Báo cáo khoa học: Modulation of glucocorticoid receptor-interacting protein 1 (GRIP1) transactivation and co-activation activities through its C-terminal repression and self-association domains pptx

We provide evidence that the GRIP1 C-terminal region may be involved in regulating its own transactivation and nuclear receptor co-activation activities through primary self-association

protein 1 (GRIP1) transactivation and co-activation

activities through its C-terminal repression and

self-association domains

Pei-Yao Liu1, Tsai-Yuan Hsieh2, Wei-Yuan Chou1and Shih-Ming Huang1

1 Department of Biochemistry, National Defense Medical Center, Taipei, Taiwan

2 Department of Medicine, Division of Gastroenterology, Tri-Service General Hospital, National Defense Medical Center, Taipei, Taiwan

Members of the nuclear receptor (NR) superfamily are

ligand-inducible transcription factors This family

includes the receptors for steroids, thyroid hormone

and vitamin D, as well as orphan receptors for which

no ligands have yet been identified [1,2] Each receptor

has two activation functions (AFs), namely hormone

independent (AF-1) and hormone dependent (AF-2)

The relative importance of AF-1 and AF-2 varies

between different NRs and is influenced by ligand, cell

type and the target gene promoter [3,4] The mechanism

by which DNA-bound NRs regulate transcription

appears to involve the recruitment of co-regulatory proteins, including co-activators and co-repressors [5–8] Co-activators are not usually DNA-binding pro-teins, but are recruited to the promoter through protein–protein contact with transcriptional activators Transcriptional co-repression can involve competition for limiting factors, displacement of positive factors, or histone deacetylation to generate a chromatin structure that limits promoter accessibility [7,8] Therefore, the latest working model regarding transcriptional regula-tion by NRs is an initial associaregula-tion with transcripregula-tional

Keywords

co-activation; GRIP1; HDAC1; nuclear

receptor; transactivation

Correspondence

S.-M Huang, National Defense Medical

Center, Department of Biochemistry, 161,

Section 6, MinChuan East Road, Taipei,

Taiwan 114

Fax: +886 287924057

Tel: +886 227937318

E-mail: shihming@ndmctsgh.edu.tw

(Received 11 December 2005, revised

13 March 2006, accepted 16 March 2006)

doi:10.1111/j.1742-4658.2006.05231.x

Glucocorticoid receptor-interacting protein 1 (GRIP1), a p160 family nuc-lear receptor co-activator, possesses at least two autonomous activation domains (AD1 and AD2) in the C-terminal region AD1 activity appears

to be mediated by CBP⁄ p300, whereas AD2 activity is apparently mediated through co-activator-associated arginine methyltransferase 1 (CARM1) The mechanisms responsible for regulating the activities of AD1 and AD2 are not well understood We provide evidence that the GRIP1 C-terminal region may be involved in regulating its own transactivation and nuclear receptor co-activation activities through primary self-association and a repression domain We also compared the effects of the GRIP1 C terminus with those of other factors that functionally interact with the GRIP1 C ter-minus, such as CARM1 Based on our results, we propose a regulatory mechanism involving conformational changes to GRIP1 mediated through its intramolecular and intermolecular interactions, and through modulation

of the effects of co-repressors on its repression domains These are the first results to indicate that the structural components of GRIP1, especially those of the C terminus, might functionally modulate its putative transacti-vation activities and nuclear receptor co-activator functions

Abbreviations

ACTR, activator for thyroid hormone and retinoid receptors; AD, activation domain; AF, activation function; AR, androgen receptor; CARM1, co-activator-associated arginine methyltransferase 1; CoCoA, coiled-coil co-activator; ER, estrogen receptor; GAC63, GRIP1-associated co-activator 63; GAL4DBD, Gal4 DNA-binding domain; GRIP1, glucocorticoid receptor-interacting protein 1; GST, glutathione S-transferase;

HA, hemagglutinin; HAT, histone acetyltransferase activity; HDAC1, histone deacetylase 1; HMT, histone methyltransferase; NR, nuclear receptor; RLU, relative light unit; SRC-1, steroid receptor co-activator 1; TR, thyroid receptor; TSA, trichostatin A.

co-repressors, followed by recruitment of co-activators

in response to ligands and other signals [9]

There are at least three families of NR co-activators:

CBP⁄ p300; the p160 family; and p ⁄ CAF [7,10,11] The

best characterized of these is a family of three

structur-ally related, but geneticstructur-ally distinct, 160 kDa proteins

called the NR co-activators or p160 co-activators

[12–18] These three proteins are steroid receptor

co-activator 1 (SRC-1), glucocorticoid

receptor-inter-acting protein 1 (GRIP1, also called TIF2), and

activ-ator for thyroid hormone and retinoid receptors

(ACTR) (also called RAC3, pCIP, AIB1 and

TRAM1) These co-activators bind directly to the

DNA-bound NRs and apparently function by

recruit-ing secondary co-activators, such as CBP⁄ p300,

co-activator-associated arginine methyltransferase 1

(CARM1), or related proteins, and possibly by

acety-lating or methyacety-lating histones or other proteins

involved in the transcription machinery [5,10,19–21]

Two separate domains of p160 co-activators can bind

to AF-1 and AF-2 of NRs The p160 co-activators

contain at least three NR-interacting boxes or LXXLL

motifs (where L stands for leucine and X can be any

amino acid) in their central regions, which interact

directly with the highly conserved AF-2 domain of

NRs [22] The C-terminal region of the p160

co-activa-tor can interact with the AF-1 domain of some NRs

and enhance their AF-1 activities in the absence of

lig-ands [23–25]

Recent studies have identified three activation

domains (which transduce the activation signal) in the

p160 co-activator [23,26,27] The enhancement of NR

activity by the p160 co-activator depends on the

CBP⁄ p300 family, which is necessary for the function

of activation domain 1 (AD1) (amino acids 1075–1083

in GRIP1) [26] AD1 receives an activating signal from

DNA-bound NRs and recruits CBP⁄ p300 [23,27]

CBP⁄ p300 may activate the transcription machinery

through its histone acetyltransferase (HAT) activity,

which acetylates histones and other proteins involved

in transcription [28] The second activation domain of

the p160 co-activator, AD2, is located in its far

C-ter-minal region (amino acids 1305–1462 in GRIP1)

[23,26] The mechanism of signalling by AD2 may

involve the weak HAT activity found in two p160

fam-ily members (SRC-1 and ACTR), but not in GRIP1

[14,20] The importance of HAT activity for p160

co-activator function has not been established, and no

efficiently acetylated substrates have yet been reported

CARM1 is a protein with histone methyltransferase

(HMT) activity It mainly binds to the C-terminal

region of GRIP1 to stimulate its AD2 transactivation

function [19] Furthermore, CBP and CARM1 also

support synergistic cross-talk through their HAT and HMT specificities for histones and other transcrip-tional factors [19,21] A third activation domain,

AD3, was recently identified in the highly conserved N-terminal bHLH-PAS domain of p160 co-activators

by recruitment of secondary co-activators, including coiled-coil co-activator (CoCoA) and GRIP1-assoc-iated co-activator 63 (GAC63) [29,30] As CoCoA and GAC63 have no obvious sequence homology, the nature of their downstream targets and the specific components of the transcriptional machinery remain unknown

The mechanisms by which the p160 co-activators function in NR transcriptional activation, and how they are regulated, are not fully understood, and their components have not been identified in detail It remains to be established whether the functions of the p160 co-activator are modulated by post-translational events, such as self-association, protein modification,

or subcellular localization In this article, we present several lines of evidence that demonstrate the func-tional roles of the GRIP1 C terminus in the regulation

of its own transactivation and of NR co-activator activities, which are mediated through its repression and self-association properties Hence, our results pro-vide insights into the regulatory mechanisms control-ling the functional activities of GRIP1 They extend our understanding of the importance of the structural status of GRIP1 in modulating NR functions

Results

Autoregulation of GRIP1 AD activities by its C-terminal region

Previous studies have demonstrated that deletion of the AD1 or AD2 domain of GRIP1 results in selective loss of its co-activator functions in the NR system, affecting specific primary or secondary co-activator functions [23,26] We were interested in establishing whether this involved the structural components of GRIP1 Therefore, we created various GRIP1 frag-ments fused with the yeast Gal4 DNA-binding domain (Gal4DBD) and monitored Gal4-responsive reporter (GK1 reporter) luciferase activity in HeLa cells to assess the transactivation activity of the fragments (Fig 1A,B) In general, the reporter activity of full-length GRIP1 (amino acids 5–1462) was negatively regulated by its structural component (Fig 1, histo-gram 2, compare A and B) We performed western blotting analysis to examine the expression levels of Gal4 fusions and GRIP1 fragments and found poor expression of full-length GRIP1 (Fig 1C, lane 2),

which is also evident from Figs 2B and 5C Although the expression of GRIP1 fragments varied, their trans-activation activities were primarily determined by structural components For example, a C-terminal truncated GRIP1 (amino acids 5–1121) showed greater reporter activity than one N-terminal truncated GRIP1 (GRIP1-563–1462) (Fig 1A, compare histogram 3 with histogram 4), suggesting a repression region in its

C terminus Subsequently, the region encompassing amino acids 1122–1304 was identified as the major repression region in the GRIP1 C terminus (Fig 1A, compare histogram 5 with histogram 6) Furthermore, GRIP1-1013–1121 induced maximal AD1 activity (Fig 1B, histogram 9, compare A and B), which sug-gests that amino acids 563–1012 also constitute a repression region for AD1 activity (compare histogram

5 with histogram 9 of Fig 1B) Similar patterns of transactivation activity were exhibited by these GRIP1 fragments in human embryonic kidney 293 cells, and the identities of AD1, AD2, and at least two repres-sion regions in amino acids 563–1012 and 1122–1304, were consistent with our findings derived from HeLa cells (data not shown)

HDAC1 is involved in the GRIP1 repression complex

Having established the existence of a repression prop-erty of GRIP1, we investigated whether deacetylase activity mediated through the histone deacetylase (HDAC) family was involved in the repression effect First, we treated HeLa cells with 100 ngÆmL)1 trichost-atin A (TSA), an inhibitor of HDAC activity [31], and monitored the changes in reporter activity of Gal4DBD fused with various GRIP1 fragments after

16 h of TSA treatment (Fig 2A) TSA enhanced the reporter activity of the GRIP1 C-terminal fragment (amino acids 1122–1462) (4.5-fold) and suppressed that

of full-length GRIP1 (Fig 2A) We then used glutathi-one S-transferase (GST) pull-down analysis to examine which of the co-repressor proteins, HDAC1, HDAC4, mSin3a or SMRT-a, were involved in the repression complex We found that HDAC1, mSin3a and

SMRT-a interSMRT-acted physicSMRT-ally with two C-terminSMRT-al frSMRT-agments (amino acids 1122–1462 or 1305–1462) (data not shown) In addition, we examined the GRIP1–HDAC1 complex using a co-immunoprecipitation assay in COS7 cells We detected the GRIP1–HDAC1 complex

by immunoprecipitating GRIP1 using a hemagglutinin (HA) antibody or by immunoprecipitating HDAC1 using a myc antibody (Fig 2B) HA antibody immuno-precipitation identified three HDAC1-interacting regions, in GRIP1 residues 563–1121, 5–765, and

AD2 AD2

1122 1305 1462 1462

AD1 AD2

AD1

AD1 AD2

1462 563

AD1 1121 563

AD1 1304 563

2 3 4 5 6 7 8

Luciferase Activity

5 (RLU 10 )

Luciferase Activity (RLU 10 )5

0 1 2 3 120120 160

1x 858x 29000x 290x

1 5 9 10

AD1 1013

1013 1121 1121 AD1 AD2 1462 1013

AD1 1121 563

[Gal4DBD; pM vector]

Luciferase Activity

4 (RLU 10 )

Luciferase Activity (RLU 10 )4

1x 13x 277x 15x

10x

858x 16x

2x

7 8 9 10 10

2

WB anti-Gal4DBD

WB anti-HuR

Mr

170 130 100 72 55

72 55 40 33 24

7 8 9 10 10

2

A

B

C

Fig 1 Modulation of glucocorticoid receptor-interacting protein 1

(GRIP1) transactivation activity (A, B) Expression vectors (0.5 lg)

for the indicated fragments of GRIP1 fused to the Gal4

DNA-bind-ing domain (Gal4DBD) were transiently transfected into HeLa cells

together with the GK1 reporter gene (0.5 lg), which encodes

erase and is controlled by the Gal4 response element The

lucif-erase activity of transfected cell extracts was determined.

Numbers beside the bars indicate fold activation compared with

that of the Gal4DBD alone RLU, relative light units These data are

the average of three experiments (mean ± SD; n ¼ 3) (C) COS-1

cells were co-transfected with various Gal4DBD.GRIP1 fragments

(2 lg) in a six-well plate Cell lysates were subjected to western

blotting analysis and then immunoblotted with anti-Gal4DBD (upper

panel) to determine the GRIP1 expression level and anti-HuR

(bottom panel) to determine the loading control Results shown are

representative of three independent experiments.

1122–1462 (Fig 2B, compare lanes 1, 4, 6, and 7) The

myc immunoprecipitation also contained these GRIP1

fragments (data not shown) Our results with TSA

(Fig 2A) suggested that the HDAC family might be

involved in repression through a

deacetylase-independ-ent pathway Hence, we used a mutant HDAC1 protein

that lacks deacetylase activity and found that the

parti-ally repressive effect on the Gal4 reporter activity was

the same as with wild-type HDAC1 for both full-length

GRIP1 (amino acids 5–1462) and C-terminal GRIP1

(amino acids 1122–1462) in HeLa cells (Fig 2C,

compare the histograms with open and grey columns)

Homo-oligomerization of GRIP1

We examined whether the GRIP1 C terminus can

inter-act inter- or intramolecularly with full-length GRIP1 to

modulate its transactivation response to other GRIP

C-terminal interacting proteins, such as CARM1,

Zac1 and ACTN2 [19,32,33] A co-immunoprecipitation

assay in COS-7 cells showed that Gal4DBD fused to

the full-length GRIP1 (amino acids 5–1462) complexed

strongly with HA.GRIP1-563–1462 and weakly with

HA.GRIP1-5–765 or HA.GRIP1-563–1121 (Fig 3A,

lanes 7, 5 and 8, respectively) Our

co-immunoprecipi-tation analysis suggested that the primary region of

GRIP1 self-association is located at its C terminus,

within amino acids 1122–1462 (Fig 3A, compare lanes

5–8) In a parallel experiment, we were unable to detect

any HA-tag signal by immunoprecipitation using a

mouse anti-IgG antibody (Fig 3A, lanes 9–12) Based

on the results in Fig 3A, we used GST pull-down

assays to confirm this potential self-association motif with different C-terminal regions of GRIP1 (amino acids 1122–1462, 1305–1462, 1122–1304, 1305–1398, 1305–1462 and 1399–1462) These regions were fused to GST and the fusion proteins were immobilized on agarose beads Their ability to bind to a synthesized

AD2

[Gal4DBD; pM vector]

1

2

3

1

2

3

2x

0.3x

4.5x

HDAC1.myc

WB by α-myc

WB by α-myc

WB by α-myc

WB by α-myc

IP by α-HA

IP by α-HA

Input (5%)

HA HA.GRIP15-1462 GRIP15-1121 HA

HA.GRIP1563-1121 HA.GRIP1563-1462 HA.GRIP11122-1462 HA.GRIP15-765 HDAC1.myc

HA HA.GRIP1 GRIP1

5-1462 5-1121

HA

HA.GRIP1 HA.GRIP1 HA.GRIP1 HA.GRIP1 HDAC1.myc

563-1121 563-1462 1122-1462 5-765

+ + + + + + +

+ +

+ + + + +

WB by α-HA

WB by α-HA

66 46 30

97.6220 kDa

1 2 3 4 5 6 7 HDAC1.myc

0 1 2 3

GRIP1

5-1462

Gal4DBD.

GRIP1

5-1462

Gal4DBD GRIP1

1122-1462

Gal4DBD GRIP1

1122-1462

4

none HDAC1 wt HDAC1 mt

A

B

C

Fig 2 GRIP1 physically and functionally interacts with histone

deacetylase 1 (HDAC1) (A) Expression vectors (0.5 lg) for the

indi-cated fragments of GRIP1 fused to the Gal4 DNA-binding domain

(Gal4DBD) (pM vector) were transiently transfected into HeLa cells

along with the GK1 reporter gene (0.4 lg) in the absence or

pres-ence of 100 ngÆmL)1trichostatin A (TSA) for 16 h Numbers above

the bars indicate fold activation compared with that of no TSA

treatment (B) COS-7 cells were co-transfected with various

Gal4DBD.GRIP1 fragments (5 lg) and with HDAC1.myc (5 lg) in a

100 mm Petri dish Cell lysates were subjected to

immunoprecipi-tation with anti-myc (upper panel) immunoglobulin and then

immu-noblotted with anti-hemagglutinin (middle panel) and anti-myc

(bottom panel) immunoglobulin for the loading control for GRIP1

and HDAC1 proteins Results shown are representative of three

independent experiments (C) Expression vectors (0.4 lg) for the

indicated fragments of GRIP1 fused to the Gal4DBD were

transi-ently co-transfected into HeLa cells, together with the GK1 reporter

gene (0.2 lg) with 0.2 lg of wild-type pcDNA3.HDAC1.flag (open

column) or the enzyme-dead HDAC1 mutant (grey column) The

luciferase activity of the transfected cell extracts was determined.

These data (A,C) are the average of three experiments (mean ±

SD; n ¼ 3).

GRIP1 C-terminal fragment (amino acids 1122–1462)

was measured in vitro (Fig 3B) The results indicated

that amino acids 1305–1398 constitute the primary

self-association region in the GRIP1 C terminus (Fig 3B, compare lanes 6–10) The amount of protein pulled down by GST–GRIP1-1305–1462 was greater than that pulled down by GST–GRIP1-1122–1462 (Fig 3B, compare lane 3 with 4) GST–GRIP1-1305–1398 was subsequently used to identify whether other GRIP1 regions interact with this C-terminal region in vitro GST–GRIP1-1305–1398 pulled down full-length GRIP1 (amino acids 5–1462) and C-terminal GRIP1 fragments (amino acids 1122–1462) but not N-terminal (amino acids 5–765) or central (amino acids 563–1121) GRIP1 fragments (Fig 3C) Thus, our in vivo and

in vitro results suggest that GRIP1 might form at least

a homodimer through its C-terminal region

Enhancement of GRIP1 AD1 and AD2 activities by

an exogenously overexpressed GRIP1 C terminus The recent identification of CARM1, Zac1 and ACTN2 using GRIP1 amino acids 1122–1462 as bait suggests that the GRIP1-dependent co-activation func-tion of these factors might be mediated through a pro-tein–protein interaction with the GRIP1 C terminus [19,32,33] Hence, we examined the effect of exogen-ously overexpressed full-length GRIP1 (GRIP1-5– 1462), a C-truncated fragment (GRIP1-5–1121) and a C-terminal fragment (GRIP1-1122–1462), on GRIP1 transactivation activity We measured GRIP1 transac-tivation using the Gal4 reporter activities of full-length GRIP1 and a C-terminal GRIP1 fragment (amino acids 1122–1462) fused with the Gal4DBD vector (Fig 4) The full-length and C-terminal GRIP1 frag-ments expressed various levels of enhanced reporter activities in the presence of all Gal4DBD-fused GRIP1 fragments (Fig 4A,B) The C-terminal fragment, GRIP1-1122–1462, expressed greater enhancement than full-length GRIP1 only on the Gal4 reporter activity fused with full-length GRIP1, not C-terminal GRIP1 (Fig 4, compare histogram 2 with histogram 4) This suggests that GRIP1-1122–1462 might mediate its enhancement effect on full-length GRIP1 both through its C terminus and through other regions A C-truncated GRIP1 had no or a little enhancement effect on the Gal4 reporter activities (Fig 4, compare histogram 1 and histogram 3)

We then used a series of C-terminal truncations to explore the importance of the GRIP1 C-terminal region

in the regulation of GRIP1 transactivation activity (Fig 5) The results suggested that residues 1161–1280 constitute the primary repression region for AD1 trans-activation activity (Fig 5A, compare histograms 6–9)

We also found that GRIP1-truncated fragments associ-ated with full-length GRIP1 in a sequence-dependent

1122 1462

1121 563

GRIP1

1 2 3 4 5 6 7 8 9 10 1011 11 12

HA.GRIP1 563-1462

HA.GRIP1 563-1462

HA.GRIP1 5-765

HA.GRIP1 5-765

HA.GRIP1 5-1 121

HA.GRIP1 5-1 121

HA.GRIP1 563-1 121

HA.GRIP1 563-1 121

Gal4DBD.GRIP1 5-1462

Gal4DBD.GRIP1 5-1462

+ + + + + + + +

+ + + + + + + +

+ + + + + + + +

Input 5%

IP by α-Gal4DBD

IP by -Gal4DBD α

IP by α-IgG

IP by -IgG α

WB by α-HA

WB by α-HA

NS

97.6 66 46

1305-14621305-1462 1399-14621399-1462

GST-GRIP1

GRIP1 1122-1462

GRIP1 1122-1462

GST-GRIP1

A

B

C

Fig 3 GRIP1 forms a homodimer under in vitro and in vivo

conditions (A) COS-7 cells were transfected with the Gal4

DNA-binding domain (Gal4DBD) GRIP15)1462 (5 lg) in the

pre-sence of HA.GRIP15)765, HA.GRIP15)1121, HA.GRIP1563)1462, or

HA.GRIP1563)1121(5 lg, in a 100 mm Petri dish) Cell lysates were

subjected to immunoprecipitation with anti-Gal4DBD (lanes 5–8) or

control (normal mouse IgG) (lanes 9–12) immunoglobulin and then

immunoblotted with anti-HA immunoglobulin (B) The protein for the

GRIP1 C-terminal region (amino acids 1122–1462) was translated

in vitro and incubated with bead-bound glutathione S-transferase

(GST)–GRIP1 (amino acids 1122–1462, 1305–1462, 1122–1304,

1305–1398, 1305–1462, and 1399–1462) fusion proteins or with GST

alone; bound proteins were eluted, separated by SDS ⁄ PAGE, and

visualized by autoradiography (C) The proteins for the GRIP1

frag-ments were translated in vitro and incubated with bead-bound GST–

GRIP11305)1398 fusion protein or GST alone; bound proteins were

eluted, separated by SDS ⁄ PAGE, and visualized by autoradiography.

Results shown are representative of three independent experiments.

manner in the mammalian two-hybrid analysis

(Fig 5B), and that amino acids 1350–1400 constituted

the primary association site of GRIP1 (Fig 5B,

com-pare histogram 4 with histogram 5) Hence, the enhancement effect on transaction activities of these C-terminal truncations by exogenous full-length or

1 2 3 4

1 2 3 4

[ pSG5.HA vector]

1122 1462 AD1 AD2

AD2 AD1

Gal4DBD

1122

1122 1462 1462

AD2

1x

69x

2.8x

111x

1x

39x

1.3x

13x

1 2 3 4

Luciferase Activity

4

(RLU 10 )

Luciferase Activity (RLU 10 ) 4

Luciferase Activity

3

(RLU 10 )

Luciferase Activity (RLU 10 ) 3

Fig 4 The C-terminal region of GRIP1 is the primary regulatory region for GRIP1 transactivation activities Expression vectors (0.4 lg) for the indicated fragments of GRIP1 (A, amino acids 5–1462; and B, amino acids 1122–1462) fused to the Gal4 DNA-binding domain (Gal4DBD) were transiently transfected into HeLa cells together with the GK1 reporter gene (0.2 lg) in the presence of 0.2 lg of pSG5.HA vector and the indicated fragments of GRIP1 in the pSG5.HA vector The actual luciferase activities measured for each histogram were as follows: for Gal4DBD.GRIP1 5 )1462, 3.3· 10 3 ± 5 relative light units (RLU) and for Gal4DBD.GRIP1 1122 )1462, 1.7· 10 2 ± 18 RLU Numbers above the bars indicate fold activation compared with that of the ratio related pM.GRIP1 to pM vector These data are the average of three experi-ments (mean ± SD; n ¼ 3).

Luciferase Activity

3

(RLU 10 )

Luciferase Activity (RLU 10 ) 3

0 3 6 9 12 15 1

2 3 4 5 6 7 8 9 10

1x

2.7x

3.3x

2.1x

2.6x

4x

17x

28x

46x

39x

1 2 3 4 5 6 7 8 9 10

Gal4DBD

1 2 3 4 5 6 7 8 9 10

0 5 10 15 20 25 pVP16.GRIP1/pVP16

10

Gal4DBD.GRIP1 fragment

WB anti-Gal4DBD

WB anti-HuR

Mr

Mr

170 130

C

Fig 5 Residues 1161–1280 are the primary

repression region in the GRIP1 C terminus.

Expression vectors (0.4 lg) for the

trun-cated fragments of GRIP1 fused to the Gal4

DNA-binding domain (Gal4DBD) were

transi-ently transfected into HeLa cells together

with the GK1 reporter gene (0.2 lg) (A) in

the presence of 0.2 lg of pVP16 vector or

pVP16.GRIP1 (B) Luciferase activity of the

transfected cell extracts was determined.

Numbers beside the bars indicate fold

acti-vation compared with that of the Gal4DBD

vector These data are the average of three

experiments (mean ± SD; n ¼ 3) (C) COS-1

cells were co-transfected with various

Gal4DBD.GRIP1 fragments (2 lg) in a

six-well plate Cell lysates were subjected to

western blotting analysis and then

immuno-blotted with anti-Gal4DBD (upper panel) for

GRIP1 expression and anti-HuR (bottom

panel) immunolglobulin for the loading

con-trol Results shown are representative of

three independent experiments.

C-terminal GRIP1 also depended on the sequence

con-stitution in the C-terminal region (data not shown)

Furthermore, the low expression levels of GRIP1

frag-ments, such as amino acids 5–1462, 5–1430 and 5–1400

(Fig 5C), suggest that the expression level was not the

primary factor because the GRIP1-5–1200 induced

higher transactivation activity than GRIP1-5–1350

(Fig 5A,C, compare histograms and lanes 5 with 8)

GRIP1 C terminus functions as a

GRIP1-dependent NR co-activator in HeLa cells

Because the C-terminal region of GRIP1 is involved in

the repression of transactivation activity and

self-association of GRIP1 (Figs 1–5), we examined the

relationship between transactivation and co-activation

of GRIP1, using a series of C-truncations to monitor

its co-activator functions in the androgen receptor (AR), estrogen receptor (ER) and thyroid receptor (TR) systems (Fig 6) Our previous study suggests that GRIP1 AD2 activity is necessary for its co-activation

in the AR system, AD1 activity is necessary for its co-activation in the TR system, and cross-talk between

AD1and AD2 activities is necessary for maximal co-activation in the ER system [26] We next examined whether the GRIP1 C terminus itself functions as a secondary (or GRIP1-dependent) co-activator, in a manner similar to that of CARM1, in NR transcrip-tional activation The exogenously co-transfected GRIP1 C terminus, or CARM1 with GRIP1, further enhanced the co-activator function of GRIP1 on var-ious NR transcriptional activations, including AR, ER and TR (Fig 6) In the AR system, the GRIP1 C ter-minus had a stronger enhancement effect than

pSG5.HA

1 2 3 4 5 6 7 8

1 2 3 4 5 6 7 8

1 2 3 4 5 6 7 8

1

2

3

4

5

6

7

8

none

CARM1

WB anti-HA

WB anti-HuR

170 130 100 72

Fig 6 The GRIP1 C terminus serves as the GRIP-dependent nuclear receptor (NR) co-activator HeLa cells were transfected with the repor-ter plasmid [0.25 lg of MMTV-LUC vector for androgen receptor (AR) (A), EREII-LUC vector for estrogen receptor (ER) (B), and MMTV[TRE]-LUC vector for thyroid receptor (TR) (C)] and the NR expression vector [0.15 lg of AR (A), 0.04 lg of ER vector (B) and 0.04 lg of TR vector (C)] Transfected cells were grown with 100 n M dihydrotestosterone (A), 100 n M estradiol (B) or 100 n M 3,5,5¢-triido- L -thryonine (C) Expres-sion vectors (0.35 lg) for the indicated fragments of GRIP1 fused to the pSG5.HA were transiently transfected into HeLa cells together with GRIP11122)1462(open column) or CARM1 (grey column) The luciferase activity of transfected cell extracts was determined Numbers beside the bars indicate fold activation compared with that of the pSG5.HA vector alone without co-activator co-transfection These data are the average of three experiments (mean ± SD; n ¼ 3) (D) COS-1 cells were co-transfected with various HA.GRIP1 fragments (2 lg) in a six-well plate Cell lysates were subjected to western blotting analysis and then immunoblotted with anti-HA (upper panel) for GRIP1 expression and anti-HuR (bottom panel) immunoglobulin for the loading control The results shown are representative of three independent experiments.

CARM1 (Fig 6A, compare open with grey columns),

whereas CARM1 had a stronger effect on TR

tran-scriptional transactivation than the GRIP1 C terminus

(Fig 6C, compare open with grey columns) No

GRIP1-dependent TR co-activator effect by the

GRIP1 C terminus was observed in GRIP1 fragments

containing amino acids 1161–1462 (Fig 6C, compare

histograms 2–6, open columns) In the ER

transcrip-tional system, the particular sequence that was

trun-cated determined the effectiveness of the GRIP1

C terminus or CARM1 on GRIP1 co-activator

func-tion (Fig 6B, compare open and grey columns) The

expression levels of various HA-tag fused GRIP1

frag-ments were similar to those of the respective

Gal4DBD-tag fused GRIP1 fragments, including poor

full-length GRIP1 expression (Fig 6D, lane 2) The

protein level of GRIP1 fragments was not the primary

factor for NR co-activator function, because amino

acids 5–765 could not serve as a NR co-activator, even

when present at a higher level (Fig 6, compare

histo-grams, and lane 2 with lane 8)

Discussion

Autoregulation of GRIP1 transactivation activity

To date, some of the functions of the N- and

C-ter-mini of p160 co-activators were unclear Recently,

Stallcup’s laboratory identified two new GRIP1

N-ter-minal interacting proteins, CoCoA and GAC63

[29,30] In this study, we investigated the regulation of

GRIP1 transactivation and co-activation activities by

its own C terminus through the repression and

self-association motifs Our work showed that the major

masking effect of the GRIP1 C terminus on GRIP1

transactivation functions could be overcome by

exo-genous co-expression of the GRIP1 C terminus, but

not by the GRIP1 N-terminal fragment (Fig 4) The

enhancement of GRIP1 transactivation activities of

AD1 and AD2 might be mediated either through

trun-cation or overexpression of its C-terminal region

(Figs 1, 4 and 5) These effects differed from those

induced by other general GRIP1-dependent

co-activa-tors, such as CBP and CARM1 Generally, CBP and

CARM1 regulate the co-activator functions of the

p160 co-activator in NR systems both through

pro-tein–protein interaction and through their catalytic

effects (acetylation and methylation, respectively) on

histones or other transcriptional factors [19,34–36]

There are no reports showing that the C-terminal

region of GRIP1 has specific enzymatic activity in

modulating basal transcriptional machinery In

addi-tion, the effects of CBP or CARM1 on GRIP1 AD1

or AD2 activity differed from those of the GRIP1

C terminus (data not shown)

The repression region of the GRIP1 C terminus might recruit the co-repressor family (Fig 2B and data not shown) The deacetylase inhibitor, TSA, only func-tioned with the GRIP1 C-terminal fragment (amino acids 1122–1462), and not with full-length GRIP1, sug-gesting the existence of a mechanism that is different from the deacetylase activity of HDAC1 (Fig 2A) The similarity between the repression effect on GRIP1 transactivation function by wild-type HDAC1 and its enzyme-dead mutant suggested that a protein–protein interaction was involved, not deacetylase activity (Fig 2B,C) GRIP1 associated with its C-terminal region in the co-immunoprecipitation analysis and GST pull-down, but it complexed with the N-terminal and central regions only in the co-immunoprecipitation analysis, not in the GST pull-down analysis (Fig 3) These findings supported the idea that the conforma-tional change of GRIP1 might have resulted from inter- and intramolecular interactions within its C-ter-minal and other regions Hence, the modulation of GRIP1 transactivation and co-activation activities through its C terminus or other exogenous factors (HDAC1 or CARM1) might be mediated through pro-tein–protein interaction, which change the local con-formation of GRIP1 or have downstream effects on basal transcriptional machinery for expressing full GRIP1 co-activator function Figure 7 shows a work-ing model based on our findwork-ings

The functional roles of the GRIP1 C-terminal region

In Figs 1–6, we present several lines of evidence to sup-port the concept that the GRIP1 C-terminal region is involved in the modulation of self-transactivation (AD1 and AD2) and co-activator (AR, ER and TR) functions

in HeLa cells The outcome of the relationship between GRIP1 transactivation and co-activator functions var-ies according to the system under investigation (Figs 5 and 6) In the AR transcriptional system, the GRIP1 co-activator function was destroyed when GRIP1-trun-cated fragments expressed higher transactivation activ-ity because of the loss of intact AD2-dependent function In contrast, GRIP1 transactivation and co-activation activities were correlated in the TR transcrip-tional system and the relationship was independent in the ER transcriptional system We also found that the GRIP1 co-activator function depends not only on the existence of a repression domain or a protein–protein interaction with identified and unidentified factors, but also on the GRIP1 conformation under specific

conditions (Figs 5 and 6) Hence, the linking of

repres-sion and self-association motifs to the GRIP1

confor-mation demonstrated in this study might be explained

by the effect of the co-expressing GRIP1 C terminus on

GRIP1 transactivation and co-activation activities

Our western blotting analysis showed that the

amount of protein expressed by the exogenous GRIP1

fragment was also tightly regulated by its structural

component These findings are consistent with a recent

study conducted by the Hager laboratory, which

dem-onstrated that the C terminus of GRIP1 is essential for

the formation of discrete nuclear foci and 26S

protea-some degradation in gene regulation [37] Similarly to

the regulatory mechanism reported in p53 studies

[38,39], GRIP1 might form a more active

conforma-tion, determined by its relative concentration in cells

The relative concentration of GRIP1 might depend on

its homo-oligomerization status, which is mainly

deter-mined by the involvement of its C-terminal region in protein–protein interactions, including self-association, repression by HDAC1 and other proteins, 26S protea-some degradation, or translocalization Taken together, the effect of GRIP1 C-terminal interacting proteins as

a GRIP1-dependent secondary co-activator might, in part, be mediated through conformational change of the GRIP1 C terminus and subsequent exposure of a working surface, with extra downstream signalling for its transactivation and NR co-activator functions

Experimental procedures

Plasmids

The pSG5.HA vectors coding for full-length GRIP1 (codons 5–1462), other GRIP1 fragments (codons 5–1121 and 1122–1462), and HA.CARM1 have been described

1

1013 1122 1305 1398

1462

1013 1122 1305 1398

1462

1

1305

1398

1462 1122

1013 1122 1305 1398

1462 1122

1305 1398

4

1 62 4

1 62

1122 1305 1398

1 62 4

1 62 4

1013 1122 1305 1398

1462

1

1013 1398

1013 1122 1305 1398

1462 1122 1305

1398

1 2

1 2

1

1013

1122 1305

1398

14 62

AD1

Repression domain

Association domain

AD2

AD3

1305

1305 1398

1462 1122

+

?

+

1 62 4

1 62 4 1

1

1

III

IV

V

1122 1

1013

1122 1305

1398

1462

+

?

1

1013

1122 1305

1398

1462

+

Fig 7 Dynamic model of the potential GRIP1 conformational change mediated through its C terminus We propose that either monomeric (I) or dimeric (or higher oligomeric) (II) GRIP1 might form a distinct conformation in cells One repression (grey circle) and association (dotted circle) are defined in this study AD1 (slant circle), 2 (closed circle), and 3 (open circle) have been previously reported [23,26,28] The expo-sure of any GRIP1 C-terminal interacting protein, including the GRIP1 C terminus in this model, might alter GRIP1 conformation I through intramolecular interaction into conformation III or conformation II through intermolecular interaction into conformation IV (first effect) In this study, the exogenous GRIP1 C terminus dramatically enhanced GRIP1 transactivation activity through the repression and association domains (or the titration of co-repressors), resulting in conformational changes from conformation I (or II) into III or IV In contrast, the extra downstream signal (second effect) of other GRIP1-dependent co-activators might be required for some full GRIP1 NR co-activator functions, for example, the methyltransferase activity of co-activator-associated arginine methyltransferase 1 (CARM1) in this study The question mark indicates that further analyses are necessary to identify the involvement of the GRIP1 N terminus or the status of oligomerization in cells.

previously [19]; GRIP1-563–1121 was constructed by inserting

an EcoRI–SalI fragment of the appropriate PCR-amplified

GRIP1 cDNA into the EcoRI and XhoI sites of the

pSG5.HA vector GRIP1-5–765 and GRIP1-1305–1462

were constructed by inserting EcoRI–XhoI fragments

vector; GRIP1-563–1462 was constructed by inserting an

diges-tion Vectors encoding Gal4DBD fused to various GRIP1

fragments were constructed by inserting EcoRI–SalI

frag-ments of the appropriate PCR-amplified GRIP1 cDNA or

pSG5.HA.GRIP1s into the EcoRI and SalI sites of the pM

vector (Clontech, Mountain View, CA, USA), a vector for

expression of Gal4DBD fusion proteins from a

constitu-tive SV40 early promoter C-terminal truncations of

fragments of the appropriate truncated PCR-amplified

GRIP1 (amino acids from 750 to indicated numbers) into

C-terminal truncations of pSG5.HA.GRIP15)1462were

con-structed by inserting EcoRI–SalI fragments of the indicated

pM.GRIP1 truncations into the EcoRI and XhoI sites

pCDNA3.1.HDAC1.myc [40] were gifts from M.A Lazar

(University of Pennsylvania, Philadelphia, PA, USA), and

pCDNA3.HDAC1.flag wild type and H141A mutant were

gifts from T.P Yao (Duke University, Durham, NC,

USA) [41] Reporter genes MMTV-LUC, EREII-LUC

[GL45], MMTV[TRE]-LUC, and GK1, were as described

previously [42,43] The expression of NRs in mammalian

cells and⁄ or in vitro, vectors pSVAR0for human AR [44],

pHE0 for human ERa [43] and pCMX.hTRb1 [9] for

human TRb1, were as described previously

Bacterial expression vectors for GST fused to various

GRIP1 fragments (codons 1122–1462, 1305–1462, 1122–

1304, 1305–1398, 1305–1462 and 1399–1462) were

con-structed by inserting the appropriate PCR fragment into

pGEX-4T1 expression vector (GE HealthCare, Chicago,

IL, USA) via EcoRI–XhoI sites

Cell culture and transient transfection assays

HeLa, COS-7 and COS-1 cells were grown in Dulbecco’s

modified Eagle’s medium (DMEM) supplemented with

10% charcoal⁄ dextran-treated fetal bovine serum The cells

in each well (a six- or a 24-well plate) were transfected with

SuperFect Transfection Reagent (Qiagen, Hilden,

Ger-many) or jetPEI (PolyPlus-transfection, Illkirch, France),

according to the manufacturer’s protocol; total DNA was

adjusted to 2.0 lg (six well) or 1.0 lg (24-well) by addition

of the empty vector pSG5.HA Luciferase assays were

per-formed with the Promega Luciferase Assay kit (Madison,

WI, USA), and the measurement is expressed numerically

as relative light units (RLU) Luciferase activities are shown

as the mean and SD from two transfected sets The results shown are representative of at least three independent experiments Because some co-activators, including GRIP1 and CARM1, enhance the activities of so-called constitutive promoters two- to ninefold, internal controls by co-trans-fection of constitutive b-galactosidase expression vectors were not used to normalize luciferase data However, inter-nal controls were used strategically to show that variation

in transfection efficiency was not a factor in the key results (data not shown)

Immunoprecipitation and immunoblots

For analysis of the homo-oligomerization of GRIP1 and the physical interaction between GRIP1 and HDAC1, these expression vectors were transfected into COS-7 cells After transfection, cells were lysed in RIPA buffer (100 mm

Tri-ton 100) at 4
C Lysates were subjected to immunoprecipi-tation with antibodies against Gal4 DBD or HA for 3 h, followed by adsorption to Sepharose-coupled protein A⁄ G (Santa Cruz Biotechnology, Santa Cruz, CA, USA) for 3 h

analysed with immunoblots For determination of total protein levels of Gal4DBD- or HA-GRIP1 fragments, aliquots of cell lysates were subjected to direct immuno-blots Immunoblots were performed as previously described [23] using 10% of the extract from lysates for immunopre-cipitation and monoclonal antibodies 3F10 against the HA epitope (Roche, Mannheim, Germany), RK5C1 against Gal4DBD, 3A2 against HuR, and normal mouse IgG (Santa Cruz Biotechnology)

Protein–protein interaction assays

For GST pull-down assays, 35S-labelled proteins were pro-duced using the TNT T7-coupled reticulocyte lysate system (Promega, Madison, WI, USA) GST fusion proteins were produced in Escherichia coli BL21, eluted, and analysed by gel electrophoresis, as previously described [23]

Acknowledgements

We thank Dr W Feng (University of California, USA) for expression vectors and reporter genes for TR;

P Webb and P J Kushner (University of California, USA) fro expression vectors and reporter genes for ER; A O Brinkmann (Erasmus University, Rotter-dam, the Netherlands) for AR expression vector;

M A Lazar (University of Pennsylvania, USA) for pCDNA3.1.HDAC1.myc; and T P Yao (Duke University, USA) for pCDNA3.HDAC1.flag (wild-type and H141A mutant) expression vectors This work