| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Affymax Research Institute Santa Clara, California 95051
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
-helices forming a 3-layered structure containing a ligand-
binding cavity that is almost entirely buried within the helices. While
several minor structural changes occur upon ligand binding, the highly
conserved carboxy-terminal helix 12, which has been shown to be
associated with the ligand-dependent AF-2 function (12, 13, 14, 15), appears to
occupy quite different positions in apo-, holo-, and antagonist-bound
receptors (10, 11). In agonist-bound receptors, helix 12 appears to cap
off the ligand-binding pocket, thus stabilizing the holostructure and
forming a new potential protein interaction face (7).
Several studies have indicated that the activity of individual NRs can
both modulate, and be modulated by, other transcription factors,
including other NRs, AP-1, and NF
B (16, 17, 18, 19, 20, 21, 22), and that stimulation of
the protein kinase A pathway can affect NR signaling (23, 24, 25).
These findings and others have led to the search for transcriptional
coactivator and corepressor molecules which, by forming large,
multicomponent complexes with NRs, would be able to integrate and
transmit diverse cellular signals to the basal transcriptional
machinery. Several classes of NR cofactors have been identified to date
(reviewed in Ref. 26). One class of coactivator molecules consists of
the CREB-binding protein (CBP) and the related factor p300 (27, 28, 29)
which, in addition to functioning in the transcription of both CREB-
and AP-1-responsive genes (30), have been shown to function in
NR-mediated transcription, and thus have been classified as integrators
of multiple signal transduction pathways (31, 32). A second class of
proteins of 160 kDa, termed steroid receptor coactivators (SRC) or
nuclear receptor coactivators (NCoA), include SRC-1/NCoA-1,
TIF2/GRIP/NCoA-2, and RAC3/p/CIP/ACTR/AIB1/TRAM-1 (see Ref. 33 and
references therein). These proteins have been shown to interact with NR
LBDs and with CBP and p300, and to be directly involved in
ligand-dependent transactivation by NRs (31, 34, 35, 36, 37, 38, 39). The SRC family
members, as well as CBP and p300, contain highly conserved amino acid
sequences of the form LXXLL (where L is leucine and X is any amino
acid), which likely form amphipathic -helices (37, 40). Three copies
of this motif, also called leucine-charged/helical domains or NR boxes,
are located in a region of the SRC members capable of hormone-regulated
interaction with NRs (nuclear receptor interaction domain, NRID), and
additional motifs may allow the assembly of larger multicomponent
coactivator complexes. The presence of multiple copies of these NR
boxes within coactivators suggests a mechanism whereby signal input and
cellular context are varied to allow multiple combinatorial coactivator
complexes to be assembled to facilitate flexible but precise
transcriptional control (41). Multiple NR-boxes within the NRID
of SRC family members could reflect the ability of these proteins to
interact with many different NRs or reflect a requirement for multiple
cooperative interactions (perhaps with receptor dimers) for efficient
association. Inactivation of the SRC-1 gene by gene targeting (42) has
shown that there may be at least partial functional redundancy between
the SRC family members. The molecular nature of this NR/coactivator
interaction mediated by these LXXLL motifs has been a topic of
recent intense investigation. Studies thus far have focused mainly
on the functional requirements of the three NR boxes within the NRID of
the SRC family members. Several studies have shown that different SRC
family members may have receptor preferences (43, 44) and that
different receptors have differing requirements for one or more intact
LXXLL motifs (37, 44, 45, 46, 47). Thus, the stoichiometry of
coactivator-receptor interactions may vary significantly for different
receptors. In addition, several studies have attempted to determine the
residues in and around the LXXLL motifs that are required for receptor
interaction. Mutational analysis has shown that the leucine residues
are critical for interaction, while residues upstream and downstream of
the core motif have more variable contributions (37, 45, 46, 47, 48). In
addition, recent studies (47, 49) have suggested that different ligands
for the same NR may change the coactivator binding specificity, thus
potentially adding a further refinement to transcriptional control
through a given receptor.
The use of large random and focused peptide libraries displayed either on phage (reviewed in Refs. 50, 51) or by the peptides-on-plasmids method (52, 53), have allowed the selection of peptides that interact specifically with a variety of molecules including antibodies (54), cytokine and other cell surface receptors (55, 56, 57), SH2 and SH3 domains (58, 59, 60, 61), and proteases and other enzymes (62, 63, 64). In the work presented here, we have chosen to take a different approach to studying the requirements of NR-coactivator interactions by isolating novel peptides that interact with NR LBDs in a receptor and ligand-dependent manner. We show, by mammalian-two-hybrid (M2H) assay, that the estrogen receptor-ß (ERß) and thyroid hormone receptor-ß (TRß) have significantly different NR box preferences when analyzed with fragments of SRC-1. Screening random and focused LacI-fused peptide libraries with ERß, and a focused headpiece dimer (HpD) library (65) with ERß and TRß, allowed the isolation of peptides with both receptor-selective and ligand-dependent properties, as well as peptides that interact well with both receptors. Sequence analysis of these peptides revealed several interesting differences between the peptides preferred by ERß and those preferred by TRß. With a coactivator-dependent scintillation proximity assay (SPA), we determined that the affinities of selected peptides for these two receptors are as high, and in some cases higher, than natural NR box peptides. Expression of selected peptides in mammalian cells resulted in receptor-selective dominant inhibitory activity toward wild-type receptors. Finally, based on observed sequence preferences, we made directed mutations in these peptides to alter their receptor specificity.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
-helical LXXLL
structure where L is leucine and X is any amino acid (37, 40). The aim
of our initial studies was to determine whether one or more than one of
these NR boxes is necessary for efficient interaction with
ligand-activated NRs. Second, we sought to determine whether there are
receptor-specific preferences for NR box interactions. As a model
system, we chose ERß and TRß because they represent both those NRs
that typically homodimerize and bind to palindromic response elements
(REs,) and also those that are capable of heterodimerization with the
retinoid X receptor and bind to direct repeat REs.
In a cell-based M2H assay, we analyzed various combinations of NR boxes
within the central SRC-1 NRID and the first NR-box of CBP (Fig. 1A
) for interaction with ERß and TRß.
Recent observations (37, 47, 48) have indicated that NR box 2 is
sufficient for the interaction of SRC-1 with ER and that NR box 3
does not interact with ER significantly. In contrast to these
observations, we found that ERß prefers the presence of at least two
motifs (1+2 > 2+3) for interaction comparable to that observed
when all three motifs are present (Fig. 1B). NR box 2, however, does
appear most important for this interaction. TRß requires both the NR
boxes 2 and 3 for efficient interaction while 1+2 interacts very poorly
(Fig. 1C). Thus, the NR box preference for TRß is substantially
different from that of ERß. The interaction of the amino terminus of
CBP with these receptors was minimal as compared with the interaction
with the SRC-1 NRID (Fig. 1, B and C). We conclude that ERß and TRß
interact minimally or not at all with the amino terminus of CBP.
|
) to maltose binding protein (MBP), purified
this protein, and demonstrated ligand-enhanced interaction with ERß
(Fig. 2C) and TRß (data not shown) in an enzyme-linked immunosorbent
assay (ELISA) format.
|
). Sequences of clones derived from the
focused library (Fig. 2D) revealed patterns quite similar to natural NR
box peptides. Charged residues occurred in many positions but notably
at +2 and +3 between the Leu residues as in NR box 2 and 3 of the SRC
family members. In addition, clones frequently contained Lys, Ile, and
Leu at the -1 position as found in SRC NR boxes. More importantly,
sequencing of only a few of the random 15 mer-derived peptides revealed
that almost all contained LXXLL motifs containing appropriately
positioned charged and hydrophobic residues. These results indicated
that this method could be used to select appropriate interacting
peptides and that the sequence motif most easily selected from a random
library closely resembles the natural sequence.
To attempt to rank order peptides more effectively, we transferred the
fourth round LacI panning output peptides, in bulk, to an MBP fusion
vector and assayed individual clones for their interaction with ERß
in the presence of E2 by ELISA. Display of the
peptides on MBP should make a monovalent rather than multivalent
reagent, leading to a better correlation between affinity of the
peptide for the receptor and the observed ELISA signal (53, 65). Many
of these clones interacted specifically with ERß, although they gave
lower apparent signals compared with MBP SRC-1 NRID, which contains
three NR boxes (Fig. 2B). Rare clones (M-23 for instance)
interacted with GST alone. To determine the ligand dependence of these
interactions, we assayed several clones with and without
E2. In each case tested, the ERß-specific
binding was almost entirely dependent on E2,
mimicking the interaction with natural coactivator molecules (Fig. 2C).
These results indicate that peptide libraries can be used as a rich
source of diversity for identifying NR interacting motifs and may have
utility in identifying either receptor and/or ligand-selective
reagents.
Selection of Receptor-Selective Peptides from HpD Libraries
To improve the affinity of selected peptides and to attempt to
differentiate selectivities between receptor types, we constructed a
focused X7LXXLLX7 HpD (65)
peptide library containing 1.8 x 109
members. The use of HpD-based libraries can result in the isolation of
peptides with higher affinity than those derived from intact LacI-based
libraries. We panned this focused library on immobilized GST-ERß and
GST-TRß fusion proteins. After four rounds of enrichment, we
transferred the pooled population of selected peptides to an MBP fusion
vector for ELISA analysis. Random peptide clones showed specific
binding in the presence of appropriate ligand, to the receptor on which
they were selected (Fig. 3, A and B).
Again, a few clones showed specificity to GST (M-169,
M-654, M-682). To determine whether these
peptides display receptor selectivity, we tested several clones against
both receptors by ELISA (Fig. 3C). Four ERß-selected peptides, tested
as MBP fusions, displayed remarkable selectivity for ERß with little
or no TRß cross-reactivity. Similarly, three (M-607,
M-683, and M-693) of four tested TRß-selected
peptides displayed a high degree of selectivity for TRß. One peptide,
M-655, appeared to bind both TRß and ERß with roughly
equal efficacy. Thus, the selection of receptor-selective NR-box
mimetics is possible using the peptide library approach.
|
). As with the peptides isolated with the LacI-based
libraries, these peptides, isolated in association with ERß, contain
charged residues in the +2 and/or +3 positions. Peptides with the
highest ELISA signal and affinity (see below and Table 1) for ERß contain a basic residue (R
or K) at the +3 position mimicking the natural SRC-1 NR box 2 peptide
(Fig. 4C). In contrast, the peptides isolated in association with TRß
frequently contain Arg in the +2 position, but never at the +3
position, similar to the natural SRC-1 NR box 3 peptide. The M2H
results above are consistent with these peptide results in that adding
NR box 3 to NR box 2 (SRC2+3) results in a synergistic effect upon
apparent association with VP16-TRß while the addition of NR box 1 to
NR box 2 (SRC1+2) results in no synergism (Fig. 1c). This suggests that
while NR box 3 alone is ineffective, it contributes significantly to
the interaction of TRß with SRC-1. A second striking feature of the
TRß-selected peptides is the heavy bias toward the presence of Pro at
the -2 position with hydrophobic residues (L, I, M) at the -1
position. While proline is not in general found at this position in
natural NR boxes, it does not preclude binding to ERß as some
ERß-selected peptides contain Pro in this position and at least one
TRß-selected peptide binds effectively to ERß as well (see below).
The bias for hydrophobic residues at the -1 position by ERß is not
as strong, allowing basic residues (K, R) to be selected as well. This
may reflect the M2H results indicating a relative preference for NR-box
1 (over NR box 3) by ERß and NR box 3 (over NR box 1) by TRß (Fig. 1, B and C). The selection of specific residues outside the -2 to +5
positions are more difficult to define for either receptor. Both ERß
and TRß select for both positively and negatively charged residues as
well as a number of Ser residues in the +6 to +12 region; however, it
is also apparent that peptides extending only to the +6 or +7 position
can be efficiently selected by these receptors. No apparent consensus
appears defined in the -7 to -3 region of the selected peptides.
|
|
) for further
analysis. In addition, MBP fusions to several of the SRC-1 fragments
used in the initial M2H studies (Fig. 1A) were constructed. Finally,
for comparison to natural NR boxes, we made MBP fusions to eleven 19
mer peptides derived from the three SRC-1 family members (SRC-1, TIF-2,
and RAC-3), and from CBP and p300 (Fig. 4A). These MBP-peptide fusions were
constructed to mimic the format of the focused
X7LXXLLX7 HpD-derived
peptides and were purified by affinity chromatography.
We set up a homogeneous binding assay based on SPA using immobilized
receptor LBDs and iodinated SRC-1 NRID (SRC1+2+3) as the tracer.
Control experiments using unlabeled tracer as the competitor are shown
in Fig. 4B. Even in the absence of E2, ERß
interacts specifically and saturably with the NRID (Fig. 4B, top
inset). The IC50 for SRC1+2+3 is 21
nM without ligand but decreases 
5 fold to 4.3
nM upon the addition of E2.
In addition, as expected from the ELISA experiments (see Fig. 2C), the
amount of tracer bound to the immobilized receptor increases
dramatically with ligand addition. TRß does not interact measurably
with the NRID in the absence of ligand (Fig. 4B, bottom),
consistent with our observation of much lower basal ELISA signals with
TRß as compared with ERß (Fig. 3C). Addition of ligand to the TRß
SPA results in the expected association of the
125I NRID and an IC50 of
5.1 nM for unlabeled SRC1+2+3.
We next used the various MBP fusions described above as competitors in
both the ERß/NRID and TRß/NRID SPAs in the presence of
E2 and T3, respectively
(Exp. 1, Table 1). To ensure that affinity and selectivity trends we
observed were not artifacts of any one protein preparation, we
reisolated a second batch of several of the MBP fusions and repeated
all measurements (Exp. 2, Table 1). Although there are some
quantitative differences between the two protein preparations, the
results lead to the same conclusions in each case. As expected, the
IC50 values for SRC1+2+3 and SRC2+3 are nearly
identical (3–11 nM) for the TRß assay, and these protein
fragments are more potent than any natural or synthetic, single NR
box-containing peptide, assayed on this receptor. SRC1+2 and SRC2,
either as a peptide (19 mer) or fragment (56 mer), are less potent,
consistent with the M2H assays above. Homologous peptides derived from
the three SRC-1 family members behave in a similar fashion with TRß.
The order of affinity of the three NR boxes in each case is 2 >
3 > 1 suggesting that, at the level of NRID-AF2 interactions,
TRß displays no preference for any one of the three SRC-1 family
members. Consistent with M2H experiments, the amino-terminal NR box of
CBP has relatively poor affinity for TRß (as does the similar
sequence from p300). While many of the TRß-selected peptides
(M-668, M-673, M-675, and
M-683 for instance) have IC50 values
similar to the best natural peptides, e.g. SRC2, the
ERß-selected peptides have low affinity for TRß. The fold
selectivity of the TRß-selected peptides for the TR is always >2,
except for M-655, which, by ELISA (Fig. 3C), binds to both
TRß and ERß. In contrast, most of the natural coactivator-derived
peptides have receptor selectivities of only 1- to 2-fold. The observed
>1- to 2-fold selectivity of SRC3 for TRß is consistent with the M2H
results (Fig. 1C).
The interactions of the natural coactivator-derived peptides or
SRC-1 fragments with ERß are slightly different than with TRß, as
expected. While SRC1+2+3 still has the lowest
IC50, there is little difference between the
values obtained for SRC1+2 and SRC2+3 on ERß, in contrast to those
obtained for TRß. SRC2, in fragment or peptide form, has lower
affinity than the fusion proteins containing more than one NR box as
predicted by M2H assay (Fig. 1B). In addition, the order of affinity of
the three NR boxes in each SRC-1 family member for ERß is 2 >
3 = 1. From the data presented in Table 1, it is tempting to
predict that TIF-2 and RAC-3 would be more effective coactivators of
ERß-dependent transcription than SRC-1, as the affinities of the 1
and 3 NR boxes of these proteins are significantly higher than those of
the corresponding NR boxes of SRC-1. While previous studies using
yeast-two-hybrid systems have shown that an ER LBD is able to
interact with the amino terminus of CBP (31) or with the amino-terminal
CBP NR-box [albeit poorly compared with the SRC NR boxes (40)], our
data suggest that these interactions are of quite low affinity (Fig. 1B
and Table 1). We speculate that the type of assay used to determine
NR-coactivator interactions has a significant effect on the
interpretation of the results.
Peptides we selected using ERß as the panning reagent show a
relatively high degree of selectivity for ERß over TRß (Table 1).
As with the TRß-selected peptides and TRß, the affinity of the
ERß-selected peptides for ERß is similar if not slightly higher
than the best natural NR boxes (compare M-180 and
M-193 with TIF2, Table 1). In addition, similar to the
TRß-selected peptides, these individual peptides do not have
affinities for ERß that are as high as any of the SRC-1 fragments
containing more than one LXXLL motif. The experiments above demonstrate
that screening peptide libraries with individual receptors is a
successful method for obtaining receptor-selective peptides with
affinities equal to or greater than the relatively nonselective natural
NR box-derived sequences.
Dominant Negative Activities of Receptor-Selective Peptides
We next asked the question whether peptides that showed
receptor-selective binding in vitro would also show
selective binding, and therefore competition for natural coactivators,
in mammalian cells. Expression of a Gal4-NRID fusion protein, but not
Gal4 alone, in CHO cells in the presence of wild-type ERß and an ERE
reporter construct dramatically inhibited estradiol-induced
transcription, reflecting a potent dominant inhibitory activity (Fig. 5A). We tested three Gal4-peptide
fusions in this system: The relatively nonselective peptide, SRC2; the
ERß-selective peptide, M-193; and the TRß-selective
peptide, M-668 (Fig. 6A).
Using 4 ng/well of expression plasmid, both SRC2 and M-193
peptides inhibited estradiol-induced transcriptional activity as well
as the SRC123 (NRID) protein, while M-668 was less effective
(Fig. 5B, top). Reducing the amount of protein expressed
(using 2 ng/well, Fig. 5B, bottom) showed that
M-193 peptide is more potent than either SRC2 or
M-668 as a dominant inhibitor of ERß-directed
transcription. These data are consistent with the observed in
vitro affinities of the peptides for ERß (Table 1). High level
expression (2–4 ng/well plasmid) of Gal4 vector alone inhibited
transcription by TRß on a TRE-dependent reporter (data not shown). We
therefore used low level expression of the peptide fusions to show that
M-668 is as potent as SRC2 and that M-193 is the
least potent dominant negative inhibitor of TRß, as expected from the
binding affinities (Fig. 5C and Table 1). Therefore, receptor-selective
binding results in receptor-selective transcriptional inhibition by
these peptides.
|
|
), yet contains
several features quite similar to the M-193 ERß-selective
peptide such as lack of a proline at the -2 position, presence of a
basic residue at the +3 position, and more than one acidic residue
carboxy-terminal to the core LXXLL motif. We therefore introduced
changes (Fig. 6A) directed at specifically enhancing the interaction of
this peptide with TRß. Changing the +2 and +3 residues from His-Arg
to Arg-Ser (K RS ED), found in several TRß-selected clones
(Fig. 3D), resulted in an increase in activity with TRß but a
moderate drop in activity with ERß (Fig. 6B). Substitution of the Lys
at -2 with Pro (P HR ED) resulted in increased activity
with TRß, as expected, but also an increase in activity with ERß,
especially in the absence of hormone. This is consistent with the
finding of Pro in this position of some ERß-selected peptides (Fig. 3D) and with dual specificity peptides such as M-655 (Fig. 3, C and D). Subsequent removal of the carboxy-terminal acidic residues
from P HR ED and substitution with Arg and Ser (P
HR RS) further increased activity with TRß but decreased
activity with ERß as compared with P HR ED (Fig. 6B).
Peptide M-668, the most TRß-selective peptide we
identified, contains features that, either alone or in combination,
could bias its binding away from ERß including a lack of a basic
residue at the +3 position, and a lack of an acidic residue at
the +7 position. M-668 also contains a Pro at the -2
position, rarely found in ERß-selected peptides, but apparently not
detrimental to ERß binding (see above). Substituting this Pro for Ser
(S TM G), found in several ERß-selected peptides (Fig. 3D), resulted in loss of more than 80% of the interaction with TRß
while causing little change in the interaction with ERß (Fig. 6C).
Subsequently placing a basic residue, Arg, in the +3 position and an
acidic residue, Glu, in the +7 position (S TR E)
resulted in little further change in binding to either receptor,
suggesting that additional features of the peptide preclude a more
effective interaction with ERß.
The ERß-selective peptide, M-193, contains
features that are common in the ERß-selected peptides but
uncommon in the TRß-selected peptides. These include lack of a
proline at the -2 position, presence of Arg at the +3 position, and
the presence of acidic residues at the +7 or +8, and +12 positions.
Changing the two Glu residues at positions +8 and +12 to Arg (S WR
RR) dramatically reduced both basal and estradiol-induced
interactions with ERß (Fig. 6D). As expected, introduction of a Pro
residue at the -2 position (P WR EE) slightly increased
interaction with TRß but had relatively little effect on interaction
with ERß (Fig. 6D). Combining these changes with a shift in the
position of the basic residue from +3 to +2 (P RS RR)
resulted in loss of most of the interaction with ERß but some
increased interaction with TRß relative to the wild-type 193 peptide
(Fig. 6D). The results above suggest that directed mutations can alter
the receptor specificity of natural or selected interacting
peptides.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
/RXR, suggest that
each partner binds to one coactivator molecule (46), other similar
studies with RAR/RXR (74) support a model where, again, two LXXLL
motifs within one coactivator molecule bridge across the heterodimer.
In addition, the potential for more than one site of coactivator
interaction within full-length receptors exists, e.g. with
both the AF1 and AF2 domains (34, 72). Therefore, while the exact
reasons for multiple LXXLL motifs within coactivator molecules are
still not well understood, this feature presumably allows for a limited
set of coactivators to interact with multiple NRs and/or allows for
cooperative interactions between several coactivators and receptor
monomers or dimers to form transcriptionally active complexes.
In the study presented here, we have focused on the NR box sequence
preferences for hormone-dependent coactivator interactions with ERß
and TRß. Using a M2H assay, we analyzed interactions with fragments
of the previously defined minimal NR interaction domain containing
three NR boxes (37, 40) and found that these two receptors display
different preferences for optimal interaction. While the second NR box
from SRC-1 (SRC2) alone interacts somewhat with both receptors, the
first and third NR boxes are needed but contribute to optimal binding
differently for these two receptors. Interestingly, while our results
(not shown) and those of others (37, 47, 48) indicate that ER
requires only the SRC2 NR box for optimal interaction, it appears that
ERß requires the addition of a second NR box (preferably SRC1) for
efficient interaction with the coactivator. While the M2H data for
ERß are consistent with some sort of cooperative interaction when two
NR boxes are present, it is difficult at this time to propose that this
represents a fundamental difference in the way ER and ERß interact
with this class of coactivator. The data do, however, support the
notion that the two receptors have differing sequence preferences for
interacting NR boxes. In further support of this, Chang et
al. (75) have recently identified a set of LXXLL peptides
containing a Pro in the -2 position, similar to our TRß-selected
peptides, that bind well to both TRß and ERß (like peptide
M-655) but that bind poorly to ER. Like ERß, TRß
interacts most strongly with the second NR box on its own but also
requires two NR boxes, SRC2+3 in particular, for strong interaction.
These observations are consistent with recent findings of others using
GRIP, which show that mutation of NR box 3 reduces coactivator function
in mammalian cells (44, 45).
We used both random and focused libraries to select novel
ERß-interacting peptides. Sequence analysis of peptides derived from
a random 15 mer library revealed the presence of LXXLL motifs. This
finding is, perhaps, not surprising given that previous studies have
demonstrated the importance of the Leu residues for coactivator binding
(37, 40). It does, however, demonstrate that, at least in the presence
of an agonist ligand, the LXXLL binding site on the receptor is
dominant over other possible binding sites in its ability to enrich
peptide ligands. Consistent with this notion, in a recent study using
E2-bound ER as the target and random peptide
libraries presented by phage display, LXXLL motifs were also isolated
(76).
The position of the LXXLL motif within the random 15 mer clones varies.
It is not, however, found in a position that would allow the peptide to
be shorter than extending to the +8 or -2 positions. This is
consistent with recent peroxisome proliferator-activated receptor-
(PPAR) and TRß cocrystallization studies with NR box peptides
showing that the conserved Glu in helix 12 and Lys in helix 3 form
specific hydrogen bonds with residues contained within this peptide
length (10, 45). Using a focused, LXXLL-containing library, we
found strongly ERß interacting peptides with higher frequency (Fig. 2A). When all of the selected LacI-coupled peptides were transferred to
an MBP expression vector, the frequency of strongly interacting,
receptor-specific clones dropped, consistent with the notion that
weakly interacting peptides may not be detected unless they are
displayed as a multivalent reagent.
We show by both ELISA and SPA that the SRC-1 NRID can interact with
ERß in the absence of hormone but that E2
increases its affinity for ERß by 5-fold. Similarly, many of the
ERß-selected peptides also interact with ERß in the absence of
estradiol, especially those selected from HpD libraries (Figs. 2c and 3c). Interestingly, neither the SRC-1 NRID nor the TRß-selected
peptides interact with TRß in the absence of
T3. This likely represents a fundamental
difference in the conformation these two receptors assume in the
nonliganded form and is consistent with the ability of the nonliganded
TRs to bind corepressor molecules and repress transcription in intact
cells (77) while ERs appear to bind corepressors tightly only in the
presence of antagonists (78). Importantly, peptides we selected using
ERß and LacI fused libraries were, in addition to having some
constitutive interaction, also responsive to E2,
demonstrating that they have properties very similar to natural
coactivators.
Analysis of natural NR box peptides by competition SPA is consistent
with the M2H studies and supports the view that NR box 2 interacts most
strongly followed by NR box 3>1 for TRß, and NR box 1= 3 for ERß.
Some differences between the NR-boxes of the three SRC family members
are evident. Ding et al. (44) have suggested that TIF2/GRIP1
may interact with ER more strongly than SRC-1. Our SPA data are not
inconsistent with this but suggest that RAC-3, in addition, might
interact more strongly with ERß (Table 1).
We selected peptides that bound preferentially to ERß or TRß in a
ligand-dependent manner. The affinities of selected peptides for TRß
and ERß were, in general, similar to the best natural
coactivator-derived peptides, yet showed a much more consistent degree
of selectivity toward the receptor used to isolate them. Although
peptides with much higher affinity were not isolated in the current
work, the use of specific experimental methods such as more stringent
wash conditions or inclusion of low-affinity peptides, could increase
the selection or identification of higher affinity peptides. In
addition, most selected peptides that interacted well with ERß also
interacted with ER, and those that interacted well with TRß also
interacted with TR (data not shown). Thus, it appears that it may be
more difficult to find receptor subtype-specific LXXLL-containing
peptides. However, one recent study has identified non-LXXLL-containing
peptides with binding preference for either ER or ERß (76), and
another study has identified a single LXXLL-containing peptide with
specificity for ERß (75).
An important aspect of these studies is the ability to demonstrate that
peptides specifically selected to bind to one receptor LBD over another
in vitro should show selective activities on wild-type
receptors in cell-based assays. Our dominant negative results suggest
that this is indeed true. Interestingly, Norris et al. (79)
have demonstrated hormone-specific dominant negative activity of
phage-selected peptides using ER. The ability to derive reagents
that specifically inhibit the transcriptional activity of one type of
NR, as demonstrated here, one subtype of NR (75), or the
transcriptional activity of one type of ligand on a specific NR (79)
may be useful in elucidating complex hormonal responses in cell-based
systems or even whole animals.
Recent studies have begun to address the questions of 1) which residues
within and around the conserved leucines within NR boxes are important
for receptor binding, and 2) which residues impart binding specificity
for one receptor over another. Using ER, McInerney et al.
(47) found that residues N-terminal to the core LXXLL motif of
SRC-1/NCoA-1 NR box 2 were not important for receptor interaction while
C-terminal residues, especially positions +12 and +13, were very
important. Conversely, Mak et al. (80) found that a NR box 2
peptide extending only to residue +7 but containing important
N-terminal residues at positions -2 to -4 bound efficiently to ER.
Although the assay methods were different, it is not immediately clear
what is the source of this discrepancy. Darimont et al. (45)
found that sequences flanking the core NR box 2 of GRIP (although the
hydrophobic core was also necessary) determined binding affinity for
TRß, while the core motif itself determined binding affinity for the
glucocorticoid receptor. We demonstrate that peptides selected from a
focused (LXXLL) library by TRß contain several characteristics that
differentiate them from those selected by ERß. These include Pro at
the -2 position with Leu, Ile, and Met at the -1 position, basic
residues at the +2 position but not +3 position, and a somewhat higher
proportion of basic residues in the +7 to +12 positions. ERß-selected
peptides may contain these features but also permit basic residues in
the -1 and +3 positions, tend to have a higher proportion of acidic
residues in the +7 to +12 positions, and have a high frequency of Ser,
Thr, and Gly rather than Pro at the -2 position.
It is unclear what the roles of residues +2 and +3 are for either
affinity or selectivity of NR box interactions with NRs. Cocrystal
structure analysis (10) indicates that the side chains of these
residues are solvent exposed, and Darimont et al. (45) have
shown that peptide interactions are unaffected by mutation of these
residues. Our data show that charged residues are selected frequently
here and suggest that different receptors may have preferences for the
location of the charge at the +2 and +3 positions. Altering the
position of the basic residue from +3 to +2 in certain peptides (SRC2
for instance) can decrease interaction with ERß and increase
interaction with TRß. Further specific mutations and perhaps
cocrystallization studies will be necessary to resolve these issues.
Studies with ER and RAR (47) suggest that mutation of residues
C-terminal to the core LXXLL motif can differentially affect
interactions with these two receptors, implying that specific
interactions with the receptor occur in this area. Cocrystal structure
analysis with PPAR (10, 47) and TRß (45) show that several
interactions are present, most notably between conserved residues in
the C terminus of receptor helix 3 and residues +4 to +7 of either
SRC-1 or GRIP NR box 2. These interactions form one half of what has
been termed a "charge clamp" (10), which may serve to orient the
helix on the hydrophobic face and to limit the length of the -helix.
Our results support a view that charged residues C-terminal to the core
motif can affect receptor affinity and selectivity. Switching two
negatively charged residues in the +7 and +12 positions with one basic
residue and a serine in peptide SRC-1 NR box 2 decreases interaction
with ERß while increasing interaction with TRß. It will be
interesting to determine whether these changes alter receptor
specificity by changing specific side chain interactions or rather by a
more global change in the motif structure.
Using ER with SRC-1 NR boxes and TRß with GRIP NR boxes, Mak
et al. (80) and Darimont et al. (45),
respectively, have shown that residues N-terminal to the core LXXLL
motif, especially three basic residues -4 to -2 in the case of ER,
are involved in the observed high-affinity interactions. We noted that
ERß- and TRß-selected peptides usually contained at least one basic
residue in either the -4 or -3 position but that no consensus is
evident beyond that. The strong selection for proline in the -2
position by TRß suggests that residues N-terminal to the core motif
[aside from the -1 position, which appears to be involved in the
charge clamp (10)] may actually interfere with high-affinity
interactions with this receptor if forced to be in proximity to the
receptor by a rigid, extended -helix. While it is clear that certain
combinations of residues are permissible, as evidenced by the ability
of SRC-1 NR box 2 to interact with TRß, proline in this position may
allow potentially interfering residues to bend away from the surface of
the receptor. ERß appears to be much less sensitive to negative
effects of N-terminal residues as proline is not strongly selected in
the -2 position; however, we show that substitution for proline here
can enhance interactions with both TRß and ERß. Recently, Paige
et al. (76) have presented several sequences obtained from
panning phage libraries with ER in the presence of estradiol.
Although these peptides are presented to the receptor as N-terminal
rather than C-terminal fusions as in our study, their sequences show
several characteristics in common with our ERß-selected peptides.
These include Ser, Gly, and Pro in the -2 position, a mix of basic and
hydrophobic residues in the -1 position, acidic residues in the +7 and
+8 position, and charged residues in both the +2 and +3 positions. As
these peptides were selected using phage display, their relative
affinities for the receptor compared with natural motifs or our
HpD-selected sequences are not known.
Starting with the NR box 2 from SRC-1, we were able to make directed mutations that moderately decreased interaction with ERß and substantially increased interaction with TRß. Since the direction for these changes was based on sequences of newly isolated peptides, it is apparent that the methodology predicts real differences in individual receptor preferences. Attempts to alter a TRß-selective peptide to ERß selectivity have not been successful so far, perhaps due to the short length of the particular peptide tested. We did, however, achieve modest success in switching an ERß-selective peptide to TRß selectivity. Experimentation with specific changes on several more starting peptides will, no doubt, further our interpretation of these results.
In summary, these studies demonstrate that receptor-selective and ligand-dependent peptides can likely be selected for any given NR. The information gained can potentially be used to predict what sequences, and therefore presumably what coactivator molecules, might best interact with a given NR. The results thus suggest a general method for the development of reagents for studying orphan receptors and for characterizing ligand-dependent receptor conformations using peptide libraries.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Protein Expression and Purification
Fusions to GST were constructed in expression vectors derived
from pGEX-3X (Amersham Pharmacia Biotech, Piscataway, NJ).
Receptor LBDs of ERß and TRß fused to GST contain the identical
amino acids as described in the above section. In addition, a GST-fused
full-length TRß construct was made and the resulting fusion protein
used for peptide library screening (see below). Coactivator fragments
and peptides were fused to MBP in derivatives of the vector pMALc2
(New England Biolabs, Inc., Beverly, MA) as described
(53). SRC-1 fragments contain the following amino acids: SRC1+2+3,
597–781; SRC1+2, 597–721; SRC2+3, 646–781; and SRC2, 646–701. In
addition, a SRC1+2+3 construct containing the additional amino acids
CLEPYTACD [antibody 179 epitope, (69)] at the carboxy terminus was
constructed and used for radiolabeling (see below). Peptides derived
from the NR-boxes of the coactivators SRC-1, TIF-2, RAC-3, CBP, and
p300 were constructed as 19 mers in the form
X7LXXLLX7 (sequences
depicted in Fig. 4C) and fused to MBP in the vector pELM3 (53). Fusion
proteins were expressed in E. coli strain ARI814 (65) using
400 µM isopropylthiogalactoside
added at OD600 0.5, followed by growth at 37
C for 3 h. For GST fusions, sonication buffer was PBS, 1
mM EDTA, 0.1 mM
phenylmethylsulfonylfluoride (PMSF), protease inhibitors 10 µg/ml
leupeptin and 5 µg/ml aprotinin, and 5 mM
dithiothreitol (DTT). For MBP fusions, sonication buffer was 10
mM Tris, pH 7.4, 1 mM EDTA,
200 mM NaCl, 0.1 mM PMSF,
protease inhibitors as above, and 5 mM DTT.
Harvested bacteria were frozen and thawed, incubated with 200 µg/ml
lysozyme in sonication buffer (15 min, 4 C), sonicated, and a cleared
lysate produced by centrifugation (Beckman Coulter, Inc.
JA20 rotor, 14,000 rpm, 15 min, 4 C). GST fusions were purified on
glutathione sepharose resin (Amersham Pharmacia Biotech),
and MBP fusions on amylose resin (New England Biolabs, Inc.). GST fusion proteins were eluted with 50
mM Tris-HCl, pH 8.0, 10 mM
glutathione, and MBP fusions with sonication buffer containing 10
mM maltose. Purified proteins were characterized
by PAGE.
Library Construction, Panning, and Analysis of Individual Clones
by ELISA
Detailed procedures for construction, panning, and analysis of
LacI (peptides-on-plasmids) and HpD libraries have been described
previously (52, 53, 65). The three libraries used in the current work
consist of a random 15 mer (X15), X = (NNK)
where N = A, C, G, T and K = G, T, and a focused 19 mer
(X7LXXLLX7), both fused to
LacI, and a focused 19 mer
(X7LXXLLX7) fused to the
HpD. Panning was performed by coating 24 (first round) or 6 (subsequent
rounds) microtiter wells each with 2 µg of either GST ERß-LBD or
GST TRß-full length, followed by appropriate BSA blocking. Four
rounds of panning were performed essentially as described (53) except
that 2–5 µg/ml of purified soluble GST were added to each lysate to
reduce the probability of enriching for peptides specific for GST.
ERß was panned using all three libraries in the presence of 1
µM E2 and TRß panned with only
the focused, HpD library, in the presence of 1 µM
T3. Sequencing of LacI and MBP fusions was by the
dideoxy chain termination technique using double stranded plasmid DNA
and appropriate primers. Analysis of clones by ELISA, using crude
lysates, was performed with ERß-LBD and both TRß-LBD and TRß-full
length fusions without significant differences observed between these
latter two (data not shown). Briefly, ELISA plates were coated with 10
µg/ml, 100 µl/well of GST ERß-LBD, GST TRß-LBD, or GST
TRß-full length receptor in PBS, 1 mM DTT (shaking for
1 h at room temperature). Plates were washed three times (200 µl
PBS, 0.05% Tween 20, 1 mM DTT, 4 C), and blocked in assay
buffer (200 µl PBS, 0.1% BSA, 0.05%Tween 20, 1 mM DTT,
4 C) for 1 h with shaking. After removal of the blocking buffer,
100 µl/well of MBP or LacI fusion protein-containing bacterial
lysates diluted 1:50 or 1:20, respectively (53), in assay buffer were
added to the plates and incubated with or without 1 µM
hormone (shaking at 4 C for 1 h). Plates were washed as above, and
100 µl/well of polyclonal rabbit anti-MBP antiserum (New England Biolabs, Inc.) diluted 1:10,000 or anti-LacI antiserum
diluted 1:15,000 plus alkaline phosphatase-conjugated goat antirabbit
IgG (Tago) diluted 1:2000 in assay buffer (minus DTT) were added
concurrently to plates (shaking at 4 C for 45 min). After washing as
above (but without DTT), colorimetric development
(p-nitrophenyl phosphate, Bio-Rad Laboratories, Inc.) was performed and measured at
OD405nm.
Iodination of MBPSRC1+2+3 Protein
Purified MBPSRC1+2+3 protein containing a carboxy-terminal
antibody 179 epitope (see above) was iodinated using IODO-BEADS reagent
(Pierce Chemical Co., Rockford, IL) according to the
manufacture’s instructions. Briefly, 100 µg of protein in 100 µl
volume were mixed with 100 µl BupH phosphate buffer (Pierce Chemical Co.), one IODO-BEAD, and 5 µl carrier-free iodine-125
(3.7GBq/ml, 100 mCi/ml, (Amersham Pharmacia Biotech) . The
labeling reaction proceeded at room temperature for 5 min and was then
terminated by removing the reaction mix from the IODO-BEAD. The labeled
protein was purified on a Sephadex G-50 column immediately. Aliquots
taken from the unpurified reaction and each of the fractions collected
from the G-50 column were counted with scintillation fluid. Protein
recovery was estimated by ELISA with unlabeled protein as a control.
Typical specific activity of the iodinated MBPSRC1+2+3 was between
166–246 Ci/mmol protein. Autoradiography of the protein analyzed by
PAGE showed a major labeled band at approximately 55 kDa.
Competition SPA
Yttrium silicate SPA beads (60 µg/well) coated with the
polycationic polymer, polylysine (Amersham Pharmacia Biotech) were batch bound with 250 ng/well of either purified
GST ERß-LBD or TRß-LBD in coating buffer [Tris-buffered saline
(TBS) pH 7.4, 1 mM DTT, room temperature, 15 min with
shaking]. Receptor-bound beads were pelleted and then resuspended in
assay buffer (TBS, 0.1% BSA, 1 mM DTT) to remove unbound
receptor. Purified MBP fusion proteins or fusion peptides were added at
varying concentrations to wells and a fixed amount (20,000 cpm) of
iodinated MBPSRC1+2+3 was subsequently added in assay buffer with or
without either 1 µM E2 or 1
µM T3 final as appropriate for a
total assay volume of 100 µl/well. Assays were incubated for 1 h
at 4 C in 96-well OptiPlates (Packard) with shaking and subsequently
read on a TopCount (Packard). The IC50 values
were determined with a nonlinear least squares regression to a
four-parameter logistic equation (70).
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Received for publication December 1, 1999. Revision received February 9, 2000. Accepted for publication February 14, 2000.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
ligand-binding domain
bound to all-trans retinoic acid. Nature 378:681–689[CrossRef][Medline]
. Nature 375:377–382[CrossRef][Medline]
. Nature 395:137–143[CrossRef][Medline]
down-regulate the transforming growth factor-ß 1 promoter by
antagonizing AP-1 activity. Mol Endocrinol 7:47–56 B and
the glucocorticoid receptor. Proc Natl Acad Sci USA 91:752–756-mediated induction of intercellular adhesion molecule-1 is
inhibited by dexamethasone. Functional analysis of the human
intercellular adhesion molecular-1 promoter. J Biol Chem 269:6185–6192 B activation. J Biol Chem 271:1651–1659
n H, Paige LA, Hamilton PT,
Kenan DJ, Fowlkes D, McDonnell DP 1999 Dissection of the LXXLL nuclear
receptor-coactivator interaction motif using combinatorial peptide
libraries: discovery of peptide antagonists of estrogen recepors
and ß. Mol Cell Biol 19:8226–8239 and ERß. Proc Natl Acad Sci USA 96:3999–4004This article has been cited by other articles:
 |
|
 |
|
  A. Koppen, R. Houtman, D. Pijnenburg, E. H. Jeninga, R. Ruijtenbeek, and E. Kalkhoven Nuclear Receptor-Coregulator Interaction Profiling Identifies TRIP3 as a Novel Peroxisome Proliferator-activated Receptor {gamma} Cofactor Mol. Cell. Proteomics, October 1, 2009; 8(10): 2212 - 2226. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. L. Privalsky, S. Lee, J. B. Hahm, B. M. Young, R. N. G. Fong, and I. H. Chan The p160 Coactivator PAS-B Motif Stabilizes Nuclear Receptor Binding and Contributes to Isoform-specific Regulation by Thyroid Hormone Receptors J. Biol. Chem., July 17, 2009; 284(29): 19554 - 19563. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. Wang, H. Li, L. B. Moore, M. D. L. Johnson, J. M. Maglich, B. Goodwin, O. R. R. Ittoop, B. Wisely, K. Creech, D. J. Parks, et al. The Phytoestrogen Coumestrol Is a Naturally Occurring Antagonist of the Human Pregnane X Receptor Mol. Endocrinol., April 1, 2008; 22(4): 838 - 857. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Lodrini, T. Munz, N. Coudevylle, C. Griesinger, S. Becker, and E. Pfitzner p160/SRC/NCoA coactivators form complexes via specific interaction of their PAS-B domain with the CID/AD1 domain Nucleic Acids Res., April 1, 2008; 36(6): 1847 - 1860. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P.-J. Tai, Y.-H. Huang, C.-H. Shih, R.-N. Chen, C.-D. Chen, W.-J. Chen, C.-S. Wang, and K.-H. Lin Direct Regulation of Androgen Receptor-Associated Protein 70 by Thyroid Hormone and Its Receptors Endocrinology, July 1, 2007; 148(7): 3485 - 3495. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Folkertsma, P. I. van Noort, A. de Heer, P. Carati, R. Brandt, A. Visser, G. Vriend, and J. de Vlieg The Use of in Vitro Peptide Binding Profiles and in Silico Ligand-Receptor Interaction Profiles to Describe Ligand-Induced Conformations of the Retinoid X Receptor {alpha} Ligand-Binding Domain Mol. Endocrinol., January 1, 2007; 21(1): 30 - 48. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. N. Duong, S. Elliott, D. E. Frigo, L. I. Melnik, L. Vanhoy, S. Tomchuck, H. P. Lebeau, O. David, B. S. Beckman, J. Alam, et al. AKT Regulation of Estrogen Receptor {beta} Transcriptional Activity in Breast Cancer. Cancer Res., September 1, 2006; 66(17): 8373 - 8381. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. J. Dubbink, R. Hersmus, A. C. W. Pike, M. Molier, A. O. Brinkmann, G. Jenster, and J. Trapman Androgen Receptor Ligand-Binding Domain Interaction and Nuclear Receptor Specificity of FXXLF and LXXLL Motifs as Determined by L/F Swapping Mol. Endocrinol., August 1, 2006; 20(8): 1742 - 1755. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. M. R. Moore and R. K. Guy Coregulator Interactions with the Thyroid Hormone Receptor Mol. Cell. Proteomics, April 1, 2005; 4(4): 475 - 482. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C M Klinge, S C Jernigan, K A Mattingly, K E Risinger, and J Zhang Estrogen response element-dependent regulation of transcriptional activation of estrogen receptors {alpha} and {beta} by coactivators and corepressors J. Mol. Endocrinol., October 1, 2004; 33(2): 387 - 410. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. M. R. Moore, S. J. Galicia, A. C. McReynolds, N.-H. Nguyen, T. S. Scanlan, and R. K. Guy Quantitative Proteomics of the Thyroid Hormone Receptor-Coregulator Interactions J. Biol. Chem., June 25, 2004; 279(26): 27584 - 27590. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. A. Iannone, C. A. Simmons, S. H. Kadwell, D. L. Svoboda, D. E. Vanderwall, S.-J. Deng, T. G. Consler, J. Shearin, J. G. Gray, and K. H. Pearce Correlation between in Vitro Peptide Binding Profiles and Cellular Activities for Estrogen Receptor-Modulating Compounds Mol. Endocrinol., May 1, 2004; 18(5): 1064 - 1081. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A.-M. Leduc, J. O. Trent, J. L. Wittliff, K. S. Bramlett, S. L. Briggs, N. Y. Chirgadze, Y. Wang, T. P. Burris, and A. F. Spatola Helix-stabilized cyclic peptides as selective inhibitors of steroid receptor-coactivator interactions PNAS, September 30, 2003; 100(20): 11273 - 11278. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Liu, K. S. Knappenberger, H. Kack, G. Andersson, E. Nilsson, C. Dartsch, and C. W. Scott A Homogeneous in Vitro Functional Assay for Estrogen Receptors: Coactivator Recruitment Mol. Endocrinol., March 1, 2003; 17(3): 346 - 355. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Y. Wu, W. W. Chin, Y. Wang, and T. P. Burris Ligand and Coactivator Identity Determines the Requirement of the Charge Clamp for Coactivation of the Peroxisome Proliferator-activated Receptor gamma J. Biol. Chem., February 28, 2003; 278(10): 8637 - 8644. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Ko, G. R. Cardona, T. Iwasaki, K. S. Bramlett, T. P. Burris, and W. W. Chin Ser-884 Adjacent to the LXXLL Motif of Coactivator TRBP Defines Selectivity for ERs and TRs Mol. Endocrinol., January 1, 2002; 16(1): 128 - 140. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. S. Bramlett, Y. Wu, and T. P. Burris Ligands Specify Coactivator Nuclear Receptor (NR) Box Affinity for Estrogen Receptor Subtypes Mol. Endocrinol., June 1, 2001; 15(6): 909 - 922. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. M. Hall, C.-y. Chang, and D. P. McDonnell Development of Peptide Antagonists That Target Estrogen Receptor {beta}-Coactivator Interactions Mol. Endocrinol., December 1, 2000; 14(12): 2010 - 2023. [Abstract] [Full Text]
|
|
|
|
|
|
A. Warnmark, T. Almlof, J. Leers, J.-A. Gustafsson, and E. Treuter Differential Recruitment of the Mammalian Mediator Subunit TRAP220 by Estrogen Receptors ERalpha and ERbeta J. Biol. Chem., June 22, 2001; 276(26): 23397 - 23404. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
