Cross Talk in Hormonally Regulated Gene Transcription through Induction of Estrogen Receptor Ubiquitylation

Estrogen tightly regulates the levels of circulating gonadotropins,but a direct effect of estrogen receptor alpha (ER ${alpha}$ ) on the mammalianLHß gene has remained poorly defined. We demonstratehere that ER ${alpha}$ can associate with the LHß promoter throughinteractions with Sf-1 and Pitx1 without requiring an estrogenresponse element (ERE). We show that gonadotropin-releasinghormone (GnRH) promotes ER ${alpha}$ ubiquitylation and also degradationwhile stimulating expression of ubc4. GnRH also increases theassociation and lengthens the cycling time of ER ${alpha}$ on the LHßpromoter. The ER ${alpha}$ association and transactivation of the LHßgene, as well as ER ${alpha}$ degradation, are increased following ubc4overexpression, while the effects of GnRH are abated followingubc4 knockdown. Our results indicate that ER ${alpha}$ ubiquitylationand subsequent transactivation of the LHß gene canbe induced by increasing the levels of the E2 enzyme as a resultof signaling by an extracellular hormone, thus providing a newform of cross talk in hormonally stimulated regulation of geneexpression.

INTRODUCTION

Top

Introduction

Estrogen is a growth factor that stimulates cell growth anddifferentiation in diverse tissues. One of its most centralroles, however, is regulating the production of gonadotropinsin the pituitary gland, which is exerted via negative and positivefeedback acting at the brain and pituitary. It is the high levelof estradiol (E₂) at the end of the follicular phase that workssynergistically with the hypothalamic gonadotropin-releasinghormone (GnRH) and leads to the preovulatory luteinizing hormone(LH) surge that initiates ovulation.

The synergistic effects of GnRH and E₂ on LHß geneexpression have been explained by a number of mechanisms, someof which increase the gonadotropes' sensitivity to GnRH. Theseinclude E₂/estrogen receptor (ER) effects on Egr-1, calciuminflux, and mitogen-activated protein kinase (MAPK), which comprisespart of the GnRH signaling pathway (4, 6, 8, 45). Conversely,the effects of ER ${alpha}$ on estrogen-responsive promoters are enhancedby GnRH, possibly involving MAPK phosphorylation of ER ${alpha}$ , whichfacilitates protein-protein interactions and transactivation,even in the absence of a ligand (9, 19, 33, 36, 53).

Despite evidence that E₂ enhances LHß gene transcriptiondirectly, the molecular mechanisms involved have remained elusive(3, 15, 25). This contrasts with the well-studied conservedtripartite element comprising Sf-1, Pitx1, and Egr-1 responseelements on the proximal promoter, which likely mediates theGnRH response of all mammalian LHß genes (3, 11, 20,46, 58). Notably, of these, only the rat LHß genecontains a possible estrogen response element (ERE). This element,located 1,159 bp upstream of the transcriptional start site,was shown to bind recombinant ER in gel shift assays (55) butbears little resemblance to the consensus ERE motif and is notfound on the homologous genes of other mammals. Thus, any directactions of the liganded ER on the LHß gene promoter,if they occur, are likely to involve other DNA binding factors,as shown, for example, for Sp1 and AP-1 (reviewed in reference19). This contrasts with the situation in teleost fish, in whichall of the isolated LHß gene proximal promoters docontain a near-consensus ERE, and in chinook salmon this wasshown to mediate estrogen/ER ${alpha}$ responsiveness (27, 31).

ER ${alpha}$ transactivation appears to involve ubiquitylation which signalsproteasome-mediated degradation. A direct correlation betweenthe rate of ER ${alpha}$ degradation and its transcriptional activityhas been noted (32, 48, 62), although other studies have suggestedthat it causes down-regulation, as inhibition of the degradationcaused an increase in target gene expression (13). How thesetwo seemingly opposing observations can be reconciled is notyet clear, although Fan et al. (13) suggest promoter contextis likely crucial and may relate to the degree that ER ${alpha}$ availabilityis a limiting factor in transcription of a particular gene.

The ER ${alpha}$ is reportedly ubiquitylated after the first round oftranscription, allowing its release from the promoter, whichmay be essential for subsequent ER ${alpha}$ /E₂-mediated transcription.In this way, the ER ${alpha}$ cycles on and off the promoter for as longas stimulation is present (37, 48, 54). This cycling reportedlyalso occurs in the absence of a ligand, but the cycling timesare much shorter; it was suggested that the cycle time is limitedby the formation of an active complex which promotes transcriptionand thus presumably also ER ${alpha}$ ubiquitylation (48).

A prevalent role of ubiquitin in mediating regulation of transcriptionfactor activity is recognized: the ubiquitylation is thoughtto comprise a mechanism to switch off transcription until itis reinitiated, facilitating precise regulation of activatorfunction (38, 50). Such a mechanism could regulate transcriptionfactors whose activity is tightly controlled, allowing for aprecise switching mechanism and rapid reinitiation of transcriptionin the case of continuous stimulation (30, 40). Furthermore,a possible direct association between transactivation and ubiquitylation,in which the activation domains of certain factors overlap withelements that signal destruction, has been noted (49, 50). However,the exact mechanisms through which ubiquitylation is effectedat the promoter have yet to be elucidated, while additionalroles for this modification in regulating transactivation, otherthan through protein degradation, have also been proposed (5,38). Evidence for nonproteasomal effects of ubiquitylation intranscription include the finding that the APIS complex of the19S base interacts with the Gal4 activation domain and appearsto be involved in transcriptional activation independently ofdegradation, perhaps through enhancing initiation and/or elongation(14, 16, 43).

Our hypothesis for this work was that ER ${alpha}$ interacts with GnRH-mediatedsignals at the promoter to activate transcription of the LHßgene, and our aim was to elucidate the nature of this interactionand its effects. This led us to discover that GnRH exposureincreases levels of ubiquitin-conjugating enzyme 4 (ubc4) andcauses ubiquitylation of ER ${alpha}$ while enhancing its transactivationof the LHß gene and being involved in its degradation.We further demonstrate that the ER ${alpha}$ effect on LHß geneexpression is direct, showing that ER ${alpha}$ is associated with theLHß gene proximal promoter in its native chromatinsetting, and propose the mechanism of its recruitment.

MATERIALS AND METHODS

Top

Materials and Methods

Western blotting. Cells were lysed in extraction buffer (1x phosphate-bufferedsaline [PBS], 2% sodium dodecyl sulfate [SDS]) and stored at–80°C until required. The proteins (30 µg/lane)were resolved by electrophoresis on 12% SDS-polyacrylamide gelsat 100 V for 2 h and subsequently transferred to an Immuno-Blotpolyvinylidene difluoride membrane (Bio-Rad, Hercules, CA) intransfer buffer (39 mM glycine, 48 mM Tris, 0.037% SDS, and20% methanol) at 30 V for 4 h at 4°C. The membrane was blockedwith 10% nonfat milk in Tris-buffered saline-Tween (TBST) (50mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.1% Tween 20) for 1 h atroom temperature (RT), washed with TBST for 30 min at RT, andthen incubated for 1 h at RT with antisera to ER ${alpha}$ or ubc4 (H-184or MC-20 for ER ${alpha}$ and SC-15000 for ubc4; Santa Cruz Biotechnology,Inc., Santa Cruz, CA) and GAPDH (sc-20357; Santa Cruz Biotechnology)diluted 1:1,000 in TBST plus 1% nonfat milk. After washing for30 min, the membrane was incubated with goat or bovine horseradishperoxidase-conjugated antisera to rabbit or goat immunoglobulinG (0.4 µg/ml; Santa Cruz) for 1 h at RT and rinsed inTBST for 30 min at RT. The immunoreactive protein was detectedusing the Super Signal Pico West chemiluminescent system (PierceChemical Co., Rockford, IL), followed by exposure to Cl-XPosurefilm (Pierce Chemical Co.) for 1 to 5 min.

Cell culture and transient transfections. Cells were cultured in six-well plates and transiently transfectedfor chloramphenicol acetyltransferase (CAT) assays as describedpreviously (35) using either FuGENE 6 (Roche Diagnostics, Basel,Switzerland) or GenePORTER 2 (Gene Therapy Systems, San Diego,CA). Alternatively, for luciferase assays, cells were grownin 96-well plates, and vectors for rLHß-luc (200 ng),the transcription factors (80 ng for Sf-1 and ER ${alpha}$ ; 10 ng forPitx1), and simian virus 40-Renilla (2 ng) were transfectedusing GenePORTER 2 after the total amount of DNA had been equilibratedusing pWhitescript. Cells were incubated for 40 to 48 h aftertransfection before harvest. For cells exposed to GnRH (Buserilin)or E₂ (Sigma, St. Louis, MO), these were added to the mediumto a final concentration of 10 nM 24 h before harvest unlessotherwise stated. MG132 (Z-Leu-Leu-Leu-al; Sigma) was dissolvedin dimethyl sulfoxide and added to cells to a final concentrationof 10 µM 16 h before harvest.

CAT and ß-galactosidase (ß-Gal) assays wereperformed as described previously (27), and normalized activitywas calculated as a ratio to the basal levels of activity incells transfected with the control reporter constructs alone.Firefly luciferase and Renilla levels were measured using theDual-Glo system (Promega) and the Veritas Microplate luminometer(Turner Biosystems), and reporter gene activity was calculated,after normalization, as activity (n-fold) over that in untreatedcontrols.

Plasmid constructs. The csLHß-CAT and rLHß-CAT constructs havebeen described previously (28, 64). The rLHß-luc constructwas created by ligating the 5' flanking region of the rLHßgene from the above construct into the pGL3 plasmid (Promega)containing the firefly luciferase reporter gene. Site-directeddeletions were carried out using the QuikChange site-directedmutagenesis kit (Stratagene, La Jolla, CA) using primers comprising25 bp of the region flanking the response element on both sidesand the complementary sequence. All constructs were verifiedby sequencing.

The ubc4 expression vector was created by amplifying the codingsequence (accession no. BC003923) from LßT2 cell cDNAusing primers to include BamHI and EcoRI restriction sites (5'-GACGGATCCATGGCTCTGAAGAGAATC-3'and 5'-GCAGAATTCTTACATCGCATACTTCTG-3'). The amplified and digestedfragment was cloned into pCS2.

The small interfering RNA (siRNA) constructs for ubc4 and thegreen fluorescent protein (GFP) control were constructed inthe pSUPER vector (Oligoengine, Seattle, WA) to include thespecific 19-bp target sequence for the siubc4 construct (5'-GGCTACAATAATGGGGCCA-3')and the sequence for the siGFP construct (5'-GAACGGCATCAAGGTGAAC-3').All constructs were verified by sequencing.

The expression vectors for ER ${alpha}$ , Pitx1, and Sf-1 were gifts fromF. Pakdel (Rennes, France), J. Drouin (Montreal, Canada), andK. Parker (Durham, NC), respectively.

RNA extraction and cDNA synthesis. RNA was extracted using TRIzol (Invitrogen, Carlsbad, CA), andthe total RNA (5 µg) was reverse transcribed using SuperscriptII reverse transcriptase (Invitrogen) and random oligo(dT) primers(5 mM; New England Biolabs, Beverly, MA).

Construction and screening of subtractive hybridization libraries. Subtractive hybridization was carried out using the PCR-Selectsubtraction kit (BD Biosciences Clontech, Palo Alto, CA) usingmRNA from unstimulated LßT2 cells and those exposedto GnRH for 8 h. The secondary PCR products were cloned intopGEM-T-Easy Vector (Promega, Madison, WI), and the up-regulatedsequences were enriched by PCR. The library was plated, andindividual colonies were picked for sequencing and identificationagainst GenBank using BLAST (http://www.ncbi.nlm.nih.gov/BLAST).

Real-time PCR. Real-time PCR was carried out using the SYBR green I dye withthe ABI Prism 7700 sequence detector (Perkin-Elmer Applied Biosystems).Initially, all PCR products were analyzed on agarose gels toensure specificity of the amplification. PCRs were performedin a 20-µl volume, containing the PCR Master Mix, forwardand reverse primers (for ubc4, 0.1 µM [each] 5'-AATGACCTGGCTCGAGATC-3'and 5'-GCAACCTTAGGCGGTTTGA-3'; for ß-actin, 5'-GCCATGTACGTAGCCATCCA-3'and 5'-ACGCTCGGTCAGGATCTTCA-3'), and 1 µl cDNA template.The samples were heated to 95°C for 10 min, followed by40 cycles of 95°C for 15 s and 60°C for 1 min. The templateswere serially diluted fivefold in order to ensure that the dynamicranges of both the target and reference were similar, and thecomparative cycle threshold (C_T) method was used to comparemRNA levels in the various samples. All samples were assayedin duplicate.

Stable transfection and enrichment of tagged proteins. The mouse ubiquitin coding sequence was PCR amplified from LßT2genomic DNA extracted with DNAzol (Invitrogen) using primersdesigned to include NotI and PacI restriction sites (5'-CCTGAGAGCGGCCGCATGCAGATC-3'and 5'-CCGATTAATTAATTAGCCACCCCTC-3'). The digested product wascloned into pIVEX2.4b (Novagen). The sequence encoding the Histag and ubiquitin was PCR amplified to include BamHI and XhoIrestriction sites (CTTTAAGAAGGATCCGCCACCATGGCTGG and 5'-CTCGCTCGAGATCGACTTGCCATG-3')and subcloned into pCMV (Stratagene).

LßT2 cells were transfected with the His-tagged ubiquitinexpression vector using GenePORTER 2. After 48 h, cells weresplit into 96-well plates with around 500 cells per well. Stablytransfected cells were selected by gentamicin (150 µg/ml)for 2 weeks, after which individual clones were transferredto six-well plates. After 1 week, these were selected for verificationof expression by Western blotting using antisera to the Histag.

The stably transfected cells were lysed by boiling in PBS bufferwith 2% SDS, and lysates were incubated at 100°C for 15min prior to centrifugation (14,000 rpm, 15 min). The extractedprotein (500 µg) was loaded onto the His tag binding column(Amersham Biosciences, Piscataway, NJ) preequilibrated with20 mM imidazole in phosphate-NaCl buffer (20 mM phosphate and0.5 M NaCl, pH 7.4). After binding for 15 min at RT, the unboundproteins were removed by centrifugation (735 x g for 1 min).The column was washed with 240 mM imidazole in the phosphate-NaClbuffer and incubated for 15 min, after which bound proteinswere eluted in 100 µl of 480 mM imidazole. Proteins wereprecipitated with trichloroacetic acid and resuspended in 25µl loading buffer before separation on SDS gels and Westernanalysis, as described above.

Two-hybrid assays. The reporter gene was the Gal4 response element of pG5CAT (Clontech),which was subcloned into the pGL3 firefly luciferase plasmid.The ER ${alpha}$ cDNA was cloned into the Gal4 DNA binding domain (DBD)-encodingplasmid (pM), and the ER ${alpha}$ , Sf-1, and Pitx1 cDNAs were clonedinto the activation domain-encoding plasmid (pVP16; both fromClontech). Cells were plated at 2 x 10⁴ cells/well in 96-wellplates before transfection using 2 µl GenePORTER 2 perwell, 0.15 µg of the pM and pVP16 fusion constructs, and0.1 µg of the reporter gene, with 0.01 µg of simianvirus 40-Renilla luciferase as internal control. As marked,ubc4 expression vector was cotransfected at 0.1 µg, andtotal DNA amounts were equilibrated using pWhitescript. Cellswere incubated for 40 h before harvest, with E₂ (Sigma) addedto the medium (0.1% volume) for the last 24 h. Luciferase activitywas measured and normalized, and reporter gene activity wascalculated as activity over basal levels (n-fold) generatedfrom transfection of the unfused pM and pVP16 constructs.

Plasmid immunoprecipitation (PIP). ${alpha}$ T3-1 cells (60% confluent in 100-mm plates) were transfectedwith 6 µg DNA and incubated for 24 h before cross-linkingby formaldehyde (1%) for 10 min at RT on a shaking platform.Cross-linking was arrested by addition of glycine (0.125 M)for 5 min at RT. Cells were washed twice with 3 ml PBS containingprotease inhibitors (1 mM phenylmethylsulfonyl fluoride, 1 µg/mlleupeptin, 1 µg/ml pepstatin A) before harvest. Cellsfrom each plate were divided into four samples, centrifuged(8,000 rpm, 10 min, 4°C), resuspended in 500 µl SDSlysis buffer (1% SDS, 10 mM EDTA, 50 mM Tris-HCl, pH 8.1, proteaseinhibitors), and incubated on ice for 10 min. Cell debris wasremoved (13,000 rpm, 10 min, 4°C) and the supernatant transferredto fresh tubes; a 20-µl aliquot was kept for DNA inputquantitation. The remainder was diluted with chromatin immunoprecipitation(ChIP) dilution buffer (0.01% SDS, 1.1% Triton X-100, 1.2 mMEDTA, 16.7 mM Tris-HCl, pH 8.1, 167 mM NaCl, protease inhibitors)before preclearance with protein A Sepharose (PAS) (40 µlof 50% [vol/vol]) for 2 h at 4°C on a rotating platform.The PAS was removed (7,000 rpm, 1 min, 4°C) for incubationof the supernatant with ER ${alpha}$ antisera (H-184; Santa Cruz) overnightat 4°C on a rotating platform.

Proteins were collected by incubation with 40 µl of 50%(vol/vol) PAS (1 h, 4°C on a rotating platform), centrifugation(700 rpm, 1 min, 4°C), and sequential washes (4 min on arotating platform) with each of the following buffers (1 ml):low-salt wash buffer (0.1% SDS, 1% Triton X-100, 2 mM EDTA,20 mM Tris-HCl, pH 8.1, 150 mM NaCl), high-salt wash buffer(0.1% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM Tris-HCl, pH 8.1,500 mM NaCl), and LiCl wash buffer (0.25 M LiCl, 1% NP-40, 1%deoxycholate, 1 mM EDTA, 10 mM Tris-HCl, pH 8.1). This was followedby two washes with Tris-EDTA. After each wash, the proteinswere collected by centrifugation (700 rpm, 1 min at 4°C).The DNA was eluted twice in 250 µl elution buffer (1%SDS, 0.1 M NaHCO₃) and vortexed for 15 min at RT, and the eluatewas collected by centrifugation (700 rpm, 1 min). Reverse cross-linkingwas carried out on the 500 µl combined eluates by incubatingat 65°C for 4 h with addition of 5 M NaCl (40 µl)and RNase (2 µl of 100 ng/µl). Further incubationwas carried out for 1 h at 45°C in the presence of 10 µlof 0.5 M EDTA, 20 µl of 1 M Tris-HCl, pH 6.5, and 2 µlof 10-mg/ml proteinase K. The precipitated DNA was recoveredby phenol-chloroform and extracted with isopropanol-sodium acetate.Reverse cross-linking and extraction were carried out similarlyfor the input DNA.

Precipitated and input DNA served as template for the PCRs usingprimers to the CAT3 (AAGTTGGGTAACGCCAGGGT and ACAAGGGTGAACACTATCCC)or CAT6 (GCGACACGGAAATGTTGAAT and ACAAGGGTGAACACTATCCC) vectoror the csLHß (TGTGTACAGCAGGGATCTAA and TTCAGGGTTGGATGCTCTGC)or rat LHß (ACGCAAACTCCAACATTAGA and TCCCTACCTTGGGCACTTGG)proximal promoter. Amplified products were separated on ethidiumbromide-stained agarose gels.

ChIP. LßT2 cells (80% confluent in 60-mm plates; approximately5 x 10⁶ cells) were treated with E₂ (10 nM) for 1.5 h beforeaddition of GnRH (100 nM). Cross-linking was carried out andarrested, and cells were rinsed and collected as for PIP butin batches of 4 x 10⁶ cells. Cells were centrifuged (8,000 rpm,10 min, 4°C), resuspended, and incubated on ice for 10 minin 800 µl SDS lysis buffer. Samples were sonicated witha MISONIX XL2020 sonifier as follows: setting 2, 10 s at 30%duty cycle; setting 1, 10 s at 30% duty cycle with a 30-s intervalon ice. Cell debris was removed (13,000 rpm, 10 min, 4°C)and the supernatant transferred to fresh tubes. A 20-µlaliquot was removed, and the remainder was diluted and processedas described above. Regions of the proximal promoters of LHßor ß-actin genes were amplified from input and precipitatedsamples as in the PIP, using primers as described above or thefollowing primers for ß-actin: 5'-GCCATGTACGTAGCCATCCA-3'and 5'-ACGCTCGGTCAGGATCTTCA-3'. Alternatively, quantificationwas carried out using real-time PCR with primers on the mouseLHß proximal promoter (5'-GTGAAGCCCACCCACCACGC-3'and 5'-CCTTGGGCACCTGGCTTTAT-3') using ABI Prism 7700 and SYBRgreen. Each sample was assayed in duplicate, and the amountof precipitated DNA, based on average C_T, was calculated inrelation to that in the input sample.

Statistical analysis. One-way analysis of variance (ANOVA) followed by Bonferroni'sor Welch's t test were employed to determine means that werestatistically different. Differences were considered significantwhen P was <0.05. All experiments were repeated at leastthree times, and representative individual experiments or compiledresults are presented.

RESULTS

Top

Results

GnRH leads to ER ${alpha}$ ubiquitylation and proteasome-dependent degradation in gonadotropes, which is mediated by stimulation of ubiquitin-conjugating enzyme 4. LßT2 gonadotrope cells were exposed to GnRH for 0to 24 h, after which cells were harvested and lysed and Westernanalysis was carried out to detect changes in ER ${alpha}$ protein levels.The ER ${alpha}$ protein level was slightly reduced after 2 h of GnRHexposure, barely detectable after 8 h, and undetectable after24 h; a similar decrease in the levels of the GAPDH internalcontrol did not occur (Fig. 1A).

View larger version (32K):
[in this window]
[in a new window]
FIG. 1. GnRH exposure of gonadotropes causes a reduction in ER ${alpha}$ protein levels via activation of ubc4. (A) LßT2 cells were cultured and exposed to GnRH for 0 to 24 h. The cells were lysed, and the cell extract was run on 12% SDS gel before Western analysis using antisera to ER ${alpha}$ (upper panel) or GAPDH (lower panel) as control for protein loading. (B) LßT2 cells were cultured, some were exposed for 24 h to 10 nM E₂ or GnRH in the absence or presence of MG132, and levels of ER ${alpha}$ (upper panel) and GAPDH (lower panel) proteins were assessed as described for panel A. (C) Real-time PCR was carried out on LßT2 cells following GnRH treatment (10 nM; 0 to 8 h), and mRNA levels of ubc4 and ß-actin were measured and are expressed as the levels (n-fold) over those in untreated cells. (D) ubc4 (upper panel) and GAPDH (lower panel) protein levels were measured as in panel A in control cells, in cells treated with GnRH, and in cells in which ubc4 was overexpressed. (E) Protein levels of ubc4 (top panel), ER ${alpha}$ (middle panel), and GAPDH (lower panel) were measured in LßT2 cells following ubc4 overexpression or transfection of a construct to block its expression (siubc4) and with or without GnRH treatment.

In order to test the role of the proteasome in the inhibitoryeffect of GnRH on ER ${alpha}$ protein levels, LßT2 cells wereexposed to GnRH or E₂ for 24 h in the presence or absence ofthe proteasome inhibitor MG132. The GnRH and E₂ treatments bothreduced the level of the ER ${alpha}$ , but in both cases addition of MG132abated this effect (Fig. 1B).

In order to detect global transcriptional activation in gonadotropesresulting from GnRH exposure, subtractive hybridization wascarried out in control and GnRH-treated LßT2 cells.One of the clones in the subtracted library contained the codingsequence of ubiquitin-conjugating enzyme 4 (ubc4). The effectof GnRH on the mRNA levels of this gene were confirmed in real-timePCR, which showed a continuing increase in transcript levelsbetween 1 and 8 h of exposure to GnRH, reaching nearly ninefoldthat in the control cells by 8 h, while levels of ß-actinremained unchanged (Fig. 1C). Western analysis of LßT2cells confirmed that ubc4 protein levels are higher in LßT2cells following GnRH treatment and that they could be substantiallyreduced by transfection of a construct (siubc4) designed toinhibit ubc4 expression (Fig. 1D and E). The transfection ofthis siubc4 construct increased ER ${alpha}$ protein levels in cells exposedto GnRH (normalized values, 0.19 ± 0.02-fold over untreatedcontrols [GnRH treated only] versus 1.23 ± 0.15-foldover untreated controls [GnRH treated and siubc4 transfected]),while overexpression of ubc4 reduced the ER ${alpha}$ protein levels (normalizedvalue, 0.54 ± 0.13 of untreated controls) (representativegel shown in Fig. 1E). Transfection of the siGFP pSUPER plasmidas a negative control did not alter ER ${alpha}$ protein levels (normalizedvalue, 0.99 ± 0.03-fold of untreated controls) (not shown).These results indicate that the effect of GnRH in reducing ER ${alpha}$ protein levels could involve its increase in ubc4, which targetsubiquitylation of ER ${alpha}$ .

In order to examine the ubiquitylation status of ER ${alpha}$ followingGnRH exposure, a stably transfected LßT2 cell lineexpressing His-tagged ubiquitin was produced. Some of thesecells were exposed to GnRH, and the sample was enriched forHis-tagged proteins for subsequent Western analysis using ER ${alpha}$ antisera; unenriched cell extract was run alongside for comparison.The immunoreactive ER ${alpha}$ in the enriched samples differed fromthat of the unenriched samples by an additional ca. 10 kDa andwas evidently more abundant in GnRH-treated cells, apparentlyas a result of its increased monoubiquitylation (Fig. 2A). Atime course study revealed a maximal effect of GnRH on the ubiquitylatedER ${alpha}$ after 2 h, while it was virtually undetectable after 8 hof exposure. This corresponded with the levels of unubiquitylatedER ${alpha}$ , which dropped dramatically between 2 and 4 h of GnRH exposure,as seen in the unenriched cell lysate (Fig. 2B).

View larger version (40K):
[in this window]
[in a new window]
FIG. 2. GnRH and ubc4 stimulate ER ${alpha}$ monoubiquitylation. (A) LßT2 cells stably transfected with a His-tagged ubiquitin (Ub) were exposed to GnRH for 2 h. After lysis, the unenriched (10 µg protein/lane) and His tag-enriched (from 500 µg protein) lysates from control and GnRH-treated cells were run simultaneously, and Western analysis was used to detect the ER ${alpha}$ proteins. (B) The stably transfected LßT2 cells were exposed to GnRH for 0 to 8 h, and the cell lysate was analyzed by Western analysis as above, before (top panel) or after (bottom panel) enrichment for His-tagged proteins. GAPDH levels in the unenriched cell lysate are also shown (middle panel) as loading control. C, control untransfected LßT2 cells.

The liganded ER ${alpha}$ transactivates LHß directly in synergy with Sf-1 and Pitx1 without requiring a consensus ERE. In order to understand how GnRH-mediated ubiquitylation of ER ${alpha}$ affects its transactivation in the gonadotrope, we first soughtto elucidate the effects of ER ${alpha}$ on the LHß gene promoter,both directly and through synergistic interaction with othertranscription factors known to mediate GnRH effects on thisgene in the gonadotrope (11, 34, 45). Promoter regions fromthe LHß of rat (rLHß) and Chinook salmon(csLHß; the proximal 289 bp), ligated to CAT reportergenes, were transfected into COS 1 cells to test their responsivenessto ER ${alpha}$ and E₂. Only the csLHß gene promoter respondedto ER ${alpha}$ and E₂ alone, but the activity of both promoters increaseddramatically when ER ${alpha}$ was transfected together with Sf-1 andPitx1 expression vectors, reaching levels 40- to 60-fold thosein unstimulated cells. This effect was dependent on the presenceof E₂ (Fig. 3A and B). The levels of reporter gene activityin cells transfected with the promoterless CAT vector were unalteredby ER ${alpha}$ overexpression (not shown). The near-consensus ERE onthe csLHß proximal promoter was subsequently removed,and the synergistic effect was still apparent; an inhibitoryrole for this sequence in basal transcription in these cellswas also revealed (Fig. 3C).

View larger version (34K):
[in this window]
[in a new window]
FIG. 3. The liganded ER ${alpha}$ transactivates two vertebrate LHß gene promoters in synergy with Sf-1 and Pitx1. (A) The rLHß promoter-CAT6, (B) the csLHß 289-bp proximal promoter-CAT3, (C) the longer 3.3-kbp csLHß promoter-CAT3 with or without deletion of the ERE (at –260 bp), or (D) the rLHß promoter-luciferase constructs were transfected into COS 1 cells alone or with ER ${alpha}$ , Sf-1, and/or Pitx1 expression vectors. For panels A and B, as indicated, cells were exposed to 10 nM E₂ at the time of transfection; for panels C and D, all cells transfected with ER ${alpha}$ were exposed to E₂. Cells were harvested, reporter gene levels were assayed, and normalized values are expressed as the levels (n-fold) over those in control cells. (A to C) Statistical analysis (ANOVA followed by Bonferroni t test) compared mean values separately for groups with and without E₂ (for panels A and B) or for intact and ERE-deleted promoter constructs (for panel C), and similar values (P > 0.05) are designated with the same letter in uppercase (for control and intact constructs) or lowercase (for E₂-treated or ERE deletion constructs). (D) Welch's t test was used to compare means of treatments with and without cotransfection of ER ${alpha}$ plus E₂. *, P < 0.05; **, P < 0.005. Means ± standard errors of the mean (SEM) are shown (n = 4 to 6).

Further experiments were carried out to clarify the relativeimportance of Sf-1 and Pitx1 for the synergy with ER ${alpha}$ and E₂on the rLH promoter using the rLH-luc construct. The levelsof reporter gene activity in cells transfected with the promoterlesspGL3 vector are increased somewhat by ER ${alpha}$ and E₂ or by Pitx1overexpression; these levels of increase were thus subtractedfrom values obtained with the rLH-luc construct. Although ER ${alpha}$ overexpression alone failed to increase basal rLHßpromoter activity, it increased the effects of Sf-1 or Pitx1by six- or twofold, respectively, while increasing the combinedeffects of Sf-1 and Pitx1 by as much as sevenfold (Fig. 3D).Similar experiments on the csLHß promoter constructhave been carried out previously, revealing the synergisticaction of ER ${alpha}$ with Sf-1 alone, which is increased further withthe addition of Pitx1, while ER ${alpha}$ does not act synergisticallywith Pitx1 alone (27, 35).

The mechanism of association of endogenous ER ${alpha}$ with these LHßproximal promoters in the gonadotropes was demonstrated usingPIP. For this, the same LHß-CAT constructs were transfectedinto mouse gonadotrope ${alpha}$ T3-1 cells for subsequent cross-linkingand pull-down of the ER ${alpha}$ and associated plasmid DNA. In cellstransfected with the rLHß-CAT construct, but not whenthe empty CAT vector was used, PCR successfully amplified regionsof the construct precipitated using ER ${alpha}$ antisera. Deletion ofthe distal nonconsensus ERE (at bp –1159) (Fig. 4A) revealedthat this element is not required for association of ER ${alpha}$ withthe rLHß gene promoter, and deletion of the distalSf-1 binding site (at bp –119) alone or together withthe ERE also failed to abolish ER ${alpha}$ binding to the rLHßgene promoter. However, removal of both proximal (at bp –51)and distal Sf-1 binding sites abolished ER ${alpha}$ association withthe promoter; this was in accordance with the binding of Sf-1,which was prevented only by the removal of both Sf-1 bindingsites (Fig. 4B). Subsequently, a similar experiment to assessthe role of the Pitx1 response element (RE) in association ofthe ER ${alpha}$ with the rLHß gene promoter was carried out.Removal of this element revealed that the ER ${alpha}$ association isconsiderably reduced compared to the intact control promoterconstruct but is not abolished (Fig. 4C).

View larger version (51K):
[in this window]
[in a new window]
FIG. 4. Association of ER ${alpha}$ with the rLHß promoter requires the Sf-1 REs. (A) Locations of response elements on the rLHß gene promoter for Sf-1, Pitx1, Egr-1, and a distal ERE are shown. (B) The intact rLHß-CAT construct or that lacking the ERE and/or proximal, distal, or both Sf-1 REs was transfected into ${alpha}$ T3-1 cells; also transfected was the empty CAT6 vector. After 24 h, cells were fixed by formaldehyde before lysis and precipitation of plasmids and associated proteins using antisera to ER ${alpha}$ or to Sf-1. The cross-linking was reversed and PCR amplification carried out using primers targeting the respective vectors. The top row in each pair (input) shows amplification of the target sequences in the same input samples, and the lower row in each pair (PIP) shows those after precipitation. All treatments were carried out in duplicate, with and without the antisera. (C) Similarly, the intact rLHß-CAT construct and that lacking the Pitx1 RE were transfected into ${alpha}$ T3-1 cells, and PIP was carried out in duplicate using antisera to ER ${alpha}$ . Amplification from the input sample is shown, as are those resulting from precipitation with and without antisera. Also shown are nontransfected controls (right column).

Similar experiments on the csLHß gene promoter revealedthat the near-consensus ERE (at bp –260) (Fig. 5A) isdispensable for ER ${alpha}$ association with the promoter, but removalof the single Sf-1 binding site at bp –154 abolished bothER ${alpha}$ and Sf-1 binding (Fig. 5B). Removal of the Pitx1 RE did notappear to alter association of the ER ${alpha}$ with the csLHßpromoter construct (Fig. 5C).

View larger version (39K):
[in this window]
[in a new window]
FIG. 5. Association of ER ${alpha}$ with the csLHß promoter requires the Sf-1 RE. (A) Locations of response elements on the csLHß gene promoter for Sf-1, Pitx1, and ER are shown. (B and C) PIP assays were carried out using the intact csLHß-CAT construct or (B) that lacking the ERE or the Sf-1 RE; also transfected was the empty CAT3 vector; (C) alternatively, intact and Pitx1 RE-deleted csLHß-CAT constructs were used. Precipitation and other experimental details and presentation of the data are as in Fig. 4.

Subsequently, reporter gene experiments were carried out inLßT2 gonadotropes to verify the functionality of theseREs in the response of the csLHß promoter to ER ${alpha}$ overexpression.The csLHß gene promoter shows a clear response toER ${alpha}$ overexpression, and the CAT activity was dramatically reducedwith removal of the ERE or Sf-1 REs, while deletion of Pitx1RE had no effect. In contrast with the PIP studies, however,some induction of promoter activity in response to ER ${alpha}$ was evidentin the absence of the Sf-1 RE, presumably resulting from thehigh levels of ER ${alpha}$ , which was overexpressed in this study, sofacilitating its binding directly to the ERE; this was abolishedwhen both REs were removed (Fig. 6A). Notably, the effect ofERE deletion to reduce slightly the basal promoter activitydiffered from that seen in the COS 1 cells, where the deletiongave rise to higher levels of promoter activity (e.g., Fig.3C), presumably because it binds an inhibitory factor in COS1 cells. In contrast to the csLHß gene promoter activity,the rLHß gene promoter is not responsive to ER ${alpha}$ expressionalone but is responsive only when Sf-1 and Pitx1 are also overexpressed;this additive effect of ER ${alpha}$ on stimulation by Sf-1 and Pitx1requires both the Sf-1 and Pitx1 REs (Fig. 6B).

View larger version (27K):
[in this window]
[in a new window]
FIG. 6. Activation of the LHß gene promoter by ER ${alpha}$ requires Sf-1 REs. (A) The intact csLHß-CAT construct and the same construct lacking the ERE, Sf-1 RE, or Pitx1 RE or having combinations of these deletions were transfected into LßT2 cells. (B) Alternatively, the intact rLHß-CAT construct or the same construct lacking the distal ERE, both Sf-1 REs, or the Pitx1 RE were transfected into COS 1 cells, all with or without cotransfection of the ER ${alpha}$ expression vector and exposure to E₂. After 48 h, cells were harvested and CAT/ß-Gal levels assayed and presented as in Fig. 3. Statistical analysis (ANOVA followed by Bonferroni t test) compared mean values separately for groups with or without ER ${alpha}$ , and similar values (P > 0.05) are designated with the same letter in upper- or lowercase. Additional analysis compared means for the same promoter constructs with or without ER ${alpha}$ , and only those significantly different are shown. *, P < 0.05. Means ± SEM are shown (n = 3 to 4).

GnRH-induced ubc4 enhances ER ${alpha}$ transactivation of the LHß gene. In order to verify the role of ubc4 in mediating the GnRH andER ${alpha}$ effects on LHß gene transcription, LßT2cells were transfected with the csLHß-CAT constructand exposed to GnRH, and some were cotransfected with ER ${alpha}$ andexposed to E₂. Some of these cells also had the siubc4 constructcotransfected. GnRH or ER ${alpha}$ (with E₂) alone increased promoteractivity by around 7- or 12-fold over controls, while a combinationof both treatments increased activity to nearly 35-fold thatin control cells. Cotransfection of siubc4 led to a drop inpromoter activity in all cells treated with GnRH or a smallerdrop in cells in which ER ${alpha}$ was overexpressed alone (Fig. 7A);transfection of the siGFP pSUPER plasmid as a negative controldid not alter transactivation of the LHß gene by ER ${alpha}$ ,with or without GnRH (P > 0.1) (not shown). Conversely, theeffect of ER ${alpha}$ overexpression was enhanced by cotransfection withthe ubc4 expression vector in both gonadotrope and nongonadotropecells (Fig. 7B and C). In the latter, ubc4 overexpression alsoincreased the already synergistic effect of ER ${alpha}$ with Sf-1 andPitx1 (Fig. 7C).

View larger version (33K):
[in this window]
[in a new window]
FIG. 7. ubc4 is involved in mediating the effect of GnRH on the LHß gene and increases ER ${alpha}$ transactivation and apparent dimerization and interaction with Sf-1 and Pitx1. (A and B) LßT2 or (C) COS 1 cells were transfected with the csLHß promoter-CAT construct with or without ER ${alpha}$ , Pitx1, and Sf-1 expression vectors and/or exposed to GnRH, as marked. All cells cotransfected with ER ${alpha}$ were also exposed to E₂. Some of the cells were also transfected with the siubc4 construct or ubc4 expression vector. After 48 h, cells were harvested and CAT/ß-Gal levels were assayed; results are presented as in Fig. 3. (D) Two-hybrid assays were carried out in LßT2 cells using a Gal4 DBD-ER ${alpha}$ fusion construct with VP16 AD fused to ER ${alpha}$ , Pitx1, or Sf-1, the Gal4-responsive reporter gene, and a Renilla luciferase construct for internal control; all cells were exposed to E₂ (10 nM) at the time of transfection. Luciferase activity was measured 40 h after transfection and is calculated, after normalization and subtraction of any activity of the fusion constructs alone, as the level (n-fold) over the levels in the negative control, in which unfused Gal4 DBD and VP16 AD constructs were transfected (basal level). (A to D) Statistical analysis (ANOVA followed by t test) compared pairs of means for each treatment group, with and without cotransfection of siubc4 or ubc4 expression vector; statistically different means are designated with the following symbols: *, P < 0.05; **, P < 0.01; or ***, P < 0.001. Other values are not significant (N.S.). Means ± SEM are shown (n = 3 to 7).

In order to test for an effect of ubc4 on ER ${alpha}$ protein-proteininteractions, two-hybrid assays were performed in LßT2cells using a Gal4-responsive luciferase reporter gene and ER ${alpha}$ fused to the Gal4 DBD. These constructs were transfected togetherwith ER ${alpha}$ , Sf-1, or Pitx1 fused to the VP16 activation domain(AD). Some of the cells were also cotransfected with the ubc4expression vector. Background levels of reporter gene activationby Sf-1, Pitx1, or ER ${alpha}$ fused to the VP16 AD or ER ${alpha}$ fused to theGal4 DBD, where present, were subtracted from the values ofthe interacting protein; any activity by these constructs alonewas unaffected by ubc4 overexpression (not shown). Luciferaseactivity was elevated in cells transfected with ER ${alpha}$ -Gal4 DBDand Sf-1-VP16 AD, Pitx1-VP16 AD, or ER ${alpha}$ -VP16 AD, and the activitywas further increased for all of these pairs following ubc4overexpression (Fig. 7D). Similar studies were carried out totest the effect of GnRH exposure on these interactions, whichincreased generally to a lesser extent than seen following ubc4overexpression, although the interaction of ER ${alpha}$ with Pitx1 reachedover threefold that in the absence of GnRH (not shown).

GnRH and ubc4 increase the association of ER ${alpha}$ with the LHß gene promoter and lengthen its cycling time. Finally, in order to reveal the effects of GnRH and ubc4 onthe association of the endogenous ER ${alpha}$ with the native mouse LHßpromoter, ChIP assays were carried out in mouse LßT2gonadotropes. Initially, this was to evaluate changes in theassociation of endogenous ER ${alpha}$ with the LHß promoterfollowing treatment with GnRH or ubc4 overexpression or knockdownof ubc4 through transfection of the siubc4 construct; all cellsexcept the controls were exposed to E₂. The association of ER ${alpha}$ with the LHß promoter was evident in all cells exposedto E₂ except those in which the siubc4 construct was transfected.Notably, the siRNA knockdown of ubc4 abolished the clear stimulatoryeffect of GnRH on ER ${alpha}$ association with the LHß genepromoter (Fig. 8).

View larger version (37K):
[in this window]
[in a new window]
FIG. 8. GnRH treatment increases the association of ER ${alpha}$ with the murine LHß promoter through ubc4. LßT2 cells in 60-mm plates were transfected with ubc4 expression vector, siubc4 construct, or neither, as marked, all in duplicate. After 48 h, all cells except controls were treated with E₂ for 90 min before addition of GnRH (100 nM) to some of the cells, as marked. After a further 50 min, ChIP was carried out and PCR used to amplify endogenous mouse LHß gene promoter precipitated by the ER ${alpha}$ antisera and also in an aliquot from the same batch of cells which was designated as the input sample. Amplification of ß-actin was also performed as a negative control.

Subsequently, effects of these treatments on the cycling ofER ${alpha}$ on this promoter were tested using LßT2 cells treatedwith E₂ at 10-min intervals and then analyzed by ChIP and quantitativereal-time PCR. In cells treated with E₂ alone, the ER ${alpha}$ cycledon and off the proximal promoter at an interval of 30 to 40min; however, this cycling was abolished in cells which hadbeen transfected with the siubc4 construct (Fig. 9A). In cellsexposed to GnRH, the cycling time increased to around 70 min,while the ER ${alpha}$ pull-down increased the presence of the LHßgene by as much as 1,000-fold over that in the input aliquot;both of these effects were consistently seen in repeated experiments,although the magnitude of increased ER ${alpha}$ association followingGnRH treatment was not always this large. Both of these GnRHeffects were absent in siubc4-transfected cells (Fig. 9B).

View larger version (27K):
[in this window]
[in a new window]
FIG. 9. GnRH treatment lengthens the cycling times of ER ${alpha}$ on the murine LHß promoter through ubc4. LßT2 cells were transfected with (A and B) the siubc4 construct, (C) ubc4 expression vector, or neither for 48 h, after which all cells were treated with E₂ for 90 min. Some of the cells were then exposed to GnRH (100 nM) for 50 min, after which ChIP was carried out, and real-time PCR measured the association of ER ${alpha}$ with the endogenous murine LHß gene promoter. Levels of LHß gene promoter precipitated by the ER ${alpha}$ antisera were calculated in relation to the levels in the aliquot from the same batch of cells, which was designated as the input sample.

Overexpression of ubc4 had an effect that was similar to thoughless potent than that of GnRH, increasing the apparent affinityof ER ${alpha}$ for the promoter by as much as 50-fold and increasingthe amplitude of the cycles while lengthening the cycle to 50to 60 min (Fig. 9C).

DISCUSSION

Top

Discussion

We report here a novel form of cross talk between two hormonallyinduced signaling pathways that regulate LHß genetranscription, in which GnRH activates ER ${alpha}$ ubiquitylation, apparentlythrough stimulation of ubc4. We further demonstrate that thestimulatory effect of ER ${alpha}$ on this gene is mediated through interactionswith other regulatory transcription factors on the proximalpromoter without necessarily requiring an ERE. The GnRH-stimulatedubc4 increases ER ${alpha}$ binding to the LHß promoter butultimately also leads to its degradation (Fig. 10). Estrogenhas a crucial role in the estrous cycle, regulating preciselythe levels of circulating LH, whose surge is crucial in instigatingovulation. The concept that the liganded ER ${alpha}$ acts as a molecularswitch whose activity can be modified locally at the gonadotropeby GnRH-mediated ubiquitylation is in keeping with this regulatoryrole while adding a novel dimension to the molecular mechanismsregulating gonadotropin synthesis.

View larger version (32K):
[in this window]
[in a new window]
FIG. 10. Model for the role of GnRH-induced activation of ubc4 and ER ${alpha}$ ubiquitylation in the activation of LHß gene transcription. GnRH activates formation of the tripartite complex of factors comprising Sf-1, Pitx1, and Egr-1 (3, 11, 20, 58), which bind respective REs on the mammalian LHß gene promoter (distal Sf-1 and Egr-1 REs are not shown). (1) At the same time, GnRH stimulates transcription of ubc4, which leads to ER ${alpha}$ monoubiquitylation and subsequent recruitment to the DNA-bound complex. Transcription is activated as a result of recruitment of additional coactivators (black shapes) and the general transcription machinery. (2) Transcriptional activation results in recruitment of additional E2 enzymes, E3 ligases, and E4 chain elongation factors to the complex, leading to ER ${alpha}$ polyubiquitylation, removal from the LHß gene promoter, and degradation by the proteasome. (3) The LHß promoter is available for subsequent activation by ER ${alpha}$ if the liganded receptor is present.

The finding that GnRH up-regulates transcript levels of ubc4is, to our knowledge, the first report of an E2 enzyme in theubiquitin pathway being directly stimulated by a cell membrane-boundhormone. GnRH was, however, previously implicated in stimulatingubiquitylation of inositol trisphosphate receptors in the ${alpha}$ T3-1gonadotropes, which was dependent on intraluminal Ca²⁺, althoughthe molecular mechanisms involved were not described (63). Arecent report indicated that ER ${alpha}$ is degraded in MCF-7 cells throughactivation of protein kinase C and not MAPK (34), both of whichare activated in the GnRH signal transduction pathway to theLHß gene (39). Elucidation of the GnRH-stimulatedsignaling pathways leading specifically to ubc4 transcriptionin the gonadotropes will require further experimental work.

Results from the present study suggest that ubc4 is the limitingfactor for ER ${alpha}$ ubiquitylation and degradation in the gonadotrope.This was also the case for ubc4- but not ubc2-mediated ubiquitylationof various rat proteins in which immunodepletion decreased andubc4 administration increased levels of the ubiquitylated proteins(47). Although some E2 enzymes, including ubc4, are able toubiquitylate the substrate directly, the E3 enzymes are thoughtto provide a crucial role in substrate recognition (5, 61).ER ${alpha}$ is known to interact with E6AP, which accepts ubiquitin fromubc4 and was associated with ER ${alpha}$ in re-ChIP experiments (1, 48).Moreover, gene knockout of E6AP resulted in reproduction-deficientanimals with small gonads. However, the females did cycle andeven produced litters (56). Given that estrogen-mediated regulationof cyclical circulating LH levels is crucial to ovulation, itis unlikely that E6AP has a role in regulating LHßproduction. ER ${alpha}$ has been found to associate with a variety ofadditional E3 ligases: at the pS2 promoter, ER ${alpha}$ associates withMDM2 and Rpt6, while other studies have shown ubiquitylationand degradation of ER ${alpha}$ are mediated through the NEDD8 pathway(12). NEDD8 enhances ubiquitylation by the SCF (Skp1, Cul-1,ROC1, and F-box protein) E3 complex and increases recruitmentspecifically of ubc4, so increasing polyubiquitylation (24,57).

We have demonstrated the adverse effect of GnRH and ubc4 onER ${alpha}$ protein levels, which appear to be concomitant with the increasein affinity of ER ${alpha}$ to the LHß gene promoter. The ideathat ubc4 ubiquitylates ER ${alpha}$ and that this regulates its transactivationof the LHß gene concurs with a recent report thatubc4 (UbcH5) is required for ligand-mediated ER ${alpha}$ transactivationof the pS2 promoter (44). That report further suggested thatTBL1 and TBLR1 are the specific adaptor proteins responsiblefor ubc4 recruitment, which was essential for transactivationby all nuclear receptors tested, although additional more specificE2 enzymes were also recruited: ER ${alpha}$ requires specifically UbcH7,which may be recruited through interactions with SRC-1 (44,60). An earlier study demonstrated that the incubation of ER ${alpha}$ with an E1 enzyme together with both ubc4 and UbcH7 led to degradationof ER ${alpha}$ but not of progesterone or thyroid hormone receptors (41).Our Western analysis revealed only monoubiquitylated ER ${alpha}$ , suggestingthat this might be a step distinct from subsequent polyubiquitylationand proteasomal degradation; if this were not the case, we wouldexpect to see at least diubiquitylated ER ${alpha}$ before further chainelongation and recognition by the proteasome. We thus proposethat the actions of GnRH-induced ubc4 comprise a prerequisiteto subsequent degradation by additional receptor-specific E2enzymes, possibly reflecting differentially regulated mono-and polyubiquitylation. The molecular mechanisms involved indifferentiating between mono- and polyubiquitylation have yetto be defined, and although a recent report suggested the concentrationof the E3 ligase could be the determinant (29), the latter maywell also require additional enzymes, such as E4 ligases involvedin promoting chain assembly, or the inhibition of negative regulators(26, 42) (Fig. 10).

Importantly, in this study we have demonstrated clearly thatER ${alpha}$ is associated with the LHß gene proximal promoterin murine gonadotropes and, in the presence of ligand, can directlystimulate its transcription. We have shown that the recruitmentof ER ${alpha}$ to the LHß gene promoter requires the Sf-1 RE,while we also demonstrated direct physical interaction betweenER ${alpha}$ and Sf-1 or Pitx1. Our PIP studies show that the ERE is notrequired for ER ${alpha}$ binding to either promoter, while our transfectionstudies also show ER ${alpha}$ transactivation in the ERE-deleted constructs;from here we conclude that the ERE is not required for ER ${alpha}$ transactivationof the LHß genes. For the rLHß gene promoter,the Pitx1 RE is also crucial for ER ${alpha}$ transactivation and mayplay a role in ER ${alpha}$ recruitment. The csLHß operatesslightly differently due to the presence of a near-consensusERE which can bind liganded ER ${alpha}$ but is dispensable when Sf-1and Pitx1 are overexpressed, while Pitx1 is not crucial to eitherrecruitment of ER ${alpha}$ or its transactivation. Our findings are consistentwith previous studies on the mammalian LHß gene promotershowing that Pitx1's ability to interact synergistically withSf-1 is largely independent from the requirement for a functionalPitx1 RE (58). The fact that the rLHß gene promoteris unresponsive to the liganded ER ${alpha}$ unless Sf-1 and Pitx1 areoverexpressed indicates a cooperative mechanism of ER ${alpha}$ recruitmentfor activation of mammalian LHß genes requiring activationof these other transcription factors. This scenario presentsthe LHß genes as those that are not predominantlyERE driven, as described by Fan et al. (13), and by their argument,LHß gene activity would less likely be regulated byER ${alpha}$ availability, so that ubiquitylation and proteasomal degradationwould not result in desensitization. This supports our resultswhich show that the activation of ubc4 by GnRH appears to causesensitization to ER ${alpha}$ transactivation, although we have not demonstratedconclusively that this is due to the ER ${alpha}$ ubiquitylation and notvia an indirect mechanism involving additional transcriptionfactors or other signaling molecules.

Enhanced protein-protein interactions have been attributed tomonoubiquitylation (21, 38). Although we show both ER ${alpha}$ dimerizationand its interactions with Pitx1 and Sf-1 appear to increasewhen ubc4 is overexpressed, these proteins do not contain anyof the domains known to bind ubiquitin (10, 52, 61). It is possible,therefore, that the increased transactivation of the reportergene occurs as a result of increased ER ${alpha}$ interaction with a commoncoactivator complex, which is independent from but still requiresthe VP16 AD for transcription initiation in this system. Thisis feasible, given that a number of coactivators have been shownto contain ubiquitin recognition domains, while the TBLR1-ubc4complex was proposed to serve specifically to facilitate removalof corepressors and help recruit both the ubiquitin/19S proteasomeand the coactivator complex (2, 17, 44). UbcH7 was recentlyalso reported to function as a steroid hormone coactivator,acting through SRC-1 (60). Further studies will be requiredto understand whether this increased interaction is a directconsequence of the ER ${alpha}$ ubiquitylation or a result of the recruitmentof additional bridging molecules to facilitate the increasedassociation with the LHß gene promoter.

Interestingly, in the gonadotropes, a recent study showed thatthe ring finger E3, SNURF, coprecipitates with Sf-1 and specificallyenhances LHß gene transcription (7). The proposedmechanism for this enhancement was through repressing suppressionby androgen receptor which inhibits transcription of this genethrough interactions with Sf-1 (23). However, SNURF, which alsoserves as a coactivator for ER ${alpha}$ , accepts ubiquitin from ubc4and is up-regulated by estrogen exposure (18, 22, 51). It isplausible, therefore, that the effects of SNURF on LHßgene transcription are in fact through ER ${alpha}$ ubiquitylation whoseincreased association with the preinitiation complex, and specificallyPitx1, facilitates competition of the latter with androgen receptorfor binding to the putative ligand binding domain of Sf-1 (seereference 59).

Given the fact that GnRH activates a number of factors whichbind the proximal LHß promoter, we initially hypothesizedthat GnRH exposure would facilitate formation of this complexand thus would shorten the cycling time of ER ${alpha}$ on and off theLHß gene promoter. Our results show that this is notthe case and that ER ${alpha}$ is, in fact, associated with the promoterfor longer cycles following GnRH treatment. This is concomitantwith an increase in transactivation of the LHß gene,possibly suggesting a change in rate of the stochastic reactionsinvolved in the recruitment of regulatory factors to the promoter.The GnRH exposure also increases the amount of LHßgene promoter associated with ER ${alpha}$ , indicating an increase inits affinity, presumably as a result of stronger interactionwith the gene-specific and preinitiation transcriptional complexes.This is at least partly due to ubiquitylation, as ubc4 overexpressionincreases ER ${alpha}$ association with the promoter and increases synergisticactions of the three transcription factors on LHßpromoter activity. Conversely, blocking ubc4 expression reducesthe ability of ER ${alpha}$ to associate with the promoter and also virtuallyabolishes the GnRH effect on promoter activity, indicating notonly that ubc4 is required for ER ${alpha}$ activity but also that ER ${alpha}$ is crucial in mediating the GnRH effect on the LHßgene.

Our results thus present a scenario in which the level of ER ${alpha}$ ubiquitylation is directly regulated by the signaling pathwayof a cell membrane-bound hormone and so increases transactivationof the target gene. This provides a new form of cross talk inhormonally stimulated regulation of gene expression which maybe widespread while also revealing novel aspects to the molecularmechanisms regulating gonadotropin synthesis.

ACKNOWLEDGMENTS

We thank J. Drouin, F. Pakdel, and K. Parker for the Pitx1,hER ${alpha}$ , and Sf-1 expression vectors and P. Mellon for the LßT2cells.

This research was supported by the Academic Research Fund andthe Office of Life Sciences, National University of Singapore.M.K. is a recipient of a Singapore Millennium Foundation scholarship,and P.M. is a recipient of the Young Investigator Award, Officeof Life Sciences, National University of Singapore.

FOOTNOTES

* Corresponding author. Mailing address: Functional Genomics Laboratories, Department of Biological Sciences, National University of Singapore, 14 Science Drive 4, Singapore 117542, Singapore. Phone: (65) 68741882. Fax: (65) 68722013. E-mail: dbsmp{at}nus.edu.sg.

Authors contributed equally.

REFERENCES

Top