FEBS Letters 583 (2009) 2451–2456

journal homepage: www.FEBSLetters .org

Bag-1M inhibits the transactivation of the glucocorticoid receptorvia recruitment of corepressors

Wei Hong a,*, Aria Baniahmad b,*, Juan Li a, Chenglin Chang a, Weizhen Gao a, Yunde Liu a

a Department of Laboratory Medicine, Tianjin Medical University, 300203 Tianjin, Chinab Institute for Human Genetics and Anthropology, Jena University Hospital, 07740 Jena, Germany

a r t i c l e i n f o

Article history:Received 29 April 2009Revised 16 June 2009Accepted 2 July 2009Available online 15 July 2009

Edited by Robert Barouki

Keywords:Bcl-2 associated athanogene 1Nuclear hormone receptorChaperoneTranscriptionCorepressor

0014-5793/$36.00 � 2009 Federation of European Biodoi:10.1016/j.febslet.2009.07.010

Abbreviations: Bag-1, Bcl-2 associated athanogenetor; GRmt, glucocorticoid receptor K496L/I497G muthMTIIa, human metallothionein IIa; ChIP, chromatinsmall interfering RNA; IB, immunoblot; GRE, glucocorandrogen receptor; NcoR, nuclear receptor corepressfor retinoic acid and thyroid hormone receptor; HDAC

* Corresponding authors. Fax: +86 22 60357712 ((A. Baniahmad).

E-mail addresses: hongwei@tijmu.edu.cn (W. H(A. Baniahmad).

a b s t r a c t

The Bcl-2 associated athanogene 1M (Bag-1M) is known to repress the transactivation of the gluco-corticoid receptor (GR). We report here that Bag-1M inhibits the action of GR via recruitment ofcorepressors, including nuclear receptor corepressor (NcoR) and silencing mediator for retinoic acidand thyroid hormone receptor (SMRT), and histone deacetylase (HDAC)3 to the genomic responseelement of a glucocorticoid-regulated human metallothionein IIa (hMTIIa) gene. A mutant GR lack-ing the interaction with BAG-1M fails to recruit the corepressors NcoR and SMRT. RNAi-mediatedknock down of corepressors and the use of HDAC inhibitor relieved Bag-1M-induced repressionon the transactivation of the GR. In addition, Bag-1M is not involved in the degradation of the recep-tor. These findings indicate a novel mechanism by which Bag-1M acts as a corepressor and downreg-ulates the activity of the GR.

Structured summary:MINT-7216164: HDAC3 (uniprotkb:O15379) physically interacts (MI:0914) with Bag1 (uniprotkb:Q99933)by anti bait coimmunoprecipitation (MI:0006)MINT-7216183: NCOR (uniprotkb:O75376) physically interacts (MI:0914) with Bag1 (uniprotkb:Q99933)by anti bait coimmunoprecipitation (MI:0006)MINT-7216175: SMRT (uniprotkb:Q9Y618) physically interacts (MI:0914) with Bag1 (uniprotkb:Q99933)by anti bait coimmunoprecipitation (MI:0006)

� 2009 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved.

1. Introduction

The Bcl-2 associated athanogene 1 (Bag-1) proteins are cochap-erones that are involved in the regulation of nuclear receptoraction [1–5]. In human, the Bag-1 gene encodes four isoforms ofthe Bag-1 proteins (Bag-1L, Bag-1M, Bag-1S and p29), all of whichare expressed through alternative translation initiation sites fromthe same mRNA [6]. Although it was initially identified as a Bcl-2binding protein to suppress apoptosis [7], Bag-1 has been knownto interact with and regulate the activity of other proteins. For

chemical Societies. Published by E

1; GR, glucocorticoid recep-ant; Hsp, heat shock protein;immunoprecipitation; siRNA,ticoid response element; AR,

or; SMRT, silencing mediator, histone deacetylase

W. Hong), +49 3641 935502

ong), aban@mti.uni-jena.de

example, Bag-1 interacts with and inhibits the function of the tu-mor suppressor p73 [8]. Furthermore, the medium isoform, Bag-1M is increased in the hippocampus of Alzheimer’s disease pa-tients and binds to Tau protein and amyloid precursor protein(APP), and overexpressed Bag-1M induces increased level of Tauand APP that are related with the pathology and treatment of Alz-heimer disease [9,10].

The different Bag-1 proteins exert varying effects on the trans-activation function of nuclear receptors. The largest isoform, Bag-1L, enhances the transactivation of androgen receptor (AR) [11]but inhibits the action of glucocorticoid receptor (GR) [12],whereas, Bag-1M, downregulates the transactivation of GR [13].Recent study revealed that Bag-1M localized to the glucocorticoidresponse element (GRE) in a hormone sensitive manner and re-pressed the DNA binding by the GR [14]. This provides another evi-dence that molecular chaperones and cochaperones can modulatethe nuclear receptor action at the genomic response element apartfrom the previous finding that molecular chaperone p23 was pres-ent at the response element in the presence of hormone and dis-rupted receptor-mediated transcriptional activation [15].

lsevier B.V. All rights reserved.

2452 W. Hong et al. / FEBS Letters 583 (2009) 2451–2456

Moreover, the Bag-1 cochaperone also affects the protein level ofthe GR through its ubiquitin like domain that is involved in thedegradation of GR in a hormone dependent manner in the presenceof another cochaperone chromatin C-terminus of heat shock pro-tein (Hsp) 70 interacting protein (CHIP) [16].

Glucocorticoid regulates the expression of target genes involvedin many pathophysiological processes. A human methallothioneinIIa (hMTIIa) gene has been identified and used as a model to inves-tigate the action of GR [17]. Although heavy metal cations are themost potent inducers of MT in mammals, other agents also initiateincreases in MT expression [18]. For example, glucocorticoid trans-activates the expression of hMTIIa gene through the GR that bindsto the GRE located upstream of the gene [17].

Here, we report that Bag-1M is not involved in the degradationof GR through the ubiquitin–proteosome pathway as a potentialmechanism for down-regulation of GR activity. Rather, our datasuggest that Bag-1M recruits corepressors and histone deacetylase(HDAC), in a hormone dependent manner, to the GR genomicresponse element of the hMTIIa gene. In line with this, knock-downof corepressors and inhibition of HDAC activity reversed Bag-1M-mediated negative effect.

2. Materials and methods

2.1. Cell culture

COS-7 and HeLa cells were cultured in Dulbecco’s modifiedEagle’s medium (DMEM), supplemented with 10% fetal calf serumat 37 �C and in an atmosphere of 5% CO2. All culture media con-tained 100 units/ml penicillin and 100 lg/ml streptomycin.

2.2. Plasmid constructs

The plasmids expressing the wild-type human GR, the mutantGR, Bag-1M, the indicator plasmid pGL3MMTV, control plasmidpRenilla-luc and the hMTIIa reporter plasmid H1S CAT have previ-ously been reported [14]. pEGFP-C2 was commercial availablefrom Clontech, Palo Alto, CA.

2.3. Antibodies

For ChIP assay, GR (P-20), Bag-1 (FL-274), nuclear receptor core-pressor (NcoR) (H-303), silencing mediator for retinoic acid andthyroid hormone receptor (SMRT) (C-19) and HDAC3 (H-99) anti-bodies and rabbit IgG and goat IgG (all from Santa Cruz Biotechnol-ogy, Santa Cruz, CA) were used. The following antibodies were usedfor immunoprecipitation: Bag-1 (FL-274), NcoR (H-303), SMRT (C-19), HDAC3 (H-99), all from Santa Cruz. For immunoblot (IB), Bag-1(C-16, Santa Cruz), Actin (I-19, Santa Cruz), SMRT (N-20, SantaCruz) and NcoR (06-892, Upstate) antibodies were employed.

2.4. Transfection assay and luciferase reporter gene assay

COS-7 cells were transiently transfected with plasmid con-structs using Fugene 6 reagent (Roche Diagnostics, Basel, Swit-zerland) according to the manufacturer’s instructions. Theefficiency of transfection was determined as the percentage ofgreen fluorescent cells following cotransfection of pEGFP-C2, amammalian expression plasmid encoding for enhanced greenfluorescent protein. It remained stable throughout the reportedexperiments at a level of 75–80%. The reporter gene assay haspreviously been described [14]. In this assay, three independentexperiments were performed and the data were analyzed, by asoftware for the student’s t-test, to calculate the probability va-lue (P-value) to check, whether or not, a significant difference

exists between two groups of data obtained under conditionsof different treatments.

2.5. Immunoprecipitation and IB

The transiently transfected and dexamethasone treated COS-7cells were further treated with 1 mM dimethyl 3,30-dithiobispropi-onimidate (Pierce Biotechnology, Rockford, IL) for 30 min to cross-link the protein. Cells were lysed in buffer (0.5% NP-40, 20 mM Tris,pH 8.0, 100 mM NaCl, 1 mM EDTA, 10% glycerol) and preincubatedwith protein A sepharose (Amersham Bioscience, Shanghai, China).Bag-1, NcoR, SMRT or HDAC3 antibodies were used for immuno-precipitation and IB was followed with Bag-1 antibody.

2.6. ChIP and re-ChIP

ChIP and re-ChIP were performed as previously described [19].For ChIP in transfected COS-7 cells, 10% of the precleared superna-tant was taken out as input. The PCR reactions of input and immu-noprecipitated material were run with 28 cycles. The primers usedfor the PCR amplification are (in 50–30 direction): CCG GTT ACT GTGATG CTG CA (hMTIIa for), GCG GGA GGA CAC AGT GTA CC (hMTIIarev). For Re-ChIP, 20% of the elute after the first precipitation butbefore the second precipitation was obtained as input. The PCRreactions of input and immunoprecipitated material were run with32 cycles. For ChIP in HeLa cells, 20% of the precleared supernatantwas used as input and the PCR reactions were run with 28 cycles.

2.7. RNA interference

NcoR small interfering RNA (siRNA) (M-003518-01), SMRT siR-NA (M-020145-02) and the luciferase GL3 siRNA used as control,were all purchased from Dharmacon. Transfection of COS-7 andHeLa cells with the siRNA was performed using an Oligofectaminereagent (Invitrogen, Shanghai, China) at a final concentration of150 nM. The cells were finally collected for measurement of lucif-erase activity, real-time PCR and immunoblot.

2.8. Real-time PCR

The primers (in 50–30 direction): TCG GAT ACG TCA TCA GCA CC(for) and TCC CTC CTG TCC TGT ACT CGA (rev) were used for thedetection of hMTIIa gene expression. Human b-actin gene expres-sion was detected as a control by the primers: TCA CCC ACA CTGTGC CCA T (for) and CTC TTG CTC GAA GTC CAG GG (rev). TheABI PRISM 7000 sequence detection system and the SYBR GreenPCR Master Mix (Applied Biosystem, Shanghai, China) and 40cycles were used for the amplification. Three independent real-time PCRs were run and the data were analyzed, by a softwarefor the Student’s t-test, to calculate the P-value to check, whetheror not, a significant difference exists between two groups of dataobtained under conditions of different treatments.

3. Results and discussion

3.1. Bag-1M recruits corepressors and HDAC

Since we have previously shown that Bag-1M and GR are pres-ent at the GRE of a glucocorticoid-regulated human metallothio-nein IIa (hMTIIa) gene [14], it is worthwhile to investigate howBag-1M alters the action of the GR complex at the GRE. ChIP wastherefore carried out to identify the possible factors that are in-volved in the GR complex. COS-7 cells that are devoid of GR andBag-1M [12] were transfected with an hMTIIa reporter gene. Inaddition, Bag-1M, the wild-type GR or a mutant GR (glucocorticoid

W. Hong et al. / FEBS Letters 583 (2009) 2451–2456 2453

receptor K496L/I497G mutant (GRmt), which is unable to interactwith Bag-1M but mediates hormone response [21], were cotrans-fected with the reporter plasmid. No protein was found to bindthe GRE of the hMTIIa gene, in the absence of hormone but inthe presence of Bag-1M, as shown by no detectable amplifiedDNA fragment (Fig. 1A). When the cells were only transfected witheither the wild-type GR or the mutant GR, the receptors were pres-ent at the GRE after hormone treatment (Fig. 1A). In the cellscotransfected with vectors coding for Bag-1M and GR, both factorsbind to the GRE in the presence of hormone (Fig. 1A). Intriguingly,the presence of Bag-1M and the wide type GR at the GRE lead tothe recruitment of the endogenously expressed corepressors,including NcoR and SMRT, as well as histone deacetylase3(HDAC3). However, in the cells cotransfected with Bag-1M andGRmt, only GRmt was present at the GRE, indicating Bag-1M is re-cruited to the response element by interacting with specific aminoacid sequences of the GR (Fig. 1A), and suggests that NcoR, SMRTand HDAC3 require BAG-1M for recruitment to GR.

To demonstrate that the GR, Bag-1M, corepressors and HDAC3are all present at the same promoter, a re-ChIP analysis was per-formed for which the anti-Bag-1 antibody was employed for thefirst immunoprecipitation and resulted in a PCR product in thepresence of hormone (Fig. 1B). The second immunoprecipitationof the Bag-1 first immunoprecipitate using the GR, NcoR, SMRTand HDAC3 antibodies revealed a PCR product of the same genomiclocus (Fig. 1B). Control experiment showed that the Bag-1 antibodywas completely inactivated before the use of the second set of anti-bodies (Fig. 1B) and no unspecific immunoprecipitation wasoccurred by IgG (Fig. 1B). These data further confirm that Bag-1M recruits corepressors and HDAC to the GRE. We assume thatthe effect of Bag-1M may be due to its ability to interact withthe corepressors and HDAC. Accordingly, we carried out the immu-noprecipitation experiments and could show that NcoR and Bag-1antibodies could immunoprecipitated Bag-1M (Fig. 1C). Accord-ingly, SMRT and Bag-1 specific antibodies could also immunopre-cipitate Bag-1M (Fig. 1D). The use of HDAC3 and Bag-1antibodies immunoprecipitate Bag-1M as well (Fig. 1E), but Bag-1M could not be immunoprecipitated by the control IgG as shownin the IBs (Fig. 1C–E). The data suggest that BAG-1M is complexedwith SMRT, NcoR and HDAC3 on chromatin.

rIgG

gIgG

Input- - + +Dex

GR GRmt

ααGR

ααBag-1

ααNcoR

ααSMRT

Bag-1M- - + +

- + - + - + - +

ααHDAC3

A

ααBag-1

Dex - +1st IP withααBag-1

Input

B

Lysate

IB: ααBag-1

C

IB:

E

Fig. 1. Corepressors and histone deacetylase are recruited by Bag-1M to the GR target gethe wild-type GR but not the mutant GR (GRmt, K496L/I497G). COS-7 cells were transivectors encoding either the wild-type GR or GRmt, together with or without an expressiodexamethasone (10�7 M) or ethanol for 1 h and were analysed for factor occupancy at thhMTIIa gene promoter in transfected COS-7 cells with the anti-Bag-1 antibody for theimmunoprecipitation with antibodies against GR, Bag-1, NcoR, SMRT and HDAC3, respecinputs. (C–E) An aliquot from the ChIP assay was applied for immunoprecipitation with anwas performed with anti-Bag-1M antibody and IgG as a control.

While Bag-1M appears to be recruited to GRE through interac-tion with distinct amino acid sequences of the GR, the exact se-quence requirements for interacting with GR need to be furtheridentified. Several corepressor complexes associated with nuclearreceptors have been elucidated, among which the NcoR–SMRTcomplex harboring HDACs is the most notable one [20]. Bag-1Minteracts with NcoR and SMRT which in turn to bind the HDACand this may explain how these proteins are recruited to theGRE. Our previous finding revealed that Bag-1M inhibits the DNAbinding by the GR [14]. Thus, there might be two mechanisms uti-lized by Bag-1M to downregulate the transactivation of GR. One isto reduce the DNA binding by the GR and the other is to recruitcorepressors to the response element. Reduced DNA binding bythe GR was mostly seen in vitro indicating that on chromatin level,the reduced GR binding plays a subordinated role in attenuatingthe action of GR.

3.2. Corepressors and HDAC are essential for Bag-1M-mediatedinhibition of the transactivation of the GR

To further confirm whether corepressors and HDAC are in-volved in the repression of the GR activity by Bag-1M, siRNAwas employed to knock down corepressors NcoR and SMRT. Inthe cells cotransfected with GR and Bag-1M, Bag-1M repressesthe action of GR in a hormone-dependent manner (Fig. 2A, com-pare the value of lane 2 with that of lane 4, statistical analysisshowed P < 0.05), and Bag-1M-mediated repression on the trans-activation of the GR was not altered when control siRNA wasintroduced into the cells (Fig. 2A, compare lane 4 with lane 8,P > 0.05). In contrast, when siRNA of NcoR or SMRT was transfec-ted to the cells, respectively, Bag-1M-induced inhibition on thetransactivation of the GR was partially released and cotransfec-tion of the NcoR and SMRT siRNAs had a synergic effect, suggest-ing that NcoR and SMRT are involved in Bag-1M-mediatednegative regulation on the action of GR (Fig. 2A, compare lanes12, 16 and 20 with lane 8, the respective P < 0.05). None ofthe transfected siRNAs affected the ligand-dependent transacti-vation of the GR (Fig. 2A, compare lanes 6, 10, 14 and 18 withlane 2, the respective P > 0.05). In addition, no significant differ-ences could be found among the basal levels of transactivation

rIgG

- + - + - + - + - + - + - +gIgGina

ctivation

ofααBag-1

ααGR

ααNcoR

ααSMRT

ααHDAC3

re-ChIP

ααBag-1

IgG ααNcoR

ααBag-1

Lysate

IB: ααBag-1

IgG ααSMRTD

ααBag-1

Lysate

ααBag-1

IgG ααHDAC3

ne. (A) Corepressors are bound to the GRE of hMTIIa in the presence of Bag-1M andently cotransfected with a vector harbouring an hMTIIa indicator gene, expressionn vector for Bag-1M. Forty-eight hours after transfection, the cells were treated withe hMTIIa gene promoter by the ChIP assay. (B) re-ChIP assay was carried out on thefirst immunoprecipitation. These immunoprecipitates were subjected to a secondtively. IgG and inactivated antibody were used as controls. Lower panel shows theti-NcoR (C), anti-SMRT (D), anti-HDAC3 (E), and anti-Bag-1 antibodies. Immunoblot

- - + + - - + +- - - - + + + +

100200300400500600700800900

0Bag-1M

TSA

(-) Dex(+) Dex

RelativeluciferaseactivityD

B

NcoRsiRNA

controlsiRNA

-siRNA

IB: ααNcoR

IB: ααActin

- - + + - - + + - - + + - - + + - - + +- - - - + + + + - - - - - - - - - - - -

100200300400500600700800900

0Bag-1M

siNcoR

(-) Dex(+) Dex

RelativeluciferaseactivityA

con. siRNA- - - - - - - - + + + + - - - - + + + +

siSMRT - - - - - - - - - - - - + + + + + + + +

SMRTsiRNA

controlsiRNA

-siRNA

IB: ααSMRT

C

IB: ααActin

Fig. 2. Corepressors and histone deacetylase are essential for Bag-1M-mediated negative effect. (A) COS-7 cells were transiently transfected with a luciferase reporter gene,vectors encoding GR and Bag-1M. Twenty-four hours after transfection, the cells were transfected with siRNA of NcoR. Twenty-four hours after the transfection of siRNA, thecells were treated with dexamethasone (10�7 M) or ethanol for 16 h before they were harvested for the measurement of luciferase activity. The bar charts represent the meanand standard deviation of three independent experiments. (B and C) Aliquots of the transfected cells were applied for immunoblots with the anti-NcoR, anti-SMRT and anti-actin antibodies. Shown is the protein level of NcoR and SMRT. (D) Effect of trichostatin A (TSA) on Bag-1M-mediated repression of glucocorticoid receptor activity. COS-7cells were transfected as indicated in (A). Twenty-four hours after transfection, the cells were treated with dexamethasone (10�7 M) or ethanol together with or without TSAfor 16 h. Then the cells were prepared for luciferase assay. The bar charts represent the mean and standard deviation of three independent experiments.

2454 W. Hong et al. / FEBS Letters 583 (2009) 2451–2456

(Fig. 2A, when any two of the basal level values were compared,the respective P > 0.05). IBs showed that the protein level ofNcoR and SMRT was effectively knocked down, respectively, bysiRNA compared with those from the non-transfected or controlsiRNA transfected cells (Fig. 2B and C).

Since HDAC3 was also found to be present at the GRE of theglucocorticoid-regulated gene, we used trichostatin A (TSA), anHDAC inhibitor, to inhibit the activity of HDAC to investigatethe expression of the reporter gene. As previously shown, Bag-1M inhibits the transactivation of the GR, but its negative effectwas largely reversed by the use of TSA, which did not affect thetransactivation mediated by the GR (Fig. 2D, compare lane 4with lane 8, P < 0.05; compare lane 2 with lane 6, P > 0.05). How-ever, there were no significant differences among the basal leveltransactivations (Fig. 2D, when any two of the basal level valueswere compared, the respective P > 0.05). This confirmed thatBag-1M recruits corepressors that subsequently interact withHDAC to suppress the expression of the glucocorticoid-inducedreporter gene.

To further confirm, whether or not, the recruitment of corepres-sors and HDAC3 is due to the overexpressed Bag-1M in the trans-fection experiments, ChIP assays were carried out in which the

binding of the GR to genomic GRE of the hMTIIa gene in HeLa cellswas analyzed. In the absence of dexamethasone, we did not detectfactors binding to the GRE (Fig. 3A). In contrast, GR, Bag-1M, NcoR,SMRT and HDAC3 were all present at the response element in a li-gand-dependent manner (Fig. 3B). To characterize the effect ofcorepressors on GR action, we made use of HeLa cells in whichthe endogenous level of NcoR and SMRT has been knocked downby transfection of siRNA to investigate the expression of the hMTIIagene. As shown in Fig. 3C, dexamethasone treatment induced theexpression of the hMTIIa gene and transfection of siRNAs did notaffect the basal level expression of the gene (Fig. 3C, compare lanes2–5 with lane 1, the respective P > 0.05). Also, transfection of con-trol siRNA had no effect on the expression of the hMTIIa gene(Fig. 3C, compare lane 7 with lane 6, P > 0.05). In contrast, transfec-tion of either NcoR or SMRT siRNA, or a cotransfection of NcoR andSMRT siRNAs, led to increased expression of the hMTIIa gene(Fig. 3C, compare lanes 8–10 with lane 7, the respective P < 0.05),indicating these corepressors are involved in the downregulationof the activity of the GR by the endogenous Bag-1M. The endoge-nous level of NcoR and SMRT was effectively knocked down by siR-NA, as shown by the IBs, compared with those from the non-transfected or control siRNA transfected cells (Fig. 3D and E).

1

2

3

4 (-) Dex(+) Dex

- + - - - - + - - -siNcoR

Relativelevelof

hMTIIaexpression

- - + - + - - + - +con. siRNA

siSMRT - - - + + - - - + +

C

B

Dex

rIgG

ααBag-1

gIgG

ααGR

ααNcoR

ααSMRT

ααHDAC3

(+) Input

ArIgG

ααBag-1

gIgG

ααGR

ααNcoR

ααSMRT

ααHDAC3

InputDex(-)

1

2

3

4 (-) Dex(+) Dex

- + - +TSA

Relativelevelof

hMTIIaexpressionF

E

controlsiRNA

-siRNA

IB: ααActin

IB: ααSMRT

SMRTsiRNA

NcoRsiRNA

controlsiRNA

-siRNA

IB: ααNcoR

IB: ααActin

D

Fig. 3. NcoR, SMRT and HDAC3 are needed by Bag-1M to repress the transcriptionof glucocorticoid responsive gene. (A and B) ChIP experiments were performed withHeLa cells to detect the recruitment of endogenously expressed GR, NcoR, SMRT andHDAC3 in the absence (A) or presence (B) of GR agonist. (C) Transfection of NcoRand SMRT siRNAs enhances the expression of hMTIIa in HeLa cells. HeLa cellstransfected with control, NcoR or/and SMRT siRNA for 48 h were additionallytreated with dexamethasone (10�7 M) or ethanol for 16 h. Total RNA was isolatedand used for analysis of hMTIIa expression by real-time PCR. (D and E) Aliquots ofthe cells transfected with siRNA were used for immunoblots with anti-NcoR, anti-SMRT and anti-actin antibodies. Shown is the protein level of NcoR and SMRT. (F)HeLa cells were treated with dexamethasone (10�7 M) or ethanol together with orwithout TSA for 16 h. Total RNA was isolated and used for real-time PCR to analysethe hMTIIa expression. The bar charts represent the mean and standard deviation ofthree independent experiments.

MG132

1 2 3 4 5 6 7 8 9 10

IB: ααGR

IB: ααActin

IB: ααBAG-1M

Non Vec+GR BAG-1M+GR

Dex - - - + - + - + - +- + - - + + - - + +

Fig. 4. Bag-1M did not target GR for degradation. COS-7 cells were transfected withvectors encoding GR and Bag-1M, together with a vector encoding Renilla luciferaseas internal control. Forty-eight hours after transfection, the cells were treated withMG132 (10�6 M) or dimethyl sulfoxide (vehicle), 1 h later, the cells also receivedtreatment of dexamethasone (10�7 M) or ethanol for 16 h. Whole cell lysates wereprepared to measure Renilla luciferase activity for transfection efficiency and tocarry out immunoblots with anti-GR, anti-Bag-1M and anti-actin antibodies.

W. Hong et al. / FEBS Letters 583 (2009) 2451–2456 2455

Additionally, to determine whether the presence of HDAC at thehMTIIa promoter results in negative regulation of the activity ofGR, we treated HeLa cells with TSA in the absence or presence ofglucocorticoid. RNA was isolated and analyzed by real-time PCRto determine the level of transcription of the hMTIIa gene. Asshown in Fig. 3F, TSA treatment enhanced the ligand-sensitivetranscription (Fig. 3F, compare lane 4 with lane 3, P < 0.05), butdid not alter the basal transcription of the hMTIIa gene (Fig. 3F,compare lane 2 with lane 1, P > 0.05).

Taken together, Bag-1M appears to act as a corepressor whentethered to GREs by recruiting multiple factors that function to in-hibit ligand-dependent transactivation by the GR. RNAi-mediatedablation of corepressors in COS cells with transfected reporter generelieved the repression of GR by Bag-1M. In agreement with thisfinding, transfection of siRNA of corepressors into HeLa cells con-taining endogenous GR and Bag-1M enhanced the transcription

of the genomically encoded GR-regulated gene, confirming theimportance of corepressor recruitment for the inhibitory activityof Bag-1M. NcoR and SMRT are potent corepressors that are pres-ent in protein complexes containing HDACs and their corepressoractivity is mediated through HDAC-dependent mechanism.Accordingly, the HDAC inhibitor TSA largely relieved the repressionof GR by Bag-1M as demonstrated from the transfected indicatorgene as well as the endogenous glucocorticoid responsive gene,suggesting that the negative activity of Bag-1M is, at least in part,HDAC-dependent.

3.3. Bag-1M is not involved in the degradation of GR

Since Bag-1M contains a ubiquitin-like domain, we investigatedwhether Bag-1M-mediated downregulation of the action of the GRis due to its involvement in the ubiquitin–proteosome pathway forreceptor degradation. Therefore, COS-7 cells were cotransfectedwith vectors coding for GR and Bag-1M. In addition, the transfectedcells were treated with a proteosome inhibitor MG132 and thenwere lysed for IB to determine the receptor level. The data showedthat when MG132 was used to inhibit the proteosome activity, thereceptor level was enhanced implying that under basal conditionthe GR was degraded through the proteosome (Fig. 4). While inthe presence of Bag-1M, the receptor levels were not altered inthe absence or presence of MG132 (Fig. 4, compare lanes 7 and 8with 3 and 4; lanes 9 and 10 with 5 and 6), suggesting that Bag-1M did not target GR for degradation. Our data is different froma previous finding that Bag-1M is involved in the GR degradationin the presence of another cochaperone CHIP that ubiquitinatesthe receptor for subsequent degradation [16].

However, our finding is consistent with the observation thatseveral nuclear receptors including GR are ubiquitinated and de-graded in the course of their nuclear activities [21–26]. In additionto receptors, studies have revealed that coactivators, including ste-roid receptor coactivator 1 (SRC1), SRC2, SRC3, CBP and E6-associ-ated protein (E6-AP), could also be ubiquitinated and degraded,through the proteosome, in order to disassemble and reassemblecoactivator complexes, thereby promoting enhanced transcription[27]. It is conceivable that the ubiquitination and degradation ofthe p160 coactivators could possibly contributed to a preinitiationcomplex, allowing for transcription elongation to proceed. In thismanuscript, we have demonstrated that Bag-1M was not involvedin ubiquitin–proteosome-mediated degradation of the GR, but therole of Bag-1M in the ubiquitination of the p160 coactivators stillneeds to be elucidated.

It has been recognized that molecular chaperones and cochap-erones exert an important role in the modulation of the action ofnuclear receptors in addition to the formation of a complex to keep

2456 W. Hong et al. / FEBS Letters 583 (2009) 2451–2456

the receptors in an appropriate conformation for ligand binding[28,29]. Recently, heat shock protein 90 (Hsp90) has been reportedto enhance the DNA binding by the liganded vitamin D receptorand is essential for optimal vitamin D responsiveness [30].GCUNC45, an Hsp90 cochaperone, effectively blocks the progres-sion of progesterone receptor chaperoning in the presence ofHsp90beta [31]. Hsp27 was known to dysregulate the interactionof estrogen receptor with chromatin and acts as a suppressor ofestrogen-induced transcription [32,33]. Hsp27 was also shown tobe phosphorylated in an androgen-dependent manner and en-hanced the AR stability, shuttling and transactivation, thereby pro-moted the survival of prostate cancer cells [34]. In addition, ahistone chaperone termed nucleophosmin (NPM/B23) interactswith AR to promote AR binding to the androgen response elementand subsequently modulate the transactivation of the AR [35]. In-deed, our data provide additional evidence to support the notionthat chaperones modulate the activity of nuclear receptors and thismode of modulation can be fulfilled in different manners.

In conclusion, our data suggest that BAG-1M downregulatesGR-mediated transactivation by recruiting the corepressors NcoR,SMRT and HDAC3 to GR and indicates BAG-1M is a novel corepres-sor for GR to modulate its agonist-induced transactivation.

Acknowledgements

This work was supported by a grant from Tianjin Medical Uni-versity (2007 KY 11, W.H.) and a grant for the ‘‘11-5” New CenturyOutstanding Scientist from Tianjin Medical University (W.H.).

References

[1] Millar, E.K., Anderson, L.R., McNeil, C.M., O’Toole, S.A., Pinese, M., Crea, P.,Morey, A.L., Biankin, A.V., Henshall, S.M., Musgrove, E.A., Sutherland, R.L. andButt, A.J. (2009) BAG-1 predicts patient outcome and tamoxifenresponsiveness in ER-positive invasive ductal carcinoma of the breast. Br. J.Cancer 100, 123–133.

[2] Crabb, S.J., Hague, A., Johnson, P.W. and Packham, G. (2008) BAG-1 inhibitsPPARgamma-induced cell death, but not PPARgamma-induced transcription,cell cycle arrest or differentiation in breast cancer cells. Oncol. Rep. 19, 689–696.

[3] Chun, R.F., Gacad, M., Nguyen, L., Hewison, M. and Adams, J.S. (2007) Co-chaperone potentiation of vitamin D receptor-mediated transactivation: a rolefor Bcl2-associated athanogene-1 as an intracellular-binding protein for 1,25-dihydroxyvitamin D3. J. Mol. Endocrinol. 39, 81–89.

[4] Mäki, H.E., Saramäki, O.R., Shatkina, L., Martikainen, P.M., Tammela, T.L., vanWeerden, W.M., Vessella, R.L., Cato, A.C. and Visakorpi, T. (2007)Overexpression and gene amplification of BAG-1L in hormone-refractoryprostate cancer. J. Pathol. 212, 395–401.

[5] Cutress, R.I., Townsend, P.A., Sharp, A., Maison, A., Wood, L., Lee, R., Brimmell,M., Mullee, M.A., Johnson, P.W., Royle, G.T., Bateman, A.C. and Packham, G.(2003) The nuclear BAG-1 isoform, BAG-1L, enhances oestrogen-dependenttranscription. Oncogene 22, 4973–4982.

[6] Gehring, U. (2006) Activities of the cochaperones Hap46/BAG-1M and Hap50/BAG-1L and isoforms. Cell Stress Chaperon. 11, 295–303.

[7] Takayama, S., Sato, T., Krajewski, S., Kochel, K., Irie, S. and Millan, J.A. (1995)Cloning and functional analysisof BAG-1: a novel Bcl-2-binding protein withanti-cell death activity. Cell 80, 279–284.

[8] Wang, X.H., O’Connor, D., Brimmell, M. and Packham, G. (2009) The BAG-1cochaperone is a negative regulator of p73-dependent transcription. Br. J.Cancer (Epub ahead of print).

[9] Elliott, E., Tsvetkov, P. and Ginzburg, I. (2007) BAG-1 associates withHsc70.Tau complex and regulates the proteasomal degradation of Tauprotein. J. Biol. Chem. 282, 37276–37284.

[10] Elliott, E., Laufer, O. and Ginzburg, I. (2009) BAG-1M is upregulated inhippocampus of Alzheimer’s disease patients and associates with tau and APPproteins. J. Neurochem. (Epub ahead of print).

[11] Froesch, B.A., Takayama, S. and Reed, J.C. (1998) BAG-1L protein enhancesandrogen receptor function. J. Biol. Chem. 273, 11660–11666.

[12] Kullmann, M., Schneikert, J., Moll, J., Heck, S., Zeiner, M., Gehring, U. and Cato,A.C. (1998) RAP46 is a negative regulator of glucocorticoid receptor action andhormone-induced apoptosis. J. Biol. Chem. 273, 14620–14625.

[13] Schneikert, J., Hübner, S., Martin, E. and Cato, A.C. (1999) A nuclear action ofthe eukaryotic cochaperone RAP46 in downregulation of glucocorticoidreceptor activity. J. Cell Biol. 146, 929–940.

[14] Hong, W., Baniahmad, A., Liu, Y. and Li, H. (2008) Bag-1M is a component ofthe in vivo DNA–glucocorticoid receptor complex at hormone-regulatedpromoter. J. Mol. Biol. 384, 22–30.

[15] Freeman, B.C. and Yamamoto, K.R. (2002) Disassembly of transcriptionalregulatory complexes by molecular chaperones. Science 296, 2232–2235.

[16] Connell, P., Ballinger, C.A., Jiang, J., Wu, Y., Thompson, L.J. and Höhfeld, J.(2001) The co-chaperone CHIP regulates protein triage decisions mediated byheat-shock proteins. Nat. Cell Biol. 3, 93–96.

[17] Karin, M., Haslinger, A., Holtgreve, H., Richards, R.I., Krauter, P., Westphal, H.M.and Beato, M. (1984) Characterization of DNA sequences through whichcadmium and glucocorticoid hormones induce human metallothionein-IIAgene. Nature 308, 513–519.

[18] Hamer, D.H. (1986) Metallothionein. Annu. Rev. Biochem. 55, 913–951.[19] Eckey, M., Hong, W., Papaioannou, M. and Baniahmad, A. (2007) The

nucleosome assembly activity of NAP1 is enhanced by Alien. Mol. Cell. Biol.27, 3557–3568.

[20] Glass, C.K. and Rosenfeld, M.G. (2000) The coregulator exchange intranscriptional functions of nuclear receptors. Genes Dev. 14, 121–141.

[21] Wallace, A.D. and Cidlowski, J.A. (2001) Proteasome-mediated glucocorticoidreceptor degradation restricts transcriptional signaling by glucocorticoids. J.Biol. Chem. 276, 42714–42721.

[22] Garside, H., Waters, C., Berry, A., Rice, L., Ardley, H.C., White, A., Robinson, P.A.and Ray, D. (2006) UbcH7 interacts with the glucocorticoid receptor andmediates receptor autoregulation. J. Endocrinol. 190, 621–629.

[23] Tirard, M., Almeida, O.F., Hutzler, P., Melchior, F. and Michaelidis, T.M. (2007)Sumoylation and proteasomal activity determine the transactivationproperties of the mineralocorticoid receptor. Mol. Cell. Endocrinol. 268, 20–29.

[24] Zhang, P.J., Zhao, J., Li, H.Y., Man, J.H., He, K., Zhou, T., Pan, X., Li, A.L., Gong,W.L., Jin, B.F., Xia, Q., Yu, M., Shen, B.F. and Zhang, X.M. (2007) CUE domaincontaining 2 regulates degradation of progesterone receptor by ubiquitin–proteasome. EMBO J. 26, 1831–1842.

[25] Rodriguez-Gonzalez, A., Cyrus, K., Salcius, M., Kim, K., Crews, C.M., Deshaies,R.J. and Sakamoto, K.M. (2008) Targeting steroid hormone receptors forubiquitination and degradation in breast and prostate cancer. Oncogene 27,7201–7211.

[26] Kilroy, G.E., Zhang, X. and Floyd, Z.E. (2009) PPAR-gamma AF-2 domainfunctions as a component of a ubiquitin-dependent degradation signal.Obesity (Silver Spring) 17, 665–673.

[27] Yan, F., Gao, X., Lonard, D.M. and Nawaz, Z. (2003) Specific ubiquitin-conjugating enzymes promote degradation of specific nuclear receptorcoactivators. Mol. Endocrinol. 17, 1315–1331.

[28] Bohen, S.P., Kralli, A. and Yamamoto, K.R. (1995) Hold ’em and fold ’em:chaperones and signal transduction. Science 268, 1303–1304.

[29] Smith, D.F. and Toft, D.O. (2008) Minireview: the intersection of steroidreceptors with molecular chaperones: observations and questions. Mol.Endocrinol. 22, 2229–2240.

[30] Angelo, G., Lamon-Fava, S., Sonna, L.A., Lindauer, M.L. and Wood, R.J. (2008)Heat shock protein 90beta: a novel mediator of vitamin D action. Biochem.Biophys. Res. Commun. 367, 578–583.

[31] Chadli, A., Felts, S.J. and Toft, D.O. (2008) GCUNC45 is the first Hsp90 co-chaperone to show alpha/beta isoform specificity. J. Biol. Chem. 283, 9509–9512.

[32] Chen, H., Hewison, M. and Adams, J.S. (2007) Control of estradiol-directedgene transactivation by an intracellular estrogen-binding protein and anestrogen response element-binding protein. Mol. Endocrinol. 22, 559–569.

[33] Chen, H., Hewison, M., Hu, B., Sharma, M., Sun, Z. and Adams, J.S. (2004) AnHsp27-related, dominant negative-acting intracellular estradiol-bindingprotein. J. Biol. Chem. 279, 29944–29951.

[34] Zoubeidi, A., Zardan, A., Beraldi, E., Fazli, L., Sowery, R., Rennie, P., Nelson, C.and Gleave, M. (2007) Cooperative interactions between androgen receptor(AR) and heat-shock protein 27 facilitate AR transcriptional activity. CancerRes. 67, 10455–10465.

[35] Léotoing, L., Meunier, L., Manin, M., Mauduit, C., Decaussin, M., Verrijdt, G.,Claessens, F., Benahmed, M., Veyssière, G., Morel, L. and Beaudoin, C. (2008)Influence of nucleophosmin/B23 on DNA binding and transcriptional activityof the androgen receptor in prostate cancer cell. Oncogene 27, 2858–2867.