Molecular Cell, Vol. 16, 425–438, November 5, 2004, Copyright 2004 by Cell Press
Structural Basis for Androgen Receptor Interdomain
and Coactivator Interactions Suggests a Transition
in Nuclear Receptor Activation Function Dominance
dependent (Metzger et al., 1992; Tremblay et al., 1999).
Activation function 2 (AF2) in the ligand binding domain
(LBD) is a highly conserved hydrophobic cleft flanked
et al., 1998; He and Wilson, 2003) that binds the LXXLL
motifs of the steroid receptor coactivator (SRC) family
of coactivators (Onate et al., 1995; Hong et al., 1996;
Voegel et al., 1998). Some coactivators and associated
complexes such as p300/CBP have potent histone ace-
tyl transferase activity (Ogryzko et al., 1996). Hormone
binding regulates these activities by repositioning helix
12 to complete the AF2 binding surface (Moras and
Sequence conservation of NR AF2 reflects a common
function of coactivator binding. The androgen receptor
(AR) AF2 has weak activity in mammalian cells but re-
cruits SRC coactivators when highly expressed as in
recurrent prostate cancer (Gregory et al., 2001). AR AF2
nal region (He et al., 2000) and AR coregulatory proteins
(He et al., 2002b; Hsu et al., 2003). Interaction of the AR
FXXLF motif23FQNLF27with AF2 is androgen dependent
and mediates the NH2- and carboxy (C)-terminal (N/C)
interaction (Langley et al., 1998). An additional NH2-ter-
minal WXXLF binding motif433WHTLF437contributes to
the N/C interaction by binding AF2 (He et al., 2002a).
These interdomain interactions are important in regulat-
ing some but not all androgen-dependent genes in tran-
sient reporter assays.
Here we report the molecular basis for FXXLF and
LXXLL motif binding to AF2 based on a comparison of
peptide bound and peptide-free AR LBD crystal struc-
tures and site-directed mutagenesis. We demonstrate
that AR 20-30 FXXLF and TIF2-III 740-753 LXXLL motifs
bind AR AF2, but only the FXXLF motif peptide binds
with an intact primary charge clamp and better recogni-
tion conferring hydrophobic contacts than does the
LXXLL motif. Shown are key residues that differentiate
FXXLF motif binding and a functional AR mutation in
prostate cancer that recovers LXXLL motif binding. The
data suggest a transition in dominant transactivation
domains from AF2 to AF1 during NR evolution.
Bin He,1,2,3,9Robert T. Gampe, Jr.,7Adam J. Kole,7,8
Andrew T. Hnat,1,3Thomas B. Stanley,5Gang An,5
Eugene L. Stewart,6Rebecca I. Kalman,1,3
John T. Minges,1,3and Elizabeth M. Wilson1,2,3,4,*
1Laboratories for Reproductive Biology
2Lineberger Comprehensive Cancer Center
3Department of Pediatrics
4Department of Biochemistry and Biophysics
University of North Carolina at Chapel Hill
Chapel Hill, North Carolina 27599
5Gene Expression and Protein Biochemistry
Research Triangle Park, North Carolina 27709
8Department of Chemistry
Durham, North Carolina 27710
The androgen receptor (AR) is required for male sex
development and contributes to prostate cancer cell
survival. In contrast to other nuclear receptors that
bind the LXXLL motifs of coactivators, the AR ligand
main interaction with the AR FXXLF motif. Reported
here are crystal structures of the ligand-activated AR
ligand binding domain with and without bound FXXLF
and LXXLL peptides. Key residues that establish motif
binding specificity are identified through comparative
structure-function and mutagenesis studies. A mecha-
nism in prostate cancer is suggested by a functional
AR mutation at a specificity-determining residue that
recovers coactivator LXXLL motif binding. An activa-
tion function transition hypothesis is proposed in which
an evolutionary decline in LXXLL motif binding paral-
lels expansion and functional dominance of the NH2-
terminal transactivation domain in the steroid recep-
AR FXXLF Peptide-AR LBD-R1881 Structure
Human AR LBD bound to R1881 was cocrystallized with
peptide 740-753, and without peptide. The resulting
monomeric structures contain 12 ? helices and 4 small
? strands assembled into the familiar 3-layer ?-helical
structure (Figures 1A and 1B). The arrangement resem-
bles structures of the progesterone receptor (PR) (Wil-
liams and Sigler, 1998), glucocorticoid receptor (GR)
(Bledsoe et al., 2002), AR (Matias et al., 2000; Sack et
al., 2001), and other NRs (Gampe et al., 2000). R1881 is
bound in the ligand binding pocket in a mode consistent
with that of Matias et al. (2000). The characteristic A-
and D ring hydrogen (H) bonding network and other key
interactions are maintained.
tivators that regulate hormone-dependent differentia-
tion (Tsai and O’Malley, 1994; Chawla et al., 2001) and
increase gene activity by recruiting coactivators that
assist in chromatin remodeling (Glass and Rosenfeld,
2000). Steroid receptors, a subgroup of the NR super-
in sequence and its activity is receptor and cell-type
and Cellular Biology, Baylor College of Medicine, One Baylor Plaza,
Houston, Texas 77030.
Figure 1. Structures of the AR FXXLF and TIF2-III LXXLL Peptides Bound to AR-R1881
(A) Global architecture of AR 20-30 FXXLF (magenta ribbon) and R1881 (space filled atoms yellow carbon and red oxygen) bound to AR LBD
(gray ribbon) with helices 3, 3?, 4, and 12 (green ribbon) in AF2. Conserved charge-clamp residues are K720 (blue) in helix 3 and E897 (red)
in helix 12.
(B) Global architecture of TIF2-III 740-753 LXXLL (yellow ribbon) and R1881 (as in [A]) bound to AR LBD (orange ribbon) with helices 3, 3?, 4,
and 12 (green ribbon) in AF2.
(C) A ?2.0 A˚C-terminal shift of bound TIF2-III (yellow ribbon) relative to bound AR FXXLF (magenta ribbon) by superimposition of (A) and (B)
(backbone root-mean-square distance [rmsd] 0.21 A˚).
FXXLF and LXXLL Motif Binding Specificity to AR
Table 1. Crystallographic Data and Refinement Statistics
CrystalsAR R1881 AR20-30 AR R1881 TIF2-IIIAR R1881 None
I/? (last shell)
Rmsd bond lengths (A˚)
Rmsd bond angles (?)
Total nonhydrogen atoms
a ? 54.9, b ? 66.1, c ? 70.4
a ? 54.6, b ? 66.7, c ? 69.4
a ? 56.9, b ? 65.8, c ? 72.6
Rmsd is the root-mean-square deviation from ideal geometry.
aRsym? ?|Iavg ? Ii|/?Ii
bR factor ? ?|FP? FPcalc|/?Fp, where Fpand Fpcalcare observed and calculated structure factors; Rfreewas calculated from a random set of
reflections that were excluded from refinement and R factor calculations.
Excellent quality electron density accounts for AR
FXXLF residues 21–30 (Figure 2A and Table 1) and is
clearly visible as a two-turn amphipathic ? helix where
F23, L26, and F27 are directed toward the LBD surface
and Q24, N25, and Q28 are exposed to the solvent (Fig-
ure 2C). FQNLF is H bonded NH2- to C-terminal by the
conserved oxy-steroid E897 (helix 12) and K720 (helix
3), respectively, which reside in clusters of opposite
charge (He and Wilson, 2003). This result is reminiscent
of the charge clamp originally revealed in the SRC-1-
PPAR? (Nolte et al., 1998) and other coactivator bound
NR structures (Darimont et al., 1998; Shiau et al., 1998;
the E897 carboxylate and A22 and F23 amides. Back-
bone carbonyl oxygens from F27 and V30 H bond with
the side chain of K720 (Figure 2A). No interactions are
cluster. Between the charge clusters, ten mostly hy-
drophobic residues from helices 3, 4, and 12 form the
AF2 cleft floor (I898, L712, I737, and V716, Figure 2F),
which is bounded on one side by a low rise from M894,
V713, and K717, and on the other by a helix 4 ridge from
Q738, M734, V730, and Q733. Binding FXXLF increases
the distance between M734 and M894 by ?1.5 A˚,
allowing extensive hydrophobic exposure through face
on interactions from the F23 and F27 phenyl rings with
the helix 4 ridge of the AF2 cleft (Figures 2C and 2F).
Both F23 and F27 adopt similar conformers and lie in a
staggered, almost parallel orientation near M734 that
allows the planar aromatic ring of F23 to contact the
Q738 side chain (carbons C? and C?) and be enclosed
by L712, V716, M734, I737, M894, I898, and L26. All F23
aromatic side chain atoms lie between 3.5 and 3.9 A˚of
the Q738 C? and C? carbons. The planar aromatic ring
of F27 contacts the C? carbon and S? sulfur of M734
and the K720 C? and C? carbons and is enclosed by
V716, V730, Q733, and I737. All F27 aromatic side chain
atoms lie between 3.4 and 4.0 A˚of the M734 side chain,
with additional hydrophobic contact between the L26
side chain and V713, V716, and M894 from helix 12.
Superimposition of backbone heavy atoms indicates
no major rearrangement of the protein backbone or li-
sistent with an induced fit mechanism, FXXLF causes
conformational changes in AR side chains contacting
the peptide (Figure 1D). Charge-clamp residues K720
and E897 move and form H bonds with the peptide.
Side chain interactions between Q738 and F23 induce
conformational changes in V716 and in the flexible
M734. An increased distance from 3.0 to 3.9 A˚between
tions disrupt a H bond network from Q738 to Q902 and
K905 in helix 12 that was observed in the peptide-free
structure. At the start of helix 4, V730 is drawn closer
to F27. Combined interactions of F27 and the FXXLF
C-terminal backbone force K720 toward R726, whose
guanadinium group moves 4 A˚away from its location
in the peptide-free structure (Figures 2C and 2F). A
change of this magnitude may be driven in part by con-
tact between R726 and an adjacent molecule in the
crystal. In helix 12, M894 moves to contact L26 of
23FQNLF27. We conclude that M734, M894, E897, and
(D) Stereoview of AR AF2 showing conformational differences in AR 20-30 bound and unbound states. Amino acid residues of AR (gray)-
R1881-FXXLF (magenta) and AR (yellow)-R1881 without peptide are superimposed (backbone rmsd 0.21 A˚).
(E) Stereoview of AR AF2 showing conformational differences in TIF2-III 740-753 bound and unbound states. Amino acid residues for AR
(orange)-R1881-LXXLL (yellow) and AR (yellow)-R1881 without peptide structures are superimposed (backbone rmsd 0.20 A˚).
(F) Superimposed stereoview of AR AF2 from AR FXXLF (magenta)-AR (gray)-R1881 and TIF2-III (yellow)-AR (orange)-R1881 structures
(backbone rmsd 0.21 A˚). The C-terminal shifted TIF2-III fails to H bond with E897 and forces a conformational change on K720. TIF2-III L745
and L749 C?2 carbons occupy similar space to AR 20-30 F23 and F27 but fail to establish extensive hydrophobic contact afforded by F23
K720 play a prominent role in AR FXXLF binding and
undergo notable conformational change along with
R726 that does not contact the peptide (Figure 1A, 2C,
III peptide and undergo notable conformational change
(Figures 1E, 2D, and 2F).
Superimposition of the AR structures illustrates simi-
larities and differences between the AR 20-30 and TIF2-
III binding modes (Figures 1C, 1F, and 2E). Both amphi-
pathic peptides align along a similar helical axis and
shelter hydrophobic residues within the AF2 cleft. How-
ever, a ?2.0 A˚C-terminal shift along the helical axis for
III from charge-clamp residue AR E897 and preventing
H bonding. The sizeable shift and resulting absence of
H bonding distinguishes TIF2-III from AR 20-30 motif
binding to the AR LBD. TIF2-III induces conformational
changes where AR K720 is pushed away relative to the
FXXLF and peptide-free structures and R726 contrib-
utes a new distant H bond to the TIF2-III L749 backbone
oxygen and a closer one to the AR Q733 side chain.
Despite the sizeable shift of LXXLL, the respective i?1,
i?4, and i?5 leucine residues appear in register and
respectively contact many of the same residues as
mized due to the geometry and conformation of the
branched leucine side chains in relation to the AF2
TIF2 LXXLL Peptide-AR LBD-R1881 Structure
Good quality electron density defines the bound TIF2-
III 740–753 peptide as a two-turn amphipathic ? helix
with L745, L748, and L749 turned toward the LBD sur-
faceand R746and Y747exposed tothe solvent(Figures
1B and 2D, and Table 1). Backbone amides of TIF2-III
745LRYLL749are too distant to H bond with the disordered
carboxylate oxygen atoms of AR E897 in helix 12 or
other residues in the negative charge cluster (Figures
2B and 2D). Lack of an NH2-terminal backbone H bond
to the charge-clamp residue distinguishes LXXLL bind-
ing to AR from previously described coactivator bound
also prevents defining a stable electrostatic interaction
to the proximal TIF2-III N742 N?2, which further H bonds
to N?2 of AR Q738 and participates in the H bond net-
work to helix 12 Q902 and K905 described above. On
the C-terminal end of LXXLL, the backbone carbonyls
of L748, L749, and K751(A) accept a H bond from AR
K720 N?. The side chain of AR R726 moves to strongly
H bond (2.6 A˚) with the AR Q733 side chain and weakly
(3.5 A˚) with the TIF2-III L749 carbonyl. Ten residues in
helices 3, 4, and 12 interact with the branched leucine
similar conformers and lie in a staggered almost parallel
orientation near AR M734. Since fewer side chain atoms
from the branched TIF2-III L745 and L749 are directed
toward the helix 4 ridge, fewer hydrophobic contacts
are made. The single C?2 methyl of L745 makes only
two contacts with the AR Q738 side chain at C? (3.3 A˚)
and C? (3.5 A˚) and is enclosed by AR V716, M734, and
M894. Likewise, TIF2-III L749 C?2 makes two contacts
(3.9 A˚) with AR M734 C? and S? and is enclosed by AR
K720, V730, and I737. Additional hydrophobic contact
occurs between the TIF2-III L748 side chain and AR
M894 in helix 12 and is enclosed by AR V716 and V713.
Like FXXLF, LXXLL peptide binding does not impose
large changes in the global structure, but induces con-
formational changes in AR side chains that contact or
lie near the peptide (Figures 1B, 1E, and 2D). Slight
movement in AR E897 may arise from the adjacent TIF2-
III N742 side chain since it is too distant to H bond
to the peptide backbone. On the other end, AR K720
rearranges to H bond with TIF2-III L748, L749, and the
C terminus. A movement of ?2.0 A˚is observed for AR
R726 as it forms interactions with TIF2-III L749 and AR
Q733 (Figure 2) similar to the secondary charge clamp
described for GR (Bledsoe et al., 2002). However, 3.5 A˚
between the AR R726 side chain and TIF2-III L749 back-
bone oxygen indicates a weak H bond, and 4.4 A˚be-
tween TIF2-III R746 and AR D731 is too distant for a
second direct H bond. Thus, these interactions do not
qualify as a fully intact secondary charge clamp. A mod-
erate change in the flexible AR M734 side chain is seen
as the C?2 methyls from L745 and L749 make contact
with M734. TIF2-III L745 induces a slight change in AR
Q738 and M894 side chains. M734, Q738, M894, K720,
and in particular R726, contribute to binding the TIF2-
Determinants of AF2 Binding Specificity
Consistent with previous cell-based and in vitro binding
studies (He et al., 2001; He and Wilson, 2003), the AR
LBD binds AR 20-30 FXXLF (9.2 ? 0.4 ?M) with higher
affinity than the LXXLL peptide TIF2-III 740-751 (78 ?
3A and 3B). In contrast, ER? LBD preferentially binds
the LXXLL peptide (2.1 ? 0.2 ?M) over the FXXLF pep-
tide (?100 ?M).
Crystal structures and comparative sequence align-
ment (Figure 3C) suggest V730 and M734 discriminate
FXXLF andLXXLL binding.AR-V730I-M734I thatmimics
PR AF2 and AR-V730L-M734V that mimics ER? AF2
decrease binding of the AR FXXLF peptide (Figure 3D).
The ER?-like mutant also decreases binding of FXXLF
sequences from coregulatory proteins ARA54 and
ARA70 (Figure 3E). In contrast, both mutants increase
AR binding of the TIF2-III and SRC1-IV LXXLL peptides
An effect of V730 and M734 on AR FXXLF binding
Luc(Huang etal.,1999)and p21-Luc(Luet al.,1999)that
with ?90% loss of activity by the AR-23FXXAA27mutant
porters but cause little change in MMTV-Luc, a reporter
less dependent on the AR N/C interaction (He et al.,
2002a). The results support AR AF2 residues V730 and
M734 are critical for FXXLF motif binding.
Ligand Dissociation Rates Support the Role
of M734 in FXXLF Motif Binding
Mutations that disrupt the AR N/C interaction exhibit
reduced half-times (t1/2) of androgen dissociation (He et
al., 2000, 2001). [3H]R1881 dissociation from the PR-like
mutant AR-V730I/M734I (t1/2? 72 ? 11 min; Kd? 0.71 ?
0.28 nM) and ER?-like mutant AR-V730L/M734V (t1/2?
FXXLF and LXXLL Motif Binding Specificity to AR
Figure 2. Structural Details for AR FXXLF and TIF2-III LXXLL Peptide-AR LBD-R1881 Complexes
(A) 2Fo? Fcelectron density map (blue) contoured at 1.8? from 1.8 A˚data for bound AR 20-30. Except for NH2-terminal arginine, clearly
ordered electron density is observed for all peptide residues. Carbon atoms are green, oxygen red, and nitrogen blue; annotations for AR
LBD are in yellow, AR 20-30 in orange with intact charge-clamp H bonds in solid orange lines with distances.
(B) 2Fo? Fcelectron density map (blue) contoured at 1.4? from 1.9 A˚data for bound TIF2-III 740-753. Electron density is devoid for K740,
D752, and D753 and poor for L744 and K751 that were built as alanine. H bond interactions are shown with solid orange lines. Excess distance
and/or the disordered E897 carboxylate oxygens prevent description of electrostatic interactions to the TIF2-III backbone amides and the
proximal N742 side chain. Also the D731 to TIF2-III R746 distance is too long to support direct H bonding (dashed yellow lines with distances
and colors as in [A]).
(C) Surface representation of AR AF2 with bound AR 20-30. AR E897, E893, and E709 with K720, K717, and R726 (Roman font) create charge
clusters (positive in blue, negative in red) that flank FQNLF (italicized font). FXXLF is charge clamped by E897 and K720.
(D) Surface representation of AR AF2 bound to TIF2-III. AR E897, E893, and E709 and K720, K717, and R726 (Roman font) create charge
clusters (positive blue, negative red) that flank LRYLL (italicized font). TIF2-III lacks backbone H bonds to AR E897 but H bonds with K720
and AR R726 that moves left to weakly H bond with L749. TIF2-III L745 and L749 make fewer less optimal hydrophobic contacts with Q738,
M734, and V730 located in AF2 (green) helix 4 ridge and K720 as does L748 to AR V713.
(E) Superimposed surface representation of the C-terminal shift of TIF2-III (yellow) to the AR FXXLF (magenta)-AR-R1881 structure (backbone
rmsd 0.21 A˚). TIF2-III LRYLL i?1 (not visible) and i?5 leucines are shifted but in register with corresponding phenylalanines in AR 20-30.
Binding TIF2-III requires AR K720 to move (Figure 2D) allowing AR R726 to move and participate in LXXLL binding.
(F) Surface representation of peptide-free AR AF2. AR E897, E893, E709 and K720, K717, R726 present negative (red) and positive (blue)
charge clusters that flank AF2 (green).
71 ? 7 min; Kd? 1.4 ? 0.63 nM) was faster than from
wild-type AR (t1/2? 109 ? 11 min; Kd? 0.52 ? 0.08 nM)
(Figure 4C) even though equilibrium binding affinities
were not altered. A similarly fast dissociation rate from
AR-M734I (t1/2 ? 75 ? 7 min; Kd ? 0.75 ? 0.09 nM)
indicates M734 is critical for FXXLF motif binding.
increases SRC/p160 coactivator binding (Figure 5C).
The m4 mutant in which four PR AF2 residues are re-
placed by corresponding residues of AR, increases co-
activator binding, but to a lower extent than the single
mutant relative to wild-type. Similarly, GAL-GR-LBD-
I572V increases coactivator interaction (Figure 5D) as
does GAL-GR-LBDm6, but to a lower extent than the
single mutant. Evolving AF2 sequence in PR and GR
ing coactivator recruitment. The decrease is less than
that for AR where evolutionary changes favor FXXLF
A Functional AR Mutation in Prostate Cancer
at Specificity-Determining Residue V730
V730M is a functional somatic mutation that increases
AR transactivation by adrenal androgens (Culig et al.,
1993; Peterziel et al., 1995). Because mutations at V730
ing (Figure 3D, data not shown), we tested whether a
functional mutation at this site alters binding specificity
and coactivator recruitment. We found the interaction
between AR-V730M and GAL-SRC1-IV increases com-
pared to wild-type AR in the presence of R1881, dihy-
drotestosterone (DHT), androstenedione, and andro-
stanediol, but not progesterone (Figures 4D and 4E).
The small decrease in FXXLF binding agrees with ligand
dissociation studies that show AR-V730M (t1/2? 112 ?
22 min; Kd? 0.42 ? 0.15 nM) similar to wild-type AR.
In vitro binding of GST-TIF2 and GST-SRC1 LXXLL
fragments to35S-AR-LBD-V730M also increases in the
presence of DHT or androstenedione compared to wild-
type AR (Figure 4F). But there is no increase in AR-
V730M binding of AR FXXLF in the presence of DHT.
Specificity for FXXLF motif binding in the presence of
high-affinity androgens is also evident. The results sug-
gest somatic prostate cancer AF2 mutant AR-V730M
recovers LXXLL binding with minimal effect on FXXLF
FXXLF Motif Binding by PR and GR Mutants
Structural determinants of AF2 binding specificity were
tested by attempting to convert PR and GR AF2 into an
FXXLF binding site. GAL-PR-LBD and VP-AR1-660 do
not interact, but GAL-PR-LBD-L727V interacts with this
AR NH2-terminal fragment (Figure 5E) and increases
further when AR V713, V730, M734, and I898 replace
corresponding PR residues. A similar set of mutations
in GAL-GR-LBD-m6 increases binding of VP-AR1-660
compared to wild-type GAL-GR-LBD, and binding is
eliminated by the AR FXXAA mutation (Figure 5F). The
results support M734, V730, and V713 as key residues
in AR FXXLF motif binding. It is noteworthy that PR-
LBDm4 and GR-LBD-m6 retain LXXLL binding (Figures
ute to binding specificity.
Transition to AF1
Dependence of NR AF1 activity on length of the NH2-
terminal domain was measured using human NR-GAL4
DNA binding domain fusion proteins expressed in
HepG2 (Figure 6A). We found an exponential increase in
transcriptional activity with NH2-terminal domain length
(R ? 0.99, n ? 9), with a similar trend seen in HeLa, CV1,
and COS cells (data not shown). The data support that
length as it evolves in the NR family.
Declining AF2 Activity in PR, GR, and AR
AR AF2residue V713is present inER?, ER?,and steroid
receptor progenitors in sea lamprey and mollusk, but is
replaced by L727 in PR and I572 in GR (Figure 3C). In
agreement with the structures, we found that V713 is
AR-V713L and AR-V713I mimic PR and GR at this site
and reduce binding of AR FXXLF and coactivator LXXLL
motifs (Figure 5A), suggesting selective pressure main-
tained ancestral V713 for FXXLF binding.
Transition of valine to longer chain residues in PR and
GR raised the possibility that LXXLL binding and AF2
activity decreases during evolution. Based on sequence
alignment and crystal structures (Williams and Sigler,
1998), AR AF2 V713, V730, and M734 correspond to PR
L727, I744, and I748 and GR I572, L589, and M593 (Fig-
ure 3C). Transactivation by GAL-PR-LBD-L727V is
greater than wild-type (Figure 5B) and increases with
the L727V-I748V mutations that restore both ancestral
valines. GAL-GR-LBD-I572V strongly increases trans-
mutant GAL-GR-LBD-I572V-M593V, suggesting that
evolving sequence changes in PR and GR decrease
inherent activity of AF2. Inserting a set of AR AF2 resi-
dues in GAL-PR-LBDm4 and GAL-GR-LBDm6 reduced
We found that evolving sequence changes in PR and
GR that reduce AF2 activity correlate with decreased
binding of the SRC coactivators. GAL-PR-LBD-L727V
Molecular Determinants of FXXLF
and LXXLL Motif Binding
Our data indicate that differential binding affinity and
specificity are established by distinct electrostatic and
hydrophobic interactions revealed in our AR-R1881
crystal structures bound with AR FXXLF and TIF2-III
LXXLL peptides. Through an induced fit mechanism,
AR 20-30 contacts E897 with classical charge-clamp H
bonding that is absent in the bound TIF2-III. Although
new distant contacts form between the TIF2-III and AR,
the interactions apparently provide insufficient energy
to recover what might arise from close NH2-terminal
backbone H bonds with E897. More importantly, selec-
tive high-affinity binding by AR AF2 of phenylalanines
size, and complementarity contribute a substantial non-
polar binding energy.
The crystal structures show that the AR AF2 solvent-
exposed hydrophobic cleft shelters hydrophobic resi-
dues of bound amphipathic peptides from the solvent
FXXLF and LXXLL Motif Binding Specificity to AR
Figure 3. AF2 Determinants of FXXLF and LXXLL Motif Binding Specificity
(A and B) Affinities of FXXLF and LXXLL peptides for AR and ER? LBDs. Binding of ER? (?) and AR (?) LBDs to AR 20-30 ([A], FXXLF) and
TIF2-III 740-751 ([B], LXXLL) and fluoroscein-labeled peptides were measured by fluorescence polarization.
human GR P04150, human mineralocorticoid receptor NP000892, sea lamprey corticoid receptor AY028457, human ER? P03372, human ER?
NP001428, sea lamprey ER AY028456, California sea hare Aplysia californica mollusk ER AY327135, and human retinoic acid receptor ?
P10276. AR AF2 residues involved in FXXLF motif binding are shaded.
(D) Substitution of PR and ER residues in AR AF2. Two-hybrid assays in HepG2 cells with and without 10 nM R1881 used 5?GAL4Luc 3 and
10 ng/well pCMVhAR, AR-V730I-M734I, or AR-V730L-M734V, with 50 ng/well of GAL0, GAL-AR20-30 (ARFx), GAL-TIF2-738-756 (TIF2-III, 3rd
LXXLL), or GAL-SRC1-1428-1441 (SRC-IV, 4thand C-terminal LXXLL). Inset: schematic of two-hybrid assay for FXXLF motif binding by AR.
(E) Reduction in coregulator FXXLF motif binding by ER-like AR mutant. Two-hybrid assays in HepG2 cells with and without 10 nM R1881
used 10 ng/well pCMVhAR or AR-V730L-M734V with 50 ng/well GAL0, GAL-ARFx, GAL-ARA54-447-465 (ARA54Fx), or GAL-ARA70-321-
Figure 4. Promoter Dependence and Prostate Cancer Functional Mutant AR-V730M
(A and B) Dependence on the AR NC interaction. Transcriptional activity of AR and mutants was determined in HepG2 cells with and without
0.1 nM R1881 using 50 ng/well pCMVhAR (WT), AR-V730I/M734I, V730L/M734V or L26A/F27A (FXXAA) with PSA-Enh-Luc, MMTV-Luc, and
(C) Reduced AR N/C interaction and increased androgen dissociation. COS cells transfected with pCMVhAR or AR-V730M, M734I, V730I-
M734I, or V730L-M734V were incubated with 10 nM [3H]R1881 and dissociation rates measured.
(D)Increase inLXXLL bindingby prostatecancerAR mutant.Two-hybrid assaysinHepG2 cellswith andwithout 10nMR1881 used5?GAL4Luc
and 10 ng/well pCMVhAR or AR-V730M with 50 ng/well GAL0, GAL-AR20-30 (GAL-ARFx), or GAL-SRC1-1428-1441 (GAL-SRC-IV).
(E) Effects of steroids on LXXLL binding. Two-hybrid assays in HeLa cells were performed with and without 10 nM DHT, androstenedione
(AD), androstanediol (OL), or progesterone (P) using 50 ng/well GAL-SRC1-IV and 10 ng/well of pCMVhAR or AR-V730M.
(F) Increase in LXXLL binding in vitro. Partially purified GST-0, GST-TIF2-624-1141 (TIF2-M), GST-SRC1-1139-1441 (SRC-C), and GST-AR4-
52 (AR-FXXLF) were incubated with
have 30% of the reaction.
35S-AR624-919 (WT) and AR-LBD-V730M with and without 1 ?M DHT or androstenedione. Input lanes
FXXLF and LXXLL Motif Binding Specificity to AR
Figure 5. Evolutionary Decline in AF2 Activity and FXXLF Motif Binding by PR and GR Mutants
(A) Role of AR V713. Two-hybrid assays in HepG2 cells with and without 10 nM R1881 used 5?GAL4Luc3 and 5 ng/well pCMVhAR, AR-V713L,
or V713I with 50 ng/well GAL0, GAL-AR20-30 (ARFx), GAL-TIF2-738-756 (TIF2-III), or GAL-SRC1-1428-1441 (SRC-IV).
(B) Inherent transcriptional activity of PR and GR AF2 mutants. HeLa cells were assayed using 50 ng/well GAL-PR-LBD (residues 636–933)
or mutants L727V, L727V-I748V, or L727V-I744V-I748M-V912I (m4) with and without 1 nM R5020 or with 50 ng/well GAL-GR-LBD (residues
486–777) or mutants I572V, I572V-M593V, or G568E-V571L-I572V-A574V-L589V-L596I (m6) with and without 10 nM dexamethasone (DEX).
(C) Increase in LXXLL binding by PR AF2 mutants. Two-hybrid assays in HepG2 cells with or without 10 nM R5020 used 50 ng/well GAL-PR-
LBD or mutants L727V or L727V-I744V-I748M-V912I (m4) with 50 ng/well pNLVP16 (0), VP-TIF2-624-1287 (TIF2.1), VP-TRAM1-604-1297
(TRAM1.1), or VP-SRC1-568-1441 (SRC).
(D) Increase in LXXLL binding by GR AF2 mutants. Two-hybrid assays in HepG2 cells with and without 10 nM DEX used 50 ng/well GAL-GR-
LBD (residues 486–777) or mutants I572V or G568E-V571L-I572V-A574V-L589V-L596I (GR-LBDm6) with 50 ng/well VP-0, VP-TIF2.1, VP-TRAM1.1,
Figure 6. DirectCorrelationbetweenAF1Ac-
tivity and Length, and Models of NR Trans-
(A) AF1 activity of NR NH2-terminal regions.
185, RXR?1-134, PPAR?1-108, RAR?1-87,
PXR1-40, and VDR1-23 expressed as GAL4
DNA binding domain fusion proteins (150 ng/
well) were assayed in HepG2 cells using
(B) NR AF2 to AF1 transition. The data sup-
port an evolutionary transition from AF2 to
AF1 as the dominant activation function from
nonsteroid NRs to steroid receptors. Transi-
tionto AF1parallelsexpansion andsequence
ulate that the requirements for high-affinity
hormone binding impose structural con-
straints on AF2 that limit evolutionary diver-
sity in gene regulation. The NH2-terminal re-
significance during steroid receptor evolu-
tion. ER is intermediate between nonsteroid
NRs and steroid receptors.
(C) Schematic of AR gene regulation. AR
transactivation is inhibited by loss-of-func-
tion mutations that cause the androgen in-
increases from gain-of-function mutations in
prostate cancer (PC). An AR mutant in pros-
tate cancer reverts to increased binding of
SRC/p160 coactivators LXXLL motif.
shell. The 12 aromatic phenyl carbons in FQNLF present
a larger hydrophobic contact surface than does L745
and L749 of TIF2-III. Conformational changes to AR
M734 from F23 and F27 and M894 from L26 widens
AF2 with FXXLF compared to LXXLL. The gap distance
between M734 and M894 sulfur atoms is 10.4 A˚ for
FXXLF, 8.4 A˚for LXXLL, and 8.9 A˚without peptide (Fig-
ure 2F). The narrower gap may reflect a poorer hy-
drophobic match between TIF2-III LRYLL and AR AF2
and may also contribute to the C-terminal shift and lost
backbone H bonds to E897. Steric hindrance from a
motif flanking residue such as L744(A) could also not
be inferred due to the absence of electron density be-
yond the C? carbon. For the bound23FQNLF27, specific
enhancement of hydrophobic interactions may favor-
ably position AR20-30 to H bond with E897 in helix 12.
The F23 C? carbon lies ?1 A˚closer to I737 in the AF2
floor than the TIF2-III L745 C?2 methyl in the bound
745LRYLL749structure. This orientation provides exten-
sive and improved nonpolar contacts from F23 to Q738,
L26 to M894 and V713, and from F27 to M734 and V730
in the helix 4 ridge.
Prominentroles ofARM734, V730,andV713 inFXXLF
ation studies. In the PR-like mutant AR-M734I, the
smaller more rigid isoleucine increases ligand dissocia-
tion, suggesting important nonpolar interactions be-
tween M734 and F27 are altered. Faster ligand dissocia-
tion from the ER-like AR mutant V730L-M734V implies
that the larger rigid leucine and smaller rigid valine at
(E) PR AF2 mutants bind AR FXXLF. Two-hybrid assays in HepG2 cells with and without 10 nM R5020 used 50 ng/well GAL-PR-636-933 (PR-
LBD) or mutants L727V, L727V-V912I, I744V-I748M, or L727V-I744V-I748M-V912I with 50 ng/well pNLVP16 (VP-0) or VP-AR1-660.
(F) GR AF2 mutants bind AR FXXLF. Two-hybrid assays in HepG2 cells with and without 10 nM DEX used 50 ng/well GAL-GR486-777 (GR-
LBD) or mutant G568E-V571L-I572V-A574V-L589V-L596I (GR-LBDm6) with 50 ng/well VP-0, VP-AR1-660, or VP-AR1-660-FXXAA.
FXXLF and LXXLL Motif Binding Specificity to AR
F27 destabilize FQNLF interactions with AF2. Slightly
faster ligand dissociation rates also occur for PR-like
mutant AR-V730I-M734I. Increased binding of LXXLL by
these AR mutants indicates M734 and V730 contribute
to peptide recognition.
The contribution of V730 to FXXLF recognition is fur-
modates the bulky F27, establishing good complemen-
tary shape and distance separation. In contrast, ER, PR,
and GR have the larger isoleucine or leucine relative to
AR V730, preferentially binding LXXLL with its smaller
i?5 leucine. The even larger methionine in the AR pros-
tate cancer mutant V730M improves LXXLL binding but
does not greatly affect FXXLF binding. The flexible me-
thionine side chain may improve hydrophobic interac-
tions when presented with an i?5 leucine from LXXLL
or adapt to an i?5 phenylalanine from FXXLF. From the
crystal structures, sufficient space accommodates the
various conformations of a methionine side chain that
could account for this exception. Mismatching FXXLF
also lead to unfavorable contacts between side chains.
Reduced interaction between FXXLF and AR V713L or
V713I that mimic PR and GR likely result from unfavor-
able interactions between the i?4 L26 and the larger
mutated side chain. The data indicate AR M734, V730,
and V713 contribute to FXXLF recognition and prefer-
ence through optimized residue matching and comple-
mentary hydrophobic shape.
Other regions of the LBD impose allosteric effects
on coactivator binding depending on the bound ligand
(Shulman et al., 2004; Nettles et al., 2004). AR favors
FXXLF binding when bound to DHT and LXXLL with the
partial agonist androstenedione (Gregory et al., 2001).
When AR V730 and M734are changed to corresponding
residues in PR and ER, AR binding of SRC1 and TIF2
ever, transcriptional activities of the AR mutants remain
weak compared to PR or GR. Activities of PR and GR
LBDs are greater than the AR LBD when mutated at
multiple sites to mimic the AR AF2 surface, suggesting
additional determinants of NR AF2 activity.
AR AF2 preferentially binds FXXLF and other steroid
receptors bind LXXLL (Heery et al., 1997), but AF2 bind-
ing is not exclusive to a single motif. We show that AR
affinity (Chang et al., 1999). Mutated PR and GR AF2
bind both FXXLF and LXXLL motifs. Adaptability of AF2
is supported by variant sequences FXXLL (Huang et al.,
1998) and LXXIL (Li et al., 1999) that mediate coregula-
nonsteroid NRs such as RXR and RAR also tend toward
NH2-terminal expansion and increased AF1 activity, al-
though AF2 typically predominates (Nagpal et al., 1993).
ER? is evolutionarily intermediate between nonsteroid
NH2-terminal length and cell-type-dependent AF1 and
AF2 activities (Metzger et al., 1992; Tremblay et al.,
regions with potent AF1 activity. AF1 of Nurr1 mediates
strong autonomous transactivation in various cells
(Nordzell et al., 2004), whereas transactivation by its
LBD is cell-type dependent (Castro et al., 1999). Thus,
evolving NH2-terminal domains occur among liganded
and orphan NRs.
Sequence diversity in the NH2-terminal region con-
trasts the conserved LBDs, where the structural con-
surfaces to LXXLL and FXXLF-like motifs. Evolution of
AF1 as the dominant activation domain can provide
unique interaction sites for tissue and species-specific
ity in gene regulation by AR, PR, GR, and the mineralo-
corticoid receptor, which bind similar DNA response
elements. Our hypothesis for an AF2 to AF1 switch in
dominant activation domain also allows for increased
diversity in gene regulation between receptor isoforms.
In contrast to PR-B, PR-A isless active and can function
as a repressor through progesterone and estrogen sig-
naling pathways, allowing progesterone to activate and
repress gene transcription through separate isoforms
of the same receptor. In this case, the PR LBD functions
more like a regulatory domain than a transactivation
hormone, and retinoic acid that have weaker AF1 activ-
corepressors to actively repress transcription. Seques-
tering steroid receptors by heat shock proteins further
minimizes inadvertent gene activation by AF1 in the ab-
sence of hormone. For ER? in the absence of hormone,
an NH2-terminal A domain LLXXI helix competes with
the corepressor for a hydrophobic cleft in the LBD to
maintain ER? in an inactive state (Metivier et al., 2002).
Selective Advantage in Prostate Cancer
Increased AR activity in prostate cancer is associated
with increased levels of AR (Visakorpi et al., 1995) and
SRC/p160 coactivators, and autocrine signaling (Greg-
ory et al.,2004). Functional AR mutationstend to appear
mann, 2002) and contrast loss-of-function mutations
that cause androgen insensitivity (Quigley et al., 1995)
(Figure 6C). AR transactivation is typically retained in
prostate cancer and for some mutants, transactivation
increases with different steroids (Culig et al., 1993; Pe-
terziel et al., 1995; Tan et al., 1997).
Here we link an AR somatic prostate cancer mutation
to increased LXXLL motif binding and SRC coactivator
recruitment. AR-V730M retains high-affinity binding of
DHT and increased transcriptional activity by adrenal
androgens (Newmark et al., 1992; Culig et al., 1993;
Evolutionary Decline in AF2 Activity
Weak transactivation by AR AF2 (He et al., 1999) results
from evolving sequence changes that reduce LXXLL
binding. Concurrently, AR FXXLF evolved with the ex-
panding NH2-terminal AF1 and avidly binds AF2, further
limiting coactivator recruitment by competitive binding
sequence changes also decrease LXXLL binding by PR
and GR AF2, with increasing size and functional impor-
tance of AF1 (Sartorius et al., 1994). Evolutionarily older
and dissociation rate studies of [3H]R1881 were determined after
transient expression in COS cells (He et al., 2001).
Peterzielet al.,1995)(data notshown). V730Mincreases
LXXLLbinding without reducingFXXLF binding and could
impact early and late stage cancer. Increased coactivator
recruitment by a somatic mutation is a mechanism for
aberrant gene regulation that could provide a selective
growth advantage to prostate cancer cell survival.
The work was supported by Public Health Service Grant HD16910
and cooperative agreement U54-HD35014 of the Specialized Coop-
CA77739 from NCI, and NIH Fogarty International Center grant
R03TW001234 to Frank S. French. X-ray diffraction data collected
at the Industrial Macromolecular Crystallography Association-Col-
laborative Access Team (IMCA-CAT) of the Advanced Photon Source
are supported by companies of the IMCA-CAT under contract with
the Illinois Institute of Technology (IIT), executed through the IIT
Center for Synchrotron Radiation Research and Instrumentation.
Use of the Advanced Photon Source is supported by the US Depart-
ment of Energy, Basic Energy Sciences, Office of Science, under
Contract No. W-31-109-Eng-38.
NH2-terminal 6?His-tagged human AR LBD residues 663–919 with
a thrombin protease site was expressed from pET15b in E. coli
Bl21DE3. Cells were grown overnight at 17?C in nutrient-rich media
amended with 5 ?M R1881 and 1 mM isopropyl-thiogalactopyrano-
side. Overnight thrombin digestion (5 NIH units/mg protein) at 4?C
was performed on pooled, immobilized metal affinity chromatogra-
phy-purified fractions prior to cation exchange and gel filtration
chromatography. The dilute binary complex in 25 mM HEPES (pH
7.5), 0.5 M NaCl, 5 mM DTT, 0.5 mM EDTA, 0.05% ?-n-octogluco-
side, 10% glycerol, and 10 ?M R1881 was used or amended with
2- to 3-fold molar excess of AR 20-30 peptide RGAFQNLFQSV (He
et al., 2002b). The dilute binary complex dissolved in 25 mM HEPES
(pH 7.5), 0.15 M Li2SO4, 10 mM DTT, 0.5 mM EDTA, 0.05%
?-n-octoglucoside, 10% glycerol, and 10 ?M R1881 was amended
with 2- to 3-fold molar excess of TIF2-740-753 LXXLL-III peptide
KENALLRYLLDKDD (Voegel et al., 1998). Samples were filtered and
concentrated to 2–3 mg/ml prior to crystallization.
Received: April 22, 2004
Revised: July 26, 2004
Accepted: August 26, 2004
Published: November 4, 2004
Bledsoe, R.K., Montana, V.G., Stanley, T.B., Delves, C.J., Apolito,
C.J., McKee, D.D., Consler, T.G., Parks, D.J., Stewart, E.L., Willson,
T.M., et al. (2002). Crystal structure of the glucocorticoid receptor
and coactivator recognition. Cell 110, 93–105.
Brunger, A.T., Adams, P.D., Clore, G.M., DeLano, W.L., Gros, P.,
Grosse-Kunstleve, R.W., Jiang, J.S., Kuszewski, J., Nilges, M.,
Pannu, N.S., et al. (1998). Crystallography and NMR system: a new
software suite for macromolecular structure determination. Acta
Crystallogr. D Biol. Crystallogr. 54, 905–921.
Castro, D.S., Arvidsson, M., Bondesson Bolin, M., and Perlmann, T.
(1999). Activity of the Nurr1 carboxyl-terminal domain depends on
cell type and integrity of the activation function 2. J. Biol. Chem.
Chang, C., Norris, J.D., Gron, H., Paige, L.A., Hamilton, P.T., Kenan,
D.J., Fowlkes, D., and McDonnell, D.P. (1999). Dissection of the
LXXLL nuclear receptor-coactivator interaction motif using combi-
natorial peptide libraries: discovery of peptide antagonists of estro-
gen receptors alpha an beta. Mol. Cell. Biol. 19, 8226–8239.
Chawla, A., Repa, J.J., Evans, R.M., and Mangelsdorf, D.J. (2001).
Nuclear receptors and lipid physiology: opening the X-files. Science
Culig, Z., Hobisch, A., Cronauer, M.V., Cato, A.C., Hittmair, A., Rad-
mayr, C., Eberle, J., Bartsch, G., and Klocker, H. (1993). Mutant
androgen receptor detected in an advanced-stage prostatic carci-
noma is activated by adrenal androgens and progesterone. Mol.
Endocrinol. 7, 1541–1550.
P.J., Baxter,J.D., Fletterick, R.J.,and Yamamoto, K.R.(1998). Struc-
ture and specificity of nuclear receptor-coactivator interactions.
Genes Dev. 12, 3343–3356.
Gampe, R.T., Jr., Montana, V.G., Lambert, M.H., Miller, A.B.,
Bledsoe, R.K., Milburn, M.V., Kliewer, S.A., Willson, T.M., and Xu,
E.X. (2000). Asymmetryin the PPAR?/RXR? crystalstructure reveals
the molecular basis of heterodimerization among nuclear receptors.
Mol. Cell 5, 545–555.
Gelmann, E.P. (2002). Molecular biology of the androgen receptor.
J. Clin. Oncol. 20, 3001–3015.
Glass, C.K., and Rosenfeld, M.G. (2000). The coregulator exchange
in transcriptional functions of nuclear receptors. Genes Dev. 14,
Gregory, C.W., He, B., Johnson, R.T., Ford, O.H., Mohler, J.L.,
French, F.S., and Wilson, E.M. (2001). A mechanism for androgen
vation therapy. Cancer Res. 61, 4315–4319.
Crystallization and Data Collection
All crystals grew at 20?C to ?150 ?M in 2 weeks by vapor diffusion
using a 1:1 (v/v) ratio of complex to well solution. Well solution for
AR LBD-R1881 with and without AR 20-30 contained 100 mM Bis-
Tris propane (pH 7.5 or 8.5), with a 0.6–1.2 M gradient of lithium
sulfate. For TIF2-III 740-753 complexes, a solution containing 100
mM Bis-Tris propane (pH 7.9) and 0.6 M Na/K tartrate was used.
Well solutions with 20% glycerol were used to transiently mix the
crystals prior to flash freezing in liquid N2. X-ray diffraction data
were collected at ?180?C with a MAR-345 detector on a Rigaku
RU-H-2R-200 generator, an ADSC 210 detector at the IMCA-CAT,
sector 17ID or MARCCD at sector 32ID at the Advanced Photon
Source. The observed reflections were processed with the HKL2000
package (Otwinowski and Minor, 1997).
Structure Determination and Refinement
The AR-DHT structure (Sack et al., 2001) (access code 1I37) yielded
aconvincing AMoRe(Navaza,2001)molecular replacementsolution
with one AR LBD complex in the asymmetric unit. Multiple cycles
of manual model building were completed with QUANTA (Accelrys,
Inc.) and refined with CNX (Accelrys, Inc.) (Brunger, et al., 1998).
Initial AR-R1881 and AR-R1881-TIF2 structures were determined
from the final AR-R1881-AR 20-30 LBD structure, followed by itera-
tive building and refinement with CNX. Table 1 summarizes the
crystallographic and structure refinement statistics. Superimposi-
tions were performed using the homology modeling package in
Insight (Accelrys, Inc.). Figures 1 and 2C–2F were generated with
PyMol from Delano Scientific (www.pymol.org).
Cell Transfections and Biochemical Measurements
tion (Stanley et al., 2003) using 40 ?M R1881 and 17?-estradiol and
10 nM AR 20-30 (fluorescein-RGAFQNLFQSV) and TIF2-III 740-751
(fluoroscein-KENALLRYLLDK). GAL4-DNA binding domain and
VP16 activation domain fusion peptides were prepared (He et al.,
were by transfection of HepG2 and HeLa cells in 12 well tissue
culture plates using Effectene (Qiagen). MMTV-Luc, PSA-Enh-Luc
(Huang et al., 1999), and p21-Luc (Lu et al., 1999) were used at 0.25
?g/welland5?GAL4Luc3 at0.1?g/well(Heand Wilson,2003).Cells
were incubated with and without hormones for 24 hr at 37?C and
luciferase activity measured (He et al., 2001). GST fusion proteins
and35S-methionine in vitro translated AR LBD (pcDNA3-HA-AR624-
919) were analyzed (He et al., 2002a). Apparent equilibrium binding
FXXLF and LXXLL Motif Binding Specificity to AR
Gregory, C.W., Fei, X., Ponguta, L.A., He, B., Bill, H.M., French,
F.S., and Wilson, E.M. (2004). Epidermal growth factor increases
coactivation of the androgen receptor in recurrent prostate cancer.
J. Biol. Chem. 279, 7119–7130.
He, B., and Wilson, E.M. (2003). Electrostatic modulation of steroid
receptor recruitment of the LXXLL and FXXLF motifs. Mol. Cell. Biol.
He, B., Kemppainen, J.A., Voegel, J.J., Gronemeyer, H., and Wilson,
E.M. (1999). Activation function 2 in the human androgen receptor
ligand binding domain mediates interdomain communication with
the NH2-terminal domain. J. Biol. Chem. 274, 37219–37225.
He, B., Kemppainen, J.A., and Wilson, E.M. (2000). FXXLF and
WXXLF sequences mediate the NH2-terminal interaction with the
ligand binding domain of the androgen receptor. J. Biol. Chem.
He, B., Bowen, N.T., Minges, J.T., and Wilson, E.M. (2001). Andro-
gen-induced NH2- and COOH-terminal interaction inhibits p160 co-
activator recruitment by activation function 2. J. Biol. Chem. 276,
He, B., Lee, L.W., Minges, J.T., and Wilson, E.M. (2002a). Depen-
dence of selective gene activation on the androgen receptor NH2-
and carboxyl-terminal interaction. J. Biol. Chem. 277, 25631–25639.
He, B., Minges, J.T., Lee, L.W., and Wilson, E.M. (2002b). The FXXLF
motif mediates androgen receptor-specific interactions with coreg-
ulators. J. Biol. Chem. 277, 10226–10235.
Heery, D.M., Kalkhoven, E., Hoare, S., and Parker, M.G. (1997). A
signature motif in transcriptional co-activators mediates binding to
nuclear receptors. Nature 387, 733–736.
Hong, H., Kohli, K., Trivedi, A., Johnson, D.L., and Stallcup, M.R.
(1996). GRIP1,a novelmouse proteinthat servesas atranscriptional
coactivator in yeast for the hormone binding domains of steroid
receptors. Proc. Natl. Acad. Sci. USA 93, 4948–4952.
Hsu, C.L., Chen, Y.L., Yeh, S., Ting, H.J., Hu, Y.C., Lin, H., Wang,
X., and Chang, C. (2003). The use of phage display technique for
the isolation of androgen receptor interacting peptides with
(F/W)XXL(F/W) and FXXLY new signature motifs. J. Biol. Chem.
Huang, N., vom Baur, E., Garnier, J.M., Lerouge, T., Vonesch, J.L.,
Lutz, Y., Chambon, P., and Losson, R. (1998). Two distinct nuclear
receptor interaction domains in NSD1, a novel SET protein that
exhibits characteristics of both corepressors and coactivators.
EMBO J. 17, 3398–3412.
Huang, W., Shostak, Y., Tarr, P., Sawyers, C., and Carey, M. (1999).
Cooperative assembly of androgen receptor into a nucleoprotein
complex that regulates the prostate-specific antigen enhancer. J.
Biol. Chem. 274, 25756–25768.
lar NH2-/carboxyl-terminal interactions in androgen receptor dimer-
ization revealed by mutations that cause androgen insensitivity. J.
Biol. Chem. 273, 92–101.
Li, D., Desai-Yajnik, V., Lo, E., Schapira, M., Abagyan, R., and Sam-
uels, H.H. (1999). NRIF3 is a novel coactivator mediating functional
specificity of nuclear hormone receptors. Mol. Cell. Biol. 19, 7191–
crinol. 13, 376–384.
Matias, P.M., Donner, P., Coelho, R., Thomaz, M., Peixoto, C.,
Macedo, S., Otto, N., Joschko, S., Scholz, P., Wegg, A., et al. (2000).
Structural evidence for ligand specificity in the binding domain of
the human androgen receptor. Implications for pathogenic gene
mutations. J. Biol. Chem. 275, 26164–26171.
Metivier, R., Stark, A., Flouriot, G., Hubner, M.R., Brand, H., Penot,
G., Manu, D., Denger, S., Reid, G., Kos, M., et al. (2002). A dynamic
structural model for estrogen receptor-alpha activation by ligands,
emphasizing the role of interactions between distant A and E do-
mains. Mol. Cell 10, 1019–1032.
Metzger, D., Losson, R., Bornert, J.M., Lemoine, Y., and Chambon,
P. (1992). Promoter specificity of the two transcriptional activation
functions of the human oestrogen receptor in yeast. Nucleic Acids
Res. 20, 2813–2817.
Moras, D., and Gronemeyer, H. (1998). The nuclear receptor ligand-
binding domain: structure and function. Curr. Opin. Cell Biol. 10,
Nagpal, S., Friant, S., Nakshatri, H., and Chambon, P. (1993). RARs
and RXRs: evidence for two autonomous transactivation functions
(AF-1 and AF-2) and heterodimerization in vivo. EMBO J. 12, 2349–
in AMoRe. Acta Crystallogr. D. Biol. Crystallogr. 57, 1367–1372.
Nettles, K.W., Sun, J., Radek, J.T., Sheng, S., Rodriguez, A.L., Kat-
zenellenbogen, J.A., Katzenellenbogen, B.S., and Greene, G.L.
(2004). Allosteric control of ligand selectivity between estrogen re-
ceptors ? and ?: implications for other nuclear receptors. Mol. Cell
Newmark, J.R., Hardy, D.O., Tonb, D.C., Carter, B.S., Epstein, J.I.,
tor gene mutations in human prostate cancer. Proc. Natl. Acad. Sci.
USA 89, 6319–6323.
Nolte, R.T., Wisely, G.B., Westin, S., Cobb, J.E., Lambert, M.H.,
Kurokawa, R., Rosenfeld, M.G., Willson, T.M., Glass, C.K., and Mil-
burn, M.V. (1998). Ligand binding and co-activator assembly of the
peroxisome proliferator-activated receptor-gamma. Nature 395,
Nordzell, M., Aarnisalo, P., Benoit, G., Castro, D.S., and Perlmann,
T. (2004). Defining an N-terminal activation domain of the orphan
nuclear receptor Nurr1. Biochem. Biophys. Res. Commun. 313,
Ogryzko, V.V., Schiltz, R.L., Russanova, V., Howard, B.H., and Naka-
tani, Y. (1996). The transcriptional coactivators p300 and CBP are
histone acetyltransferases. Cell 87, 953–959.
Onate, S.A., Tsai, S.Y., Tsai, M.J., and O’Malley, B.W. (1995). Se-
receptor superfamily. Science 270, 1354–1357.
Otwinowski, Z., and Minor, W. (1997). Processing of x-ray diffraction
data collected in oscillation mode. Methods Enzymol. 276, 307–326.
Peterziel, H.,Culig, Z.,Stober, J., Hobisch,A., Radmayr,C., Bartsch,
G., Klocker, H., and Cato, A.C. (1995). Mutant androgen receptors
in prostatic tumors distinguish between amino-acid-sequence re-
quirements for transactivation and ligand binding. Int. J. Cancer
Quigley, C.A., De Bellis, A., Marschke, K.B., El-Awady, M.K., Wilson,
E.M., and French, F.S. (1995). Androgen receptor defects: historical,
clinical and molecular perspectives. Endocr. Rev. 16, 271–321.
Sack, J.S., Kish, K.F., Wang, C., Attar, R.M., Kiefer, S.E., An, Y., Wu,
G.Y., Scheffler, J.E., Salvati, M.E., Krystek, S.R., Jr., et al. (2001).
Crystallographic structures of the ligand-binding domains of the
agonist dihydrotestosterone. Proc. Natl. Acad. Sci. USA 98, 4904–
Sartorius, C.A., Melville, M.Y., Hovland, A.R., Tung, L., Takimoto,
G.S., and Horwitz, K.B. (1994). A third transactivation function (AF3)
of human progesterone receptors located in the unique N-terminal
segment of the B-isoform. Mol. Endocrinol. 8, 1347–1360.
tor/coactivator recognition and the antagonism of this interaction
by tamoxifen. Cell 95, 927–937.
Shulman, A.I., Larson, C., Mangelsdorf, D.J., and Rama, R. (2004).
dimers. Cell 116, 417–429.
Stanley, T.B., Leesnitzer, L.M., Montana, V.G., Galardi, C.M., Lam-
bert, M.H., Holt, J.A., Xu, H.E., Moore, L.B., Blanchard, S.G., and
Stimmel, J.B. (2003). Subtype specific effects of peroxisome prolif-
erator-activated receptor ligands on corepressor affinity. Biochem-
istry 42, 9278–9287.
Tan, J., Sharief, Y., Hamil, K.G., Gregory, C.W., Zang, D.Y., Sar, M.,
Molecular Cell Download full-text
Gumerlock, P.H., deVere White, R.W., Pretlow, T.G., Harris, S.E.,
et al. (1997). Dehydroepiandrosterone activates mutant androgen
receptors expressed in the androgen-dependent human prostate
cancer xenograft CWR22 and LNCaP cells. Mol. Endocrinol. 11,
an ancestral estrogen receptor by ligand exploitation and serial
genome expansions. Proc. Natl. Acad. Sci. USA 98, 5671–5676.
Tremblay, G.B., Tremblay, A., Labrie, F., and Giguere, V. (1999).
Dominant activity of activation function 1 (AF-1) and differential
alpha-beta heterodimeric complex. Mol. Cell. Biol. 19, 1919–1927.
Tsai, M.J., and O’Malley, B.W. (1994). Molecular mechanisms of
action of steroid/thyroid receptor superfamily members. Annu. Rev.
Biochem. 63, 451–486.
Visakorpi, T., Hyytinen, E., Koivisto, P., Tanner, M., Keinanen, R.,
Palmberg, C., Palotie, A., Tammela, T., Isola, J., and Kallioniemi,
O.P. (1995). In vivo amplification of the androgen receptor gene and
progression of human prostate cancer. Nat. Genet. 9, 401–406.
Voegel, J.J., Heine, M.J., Tini, M., Vivat, V., Chambon, P., and Gro-
nemeyer, H. (1998). The coactivator TIF2 contains three nuclear
receptor-binding motifs and mediates transactivation through CBP
binding-dependent and -independent pathways. EMBO J. 17,
Webster, N.J., Green, S., Tasset, D., Ponglikitmongkol, M., and
Chambon, P. (1989). The transcriptional activation function located
in the hormone-binding domain of the human oestrogen receptor
is not encoded in a single exon. EMBO J. 8, 1441–1446.
one complexed with its receptor. Nature 378, 392–396.
The Protein Data Bank (http://www.rcsb.org/pdb) accession num-
bers for the crystal structures presented here are 1XOW (FXXLF),
1XQ2 (LXXLL), and 1XQ3 (no peptide).