(Received for publication, September 19, 1996, and in revised form, November 4, 1996)
From the Department of Pharmacology and the § Departments of Medicine and Pathology, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania 19104
The selectivity in coupling of various receptors
to GTP-binding regulatory proteins (G proteins) was examined directly
by a novel assay entailing the use of proteins overexpressed in
Spodoptera frugiperda (Sf9) cells. Activation of G proteins
was monitored in membranes prepared from Sf9 cells co-expressing
selected pairs of receptors and G proteins (i.e. ,
1, and
2 subunits). Membranes were
incubated with [35S]guanosine
5
-(3-O-thio)triphosphate (GTP
S) ± an agonist, and the
amount of radiolabel bound to the
subunit was quantitated following
immunoprecipitation. When expressed without receptor (but with
1
2), the G protein subunits
z,
12, and
13 did not bind
appreciable levels of [35S]GTP
S, consistent with a
minimal level of GDP/[35S]GTP
S exchange. In contrast,
the subunits
s and
q bound measurable levels of the nucleotide. Co-expression of the
5-hydroxytryptamine1A (5-HT1A) receptor
promoted binding of [35S]GTP
S to
z but
not to
12,
13, or
s.
Binding to
z was enhanced by inclusion of serotonin in
the assay. Agonist activation of both thrombin and neurokinin-1
receptors promoted a modest increase in [35S]GTP
S
binding to
z and more robust increases in binding to
q,
12, and
13. Binding of
[35S]GTP
S to
s was strongly enhanced
only by the activated
1-adrenergic receptor. Our data
identify interactions of receptors and G proteins directly, without
resort to measurements of effector activity, confirm the coupling of
the 5-HT1A receptor to Gz and extend the list
of receptors that interact with this unique G protein to the receptors
for thrombin and substance P, imply constitutive activity for the
5-HT1A receptor, and demonstrate for the first time that
the cloned receptors for thrombin and substance P activate G12 and G13.
The actions of numerous cell-surface receptors on enzymes and ion
channels are mediated by heterotrimeric () GTP-binding regulatory proteins (G proteins)1 (1). Four
families of G proteins have been defined, represented by
Gs, Gi, Gq, and G12
(2). The activity of each G protein is tightly linked to the binding
and hydrolysis of GTP (3). Agonists working through 7-transmembrane
domain receptors promote the release of GDP from the
subunit and
thus an exchange for GTP present in the cytoplasm. Correlates of the
exchange are an altered conformation of the
subunit and
dissociation of the subunit from
. The
subunit as a monomer
and
as a heterodimer, or both operating conjointly, are
responsible for the regulation of effectors. G proteins can also be
activated by nonhydrolyzable analogues of GTP, for example GTP
S, a
process similarly promoted by agonists.
Considerable effort has been expended on identifying the G proteins
activated by various receptors. Notwithstanding work with purified
proteins, interactions have been defined mostly through inferences
drawn from effectors regulated. The stimulation of adenylyl cyclase is
almost always achieved through activation of Gs, for
example, while the inhibition is linked to members of the
Gi family (4). Stimulation of the phosphoinositide-specific phospholipase C- can be accomplished through Gi
(pertussis toxin (PTX)-sensitive) or through members of the
Gq family (PTX-insensitive) (5). No effector has yet been
identified for the G12 family, although
Na+/H+ exchange is a tightly linked correlate
(6). The use of effector regulation to deduce receptor·G protein
linkages, however, is less than perfect. There is often no resolution
among individual members of a particular G protein family, and the
nature of effector regulation itself has become increasingly
complicated with the appreciation that subunits released from many G
proteins can modulate the activation achieved by one (4). Effectors are
also subject to numerous forms of feedback or cross-regulation that
further limit the extent to which they faithfully mirror receptor·G
protein communication (4, 6, 7).
A number of sophisticated techniques have been used to gain either a
greater degree of resolution among G proteins regulating a particular
effector or a measurement of receptor·G protein communication directly. Antisense constructs (8, 9), PTX-resistant analogues of
Gi and Go (10), and antibodies that disrupt
interactions between receptors and G proteins (11, 12) are examples of the former. Co-purification of receptor·G protein complexes (13, 14),
agonist-promoted photoaffinity labeling with azidoanilido-GTP (15, 16),
and agonist-promoted GTPS binding (17) have been used to examine
interactions directly. None of these techniques, however, is used
widely, and many suffer from practical limitations. Even the use of
purified receptors and G proteins to assess the potential for
communication has led to some debate regarding the fidelity of
interactions (18).
Spodoptera frugiperda (Sf9) cells have recently been established as an intact cell setting for reconstitution of the human 5-hydroxytryptamine1A (5-HT1A) receptor with mammalian G protein subunits (19). Receptors endogenous to Sf9 cells have yet to be characterized, but so far have not interfered with the analysis of numerous mammalian receptors introduced through recombinant baculoviruses. The levels of G proteins endogenous to Sf9 cells, moreover, are quite low relative to those of mammalian G proteins that can be similarly introduced. Sf9 cells also carry out processing events that support the normal targeting of receptors and G protein subunits to the cell membrane, thus providing a relatively normal phospholipid milieu for interaction. In the previous study (19), the 5-HT1A receptor was demonstrated to couple to various members of the Gi family, i.e. Gi, Go, and Gz, as assessed by Gpp(NH)p-sensitive increases in affinity for radiolabeled agonists.
A valuable index of coupling is the process of G protein activation
itself. An assay of activation permits an evaluation of coupling
without resort to radiolabeled agonists and circumvents presumptions
regarding changes in agonist affinity. More importantly with respect to
many other assays, the requirement for effector activity as a means of
monitoring coupling is eliminated. In the present study, we have
investigated the selectivity of receptor·G protein coupling by
measuring directly the activation of G proteins introduced into Sf9
cells. The measurement is based on agonist-promoted binding of
[35S]GTPS to G protein
subunits isolated
subsequently by immunoprecipitation. We have used this methodology to
study the coupling of four different receptors (the 5-HT1A,
1-adrenergic, neurokinin-1 (NK1), and thrombin receptors) to members of each class of G protein. The expected
selectivity in coupling, whereas only intimated previously, was
confirmed, and novel interactions between receptors and G proteins were
identified.
Recombinant baculoviruses encoding
s-s,
i1,
q,
1, and
2 (20-22) were kindly provided by
Drs. T. Kozasa and A. Gilman at Southwestern Medical Center (Dallas),
and those for
12 and
13 were a gift from
Dr. N. Dhanasekaran at Temple University School of Medicine
(Philadelphia). Baculoviruses for the rat
1-adrenoreceptor, human NK1 receptor (23),
and human thrombin receptor were gifts from Drs. E. Ross at
Southwestern Medical Center (Dallas), T. Fong at Merck & Co., and Drs.
S. Seiler and P. Rose at Bristol-Myers Squibb (Princeton, NJ),
respectively. Those encoding the 5-HT1A receptor and
z were constructed in this laboratory (19).
Sf9 cells were
cultured as described previously (19) but with charcoal-treated serum.
For infection with recombinant baculoviruses, the cells were
subcultured in monolayer and infected with one or more viruses at a
multiplicity of infection of at least one for each virus. The medium
was replaced 16 h following infection with Sf900II optimized
serum-free medium (Life Technologies, Inc.). The cells were harvested
at 48 h and homogenized in ice-cold 20 mM HEPES (pH
8.0), 1 mM EDTA, 0.1 mM phenylmethysulfonyl
fluoride, 10 µg/ml leupeptin, and 2 µg/ml aprotinin by repeated
passage through a 26-gauge needle. The homogenate was centrifuged at
100 × g for 5 min, and the resulting supernatant
fraction was centrifuged at 16,000 × g for 30 min to
pellet the membranes. The membranes were washed and resuspended at ~3
mg/ml protein in homogenization buffer for storage at 70 °C. In
experiments where the thrombin receptor was expressed, the thrombin
protease inhibitor D-Phe-Pro-Arg chloromethyl ketone (1 µM) and N
-tosyl-Lys
chloromethyl ketone (100 µM) were included throughout the
period of infection and subsequent expression of receptor.
Membranes (20 µg of protein) from Sf9 cells expressing receptors and/or G protein
subunits were resuspended in 55 µl of 50 mM Tris-HCl (pH
7.4), 2 mM EDTA, 100 mM NaCl, 1 µM GDP, and a concentration of MgCl2
calculated to give the desired concentration of free Mg2+.
[35S]GTPS (1300 Ci/mmol, DuPont NEN) was added to a
final concentration of 30 nM, and the incubation was
allowed to proceed for 5 min at 30 °C in the absence or presence of
a selected agonist. The incubation was terminated by adding 600 µl of
ice-cold 50 mM Tris-HCl (pH 7.5), 20 mM
MgCl2, 150 mM NaCl, 0.5% Nonidet P-40
(Calbiochem), 1% aprotinin, 100 µM GDP, and 100 µM GTP. After 30 min, the extract was transferred to an
Eppendorf tube containing 2 µl of non-immune serum pre-incubated with
150 µl of a 10% suspension of Pansorbin® cells
(Calbiochem). Nonspecifically bound proteins were removed after 20 min
by centrifugation. The extract was then incubated for 1 h at
4 °C with 10 µl of a G protein
subunit-directed antiserum, pre-immune serum, or non-immune serum, all of which had been
pre-incubated with 100 µl of a 5% suspension of protein A-Sepharose.
With the exception of the
12-directed antiserum,
generated toward the peptide QENLKDIMLQ, the antisera have been
described previously (19, 24). Immunoprecipitates were collected and
washed three times in the extraction buffer, once in the buffer without
detergent, and then boiled in 0.5 ml of 0.5% SDS followed by addition
of 4 ml of Ecolite+TM (ICN, Costa Mesa, CA). The samples
were analyzed directly by scintillation spectrometry.
Immunotransfer blotting and
quantitation of z was accomplished as before (19).
Immunoprecipitation of [35S]methionine-labeled proteins
was also accomplished as before (19) but under the conditions of
extraction and immunoprecipiation used for the nucleotide-binding
assays above. Efficiency of immunoprecipitation for antiserum 6354 under these conditions was calculated by comparison to the amount of
[35S]methionine-labeled
z
immunoprecipitated by antiserum 8645 following denaturation (presumed
to be 20% (25)).
Activation of G proteins in membranes prepared from Sf9 cells was
examined first for the combination of Gz and the human
5-HT1A receptor. Gz is a member of the
Gi family and was chosen based on experiments indicating
that the binding of [35S]GTPS to
z is
strictly dependent on co-expression of receptor (see below).
Gz also exhibits the capacity to couple to the
5-HT1A receptor, as do other members of the Gi
family (19). Fig. 1 represents a set of experiments in
which Sf9 cell membranes containing Gz (i.e.
z,
1, and
2) and the
5-HT1A receptor were incubated with
[35S]GTP
S ± serotonin (5-HT) over a range of
Mg2+ concentrations.
z was subsequently
immunoprecipitated with antiserum 6354 (generated toward residues
24-33), and bound [35S]GTP
S was counted directly. As
shown in the figure, the binding of [35S]GTP
S to
z was dependent on Mg2+ and was optimum in
the range of 0.5-10 mM of the divalent cation. Binding was
enhanced by serotonin but also occurred in the apparent absence of
agonist. The use of non-immune serum instead of antiserum 6354 confirmed specificity of binding for
z, as did
pretreatment of antiserum 6354 with the peptide used for immunization
(not shown). Similar results were achieved with antiserum 2921, directed toward the C terminus of
z (residues
346-355).
Having established that Gz communicates with the
5-HT1A receptor as monitored by [35S]GTPS
binding, we examined further the requirements of the binding assay for
receptor and G protein subunits (Fig. 2). As above, a
significant degree of binding of [35S]GTP
S to
z was evident when the 5-HT1A receptor,
z, and
1
2 were
co-expressed, and inclusion of serotonin in the assay increased binding
further. Identical results were obtained when the 5-HT1A receptor and
z were expressed together but without
addition of
1
2. Co-expression of the
receptor and
1
2 without
z
confirmed that the binding was specific for
z. Omission
of the receptor demonstrated that Gz alone did not bind
[35S]GTP
S.
The amount of z following infection of Sf9 cells with
recombinant viruses encoding the 5-HT1A receptor,
z,
1, and
2 was 15-30
pmol of subunit per mg of membrane protein, and the efficiencies of
extraction and immunoprecipitation were 90 and 60%, respectively. We
calculated that 0.5 pmol of [35S]GTP
S was
immunoprecipitated with
z per mg of membrane protein (following treatment of membranes with agonist). Thus, the amount of
bound [35S]GTP
S was 5-10% of immunoprecipitated
z. Binding was not enhanced by increasing the
concentration of [35S]GTP
S nor by omitting GDP from
the assay (GDP was used to suppress agonist-independent binding).
Binding of [35S]GTP
S could be increased 2-3-fold by
increasing the time of incubation to 60 min.
The assay was next extended to G proteins from other families and to
other receptors. Fig. 3 illustrates the ability of the chosen antisera to immunoprecipitate the respective subunits. The
antisera were generated with peptides corresponding the C-terminal 10 amino acid residues of
s (1191),
q
(0945),
12 (121), and
13 (120). Results
obtained with the two antibodies specific for
z are
shown for comparison. No subunit was evident when non-immune serum was
substituted or when the Sf9 cells were not infected. The appearance of
q as two bands has been reported previously (21).
Western blots using the "consensus" antisera 8645 and 1398 reveal
that expression levels of the different subunits were within
severalfold of each other (
12 could not be analyzed by this means). Binding of [35S]GTP
S to each of the
subunits co-expressed with
1
2 but not receptors is shown as a function of Mg2+ in Fig.
4. No binding occurred for
z,
12, or
13 at any concentration of
Mg2+.
s and
q, in contrast,
bound [35S]GTP
S in a
Mg2+-dependent fashion, as did
i1 (not shown). Subsequent experiments were carried out
at 3 mM free Mg2+.
The potential for coupling of Gz to
1-adrenergic, thrombin, and NK1 (substance
P) receptors was next assessed (Fig. 5). The 5-HT1A receptor, as noted previously, caused an increase in
[35S]GTP
S binding in the absence of added agonist and
promoted a further increase when agonist (serotonin) was added. Binding
was also promoted by the thrombin and NK1 receptors, though
to a lesser extent. In the latter instances, agonists were required.
The
1-adrenergic receptor failed to activate
Gz.
A completely different order of selectivities was evident for
Gs (Fig. 6). As noted above,
s binds some [35S]GTP
S regardless of
receptor. Omission of
s revealed that the binding was
relatively specific for the introduced subunit. Activation of the
1-adrenergic receptor with isoproterenol caused a
3-4-fold increase in binding. No increase was evident without agonist. Co-expressed 5-HT1A, thrombin, and NK1
receptors, with or without agonists, had no effect. The level of
binding for the Gs/
1-adrenergic receptor
combination (~90,000 cpm) was considerably higher than that achieved
for Gz with any receptor, despite equivalent expression of
the two G proteins.
As with Gs, a significant level of receptor-independent
binding was observed for Gq (Fig. 7).
Omission of the recombinant q indicated that some of the
binding could be accounted for by
q (or a cross-reactive
subunit) endogenous to Sf9 cells. Binding to the endogenous subunit was
enhanced by activation of the NK1 receptor. However, the
signal provided by endogenous subunit was well below that achieved with
the mammalian subunit. A modest degree of [35S]GTP
S
binding was elicited by the activated 5-HT1A receptor and
was dependent on the mammalian subunit (not shown). A much higher level
of binding was attained with activated thrombin and NK1
receptors. The receptor-enhanced binding in all three instances was
agonist-dependent. The
1-adrenergic
receptor, regardless of agonist, did not promote binding.
Communication of the receptors with G13 is shown in Fig.
8. The data for G12 were identical (not
shown). In the case of these two G proteins, only two of the receptors
(the thrombin and NK1 receptors) promoted binding of
[35S]GTPS. No binding was observed without agonists.
The -fold enhancement in binding achieved by agonists, in fact, was the
highest for any of the receptor/G protein combinations (greater than
6-fold). As noted previously, GDP was included in all binding assays to suppress receptor-independent binding, an action we had confirmed for
Gz and Gs. However, GDP suppresses
[
-32P]GTP azidoanilide incorporation into
12 and
13 in platelet membranes (16).
Consistent with this report, we found that removal of GDP, while
having no effect on the negligible receptor-independent binding of
[35S]GTP
S by
12 and
13,
increased binding promoted by agonists by about 50%.
We have examined the potential of various receptors to couple to
members of each of the four families of G proteins. The method used,
receptor-promoted binding of [35S]GTPS to G protein
subunits expressed in Sf9 cells, is powerful. Purification of
receptors and G protein subunits is not required for reconstitution,
nor is the assay compromised by the heterogeneity in both types of
proteins inherent to mammalian expression systems. Importantly,
receptor·G protein coupling can be analyzed without resort to
effector activity. This latter property allows the modeling of
receptor·G protein interactions directly (but nevertheless in a
native milieu) and a definition of coupling for those G proteins whose
effectors are not yet known. We have demonstrated here that the
1-adrenergic receptor couples selectively to
Gs and not to other G proteins, that the 5-HT1A
receptor is selectively coupled to Gz (of the G proteins
tested) but shows some activity toward Gq, and that the
thrombin and NK1 receptors couple to G12 and G13, as they do to Gq and Gz, but
not Gs. Selectivities in interactions previously intimated
are now demonstrated directly, and novel interactions are
identified.
Gz was used as a representative of the Gi
family and as a prototype in the design of the assay. Among the most
important traits exhibited by Gz was an inability to bind
[35S]GTPS without co-expression of receptor. This
property was evident at all concentrations of Mg2+ and was
subsequently found to be shared with G12 and
G13. The inability to bind [35S]GTP
S under
the constraints of the assay was consistent with the low rates of
exchange of GDP for GTP
S established previously for purified
z,
12, and
13 (26-28).
Gs and Gq both displayed significant levels of
receptor-independent binding at millimolar concentrations of
Mg2+. Although GDP/GTP
S exchange has not been fully
analyzed for Gq (21), purified Gs exchanges GDP
for GTP
S quite rapidly at high concentrations of Mg2+
(29).
Gz is activated to the greatest extent by the
serotonin-activated 5-HT1A receptor. The activation by this
receptor was predicted based on the communication between the two
proteins implied previously (19) and the fact that the
5-HT1A receptor is coupled to members of the Gi
family in mammalian neurons (30, 31). We found no evidence for
activation of Gs through the 5-HT1A receptor,
as once implied (32, 33), but did document a modest activation of
Gq. The latter finding is consistent with activation of
phosphoinositide hydrolysis in several types of cells expressing the
5-HT1A receptor at high concentrations of agonist (34) and
is also reminiscent of a somewhat paradoxical affect of
q on ligand affinity previously noted in Sf9 cells (19).
Activation of Gz by the agonist-activated 5-HT1A receptor was clearly greater than the activation
achieved by the thrombin and NK1 receptors, while the
converse was true for activation of Gq.
Only a small proportion of immunoprecipitated z from Sf9
membranes containing the 5-HT1A receptor and exposed to
serotonin was complexed with [35S]GTP
S. The low level
of binding may simply be related to a normal low rate of GDP/GTP
S
exchange. Alternatively, it may reflect some instability of the
[35S]GTP
S·subunit complex through extraction and
immunoprecipitation. We were somewhat surprised that introduction of
had no influence on [35S]GTP
S binding to
z. We had determined previously that
z
alone could increase the affinity of the 5-HT1A receptor
for agonist but that a greater degree of coupling was achieved upon
co-expression with
1
2 (19). To some
extent, the apparent inactivity of
might be explained by the
fact that expression of the two additional subunits causes an
approximately 2-fold suppression in levels of
z (not
shown). However, we also suspect that
endogenous to Sf9 cells
may act catalytically with respect to the activation process.
A significant activation of Gz occurred in the presence of the 5-HT1A receptor but without serotonin. On the one hand, serotonin may have been carried over from the serum in which the cells were initially cultured. However, we employed serum-free medium during the time at which 5-HT1A receptors were expressed and washed the cells and membranes extensively. The receptor, instead, may exhibit constitutive activity. G protein-coupled receptors can convert between active and inactive conformations, a process especially evident in overexpression systems (35). Of the four types of receptors studied here, however, only the 5-HT1A receptor exhibits a readily identified activity.
The greatest degree of selectivity in the interaction of a receptor
with a G protein was encountered at the level of the
1-adrenergic receptor and Gs. Gs
was activated only by the
1-adrenergic receptor, and the
receptor had no action on any G protein but Gs. The pairing of the
1-adrenergic receptor and Gs was
obviously expected (36). Another possible interaction with
Gs, involving the NK1 receptor, was also
sought, since an earlier report had linked this receptor not only to
the activation of phosphoinositide hydrolysis but, at very high
concentrations of agonist, to the stimulation of adenylyl cyclase (37).
However, we were unable to observe any interaction between the
NK1 receptor and Gs. We suspect that the stimulation of adenylyl cyclase observed previously was indirect.
With respect to Gq, the activation by thrombin and
NK1 receptors was anticipated, though, as for all other
pairings but that of the 1-adrenergic receptor and
Gs, had not been measured directly in previous work. The
stimulation of phosphoinositide hydrolysis by thrombin in most types of
cells is largely insensitive to PTX, implying the use of a
Gq-like protein (38). Sensitivity to PTX, though partial,
is common, however, suggesting at least some contribution by
Gi. That thrombin can communicate with Gi is
quite clear from its ability to inhibit adenylyl cyclase through a
PTX-sensitive element in a large number of cells. Conjoint utilization
by thrombin of Gq and Gi is consistent with our
data, wherein Gz is used as a representative of the
Gi family. Substance P also stimulates phosphoinositide
hydrolysis in a PTX-insensitive fashion (39), and the addition of
Gq/11 to phospholipid vesicles containing NK1
receptors results in conversion of the receptors to a high affinity
state (40). Utilization of Gi (and hence Gz in
our experiments here) by NK1 receptors is less well
documented. However, the NK1 receptor appears to regulate a
large conductance Cl
channel through a member of the
Gi family (41). The list of receptors that activate
Gz can now be extended to those for thrombin and substance
P.
We were most interested in the potential of the different receptors to
activate G12 and G13. Receptors normally linked
to these G proteins are poorly characterized, and effectors have not
yet been identified. Our results clearly demonstrate that thrombin and
NK1 receptors link to the activation of G12 and
G13, while 1-adrenergic and
5-HT1A receptors do not. That thrombin should utilize these
G proteins is consistent with its ability to support incorporation of
[
-32P]GTP azidoanilide into
12 and
13 in platelet membranes (16) and with the block of
thrombin-stimulated DNA synthesis with an antibody directed toward
12 (42). Our results substantiate the capacity of the
cloned receptor for thrombin (43), as distinct from a recently deduced
second receptor (44), to achieve the activation. It is intriguing that
the two receptors that activate G12 and G13
here, the cloned thrombin and NK1 receptors, also activate
Gq and Gz. Whether activation of the latter two
G proteins conjointly is predictive of coupling to G12 and
G13 is worth pursuit. We were also interested to note that
the
1-adrenergic receptor does not couple to
G12 or G13, as several reports had indicated that
-adrenergic receptors activate Na+/H+
exchange, a correlate of G12 or G13 activation
(6), independently of Gs (45, 46). The fact that the
5-HT1A receptor does not couple to G12 and
G13 is also notable. Serotonin operating through 5-HT1A receptors is not viewed to be a complete mitogen,
but thrombin is (47). It is conceivable that the mitogenic properties
of thrombin are linked to the activation of G12 and
G13. Overexpression of
12 or
13, or expression of GTPase-deficient mutants, is linked to unregulated cell growth (6).
Sf9 cells constitute a well defined, intact cell setting upon which the
expression of mammalian receptors and G proteins can be superimposed.
Our data support the conclusion that GTPS binding is an effective
means of monitoring activation of G proteins by receptors. Our data
also define the interactions of several receptors with representatives
of each family of G protein. Novel interactions have been identified,
and their authenticity is supported by the selectivity in interactions
otherwise predicted from measurements of second messenger regulation.
We anticipate that the Sf9 reconstitution system will lend itself to
the analysis of inverse agonism and the coupling achieved by orphan
receptors. We also propose the use of the Sf9 reconstitution system for
developing or otherwise optimizing techniques to map receptor·G
protein communication in mammalian cells.