(Received for publication, July 23, 1996, and in revised form, September 20, 1996)
From the Department of Biological Chemistry, The
Johns Hopkins University School of Medicine,
Baltimore, Maryland 21205 and the ¶ Department of Biochemistry
and Molecular Genetics, University of Alabama at Birmingham,
Birmingham, Alabama 35294-0005
Estrogen receptors (ER) are ligand-inducible transcription factors regulated by Ser(Thr)-O-phosphorylation. Many transcription factors and eukaryotic RNA polymerase II itself are also dynamically modified by Ser(Thr)-O-linked N-acetylglucosamine moieties (O-GlcNAc). Here we report that subpopulations of murine, bovine, and human estrogen receptors are modified by O-GlcNAc.
O-GlcNAc moieties were detected on insect cell-expressed, mouse ER (mER) by probing with bovine milk galactosyltransferase, followed by structural analysis. Wheat germ agglutinin-Sepharose affinity chromatography also readily detected terminal GlcNAc residues on subpopulations of ER purified from calf uterus, from human breast cancer cells (MCF-7), or from mER produced by in vitro translation. These data suggest that greater than 10% of these populations of estrogen receptors bear O-GlcNAc. Site mapping of insect cell expressed mER localized one major site of O-GlcNAc addition to Thr-575, within a PEST region of the carboxyl-terminal F domain. Based upon their relative resistance to both hexosaminidase and to in vitro galactosylation, O-GlcNAc moieties appear to be largely buried on native mER. This dynamic saccharide modification, like phosphorylation, may play a role in modulating the dimerization, stability, or transactivation functions of estrogen receptors.
The estrogen receptor (ER)1 is a ligand-activated transcription factor that modulates specific gene expression by binding to estrogen response elements (ERE) (1). ER and other members of the nuclear receptor superfamily of ligand-regulated transcription factors have discrete domains responsible for ligand binding, DNA binding, nuclear localization, dimer formation or activation of transcription (1). However, the mechanisms of how gene transcription is modulated by steroid hormone binding to these receptors remains largely unknown.
Almost all members of the steroid hormone receptor superfamily are known to be phosphoproteins (2-8). ER from mouse and human become hyperphosphorylated on serine residues in the presence of estradiol (3-5, 7). Tyrosine phosphorylation of ER has also been reported (9-12). Phosphorylation of ER has been postulated to modulate dimerization, interactions with other proteins, transcriptional activity, DNA-binding activity, nuclear localization, and other functions (2).
Ser(Thr)-O-linked N-acetylglucosamine
(O-GlcNAc; 13, 14) is a form of dynamic (15-18)
intracellular protein glycosylation, which can be reciprocal with
Ser(Thr)-O-phosphorylation (for review, see Ref. 19). For
example, the phosphorylation and O-GlcNAcylation of the
carboxyl-terminal domain of RNA polymerase II (20) and of Thr-58 on the
c-myc oncogene (21, 22) are mutually exclusive events.
O-GlcNAc has the attributes of a regulatory modification analogous to phosphorylation (15, 23). O-GlcNAc is
ubiquitous in all eukaryotes, thus far examined. Unlike more
conventional types of protein glycosylation, O-GlcNAc is
nearly exclusively localized in the nucleoplasm and cytoplasm (24-27).
Other O-GlcNAc modified proteins identified to date include
nuclear pore proteins (28-31), vertebrate lens protein (32), small
heat shock proteins (18), neurofilaments (33), microtubule-associated
proteins (34), numerous other cytoskeletal components (23, 35-37),
viral proteins (38-40), tumor suppressors (41), oncoproteins (21, 22,
42), the -amyloid precursor protein (43), and a number of eukaryotic
RNA polymerase II transcription factors (44-46). Importantly, the
v-erb A oncoprotein whose gene is derived from a host cell
gene for thyroid hormone receptors is also modified by
O-GlcNAc (42). Here we show that the estrogen receptors from murine, bovine, and human sources are all modified by
O-GlcNAc and that a major O-GlcNAc site (Thr-575)
occurs within a PEST sequence of the carboxyl-terminal F domain.
UDP-[6-3H]galactose (38 Ci/mmol) was obtained from Amersham Corp. Bovine milk galactosyltransferase and aprotinin were purchased from Sigma, and galactosyltransferase (37 units/ml) was pre-galactosylated as described (24). All other chemicals were of the highest quality commercially available. SP65 (MOR1-599) (47) and baculovirus MOR1-599 (48) were supplied by Dr. M. G. Parker (Imperial Cancer Research Fund, London). PGEM-7Zf(+) that contained eight tandem copies of the consensus ERE was provided by R. A. Bambara (University of Rochester) (49). Anti-human estrogen receptor antibody (H222) was the gift of Abbott Laboratories (50).
Cell Culture and Overexpression of the Mouse Estrogen Receptor in Spodoptera frugiperda CellsSf-9 insect cells were grown
in TNH-FH medium, which consists of Grace's insect medium (Flow
Laboratories) supplemented with 10% heat-inactivated fetal bovine
serum, plated out at 70% confluence, and infected with MOR1-599
recombinant or a wild type virus at a multiplicity of infection of
5-10 plaque-forming units per cell and then incubated for 24-120 h.
Cells were harvested, collected by centrifugation, and frozen at
70 °C.
Insect cells (Sf-9) were infected
with MOR1-599 recombinant. At 3 days postinfection, insect cells were
harvested. Whole cell extracts were prepared by resuspending the frozen
pellets in the high-salt buffer (0.4 M KCl, 40 mM Tris-HCl, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride (PMSF), 50 mM
GlcNAc, pH 7.5), passing the suspension 10 times through a 27-gauge
needle, and subsequently removing debris by centrifugation at
50,000 × g for 20 min. The high speed postmicrosomal
supernatant was prepared by centrifugation at 200,000 × g for 45 min; final KCl concentration was adjusted to 175 mM KCl. The supernatant was mixed with 10 nM
17-[3H]estradiol (102 Ci/mmol; Amersham Corp.) and
heparin-agarose (Bio-Rad) previously equilibrated in TDPK-175 buffer
(40 mM Tris-HCl, 1 mM dithiothreitol, 0.5 mM PMSF, 175 mM KCl, pH 7.5). The suspension was rotated overnight at 4 °C. Heparin affinity and ERE affinity chromatography were performed as described (49).
Insect cells (Sf-9) were infected with recombinant
baculovirus containing a full-length mouse estrogen receptor (mER)
cDNA (MOR1-599), wild type virus, or mock-infected. The cells were harvested as indicated postinfection days and resuspended in an extract
buffer (20 mM Tris-HCl, pH 7.4/2 mM
dithiothreitol, 20% (v/v) glycerol, 0.2 mM PMSF,
containing protease inhibitors Pic1 and Pic2 (51)) by passing the
suspension 10 times through a 27-gauge needle. Supernatant was
separated from the low speed pellet by centrifugation at 50,000 × g for 20 min. The cytosolic fraction was prepared by
centrifuging the supernatant at 200,000 × g for 45 min. The low speed pellet was further resuspended in the extract buffer
containing 0.4 M KCl and 10 nM 17-estradiol, passing the suspension 10 times through a 27-gauge needle, and removing
debris by centrifuging at 50,000 × g for 20 min. The nuclear fraction was prepared by centrifuging the supernatant at
200,000 × g for 45 min.
Protein concentrations were estimated by the method of Bradford (52). For samples of ER eluted from the ERE-DNA-Sepharose column, protein concentrations were determined by the method of Schaffner and Weissmann (53).
Galactosyltransferase Labeling of Intact mERNuclear fractions or partially purified estrogen receptor was labeled with autogalactosylated bovine milk galactosyltransferase according to the method of Holt and Hart (24), except that the reaction mixture was incubated overnight at 4 °C.
Galactosyltransferase Labeling of mER Tryptic PeptidesMouse ER were purified by SDS-polyacrylamide gel electrophoresis and extensively digested in the gel with L-1-tosylamido-2-phenylethyl chloromethyl ketone-trypsin (ER:trypsin = 1:1) at 37 °C overnight. The mER tryptic peptides were purified by C-18 Sep-Pak column in 0.1% trifluoroacetic acid and elution with 60% acetonitrile. The acetonitrile was removed by using speed vacuum, and the mER tryptic peptides were resuspended in buffer H (0.5 M NaCl, 4% Triton X-100, 50 mM Hepes, pH 8.0). The [3H]galactose labeling was performed as described previously (24). After the [3H]galactose labeling, the samples were applied to a C-18 Sep-Pak column and eluted with 60% acetonitrile.
Coupled in Vitro Transcription/TranslationA standard 50-ml reaction, including 1 µg of plasmid SP65 (MOR1-599) and components, was incubated as described (Promega) for 2 h at 30 °C. A negative control consisted of the same components without addition of DNA (54).
Wheat Germ Agglutinin Affinity ChromatographyWheat germ agglutinin-Sepharose (E-Y Laboratories) was extensively washed with the loading buffer (10 mM phosphate, 150 mM NaCl, 0.1% Nonidet P-40, pH 7.5). Aliquots of the translation mixture were diluted with 20 volumes of the loading buffer, applied to the column, and incubated for 30 min at room temperature. After extensively washing with the loading buffer, the column was again washed with 5 volumes of the loading buffer containing 1 M galactose followed by elution of bound material with the loading buffer containing 1 M GlcNAc. Aliquots were removed for liquid scintillation counting and for 7.5% SDS-PAGE analysis.
Preparation of Calf and Human Estrogen ReceptorsCytosol (55) was prepared from calf uterine (49) or human breast cancer cells (MCF-7), and the estrogen receptors were labeled with [3H]tamoxifen aziridine (24 Ci/mmol, Amersham Corp.) as described (55). After removal of excess ligand by incubation with charcoal-dextran, an aliquot of labeled cytosol was subjected to 10 ml of Sephadex G-25 mini-column (PD-10, Pharmacia) which was previously washed with buffer (10 mM phosphate, 150 mM NaCl, pH 7.5). Labeled cytosol in phosphate-buffered saline was used for WGA chromatography.
Hexosaminidase to Remove Terminal GlcNAcTwenty microliters
of 2% SDS were added to 20 µl of the translation mixture, and then
the sample was boiled for 5 min. Subsequently, 40 µl of 2 × of
the reaction buffer (2 × solution: 80 mM Tris-HCl, 8% Triton X-100, pH 7.5) was added and mixed well. Two microliters of
0.5 units/µl jack bean
-D-N-acetylhexosaminidase (V-Labs, Inc.) was
then added and incubated for 4 h at room temperature. Boiled
hexosaminidase was added to the control mixture.
Proteins were
resolved by electrophoresis on 7.5% SDS-PAGE according to the method
of Laemmli (56). After SDS-polyacrylamide gel electrophoresis, the gels
were fixed in 10% acetic acid, 40% methanol, stained with Coomassie
Brilliant Blue, treated for 1 h with EN3HANCE (DuPont
NEN), dried under vacuum, and exposed to preflashed X-Omat diagnostic
film (Eastman Kodak Co.) at 80 °C.
Gels were electrophoresed for 14-16 h at 40 V and then silver-stained by the method of Blum et al. (57).
Western Blot AnalysisSamples resolved by SDS-polyacrylamide gel were electroblotted to prewetted polyvinylidine difluoride membranes (Millipore Corp., Bedford, MA). Blots were blocked with 3% non-fat dried milk (Carnation) in TBS-T (20 mM Tris-HCl, 150 mM NaCl, 0.2% Tween 20, pH 7.5) for 1 h at room temperature, washed with TBS-T, and probed with anti-ER antibody (monoclonal antibody H222, Abbott Laboratories) at 2 µg/ml as a primary antibody and with horseradish peroxidase-conjugated goat anti-rat IgG (Pierce) diluted 1:5000 as the secondary antibody. Each incubation was carried out at room temperature for 2 h using antibodies diluted with 1% non-fat dried milk in TBS-T. Probed blots were then washed four times for 5 min each with TBS-T. Secondary antibody was detected by chemiluminescence using reagents from the Amersham Corp.
Preparation of the DNA Affinity ResinA plasmid (a gift from Dr. Robert A. Bambara) containing eight tandem copies of the consensus ERE was grown in Escherichia coli. Plasmid DNA was isolated and digested with EcoRI and HindIII according the method of Peale et al. (49). The 337-base pair ERE fragment was purified and conjugated to Sepharose-4B as described (49, 58).
Carbohydrate CharacterizationsGel-purified
[3H]galactose-labeled mER was treated with mild base (0.1 N NaOH, 1 N NaBH4) for 24 h
37 °C (24). The reaction was stopped by adding acetic acid, and the
samples were chromatographed over a Sephadex G-50 desalting column.
PNGase F-treatment, to remove the N-linked saccharides, was
performed as described previously (24). [3H]Gal-labeled
ovalbumin, which contains only N-linked
oligosaccharides, was treated with PNGase F as a positive
control. The alkaline cleavage products were identified as
[3H]Gal1-4GlcNAcitol by a high resolution TSK gel
permeation chromatography (48) and high pH anion exchange
chromatography with plus amperometric detection (HPAEC-PAD) (Dionex)
(59).
Galactose-labeled mER was purified by 7.5% SDS-polyacrylamide gel and then digested in the gel with trypsin (Worthington L-1-tosylamido-2-phenylethyl chloromethyl ketone-treated trypsin) (ER:trypsin = 5:1) (60). The labeled glycopeptides, eluted with 60% acetonitrile in 200 mM ammonium bicarbonate, pH 7.4, were dried and then separated by RP-HPLC using a Dynamax C18 column (0.46 × 25 cm, Rainin). The column was developed at a flow rate of 1 ml/min with 120-min linear gradient of 0-60% acetonitrile in 0.15 M phosphoric acid, 100 mM NaClO4, pH 2.1 (61). The absorbance of eluant was monitored at 214 nm, and fractions were collected every minute. Aliquots of each fraction were counted. Each tritium-labeled peak was pooled and then further separated by a second and third dimension of RP-HPLC, using a 2-h, 0-40% CH3CN gradient in 0.1% trifluoroacetic acid.
Gas-phase Sequencing and Manual Edman Degradation of [3H]Galactose-labeled GlycopeptidesPurified glycopeptides from the third dimension of RP-HPLC were sequenced by automated Edman degradation in a model 470A gas-phase sequencer (Applied Biosystems, Inc.). Manual Edman degradation sequencing of glycopeptides was performed by covalent coupling of peptides to Sequelon-AA membranes (MilliGen/Biosearch of Millipore), followed by repeated phenylisothiocyanate conjugation and trifluoroacetic acid extraction as described (62), except that the product from each cycle was dried and neutralized before scintillation counting.
Overexpression of the full-length mouse
estrogen receptor (mER) was examined in insect Sf9 cells from 24 to
120 h following infection with the recombinant virus, MOR1-599
(Fig. 1). The amount of recombinant mER protein
continued to increase up to the 4th day, but the protein was degraded
very rapidly after the 5th day. The majority of recombinant mER is in
the nuclear fraction (Fig. 1, A and B).
The high-salt postmicrosomal supernatant (including cytosolic and
nuclear mER; Ref. 48) was further purified by heparin-agarose and
ERE-DNA affinity chromatography. SDS-PAGE analysis of the purified mER
reveals a major band migrating with Mr = 66,000 (Fig. 2A), which eluted from the DNA affinity
column in the same fractions as the [3H]estradiol-binding
activity. Western blot analysis demonstrates that
Mr = 66,000 and a proteolytic fragment
Mr = 50,000 were recognized by the monoclonal
antibody H222 (Fig. 2B), generated against the human ER.
Galactosyltransferase Labels Partially Purified Recombinant Mouse ER
In vitro labeling of proteins with
[3H]galactose using bovine milk galactosyltransferase is
a specific and sensitive probe for terminal GlcNAc (54, 63-65). In
Fig. 2, both silver-stained gel and immunoblot show that the ER protein
band in the nuclear extract is readily distinguished from other
proteins of wild type or noninfected cells. In
[3H]galactose-labeled nuclear fractions of insect cells
previously infected with recombinant mER baculovirus, wild type
baculovirus, or no virus, there is a distinct radiolabeled protein
Mr = 66,000 present only in nuclear extracts of
insect cells infected with baculovirus containing mER sequence (data
not shown). Immunoblots from several different experiments and sequence
analysis confirmed that the Mr = 66,000 protein
band is mER. Fig. 3 shows the results of
galactosyltransferase labeling of 20 µg of partially purified mER,
eluted from the heparin column (Fig. 3B, lane 1). Although the [3H]galactose-labeling method has high sensitivity
for detection of terminal GlcNAc residues, we were unable to detect
O-GlcNAc on any estrogen receptor purified from the ERE-DNA
affinity column.
In Vitro Translated Mouse ER Contains O-GlcNAc
Recombinant
MOR1-599 cDNA in the plasmid SP65 was transcribed, translated in a
rabbit reticulocyte lysate, and the product labeled with
[35S]methionine. The major radiolabeled band was
Mr = 66,000 (Fig. 4B).
No significant radiolabeled bands were detected on SDS-PAGE of
cell-free translations performed without the addition of plasmid (see
Fig. 4B, lanes J-N).
In vitro translated mER was subjected to WGA chromatography. After the column was extensively washed with the loading buffer and 1 M galactose (nonspecific sugar for WGA-Sepharose; Fig. 4B, lane C), bound material was eluted with GlcNAc (Fig. 4B, lane D). Mouse ER specifically bound to the GlcNAc-specific lectin (Fig. 4, A and B, lanes A-E) but did not bind to Sepharose (Fig. 4A and B, lanes F-I). Some species of mER, bound to WGA-Sepharose, were not eluted by 1 M GlcNAc but were extracted when Laemmli sample buffer was used (Fig. 4B, lane E).
Treatment of the mER with hexosaminidase reduces the binding to the WGA affinity column (Fig. 4C), confirming the carbohydrate specificity of the interaction. WGA-binding activity was reduced up to 70% if the product was first denatured by boiling in 1% SDS prior to treatment with hexosaminidase. In contrast, hexosaminidase only removed 30% of GlcNAc-specific binding activity of mER without prior denaturation.
The Recombinant, Baculovirus-expressed Mouse ER Bears O-Linked GlcNAc Monosaccharide MoietiesO-Linkage of GlcNAc to proteins through
serine or threonine hydroxyls is sensitive to alkali-induced
-elimination (13) but is resistant to cleavage by the enzyme
peptide-N-glycosidase F (PNGase F; 66, 67). More than 95%
of the radioactivity on the gel-purified
[3H]galactose-labeled ER was released by alkali-induced
-elimination (Fig. 5A). In contrast,
identical preparations of ER were resistant to PNGase F treatment and
were eluted in the Vo of the desalting column (Fig.
5B). In contrast, [3H]galactose-labeled
ovalbumin, which contains only N-linked oligosaccharides, is
quantitatively sensitive to PNGase F treatment (data not shown). The
results indicate that GlcNAc is covalently attached to the ER through
an O-glycosidic linkage.
To confirm the saccharide structure, we first chromatographed the mER
-elimination products (Vi fractions in Fig. 5A) on a TSK-gel filtration column. The released sugars
co-migrated with authentic [3H]Gal-GlcNAcitol
disaccharides (data not shown). As illustrated in Fig. 5C,
these disaccharide
-elimination products also exactly co-migrate
with Gal
1-4GlcNAcitol on a Dionex CarboPAc-MA1 column by HPAEC-PAD,
thus demonstrating that mER contains O-linked single GlcNAc
moieties.
The [3H]galactose-labeled ER was purified by
SDS-polyacrylamide gel electrophoresis, and in-gel digestion was used
to obtain tryptic glycopeptides. RP-HPLC C-18 chromatography resolved
at least five radioactive peaks (Fig. 6A).
The shoulder between peaks 2 and 3 was variable in different
preparations. Peaks 1-4 were too small to analyze further. Peak 5 was
further purified by a second round of RP-HPLC (Fig. 6B).
Fraction 69 was further purified by a third RP-HPLC run (Fig.
6C). The major radiolabeled materials in fractions 92 and 93 (Fig. 6C) were each subjected to gas-phase sequencing for 10 cycles. Both fractions yielded the beginning sequence MGVPPEEPSQ (Fig.
6C), corresponding to the first 10 amino residues of tryptic
peptide
560MGVPPEEPSQTQLATTSSTSAHSLQTYYIPPEAEGFPNTI599
of the carboxyl-terminal domain in mER. In order to identify which
amino acid residues on the glycopeptides were modified by O-GlcNAc, manual Edman degradation was performed. Both
fractions 92 and 93 (Fig. 6C, middle panel)
released the major radioactivity at cycle 16, which indicates that
Thr-575 is O-glycosylated. Released radioactivity after
cycle 16 could be due to incomplete cleavage or lower levels of
glycosylation of the adjacent serine residues. Tryptic peptide 560-599
contains potential prolidase cleavage sites at P-563, P-564, P-567,
P-589, P-590, and P-596. Fractions 92 and 93 were digested with
proline-specific endopeptidase (Fig. 6D, closed circle). A
smaller aliquot of sample was treated with boiled prolidase as a mock
treatment (Fig. 6D, open circle). Manual Edman degradation
of fraction 18 (Fig. 6D, upper panel) released radioactivity at cycle 11 (Fig. 6C, bottom
panel), consistent with preferential prolidase cleavage at P-564
and O-GlcNAcylation at Thr-575. The last bar in each manual
Edman degradation (Fig. 6, C and D, bottom
panels) represent counts remaining bound to the filter. Prolidases
are known to have selective peptide specificity (68). Other prolidase
fragments were not present at levels sufficient for further
analysis.
Trypsin Digestion of ER Exposes 2-3-Fold More GlcNAc Residues
Resistance to hexosaminidase digestion of in vitro labeled mER suggests that like certain other O-GlcNAcylated proteins (33, 69), the majority of the O-GlcNAc residues are inaccessible on native mER. We investigated this further by comparing the galactosylation of native mER to the galactosylation of an equal amount of mER tryptic fragments. Tryptic fragments of mER displayed from 2 to 3-fold more galactosylatable GlcNAc residues per mol of protein than native mER. However, the same tryptic glycopeptides were detected regardless of whether the trypsin treatment was performed before or after probing with galactosyltransferase (data not shown).
A Subpopulation of Both Calf Uterine and Human Breast Cancer Cell ER Bind to WGA-Sepharose[3H]Tamoxifen aziridine
was used to covalently label calf uterine ER at its active site
(Cys-530; Refs. 55, 70, 71). The tritiated labeled calf uterine cytosol
was subjected to WGA chromatography (Fig.
7A). Approximately 5-10%
[3H]tamoxifen aziridine-labeled ER was found in
GlcNAc-eluting fractions (compare lanes 1 and 3,
Fig. 7B). However, analysis by immunoblotting indicated that
the percentage of ER that bound to WGA-Sepharose might be substantially
higher (compare lanes 1 and 3, Fig.
7C). Virtually none of the radiolabeled material was
nonspecifically bound to the Sepharose column (data not shown).
Specificity for WGA binding was confirmed since the ER band was not
found in 1 M galactose fractions (Fig. 7B, lane
2). Low molecular weight labeled fragments are proteolytic bands
found very commonly in cell extracts even though the protease inhibitor
(PMSF) was included during homogenization (55).
We used the same method to investigate whether O-GlcNAc residues modify human ER. The postmicrosomal supernatant from MCF-7 human breast cells (55) was labeled with tamoxifen aziridine and then subjected to WGA-Sepharose chromatography. About 10% [3H]tamoxifen-labeled hER also binds WGA-Sepharose and is eluted specifically with GlcNAc (data not shown). These data indicate that a significant subpopulation of both calf ER and hER are also modified by O-GlcNAc.
These studies demonstrate that a significant subpopulation (10% or more) of estrogen receptors from mouse, bovine, or human sources are modified by Ser(Thr)-O-GlcNAc, a highly dynamic form of intracellular glycosylation that is often reciprocal with Ser(Thr)-O-phosphorylation. Given the importance of Ser(Thr)-phosphorylation in the functions of steroid receptors (2), the modification of ER by O-GlcNAc is also likely to have functional significance.
In order to purify a sufficient amount of ER protein for site mapping of O-GlcNAc glycosylation, we overexpressed the recombinant mER in the baculovirus system. Sites used by O-GlcNAc glycosylation in the baculovirus system appear to be the same as in mammalian cells (38). Recently, Greis and Hart (38) showed that the human cytomegalovirus tegument basic phosphoprotein is glycosylated in insect cells and that the glycopeptides produced by chemical or enzymatic digestion are the same as those from native basic phosphoprotein isolated from human cytomegalovirus virions. In addition, it has been shown that cytokeratins 8 and 18 expressed as recombinant baculovirus in insect cells are modified by O-GlcNAc and the major sites of glycosylation are the same as those found in human HT29 cells (17).
In initial studies, we could not detect terminal GlcNAc residues on calf ER or recombinant mER that had been purified from an ERE-DNA affinity column. We may have selectively purified the nonglycosylated forms that bind to the ERE element. The relatively low stoichiometry of O-GlcNAc on ER purified by other means may reflect the following: (i) hexosaminidase removal of O-GlcNAc during homogenization and purification (72); (ii) multiple forms of ER, with only a subset of ligand binding forms of ER actually containing O-GlcNAc; and (iii) differential accessibility of GlcNAc sites assayed by WGA binding and hexosaminidase digestion on ligand binding forms and other forms of ER. It has been postulated that O-GlcNAc is a regulatory modification analogous to phosphorylation (14, 19, 24). Since it has been shown that ERs become hyperphosphorylated in the presence of estrogen (3-7), it will be very important to further elucidate the glycosylation state of the ligand-free binding form and the effect of O-GlcNAcylation on function. Site mapping of O-GlcNAc on ER will also allow for functional studies using site-directed mutagenesis approaches.
In vitro translated recombinant mER has high affinity to estradiol and ERE-DNA (47). WGA chromatography has demonstrated that in vitro translated recombinant mER is also modified by O-GlcNAc. Some species of mER bind very tightly to WGA-Sepharose, suggesting that these species of mER contain multiple clustered O-GlcNAc residues (54). The relative sensitivities of native ER and tryptic fragments of ER to in vitro galactosylation or hexosaminidase digestion also indicate that many O-GlcNAc residues are "buried" in the native ER. O-GlcNAc is also buried on polymerized neurofilaments (33) and on the trimerized adenovirus fiber (69).
The functions of the domains A/B, C, E of the ER have been defined for
transactivation, dimerization, ligand binding, DNA binding, nuclear
localization, and interaction with other proteins, respectively. The
functional role of the F domain in the carboxyl terminus of ER remains
unknown. However, sequence analysis of the human ER reveals a PEST
region in the F domain, enriched in proline (P), glutamic acid (E),
serine (S), and threonine (T) (73, 74). PEST sequences have been
proposed as a signal for rapid intracellular breakdown of protein (74,
75). The data described here indicate that Thr-575, located within the
PEST region, is an O-GlcNAcylation site on mER. Ser-576 is
also likely O-GlcNAcylated (Fig. 8 and data
not shown), indicating that the PEST region could be multiply modified.
O-GlcNAcylation within the PEST region of F domain may play
an important role in regulating the breakdown of ER proteins.
Katzenellenbogen and colleagues (73) have deleted the last 42 amino
acids (F domain) of the hER and found that the deletion of F domain did
not affect transactivation ability, ligand binding affinity, or the
phosphorylation pattern of the receptor and suggested that F domain was
not involved in the relatively rapid breakdown of ER protein. However,
the glycosylation state of domain F in human ER has not been
characterized.
Although the functional role of O-GlcNAc has not yet been elucidated, evidence supports the hypothesis that O-GlcNAc is a post-translational regulatory modification and could regulate the protein phosphorylation by blocking the identical (20, 22) or nearby Ser/Thr sites (14, 19). ER becomes hyperphosphorylated on Ser residues in a hormone-dependent manner (3, 4, 6, 7). If O-GlcNAc and phosphate are reciprocal or their regulation is related, then the ligand-free binding forms of ER may be glycosylated and the ligand binding forms may only contain basal O-GlcNAc moieties. Our initial data suggest that the nonglycosylated form of estrogen receptor is selectively bound to an ERE-DNA affinity column, suggesting that the nonglycosylated ER may be the active ERE-binding form. While as yet inconclusive, these results are consistent with Notides and colleagues (7) showing that treatment of the ER with potato acid phosphatase resulted in the dephosphorylation of the ER and a decrease of the receptor's affinity for specific DNA sequences. Interestingly, Shaw and colleagues (41) have recently shown that O-GlcNAcylation facilitates the DNA binding of the p53 tumor suppressor. Thus, a postulate that we will explore is that dynamic O-glycosylation of the ER might play a crucial role in transcriptional regulation by modulating the activity of the ER protein. Given the substantial technical difficulties in removing O-GlcNAc via hexosaminidase digestions of native ER, site mapping and mutagenesis studies are underway to elucidate the functions of O-GlcNAc on ER.
We thank M. G. Parker for kindly sending us MOR1-599 and MOR121-339 (both in SP65 and baculovirus) and SP16. We also thank Dr. R. A. Bambara for generosity in sending us pGEM-7Zf(+)-ERE. We are grateful to Abbott Laboratories for providing the monoclonal H222 antibody. We thank Drs. Peter Maloney (Dept. of Physiology, The Johns Hopkins University), Dan Lane (Dept. of Biological Chemistry, The Johns Hopkins University), and Rodney Willoughby (Dept. of Children Infectious Diseases, The Johns Hopkins Hospital) for logistical support. We thank Selma Ebersman for proofreading and thank Rodney Willoughby and Ken Greis for critical reading of the manuscript.