(Received for publication, March 25, 1997, and in revised form, May 2, 1997)
From the § Nina Ireland Laboratory, Departments of
Psychiatry and Cellular and Molecular Pharmacology, University of
California, San Francisco, California 94143-0984 and the
Howard Hughes Medical Institute and
Division of Cardiovascular Medicine, Stanford University
Medical School, Stanford, California 94305
The -subunit of eukaryotic initiation factor
2B (eIF-2B), a guanine nucleotide exchange protein that functions in
regulation of translation, was observed to associate with the
carboxyl-terminal cytoplasmic domains of the
2A-
and
2B-adrenergic receptors in a yeast two-hybrid screen
of a cDNA library prepared from 293 cells. This protein association
was confirmed in vitro by affinity chromatography and was
shown to be specific for a subset of G protein-coupled receptors,
including the
2A-,
2B-,
2C-, and
2-adrenergic receptors, but not
the vasopressin (V2) receptor. Association of these
proteins in vivo was confirmed by specific co-immunoprecipitation of eIF-2B
with full-length
2-adrenergic receptors expressed in transfected 293 cells and by fluorescence microscopy showing co-localization of these
proteins in intact cells. Remarkably, eIF-2B
co-localized with
receptors exclusively in regions of the plasma membrane that are in
contact with the extracellular medium, but failed to associate with
membranes making cell-cell contacts. Overexpression of eIF-2B
in 293 cells caused a small (~15%) but significant enhancement of
2-adrenergic receptor-mediated activation of adenylyl
cyclase, without affecting forskolin or V2
receptor-mediated activation. These observations suggest a new role for
a previously identified guanine nucleotide exchange protein in membrane
biology and cell signaling.
G protein-coupled receptors interact with several classes of cytoplasmic proteins, including heterotrimeric G proteins, kinases, phosphatases, and arrestins (1-4). Specific roles of these protein associations in receptor signaling and regulation are now well established. These protein interactions were first inferred from their functional effects on receptor signaling and desensitization before direct physical associations of these proteins with receptors were observed biochemically (5-8). This observation raises the possibility that receptors may interact with additional cellular proteins that could play unanticipated roles in determining the efficacy or specificity of receptor-G protein coupling.
We have investigated this possibility by searching for novel protein
interactions with adrenergic receptors, focusing on the carboxyl-terminal cytoplasmic domain because mutations within this
domain have pleiotropic effects on receptor physiology
(9-12).1 Using interaction cloning and
biochemical techniques, we have observed that several subtypes of
adrenergic receptors associate specifically with
eIF-2B,2 the smallest subunit of a
cytoplasmic guanine nucleotide exchange factor. While this protein has
a well defined role in regulation of translation, it has never been
shown previously to interact with any membrane receptor.
Immunocytochemical studies suggest that receptors associate with
eIF-2B
only in restricted regions of the plasma membrane, and
functional studies suggest that this protein interaction may play a
role in the regulation of receptor-mediated signaling.
The MATCHMAKER II two-hybrid
system (Clontech) was used to screen a 293-cell
cDNA library (complexity ~2.5 × 106 total
recombinants) constructed in pACT2 (Clontech) with
the COOH-terminal cytoplasmic domains of both the murine
2A- and
2B-adrenergic receptors (13, 14).
Bait plasmids were constructed in pAS2-1
(Clontech) using the carboxyl-terminal cytoplasmic
domains from the
2A and
2B receptors,
starting from the "NPXXY-motif" in the seventh
transmembrane domain (e.g. amino acids 422-450 of the
2B receptor), which were amplified using the polymerase chain reaction. Screening of ~1 × 106 transformants
yielded two independent clones (B8 and B10), which interacted with both
the
2A and
2B receptor tails, but not
with the lamin C control (Clontech protocol). The
coding sequences of both clones were identical to human eIF-2B
(EBI
data base, accession number X95648).
The full-length coding
sequence of eIF-2B was subcloned into the mammalian expression
vector pcDNA3 (Invitrogen, San Diego, CA). This construct, referred
to as the M1 form, begins with the first methionine in the predicted
coding sequence. Amino-terminal truncations were constructed at each of
two downstream ATG codons to generate versions beginning at methionine
14 or 20 relative to the first predicted methionine and are referred to
as M14 and M20, respectively. Mutant constructs included an engineered
HindIII site followed by a Kozak consensus sequence to
promote initiation of translation at the selected residue (15, 16).
Constructs were generated using PCR amplification and subcloned into
pcDNA3, and sequences were verified using dideoxynucleotide
sequencing (Sequenase, U. S. Biochemical Corp.). In vitro
translation of these constructs was performed in the presence of
35S-labeled methionine (Amersham Corp.) using the T7 RNA
polymerase promoter and a coupled in vitro
transcription/translation system (Promega, Madison, WI).
Cytoplasmic
receptor tails were amplified by PCR and cloned into a pGEX vector
derived from pGEX-KG and expressed in Escherichia coli (17).
GST fusion proteins were prepared as described (18). The GST-fusion
protein load of individual resins was normalized by densitometric
scanning of SDS-polyacrylamide gels stained with Coomassie Blue. For
affinity chromatography of in vitro translated eIF-2B on
the different receptor tails, 30 µl of the GST-fusion protein loaded
resins (50% (v/v) suspensions) were preblocked in binding buffer (20 mM Hepes, pH 7.4, 100 mM KCl, 5 mM
MgCl2, 0.1% Triton X-100) with 10 mg/ml ovalbumin for 15 min at room temperature. In vitro translated,
[35S]methionine-labeled proteins were incubated with the
GST-fusion protein resins in binding buffer for 1 h at room
temperature. Resins were washed four times with binding buffer and
eluted with SDS-PAGE sample buffer for analysis by SDS-PAGE and
fluorography.
Carboxyl-terminally HA
epitope-tagged versions of the full-length (M1) and amino-terminally
truncated (M20) forms of eIF-2B were constructed by PCR and
subcloned into pcDNA3. These constructs were expressed in human
embryonal kidney 293 cells (ATCC), or in 293 cells stably transfected
with FLAG-tagged human
2-adrenergic receptors (19), by
transient transfection using calcium phosphate precipitation. Cells
were harvested 48-72 h after transfection and lysed in 20 mM Hepes, pH 7.4, 100 mM KCl, 5 mM
MgCl2, 1 mM CaCl2, 0.1% Triton
X-100. Receptors were immunoprecipitated with M1 monoclonal antibody,
recognizing the FLAG epitope (Eastman Kodak Co.) and protein
A-Sepharose (Pharmacia Biotech Inc.). Samples were subjected to
SDS-PAGE, transferred to polyvinylidene difluoride membranes, and
probed with anti-HA monoclonal antibody (HA.11, Berkeley Antibody Co.,
Richmond, CA). Epitope-tagged eIF-2B
was detected using horseradish
peroxidase-conjugated goat anti-mouse secondary antibody (Jackson
Immunoresearch, West Grove, PA) and enzyme-linked chemiluminescence
(ECL, Amersham).
HA epitope-tagged versions of
eIF-2B were transiently transfected into a 293-cell line stably
expressing the NH2-terminally FLAG epitope-tagged
2-adrenergic receptor. Cells were grown on glass
coverslips, fixed, and permeabilized as described (20). HA
epitope-tagged eIF-2B
was detected using the monoclonal mouse antibody HA.11 (Babco, Richmond, CA), and
2-adrenergic
receptor was detected using receptor-specific rabbit antiserum
(21).
293
cells were transiently transfected with carboxyl-terminally HA
epitope-tagged eIF-2B and/or FLAG-epitope tagged
2-adrenergic receptor or human vasopressin
V2 receptor constructs (22) using the LipofectAMINE
protocol (Life Technologies, Inc.). Transfected cells plated in 12-well
plates were labeled for 18-24 h with 4 µCi/ml
[2,8-3H]adenine (NEN Life Science Products) and incubated
in Hepes-buffered medium containing 1 mM
3-isobutyl-1-methylxanthine with or without the indicated drugs for 30 min at 37 °C. Reactions were terminated by adding 0.5 ml of 5%
trichloroacetic acid containing 1 mM ATP and 1 mM cAMP to each well. Intracellular [3H]ATP
and [3H]cAMP were separated subsequently on ion-exchange
and alumina columns as described (23). cAMP accumulation was expressed
as the ratio cAMP/(cAMP + ATP). For receptor quantification, cells were
lysed in 5 mM Tris/HCl, 2 mM EDTA, pH 7.4, using a glass homogenizer. Crude membranes were pelleted by
centrifugation, washed, and resuspended in the same buffer. For
assessment of receptor expression levels, 10-µg membrane protein
aliquots were incubated with 125I-cyanopindolol (250 pM) in 75 mM Tris/HCl, 12.5 mM
MgCl2, 1 mM EDTA, pH 7.4, at room temperature
for 3 h. Binding reactions were terminated by rapid filtration
over Whatman glass fiber filters. Specific binding was defined as the
amount of 125I-cyanopindolol binding inhibited by 1 µM (
)-alprenolol. Nonspecific binding represented <1%
of the total binding measured.
The yeast two-hybrid system (24, 25) was used to identify
candidate proteins that interact with the carboxyl-terminal cytoplasmic
tails of the 2A- and
2B-adrenergic
receptors. Screening of approximately 1 × 106
transformants resulted in the isolation of two independent clones that
interacted specifically with both the
2A- and the
2B-tails. Both of these clones encoded the same
polypeptide, a full-length form of the
-subunit of eIF-2B. eIF-2B
is a subunit of a heteropentameric guanine nucleotide exchange factor,
which has a well established function in regulating the initiation of
protein translation by mediating GTP exchange on eIF-2 (26-29).
Two products were observed by in vitro translation of this
coding sequence, with apparent molecular masses of 34 and 31 kDa (Fig.
1A, first lane). Sequence analysis revealed
two favorable Kozak translation initiation sequences (15, 16) in the
cloned cDNA (corresponding to methionines in positions 1 (M1) and 20 (M20) of the predicted amino acid
sequence) as noted previously in studies of rat eIF-2B (30). The
mobility of the two translation products is in agreement with the
molecular masses predicted for the M1 (33.7 kDa) and M20 (31.4 kDa)
forms by sequence analysis. This was further confirmed by comparing the
SDS-PAGE mobility of the translation products with those produced from
amino-terminally truncated sequences (Fig. 1A).
To examine the biochemical specificity of the interaction between
eIF-2B and receptor domains, eIF-2B
prepared by in
vitro translation was tested for specific binding to various
cytoplasmic receptor tails fused to GST. Specific binding of eIF-2B
was observed to the cytoplasmic tails of the
2A-,
2B-,
2C-, and
2-adrenergic receptor, while nonspecific binding to GST alone was negligible (Fig.
1B). eIF-2B
did not bind to the carboxyl-terminal
cytoplasmic domain of the low density lipoprotein receptor, indicating
that eIF-2B
interacts specifically with a subset of plasma membrane receptors. No binding was observed to the carboxyl-terminal cytoplasmic domain of the vasopressin (V2) receptor, suggesting further
that eIF-2B
associates specifically with a limited subset of G
protein-coupled receptors. Alignment of the carboxyl-terminal
cytoplasmic domains of the
2A-,
2B-,
2C-, and
2-adrenergic receptors revealed five identical amino acids (Fig. 2). However, only one
of those five residues was found to be identical in the vasopressin
V2 receptor. This suggests a potential role of the proximal
portion of the carboxyl terminus of the adrenergic receptors in binding of eIF-2B
.
Interestingly, while both the M1 and M20 forms of eIF-2B were
observed to bind to adrenergic receptor tails, examination of SDS-PAGE
fluorographs indicated that the M20 form bound considerably more
strongly (compare Fig. 1, A and B). These
observations suggest that the amino-terminal domain of eIF-2B
plays
a role in mediating or regulating receptor binding. Since both the M1
and M20 forms of eIF-2B
can be produced in vivo from the
same transcript (see below), it is possible that these polypeptides may
serve different physiological functions according to their different
affinities for receptors. An example of this type of specificity is the
Oskar protein in Drosophila, where different protein forms
produced by alternative start codon usage serve distinct functions in
oogenesis (31).
The ability of eIF-2B to bind to full-length
2-adrenergic receptors in vivo was examined
by co-immunoprecipitation from transfected 293 cells. The
2-adrenergic receptor was tagged in the amino-terminal
extracellular domain with a FLAG-epitope to facilitate specific
immunoprecipitation of receptors using an antibody that does not
interfere with the cytoplasmic tail. eIF-2B
was HA-tagged in the
carboxyl terminus to allow detection of proteins originating from
alternate start codon usage. Two protein products, corresponding to the
M1 and M20 forms of eIF-2B
, were observed in transfected cells. The
M1 form was by far the predominant product (Fig. 1C, lanes 1 and 2). Both forms of eIF-2B
were co-immunoprecipitated specifically from cell lysates in association with the
2-adrenergic receptor (Fig. 1C, lanes 6 and
7). Neither form of eIF-2B
was detected in control
immunoprecipitates, including those prepared from cells expressing
eIF-2B
without FLAG-tagged receptors (Fig. 1C, lanes
3-5), confirming the specificity of this protein association in vivo. Interestingly, the M20 form of eIF-2B
preferentially associated with immunoprecipitated receptors, even
though the M1 form was expressed in significant excess (Fig.
1C, compare lanes 1 and 6). These data
further confirm the specificity of the co-immunoprecipitation and
suggest that individual forms of eIF-2B
display similar binding
selectivity for full-length receptors in vivo as they do for
isolated carboxyl-terminal cytoplasmic domains in vitro.
The subcellular distribution of epitope-tagged eIF-2B was next
examined in transfected cells by fluorescence microscopy. Epitope-tagged eIF-2B
was visualized in a diffuse distribution, with
increased staining intensity near the cell periphery and no detectable
staining at regions of cell-cell contact (Fig.
3A), while
2-adrenergic
receptors were localized throughout the plasma membrane (Fig.
3B). Immunoblotting of extensively washed membrane fractions
prepared from transfected cells confirmed that a significant fraction
of eIF-2B
was membrane-associated (not shown). Optical sectioning of
antibody-labeled cells by confocal microscopy revealed eIF-2B
distributed throughout the cytoplasm and excluded from the nucleus,
consistent with the known role of this protein in regulating ribosome
function in the cytoplasm. In addition, confocal microscopy confirmed
that eIF-2B
was also closely associated with limited regions of the
peripheral plasma membrane (Fig. 3C) which contained
relatively high concentrations of
2-adrenergic receptor
(Fig. 3D). Marked co-localization of eIF-2B
(green) with
2-adrenergic receptors
(red) in these regions of plasma membrane was emphasized by
the yellow staining in the merged image (Fig.
3E). This close co-localization, which was observed even at
high magnification (Fig. 3F), suggests that these membrane microdomains may be sites of interaction between eIF-2B
and
receptors. A similar plasma membrane localization pattern for
epitope-tagged eIF-2B
was observed in cells transfected with
eIF-2B
alone or in cells co-transfected with eIF-2B
and the
vasopressin V2 receptor (not shown). These observations and
the lack of co-localization of eIF-2B
with adrenergic receptors at
regions of cell-cell contact suggest that the membrane localization of
eIF-2B
may be influenced by additional protein interactions.
Nevertheless, the co-immunoprecipitation of eIF-2B
with
2-adrenergic receptors from intact cells (Fig. 1C) suggests that these proteins physically interact at
regions of co-localization.
While eIF-2B has a well established role in regulating initiation of
translation in the cytoplasm (27, 29), the association of this protein
with receptors in the plasma membrane suggested that eIF-2B may have
additional physiological role(s). Several proteins that associate with
the carboxyl-terminal cytoplasmic tail of G protein-coupled receptors
regulate receptor-mediated signaling. The carboxyl-terminal cytoplasmic
domain of the
2-adrenergic receptor interacts with G
protein-coupled receptor kinases and arrestins, causing functional
desensitization by reducing the efficacy and potency of
agonist-dependent signaling (1-4). To examine the
possibility that eIF-2B
may also desensitize receptors, the effect
of eIF-2B
overexpression on agonist-dependent activation of adenylyl cyclase was examined in cells expressing
2-adrenergic receptors. No inhibition of signaling was
observed over a wide range of agonist concentrations, suggesting that
eIF-2B
does not increase receptor desensitization in these cells. In
contrast, overexpression of eIF-2B
actually had the opposite effect
on cell signaling. Overexpression of eIF-2B
caused a modest
enhancement of receptor-mediated activation of adenylyl cyclase at high
concentrations of agonist (not shown). In multiple experiments,
activation of adenylyl cyclase caused by the beta agonist isoproterenol
(10 µM) was enhanced by an average of 15% in cells
co-transfected with HA-tagged eIF-2B
and
2-adrenergic
receptors, when compared with cells examined in parallel that were
transfected with
2-adrenergic receptors and control
plasmid (p < 0.01, n = 11).
Radioligand binding assays using 125I-cyanopindolol
indicated that eIF-2B
overexpression had no effect on the number of
2-adrenergic receptors expressed in transfected cells
(n = 4). Furthermore, no effect of eIF-2B
overexpression was observed under the same conditions on vasopressin
V2 receptor-mediated activation of adenylyl cyclase (10 µM arginine vasopressin, n = 4), on basal
adenylyl cyclase activity, or on direct activation of adenylyl cyclase
by forskolin (10 µM, n = 4). Taken
together, these observations are consistent with the hypothesis that
eIF-2B
specifically enhances signaling by interacting directly with
2-adrenergic receptors, rather than by influencing
receptor expression or downstream signaling components. Furthermore,
the specificity of this enhancement for
2-adrenergic
receptors compared with V2 receptors is consistent with the
biochemical specificity of eIF-2B
association with isolated receptor
tails observed in vitro.
While eIF-2B caused a significant and reproducible enhancement of
adenylyl cyclase activation by
2-adrenergic receptors, the magnitude of this effect was relatively small (15% on average) in
all experiments. It is possible that endogenous levels of eIF-2B
in
293 cells may be sufficient to promote receptor signaling, so a
relatively small enhancement is observed by overexpression of this
protein. Alternatively, the effect of eIF-2B
on receptor signaling
may depend on associations with other proteins (such as other subunits
of the eIF-2B heteropentamer) that are present in limiting amounts
relative to overexpressed eIF-2B
.
Additional studies will be necessary to elucidate precisely how
eIF-2B associates with the plasma membrane and enhances receptor signaling. Interestingly, recent studies suggest that additional, as
yet unidentified, membrane-associated proteins influence the propagation of agonist-induced signals of G protein-coupled receptors in isolated plasma membranes (32, 33). In principle, it is also
possible that eIF-2B
could enhance receptor-mediated signaling by
inhibiting a known mechanism of receptor desensitization
(e.g. phosphorylation or arrestin binding). While the
present studies have focused exclusively on the functional effects of
eIF-2B
on receptor signaling, it is also possible that this protein
interaction may play an additional role in regulation of translation.
This possibility is consistent with the well established role of eIF-2B in the regulation of initiation of translation and with recent studies
indicating that various extracellular stimuli can control eIF-2B
activity (34-36).
In conclusion, we have shown that eIF-2B interacts specifically with
the carboxyl-terminal cytoplasmic domain of a subset of G
protein-coupled receptors. These results identify a novel protein
interaction with adrenergic receptors, suggesting a new role for a
previously identified guanine nucleotide exchange protein in receptor
biology.
We are grateful to Yoram Altschuler for his help in establishing the yeast two-hybrid system. We thank Peter Chu for expert technical assistance, Candace Chi for valuable contributions to the in vitro biochemical studies, and Susan Service for advice and assistance on the statistical analysis. We thank Jane Gitschier and Bruce Conklin for providing cDNAs encoding vasopressin V2 receptor and Henry Bourne, Robert Matts, and Peter Walter for valuable discussion.