Alternative O-glycosylation/O-phosphorylation of serine-16 in murine estrogen receptor beta: post-translational regulation of turnover and transactivation activity.

O-Linked N-acetylglucosamine (O-GlcNAc) is a dynamic post-translational modification abundant on nuclear and cytoplasmic proteins. Recently, we demonstrated that the murine estrogen receptor-beta (mER-beta) is alternatively O-GlcNAcylated or O-phosphorylated at Ser(16). Analyses of mER-betas containing mutations in the three adjacent hydroxyl amino acids at this locus confirmed that Ser(16) is the major site of O-GlcNAc modification on mER-beta and that mutants lacking hydroxyl amino acids at this locus are glycosylation-deficient. Pulse-chase studies in transfected Cos-1 cells demonstrate that the turnover rate of the mutant containing a glutamic acid moiety at Ser(16), which mimics constitutive phosphorylation at this locus, is faster than that of the wild type receptor. Whereas, the mutant without hydroxyl amino acids at this locus is degraded at a slower rate, indicating that O-GlcNAc/O-phosphate at Ser(16) modulates mER-beta protein stability. Luciferase reporter assays also show that the Ser(16) locus mutants have abnormal transactivation activities, suggesting that the two alternative modifications at Ser(16) on mER-beta may also be involved in transcriptional regulation. DNA mobility shift assays show that the mutants do not have altered DNA binding. Green fluorescence protein constructs of both wild type and mutant forms of mER-beta show that the receptor is nearly exclusively localized within the nucleus. It appears that reciprocal occupancy of Ser(16) by either O-phosphate or O-GlcNAc modulates the degradation and activity of mER-beta.

pcDNA3.1(Ϫ) for functional studies. All constructs were verified by automated DNA sequencing.
Altered Sites I (Promega, Madison, WI) was used to mutate mER-␤ Ser 16 . mER-␤ cDNA in pcDNA3.1(Ϫ) was subcloned into pAlter I via XbaI and EcoRI sites. The mutagenesis primers (shown in antisense; S16A, CCTTCCAGGTTACCGACGGCGGCGGGAACACTGTAGTTC; S16E, CCTTCCAGGTTACCGACCTCGGCGGGAACACTGTAGTTC) containing changes (with underline) of the Ser 16 to Ala or Glu along with the changes of two adjacent Ser 15 and Thr 17 to Ala and Val were synthesized and used, respectively, in the mutagenesis process. The mutants were named as S16A and S16E accordingly, as illustrated below in Fig. 1. The mutated cDNAs were verified by automated sequencing with serial internal primers covering the entire coding region and subcloned back into pcDNA3.1(Ϫ) and other plasmids for functional studies.
To study the subcellular localization of mER-␤, mER-␤ cDNA in pAlter I was digested with BglII and EcoRI and subcloned into GFP vector pEGFP-C1 (CLONTECH, Palo Alto, CA) in-frame. The final engineered mER-␤ cDNAs in pEGFP-C1 were verified by automated DNA sequencing. To immunoprecipitate mER-␤ expressed in mammalian cell lines, a FLAG tag encoding the peptide epitope, DYKDDDDK, was incorporated into the carboxyl-terminal end of mER-␤ cDNA using polymerase chain reaction (37). All constructs were verified by automated DNA sequencing.
Characterization of the Glycosylation of the mER-␤ O-GlcNAc Mutant-Expression and purification of mER-␤ proteins from Sf9 cells were described previously (17). The O-GlcNAc moieties on purified mER proteins were labeled with galactosyltransferase by transferring [ 3 H]galactose from UDP-[ 3 H]galactose, as described previously (38). The tritium-labeled protein was separated from unincorporated UDP-[ 3 H]galactose using a 1.5-ϫ 30-cm Sephadex G-50 column in 50 mM ammonium formate, 0.1%(w/v) SDS. For tritium images, the protein was resolved on 10% SDS-PAGE gel, and the gel was stained with Coomassie Blue R-250, impregnated with 1 M salicylic acid for 30 min, dried under vacuum, and exposed to x-ray film at Ϫ70°C for 2 days. For tryptic map comparison, the labeled proteins were digested with 2% (w/w) trypsin (sequencing grade, Roche Molecular Biochemicals, Indianapolis, IN) in 0.1 M ammonium bicarbonate (pH 7.4) at 37°C overnight. Separation of [ 3 H]galactose-labeled tryptic mER glycopeptides was achieved on a C 2 /C 18 reversed phase column (3.2 ϫ 100 mm, Amersham Pharmacia Biotech, Piscataway, NJ) using a 0 -60%(v/v) gradient of acetonitrile in 0.1%(w/v) trifluoroacetic acid over 90 min at a flow rate 0.1 ml/min on a SMART system (Amersham Pharmacia Biotech). Fractions were collected (0.2 ml/fraction) and counted.
Tissue Culture and Transfection-Estrogen receptor-deficient Cos-1 cells are maintained in phenol red-free Dulbecco's modified Eagle's medium/F-12 medium (Life Technologies Inc., Gaithersburg, MD) supplemented with 5% (w/v) charcoal/dextran-stripped fetal bovine serum (estrogen-depleted). Cos-1 cells were transfected using the liposome method (Life Technologies Inc.). At 24 h post-transfection, transfected cells were changed with fresh medium. 17␤-Estradiol (E 2 )was added to the medium at a final concentration of 20 nM, unless otherwise indicated.
Pulse-chase 35 S Labeling in Vivo-For pulse-chase analyses, Cos-1 cells (in 100-mm dishes) were transfected with wild type or mutant mER-␤. At 40 -48 h post-transfection, the cells were rinsed with methionine-cysteine-free medium once, incubated at 37°C for 30 min, then supplemented with 120 Ci of labeling mix ( 35 S Express label mix, PerkinElmer Life Sciences, Boston, MA) in the same medium. At desired time points, cells were placed into complete medium containing no label for the chase. Cells were harvested at various time points and lysed with 1 ml of lysis buffer (150 mM NaCl, 50 mM Tris, pH 8.0, 1%(v/v) Nonidet P-40, 0.2 mM phenylmethylsulfonyl fluoride, 5 g/ml aprotinin, 10 g/ml leupeptin) on ice for 1 h. The supernatants obtained after centrifugation at 10,000 ϫ g for 10 min at 4°C were stored at Ϫ80°C or used immediately for immunoprecipitation.
Immunoprecipitation-Immunoprecipitation was performed as described previously (37). Briefly, cell lysates with 5 ϫ 10 7 dpm 35 S total labeling were precleared with 20 l of a 50% (v/v) slurry of protein G-agarose at 4°C for 3 h. The precleared supernatants were then mixed with 20 l of 50% (v/v) slurry of anti-FLAG M2 beads (Sigma Chemical Co., St. Louis, MO) and rotated at 4°C for 2 h. The beads were washed, resuspended in 50 l of 2ϫ SDS-PAGE loading buffer, boiled, and loaded onto a 10%(w/v) SDS-PAGE gel. The gel was fixed and stained with Coomassie Blue R-250, destained, dried, and exposed to x-ray film.
Relative density values were determined using a digital imaging system IS1000 (Alpha Innotech Corp., San Leandro, CA).
Luciferase Activity Assays-All transfections were done with Cos-1 cells at 50 -60% confluence in 12-well plates. For each well, 0. Lysates were made by washing cells with phosphate-buffered saline buffer (pH 7.4) twice and lysed in reporter lysis buffer (Promega, Wisconsin, MI). Aliquots of lysates were assayed using commercial luciferase and galactosidase kits (Promega). Each mER-␤ sample was transfected in triplicate. Electrophoretic Mobility Shift Assay-To determine the DNA binding activity of the ER-␤ mutants, extracts of mER-transfected Cos-1 cells were made according to published methods (39). The assay was carried out in binding buffer containing 20 mM HEPES, pH 7.4, 50 mM KCl, 1 mM dithiothreitol, 10% glycerol, 50 nM estradiol, 0.5 mg/ml bovine serum albumin, 50 ng/l poly[d(I⅐C)/d(I⅐C)], and protease inhibitors. 32 P-Labeled double-stranded oligonucleotide probe (1 ng) containing a consensus estrogen response element (ERE) sequence from chicken vitellogenin (5Ј-CTAGAAAGTCAGGTCACAGTGACCTGATCATT-3Ј) was used in 20-l reaction volumes. Preincubation with all the reaction components except the labeled probe was conducted on ice for 15 min. After addition of the labeled probe, the reaction was allowed to incubate on ice for an additional 15 min. The ER⅐ERE complex was then resolved on native 6% polyacrylamide gels. The gel was fixed, dried, and exposed to x-ray film at Ϫ80°C. The antibody against ER-␤ (CalBiochem, La Jolla, CA) and excess unlabeled ERE probe were added as indicated.
Characterization of Subcellular Localization-To examine subcellular localization of mER-␤, the GFP fusion constructs of mER-␤ cDNA were transfected into mammalian cells lines, as described above. After 1 day, fresh media and 17␤-estradiol at the final concentration of 20 nM were added. Fluorescence images were recorded using a digital camera (Hamamatsu, Tokyo, Japan).

Mutation of the O-GlcNAc/O-Phosphate Locus on mER-␤-
Our previous studies demonstrated that mER-␤ is modified alternatively by O-GlcNAc or O-phosphate at Ser 16 (17). To reveal biological roles of the O-GlcNAcylation/O-phosphorylation on mER-␤, we mutated Ser 16 into either Ala or Glu along with Ser 15 and Thr 17 into Ala and Val to generate two mutants, designated mER-␤ S16A and S16E, respectively, as illustrated in Fig. 1. The S16E mutant is designed to mimic the constitutive phosphorylation of Ser 16 (40,41). Ser 15 and Thr 17 were concomitantly mutated to eliminate the possibility for promiscuous O-glycosylation/O-phosphorylation at this locus, which might occur even when the observed major acceptor site (Ser 16 1. mER-␤ mutants used in this study. Wild-type mER-␤ cDNA was mutated at the Ser 16 by changing either Ser 3 Ala or Ser 3 Glu along with changing Ser 15 (Ser 3 Ala) and Thr 17 (Thr 3 Val). The two mutants were named S16A and S16E, respectively. All constructs were engineered to incorporate a FLAG tag at their carboxyl terminus. assess the glycosylation state of mER-␤ O-GlcNAc site mutants, the protein of the S16A mutant expressed and purified from insect Sf9 cells was probed for O-GlcNAc using galactosyltransferase and UDP-[ 3 H]galactose (38). As shown in Fig.  2A, the protein of the S16A mutant was labeled approximately 5-fold less than the wild type, suggesting that the mutant S16A protein is deficient in O-GlcNAc. To further compare tryptic maps between the wild type and the S16A mutant, labeled proteins were digested with trypsin and the tryptic glycopeptides were resolved on a reverse phase column. As shown in Fig. 2B, tryptic glycopeptide maps from the S16A mutant (top panel) showed only baseline levels of radioactivity, whereas the wild type (bottom panel) showed the same major tryptic glycopeptide (retention time ϭ 42 min) and smaller labeled peaks at retention times of 20, 36, and 54 min, respectively, previously seen for mER-␤ (17). Previous analyses (17) of the smaller labeled peaks in (Fig. 2B, bottom panel) indicate that they likely result from incomplete proteolysis. This conclusion is supported by their disappearance in the mutant (Fig. 2B, top  panel). Thus, we conclude that the mER-␤ mutants at the Ser 16 locus are glycosylation-deficient.
The O-Glycosylation/O-Phosphorylation Site Mutants of mER-␤ Have Altered Turnover Rates-PEST regions in proteins, enriched with Pro, Glu, Ser, and Thr, have been proposed to be responsible for the rapid degradation of certain proteins (42,43). Our earlier studies documented that O-GlcNAcylation sites on mER-␣ (44) and mER-␤ (17) are in regions of the proteins that have high PEST scores. This observation suggests that one likely role of O-GlcNAc on ER is to modulate ER protein stability. Therefore, we directly examined the relative turnover rates in vivo of the wt and mutant mER-␤ proteins using pulse-chase analyses. Transfected Cos-1 cells were pulselabeled with 35 S-labeled amino acids in vivo for 3 h and chased for up to 6 h. Quantitative results averaged from three independent experiments are shown in Fig. 3. Compared with wt, the degradation of the S16A mutant appears to be slower, whereas the degradation of the S16E mutant appears to be much faster. Assuming rough linearity, the wt mER-␤ turned over rapidly in the presence of estrogen, with an average halflife of 7-8 h, similar to the range of ER-␣ reported previously (45). In contrast, the S16A mutant has a prolonged average half-life of about 15-16 h, and the S16E mutant has a shortened average half-life of about 4 -5 h. These findings suggest that O-phosphorylation on the Ser 16 of mER-␤ results in accelerated degradation of mER-␤ as mimicked by the S16E mutant, whereas O-glycosylation, which blocks phosphorylation, would be predicted to result in stabilization of the protein.
The O-Glycosylation/O-Phosphorylation Site Mutants of mER-␤ Have Altered Transactivation Activities-The fact that the major O-GlcNAc site on mER-␤ is located within the transactivation domain of the protein led us to examine the role of O-GlcNAc/O-phosphate at this site in ER-mediated transcriptional activation. The transactivation activities of the mutants were measured by cotransfection of mER-␤ cDNAs with an ERE-linked luciferase reporter gene in Cos-1 cells. As summarized in Fig. 4, the S16E mutant has elevated transactivation activities compared with wt-mER-␤. However, the S16E mutant is not further stimulated by estrogen, suggesting that the S16E is constitutively active under these conditions. In contrast, the S16A has only basal activity with minor stimulation by estrogen. The relatively modest level of stimulation seen in this system may reflect the lack of appropriate coactivators in the Cos-1 cells used. Nonetheless, it appears that the alternative post-translational modification of Ser 16 of mER-␤ is not only crucial to achieve normal levels of ER-mediated transac-

FIG. 2. The Ser 16 to Ala mutant is poorly glycosylated in vivo.
A, the wt and the S16A mutant mER-␤ were expressed and purified from insect Sf9 cells. Purified proteins were enzymatically labeled with UDP-[ 3 H]galactose using galactosyltransferase. [ 3 H]Galactose-labeled proteins were resolved by 10% SDS-PAGE gel electrophoresis. Gels were impregnated with 1 M salicylic acid, dried, and exposed to x-ray film at Ϫ80°C for 2 days. Scintillation counting of the bands indicated that the band from wild-type contained ϳ1.77 ϫ 10 6 dpm and that from the mutant contained ϳ3.61 ϫ 10 5 dpm. B, [ 3 H]galactose-labeled proteins were digested with trypsin. Tryptic peptides were resolved on a C 2 /C 18 reversed phase column. The column was developed with a 90min linear gradient of 0 -60% (v/v) acetonitrile in 0.1% (v/v) trifluoroacetic acid from 6 to 96 min at a flow rate of 0.1 ml/min. Eluates were collected and counted. Upper panel, the S16A mutant tryptic profile; lower panel, the wt tryptic profile. Note: Smaller labeled peaks likely result from incomplete proteolysis of the major glycosylation site. tivation but also is important to estrogen responsiveness of the receptor.

Mutation of the O-Glycosylation/O-Phosphorylation Site on mER-␤ Does Not
Affect DNA Binding-To compare the binding of the wild type and mutant forms of mER-␤ to the DNA response elements, we tested cell extracts from mER-␤-transfected Cos-1 cells using electrophoretic mobility shift assays. As shown in Fig. 5, the two mutants are able to form the same ER⅐ERE complexes as the wild type. Differences in intensity of the complexes result from the relative amount of mER-␤ in each nuclear extract as determined by Western blotting using an antibody to the FLAG epitope on each recombinant protein (data not shown). Note that the extract from the S16E mutant contains less mER-␤ protein, probably due to its rapid rate of degradation, even though the experiments were normalized for transfection efficiency (Fig. 3).
Mutant and wt mER-␤ Proteins Are Localized Exclusively in the Nucleus-Although ER-␣ is known to be mainly localized in the nucleus, the subcellular localization of the homologue ER-␤ has not been studied. To investigate the subcellular distribution of ER-␤, we fused mER-␤ to GFP. The localization of GFP-fused wt mER-␤ was then studied in several different mammalian cell lines. As shown in Fig. 6, virtually all of the receptor is restricted to the nucleus in monkey kidney Cos-1 cells, independent of the presence or the absence of its cognate ligand estrogen. The same distribution of the wt receptor was also observed for other cell lines, including HeLa and MCF-7 cells (data not shown).
Earlier studies on nuclear pore proteins suggested that O-GlcNAcylation is likely involved in nuclear transport (5, 46 -48). Because the functionality of the estrogen receptor relies on its proper subcellular localization, we also examined the subcellular localization of the two glycosylation/phosphorylation site mutants as GFP fusion proteins. Both mutants are also exclusively localized to the nucleus (data not shown), suggesting that modifications at Ser 16 are not involved in mediating mER-␤'s transport into the nucleus. DISCUSSION Since the discovery of the O-GlcNAc modification, several functional roles have been postulated, such as modulation or mediation of protein⅐protein interactions, regulation of nuclear transport, transient regulation of phosphorylation site availability, and modulation of protein turnover (1,2). In this study, we provide evidence that O-GlcNAc on mER-␤ has a reciprocal relationship with phosphorylation by capping a phosphorylation site that modulates mER-␤ degradation and is also important for the receptor's transactivational activity.
Previous studies on eukaryotic initiation factor 2-associated protein p67 and on the transcription factor Sp1 showed that O-GlcNAc removal from both p67 and Sp1 targets them for rapid degradation by the proteasome (25, 49). Studies on the mutated forms of mER-␤ not only provide additional evidence FIG. 3. The modification state of Ser 16 of mER-␤ affects the protein's turnover rate. Cos-1 cells transfected with either the wild type or mutated mER-␤ were metabolically labeled with 35 S-protein Express label for 3 h and then chased with unlabeled rich medium for various time points as indicated in the figure. Cell lysates were prepared as described under "Materials and Methods." Samples, containing ϳ5 ϫ 10 7 dpm per lysate, were immunoprecipitated with anti-FLAG antibody M2, and isolated proteins were resolved by electrophoresis on 10% SDS-PAGE gels. Gels were fixed, dried, and exposed to x-ray film for 30 min. 35  The mER-␤ cDNAs were transfected into Cos-1 cells by the liposome method. Cell extracts were prepared for the electrophoretic mobility shift assay. The vitellogenin estrogen response element (ERE) DNA probe was labeled with 32 P using the Klenow fragment of DNA polymerase. The specificity of the assay was demonstrated by cold probe competition (100-fold excess) and mER-␤ antibody blocking of the interaction. Note: mER-␤ levels are less in the S16E mutant due to its rapid degradation (see text and Fig. 3).
for O-GlcNAc regulating protein degradation but also suggest that the saccharide may act by blocking the addition of phosphate, which itself targets the protein for rapid degradation. To further understand the relative roles of O-GlcNAc versus Ophosphate at Ser 16 , we compared the wt mER-␤ to both the S16A and the S16E mutants, the latter of which mimics constitutive phosphorylation (40,41,50,51). The simplest interpretation of the data is that the S16E mutant behaved as the phosphorylated form of the protein resulting in accelerated degradation, whereas the S16A mutant behaved analogous to the "capped" glycosylated form of the protein that slowed degradation. Because many of the known O-GlcNAc sites are located near proline residues, glycosylation sites adjacent to acidic amino acids will have high intrinsic PEST scores, whereas others may be dependent upon phosphorylation to target PEST-mediated degradation (42,43). It has been suggested that phosphorylation can change Ser or Thr residues into negatively charged residues so as to convert some imperfect PST sequences into PEST degradation signals. In contrast, O-GlcNAcylation could prevent these phosphorylation effects by either competing at the same hydroxyl directly or by changing the protein conformation indirectly to mask the charged regions, as has been suggested for p53 (52).
The N terminus of mER-␤, which harbors the major O-GlcNAc site, mediates the receptor's a transactivation functions, which in turn activates target genes (53,54). Our luciferase reporter data show that the extent of transactivation is dependent upon the modification of the hydroxyl group of Ser 16 . These data suggest that O-GlcNAc/O-phosphate at this site directly plays a role in modulating mER-␤-mediated transactivation. Earlier studies suggested that O-GlcNAcylation modulates transactivation by mediating the appropriate protein⅐protein interactions of many transcription factors such as Sp1 (3,26,27). Recent in vivo studies showed that the concentration of these transcriptional activator proteins is regulated by the proteasome-mediated degradation pathway, and the rate of degradation of activators by the proteasome correlates with activation domain potency in vivo (55). Consistent with these earlier reports on other transcription factors, the alternate O-GlcNAc/O-phosphorylation of mER-␤ appears to be involved in both degradation and transactivation functions of the molecule.
Based on our in vivo [ 32 P]orthophosphate labeling studies (data not shown), Ser 16 is one of several phosphorylation sites on mER-␤. It is likely that Ser 16 is a regulatory site with typically low occupancy and rapid cycling. Earlier studies on transcription factor Sp1 have suggested that, upon glucose starvation, Sp1 undergoes rapid deglycosylation and becomes more susceptible to proteasome degradation (49). Recently, it has been reported that ER-␣ is rapidly degraded by the 26 S proteasome upon estrogen stimulation (56). O-GlcNAc transferase activity is exquisitely sensitive to concentrations of UDP-GlcNAc/UDP, which are in turn highly sensitive to energy metabolism (11,23). Thus, in energy-rich conditions, O-GlcNAc levels would be expected to increase, in turn preventing degradation of certain proteins, such as ER-␤.
Our green fluorescence protein fusion results revealed that there is no significant redistribution of mER-␤ in mammalian cells induced by the mutations, excluding the possibility that nuclear localization requires either phosphorylation or O-Glc-NAcylation at the Ser 16 of mER-␤. However, we did observe some minor nuclear pattern changes, such as a clustering of ER-␤ within the nucleus upon estrogen treatment (Fig. 6). This clustering phenomenon is similar to that seen for ER-␣ using a similar approach (57).
The occurrence of O-GlcNAcylation sites in the key regulatory regions of some oncogenes and tumor suppressors, such as c-Myc, p53, and SV-40 large T-antigen (18 -20, 52), reinforces the potential regulatory significance of this modification. If a reciprocal relationship between O-GlcNAcylation and O-phosphorylation is found to be common, then it will be important to carefully evaluate the respective roles of these distinct modifications. Generally, this will not be possible by direct sitedirected mutagenesis approaches but rather will require novel methods, such as the chemi-enzymatic synthesis of site-specifically modified glyco-and phospho-forms of these regulatory proteins (58 -60). The interplay between O-GlcNAc and Ophosphate on mER-␤ Ser 16 demonstrated in this study provides another excellent example where both of these modifications work in a coordinated manner to regulate the activity of a key regulatory protein.
FIG. 6. mER-␤ localizes exclusively to nucleus. mER-␤ cDNAs were subcloned into pGFP-C1 vector, and fusion constructs were transfected into Cos-1 cells by the liposome method. 20 nM 17␤-estradiol (E 2 ) was added at 24 h post-transfection. Images were observed at 48 h post-transfection and recorded using a Hamamatsu digital camera with settings of 5 s for fluorescent images (right column) and 0.2 s for phase images (left column). wt, cells transfected with wild-type mER-␤ fused to GFP; GFP, cells transfected with GFP construct alone; wt ϩ E 2 , cells transfected with wild-type mER-␤ grown in the presence of E 2 ; GFP ϩ E 2 , cells transfected with GFP construct alone and grown in the presence of E 2 .