com/locate/gene Gene 328 (2004) 1 – 16
16
Embed
Review Nuclear retinoid receptors and the transcription of
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
www.elsevier.com/locate/gene
Gene 328 (2004) 1–16
Review
Nuclear retinoid receptors and the transcription of retinoid-target genes
Julie Bastien, Cecile Rochette-Egly*
Institut de Genetique et de Biologie Moleculaire et Cellulaire, CNRS/INSERM/ULP, UMR 7104, 1 rue Laurent Fries, BP 10142, Illkirch Cedex 67404, France
Received 14 October 2003; accepted 2 December 2003
Received by A.J. van Wijnen
Abstract
The pleiotropic effects of retinoids are mediated by nuclear retinoid receptors (RARs and RXRs) which are ligand-activated transcription
factors. In response to retinoid binding, RAR/RXR heterodimers undergo major conformational changes and orchestrate the transcription of
specific gene networks, through binding to specific DNA response elements and recruiting cofactor complexes that act to modify local
chromatin structure and/or engage the basal transcription machinery. Then the degradation of RARs and RXRs by the ubiquitin–proteasome
controls the magnitude and the duration of the retinoid response. RARs and RXRs also integrate a variety of signaling pathways through
phosphorylation events which cooperate with the ligand for the control of retinoid-target genes transcription. These different modes of
regulation reveal unexpected levels of complexity in the dynamics of retinoid-dependent transcription.
in the LBD (Bourguet et al., 1995; Renaud et al., 1995;
Wurtz et al., 1996; Moras and Gronemeyer, 1998; Egea et
al., 2001), including repositioning of helix H11 in the
continuity of H10. The most striking effect is the swinging
of helix 12 which moves in a «mouse trap model», being
tightly packed against H3 and H4 (Fig. 3B). H12 makes
direct contacts with the ligand and seals the «lid»of the
ligand-binding pocket, further stabilizing ligand binding.
Simultaneously, the Omega loop flips over underneath H6,
carrying along the N-terminal part of H3. According to
recent studies, ligand binding would be facilitated by
cellular retinoic acid-binding protein II (CRABPII) which
upon shuttling into the nucleus and interaction with RAR/
RXR heterodimers (Delva et al., 1999) channels retinoic
acid to RARs (Budhu and Noy, 2002). This interaction of
RARs with CRABPII is stabilized by cyclin D3 (Despouy et
al., 2003).
The ligand-induced conformational changes favor the
interactions between RAR and RXR and therefore increase
their DNA affinity (Rastinejad et al., 2000; Depoix et al.,
2001). They also cause corepressor release and create a new
hydrophobic cleft formed between H3, H4 and H12 which
constitutes a surface where coactivators can bind (Fig. 5B).
According to recent studies, the surfaces involved in core-
pressor and coactivator binding partially overlap and the
ligand-induced repositioning of H12 would result in a
switch from a corepressor to a coactivator-binding surface
(Nagy et al., 1999; Perissi et al., 1999; Glass and Rosenfeld,
2000). Within RAR/RXR heterodimers bound at DR5
elements, though both liganded partners are theoretically
able to recruit coactivators, RXR is «subordinated» to its
RAR partner (Roy et al., 1995; Willy and Mangelsdorf,
Fig. 5. Three-step mechanism of retinoid receptor action. (A) In the absence of ligand, retinoid receptors bound to response elements located in the promoter of
target genes are associated with histone deacetylase-containing (HDAC) complexes tethered through corepressors and repress transcription. (B) Upon ligand
binding, the corepressors dissociate, allowing the recruitment of coactivators associated with complexes displaying histone acetyltransferase (HAT),
methyltransferase, kinase or ATP-dependent remodeling (SWI/SNF) activities that decompact repressive chromatin. (C) In the third step, the coactivators
dissociate and the SMCC mediator complex assembles. Then the mediator expedites entry of the RNA Pol II and the general transcription factors to the
promoter, resulting in transcription initiation.
J. Bastien, C. Rochette-Egly / Gene 328 (2004) 1–166
1999). This phenomenon has been attributed to the fact that
liganded RXR cannot dissociate corepressors and therefore
coactivators cannot be recruited (Germain et al., 2002).
However, in the presence of both RAR and RXR ligands,
there is synergy originating from the RAR agonist-induced
dissociation of corepressors and the subsequent cooperative
binding of coactivators to the two partners. Most impor-
tantly, the two partners synergize with each other for
transcription, not only through their AF-2 domains but also
through their AF-1 domains (Gianni et al., 2003), very
likely via their cooperation for the recruitment of coactiva-
tors (Bommer et al., 2002).
The first identified family of ligand-recruited retinoid
receptors coactivators is the SRC/p160 family which
includes SRC-1/NCoA-1, TIF-2/GRIP-1/NCoA-2/SRC-2
and pCIP/ACTR/AlB1/TRAM1/RAC3/SRC-3 (Chen,
2000; McKenna and O’Malley, 2002). Other coactivators,
p300/CBP (Vo and Goodman, 2001) and CARM-1 (Chen et
al., 1999a; Kouzarides, 2002), have been characterized, but
they are structurally and functionally distinguishable from
the SRC/p160 family. A recurring structural feature of all
these coactivators is a highly conserved alpha-helical
LxxLL motif (where L is leucine and x is any amino acid)
from a single to several copies, which is implicated in their
J. Bastien, C. Rochette-Egly / Gene 328 (2004) 1–16 7
ligand-dependent recruitment by the AF-2 domain of reti-
noid receptors. Cocrystal structures indicate that two LxxLL
motifs from a single p160 coactivator molecule interact with
the AF-2 domains of both partners. However, one molecule
of coactivator can be cooperatively recruited by each
member of the heterodimer. Most importantly, the coacti-
vators also contain domains which interact with other
coactivators. Accordingly, TIF2 possesses one domain
interacting with CBP/p300 and a second one which has
been recently shown to interact with CARM-1 (McKenna
and O’Malley, 2002).
p300/CBP and, to a lesser extent, the members of the
p160 family locally modify chromatin structure through
their histone acetyltransferase (HAT) activity which acety-
lates lysine residues located at the N-terminal tails of
histones, thereby weakening the interaction of the N-termi-
nal tails with the nucleosome DNA (Fig. 5B). Other
coactivators such as CARM-1 act through their histone
methyltransferase (HMT) activity which upon methylation
of specific arginine or lysine residues also change histone–
DNA and histone–histone contacts. Then the opening of the
chromatin environment is achieved by the recruitment
through pCIP and CBP, of larger complexes with histone
acetyltransferase (PCAF) or histone methyltransferase ac-
tivities (Kouzarides, 2000; Roth et al., 2001; Zhang and
Reinberg, 2001). Note that the efficiency of histone acety-
lation and methylation is regulated upon phosphorylation of
the nearby serines residues by associated kinases (Cheung et
al., 2000; Lo et al., 2001). Altogether, these histone mod-
ifications create tags or binding sites that form an «histone
code» read by a specialized bromodomain present in the
chromatin modifiers (Jeanmougin et al., 1997; Strahl and
Allis, 2000; Berger, 2002). This code would coordinate the
recruitment of additional HATs or HMTs for further chro-
matin decompaction. It would also allow the recruitment of
ATP-dependent chromatin remodelers (SWI/SNF) which
use the energy of ATP hydrolysis to reposition nucleosomes
at the promoter through sliding them in cis or displacing
them in trans, allowing the formation of nucleosome-free or
nucleosome-spaced regions (Kingston and Narlikar, 1999;
Vignali et al., 2000; Narlikar et al., 2002).
5. Third step: recruitment of the transcriptional
machinery
Once repressive chromatin has been decondensed, it has
been proposed that a coregulators exchange occurs, in order
to allow the RARE-bound heterodimers to participate in the
entry of RNA-Pol II and GTFs into the preinitiation com-
plex (Chen et al., 1999b; Malik and Roeder, 2000). The
current working hypothesis is that the p160 coactivators
dissociate, subsequent to their acetylation which decreases
their ability to interact with the receptors (Chen et al.,
1999b), or to their degradation by the proteasome (Yan et
al., 2003). Then the retinoid receptors become able to recruit
the transcription machinery via their association with the so-
called SMCC (Srb and Mediator protein containing com-
plex) mediator complex (Malik and Roeder, 2000; Dilworth
and Chambon, 2001; Woychik and Hampsey, 2002). The
subunit of the mediator complex that is responsible for
interaction with the AF-2 domain of liganded retinoid
receptors was identified as DRIP205 which is identical to
TRAP220 and contains a LxxLL nuclear receptor box motif.
Whether other subunits interact with the N-terminal AF-1
domain of RARs and RXRs, as described for the glucocor-
ticoid receptor (Hittelman et al., 1999), remains to be
determined.
Then the mediator expedites entry of the transcriptional
machinery to the promoter (Fig. 5C) through its interaction
(via other subunits) with the RNA Pol II holoenzyme
(Woychik and Hampsey, 2002). This process also involves
the six GTFs (Orphanides et al., 1996). The large multi-
subunit TFIID, which binds to the promoter through its
TBP, possesses associated factors or TAFIIs developing
kinase and acetylase activities (TAFII250). This will in-
crease chromatin remodeling at the promoter to permit tight
binding of the basal transcriptional machinery. Some TAFIIs
(May et al., 1996; Lavigne et al., 1999) and TFIIH also
interact with retinoid receptors (Rochette-Egly et al., 1997),
thus increasing the efficiency of the preinitiation complex
assembly (see below). Finally, the recruitment of GTFs is
also enhanced by p300/CBP and by components of the SWI/
SNF complex associated to the RNA Pol II holoenzyme
(Adelman and Lis, 2002; Orphanides and Reinberg, 2002).
Once transcription has been initiated, RNA Pol II traffics
along the gene to be transcribed. This process involves
chromatin remodeling and modifying activities endowed by
subunits of the elongation factors that track with elongating
RNA Pol II (Orphanides and Reinberg, 2000). Finally, the
equilibrium will shift in favour of histone tail deacetylation,
and methylation at residues leading to the rapid conversion
of chromatin to a repressed conformation.
In summary, in a context of chromatin where the nucle-
osomes do not impede the binding of RAR/RXR hetero-
dimers to their DNA recognition sequences (Dilworth and
first coactivators and HAT complexes resulting in histone
acetylation. Then ATP-dependent remodeling complexes are
recruited, leading to the displacement of impeding nucleo-
somes within the proximal promoter region, thus facilitating
access of the general transcription machinery to the pro-
moter. However, it cannot be excluded that the reverse order
of events, which is the recruitment of ATP-dependent
remodelers before HATs, could occur in a context of highly
condensed chromatin (Cosma, 2002). Therefore, the relative
timing and order of recruitment of chromatin modifyers and
remodellers would depend upon the nature of the promoter
and the chromatin structure in which it resides (Fry and
Peterson, 2001) in order to allow the most efficient solution
(Aalfs and Kingston, 2000; Cosma, 2002), each player
helping the other and each step facilitating another one.
J. Bastien, C. Rochette-Egly / Gene 328 (2004) 1–168
The key is that the appropriate end stage, e.g. a properly
decondensed chromatin with a functional preinitiation com-
plex posed for transcription, be attained in a timely manner.
6. Control of RAR/RXR transactivation by the
ubiquitin–proteasome system
In recent years, it has become evident that the transcrip-
tional activity of retinoid receptors, as that of most tran-
scription factors, is also regulated by the ubiquitin–
proteasome pathway. Paradoxically, both the proteolytic
and non-proteolytic activities of this system appear to
modulate transcription at different levels (Ferdous et al.,
2001; Salghetti et al., 2001; Tansey, 2001; Conaway et al.,
2002; Muratani and Tansey, 2003).
One main role of the ubiquitin–proteasome system is to
degrade transcriptional activators. In this process, following
a signal, the substrate protein is multi-ubiquitylated at a
lysine group and then targeted for destruction by the 26S
proteasome. The 19S subcomplex of the proteasome recog-
nizes the multi-ubiquitylated substrate, removes the ubiq-
uitin groups, unfolds the substrate and feeds the resulting
unstructured chain into the 20S catalytic core of the protea-
some where it is degraded (DeMartino and Slaughter, 1999).
It has been recently demonstrated that, within RAR/RXR
heterodimers bound at response elements, both partners are
degraded by the proteasome in response to retinoids (Zhu et
al., 1999; Boudjelal et al., 2000; Kopf et al., 2000; Osburn et
al., 2001; Tanaka et al., 2001; Gianni et al., 2002a, 2003).
This process involves the ubiquitylation of RARs (Zhu et
al., 1999; Kopf et al., 2000) and the recruitment of the
proteasome at the AF-2 domain through SUG-1 (Gianni et
al., 2002a) which is one of the six ATPases in the base of the
19S regulatory complex of the 26S proteasome. It has been
proposed that this degradation process would provide a
mechanism to control the magnitude and the duration of
retinoid-mediated transcription.
The importance of the ubiquitin–proteasome system in
retinoid receptors transactivation came from the particular
case of the RARg isotype (Gianni et al., 2002a). Indeed,
blocking either the ubiquitin or the proteasome systems
abrogates not only the degradation of RARg, but also
RARg-mediated transcription. The paradoxal mechanism
of how the ubiquitin–proteasome system regulates RARg/
RXR transcriptional activity is currently unknown but
several lines of evidence indicate that the ubiquitin ligases
(Imhof and McDonnell, 1996; McKenna et al., 1998), as
well as some components of the proteasome system, such as
SUG-1 (vom Baur et al., 1996), are able to bind retinoid
receptors. In addition, ubiquitin ligases belong to complexes
that are integral components of the mammalian mediator
complex associated to the Pol II transcription machinery
(Brower et al., 2002; Conaway et al., 2002) and SUG-1 also
interact with the general transcription factor TFIIH (Fraser
et al., 1997; Weeda et al., 1997; Sandrock and Egly, 2001).
Finally, the 19S subcomplex of the proteasome has been
shown to associate with transcription activators (Gonzalez et
al., 2002) and to participate to elongation (Ferdous et al.,
2001). Therefore, the ubiquitin–proteasome machineries
may play a dual role, controlling on the one hand the
functionality of RARg/RXR heterodimers through helping
the recruitment of the transcription machinery (Lin et al.,
2002), and on the other hand the ubiquitylation and the
subsequent degradation of the heterodimers. Such a dual
role may regulate the dynamic assembly/disassembly of
retinoid receptors to the promoter of the target genes, as
recently demonstrated for other nuclear receptors such as the
estrogen and androgen receptors (Freeman and Yamamoto,
2002; Kang et al., 2002; Reid et al., 2003).
It must be noted that the same conclusions could not be
made for the other RAR isotype RARa, since inhibition of
the proteasome by specific inhibitors did not abrogate, but
amplified RARa-mediated transcription (Gianni and Roch-
ette-Egly, unpublished observations), as described for the
glucocorticoid receptor (Wallace and Cidlowski, 2001).
Why the two isotypes RARa and RARg are not regulated
similarly will require further investigations. Finally, the
ubiquitin–proteasome system also targets histones and other
coregulators (Wang et al., 2002; Yan et al., 2003), therefore
increasing the complexity of retinoid-dependent transcrip-
tional control (Muratani and Tansey, 2003).
7. Regulation of RAR/RXR-mediated transcription
through phosphorylation
RARs and RXRs are substrates for a multitude of kinases
(see Figs. 1 and 6) (Rochette-Egly, 2003). Importantly,
subsequent to their interaction with TFIIH, RARs (RARa
and RARg) are phosphorylated in their N-terminal A/B
region by the cdk7 subunit of TFIIH which has a cyclin-
dependent kinase activity (Rochette-Egly et al., 1997; Bas-
tien et al., 2000). This phosphorylation process which has
been extensively studied, especially in the case of RARa,
plays a critical role in the retinoid response. Indeed, upon
mutations in a TFIIH subunit resulting in an incorrect
positioning of the cdk7 kinase relative to its substrate,
RARa is underphosphorylated and retinoid-dependent tran-
scription is decreased (Keriel et al., 2002). If phosphoryla-
tion of the AF-1 domain by TFIIH occurs when the GTFs
are recruited at the promoter, the hypothesis that the
phosphorylation of the AF-1 domain helps the ligand-
dependent recruitment of coactivators and chromatin mod-
ifyers would be rather elusive. Instead, phosphorylation
could facilitate the recruitment of components of the tran-
scription machinery and therefore stabilize the formation of
the NR transcription complex. Future efforts will undoubt-
edly reveal new «coregulators» interacting with the phos-
phorylated motif of RARa and therefore regulating RARa-
mediated transcription. However, it is not excluded that
phosphorylation might rather facilitate the dissociation of
Fig. 6. Signaling pathways activating MAP kinases (Erks, JNKs and p38), PI3K, Akt, PKA and PKC are involved in the control of retinoid-mediated
transcription. Nuclear retinoid receptors are targeted by phosphorylations in response to signaling pathways. These phosphorylations may modulate their
potentiality to recruit cofactors associated with the trascription machinery. The coactivators are also phosphorylated, modulating their activity and/or their
potentiality to recruit the chromatin modifying (HAT, HMT) and remodeling complexes. Finally, histones, the general transcription factors and the RNA
Pol II can be also phosphorylated. RTK, receptor tyrosine kinase; PI3K, phosphatidylinositol 3-kinase; Erk, extracellular signal-related kinase; MAPK,
mitogen-activated protein kinase; MEKK, MAPK kinase kinase; ASK1, apoptosis-stimulating kinase 1; AC, Adenylate cyclase; JNK, Jun amino-terminal
kinase.
J. Bastien, C. Rochette-Egly / Gene 328 (2004) 1–16 9
RARa from transcription inhibitors or help the dissociation
of RARa from the transcription machinery in order to allow
elongation to proceed. Note that, in contrast to other
transcription activators, phosphorylation of the AF-1 do-
main of RARa does not influence the ubiquitylation and the
proteasomal degradation of RARa (Kopf et al., 2000).
It must be pointed out that in the particular case of the
RARg isotype, phosphorylation by TFIIH, though neces-
sary (Bastien et al., 2000), is not sufficient. Indeed, RARg
needs to be also phosphorylated at an additional nearby
residue by p38MAPK, subsequently to its activation by
retinoids (Gianni et al., 2002a,b). Phosphorylation of these
two residues is required for both the transactivation and the
degradation of RARg.
The critical role of RARg phosphorylation has been
further dissected in our group, by using F9 cells which
represent a cell-autonomous system for analyzing retinoid
signaling (for review, see Rochette-Egly and Chambon,
2001). In these cells, the retinoid signal is transduced by
RARg/RXR heterodimers and therefore the various RA
responses are abolished in RARg null cells. Taking advan-
tage that the RA responses can be restored upon reexpres-
sion of the receptor to wild-type levels, the same strategy
has been used with RARg mutated at the phosphorylation
sites located in the N-terminal AF-1 domain. It has been
demonstrated that the integrity of these phosphorylation
sites is indispensable to the activation of a subset of RA-
target genes, for RARg degradation and for RA-induced F9
cell differentiation (Taneja et al., 1997; Kopf et al., 2000;
Gianni et al., 2002a).
Though it is assumed that phosphorylation by both
TFIIH and p38MAPK is crucial for both the transcriptional
activity and the degradation of RARg, the paradoxal mech-
anism of how phosphorylation regulates these two processes
is currently unknown. Nevertheless, some speculative mod-
els can be proposed (see Fig. 7). As phosphorylation of the
Fig. 7. Recapitulation of the effects of phosphorylations on RARg-mediated transcription of target genes. The fraction of liganded RARg2 that is bound to
cognate response elements as heterodimers with RXRa is phosphorylated by the cdk7 subunit of TFIIH at one serine residue located in the AF-1 domain. Then
the RA-induced activity of p38MAPK leads to the phosphorylation of the second serine residue. Phosphorylation of these two serines is a checkpoint
controlling on the one hand transcription and on the other hand the degradation of RARg by the 26S proteasome. How? One can propose that phosphorylation
helps the association and/or the dissociation of coregulators. Indeed, phosphorylation may allow the recruitment of ubiquitin ligase complexes which will
regulate both the transcription and the degradation of RARg through their association to the transcription machinery and through signaling RARg
ubiquitylation, respectively. However, it is not excluded that phosphorylation induces the dissociation of inhibitors, therefore making the nearby lysine residue
available for ubiquitin ligases recruited at the AF-2 domain. In that context, it has been suggested that oligo-ubiquitylation would modulate transcription, while
poly-ubiquitylation signals the switch to the degradation by the 26S proteasome.
J. Bastien, C. Rochette-Egly / Gene 328 (2004) 1–1610
two residues is also required for retinoid-induced ubiquity-
lation of RARg (Gianni et al., 2002a), one can hypothesize
that phosphorylation signals the recruitment of ubiquitin–
ligase complexes which would play a dual role. Indeed,
such complexes might control on the one hand transcription,
through their association with the transcription machinery
(see above), and on the other hand RARg degradation
through its ubiquitylation. However, it is not excluded that
phosphorylation induces the dissociation of inhibitors,
therefore making the nearby lysine residue available for
ubiquitylation by the ubiquitin ligases recruited at the AF-2
domain. Note in that context that oligo-ubiquitylation could
modulate the activity of transcription activators (Salghetti et
al., 2001), while poly-ubiquitin chains signal the switch to
the degradation by the proteasome (Gonzalez et al., 2002).
In conclusion, it is tempting to speculate that, in the case
of RARg, phosphorylation would be a checkpoint, control-
ling on the one hand transcription and signaling and on the
other hand the degradation of RARg. Future efforts will
undoubtedly reveal new «coregulators» (ubiquitin ligases,
WW domains, F-box proteins) interacting with the phos-
phorylated or unphosphorylated AF-1 domain of RARg and
therefore regulating RARg functionality.
Interestingly, RARa and RARg transcriptional activities
can be also modulated upon phosphorylation by other
kinases in response to a variety of signals. Indeed, phos-
phorylation by PKA at serine 369, between H9 and H10,
modulates positively the transcriptional activity of RARa
(Rochette-Egly et al., 1995), very likely through helping
coregulators binding and/or DNA binding of RAR/RXR
heterodimers. Finally, phosphorylation of RARa DBD
(Delmotte et al., 1999) that occurs upon activation of PKC
signaling favors dimerization and subsequently DNA bind-
ing of RARa/RXR heterodimers.
Recently, the role of RXRa phosphorylation has been
also studied. Accordingly, phosphorylation of Ser22 located
in the N-terminal AF-1 domain is required for the activation
of a subset of target genes and for the antiproliferative effect
of retinoids (Bastien et al., 2002). Recently, it has been
found that three additional residues (Ser61, Ser75 and
Thr87) located in the same N-terminal domain can be also
phosphorylated in response to retinoids (Gianni et al., 2003)
and that this phosphorylation modulates the synergy be-
tween both heterodimeric partners for maximal transcrip-
tional activity, very likely through helping the recruitment of
coregulators. In contrast, MAPK-mediated phosphorylation
of the serine residue located in the Omega loop of RXRa
between helices H1 and H3 and close to helix 12, according
to the three-dimensional structure, impairs the transcription-
al activity of RAR/RXR heterodimers (Lee et al., 2000;
Matsushima-Nishiwaki et al., 2001). It has been proposed
that phosphorylation of this residue would create conforma-
tional changes within the LBD, disrupting the interactions
with coregulators and therefore decreasing transcription.
J. Bastien, C. Rochette-Egly / Gene 328 (2004) 1–16 11
However, according to recent studies, it would make RXRa
more resistant to proteolytic degradation, therefore exerting
dominant negative inhibition (Matsushima-Nishiwaki et al.,
2001).
It must be noted that the various signal transduction
pathways also cross-talk with retinoid receptors transactiva-
tion through the phosphorylation of the coactivators and
corepressors (see Fig. 6). The phosphorylation of corepres-
sors such as SMRT correlates with an inhibition of their
interaction with RARs and their redistribution from the
nucleus to the cytoplasm (Hong and Privalsky, 2000). In
contrast, the phosphorylation of p300/CBP, pCIP, SRC-1
and TIF-2 by a variety of kinases, including MAPKs or
PKA (Fig. 7), rather enhances their enzymatic activity as
well as their efficiency to interact with retinoid receptors
and/or the HAT complexes (Font de Mora and Brown, 2000;
Rowan et al., 2000a,b; Lopez et al., 2001; Vo and Goodman,
2001). Histones are also phosphorylated, increasing mark-
edly the efficiency of HATs and HMTs to acetylate or
methylate the nearby lysine residues. The same observation
has been made for the general transcription factors (GTFs)
and RNA Pol II (Orphanides and Reinberg, 2002). All these
phosphorylation processes converge towards the formation
of an efficient transcription initiation complex and a con-
trolled maximal response.
8. Retinoid receptors cross-talk with other signaling
pathways
Due to the ability of RXRs to serve as heterodimeric
partners not only to RARs, but also to several other nuclear
receptors (PPARs, LXR) (Willy and Mangelsdorf, 1999), it
is evident that retinoids can also control, the transcription of
a wider set of hormone-responsive genes (Leid et al., 1992;
Mangelsdorf and Evans, 1995; Chambon, 1996). Moreover,
one has to consider that, as is true for many other genes, the
promoters of retinoid-target genes contain, in addition to the
cognate response elements (RAREs), other regulatory
sequences which associate together with several transcrip-
tion activators in enhanceosomes. As an example, RARs
cooperate with SF1 and Sp1/Sp3 for the transactivation of
the Oct-3/4 and CYP26 promoters, respectively (Barnea and
Bergman, 2000; Loudig et al., 2000). Similarly, in the
presence of cytokines, STAT5 cooperates with RARs to
achieve maximum transcription of some RA-target genes (Si
and Collins, 2002). Such synergistic effects very likely
result from the cooperative recruitment of coregulators,
increasing chromatin remodeling and/or entry of the tran-
scription machinery to the promoter. This would explain
why, in vivo, the mechanisms of regulation differ from one
gene to the other and depend on the cell type (Nagpal et al.,
1992; Folkers et al., 1993).
However, retinoids are also able to antagonize the
activation of a subset of heterologous genes. The best
example is that of the AP-1-regulated genes (Shaulian and
Karin, 2002). The repressive effect of retinoids on AP-1
activity has been correlated with their antitumor activity and
the importance of this cross-talk for growth control is
increasingly recognized (Altucci and Gronemeyer, 2001b).
However, the mechanistic basis of the anti-AP-1 activity of
retinoid receptors remains still elusive, despite the proposal
of several distinct mechanisms. Competition for limiting
amounts of a common coactivator (Kamei et al., 1996) has
been proposed, as well as the inhibition of the JNKs
pathway (Caelles et al., 1997; Lee et al., 1999), disruption
of the Jun–Fos dimerization (Zhou et al., 1999) or exclusion
of some components (kinases, CBP) from the AP-1 com-
plexes (Benkoussa et al., 2002). Another example of repres-
sion by retinoid receptors has been demonstrated for