(Received for publication, January 23, 1997, and in revised form, May 5, 1997)
From the Departments of Biochemistry and Molecular Pharmacology,
and Microbiology and Immunology, Kimmel Cancer Center, Thomas Jefferson
University, Philadelphia, Pennsylvania 19107 and the
Department of Molecular and Cellular Pharmacology,
University of Miami School of Medicine, Miami, Florida 33136
G protein-coupled receptor kinases (GRKs)
specifically phosphorylate and regulate the activated form of multiple
G protein-coupled receptors. Recent studies have revealed that GRKs are
also subject to regulation. In this regard, GRK2 and GRK5 can be
phosphorylated and either activated or inhibited, respectively, by
protein kinase C. Here we demonstrate that calmodulin, another mediator
of calcium signaling, is a potent inhibitor of GRK activity with a
selectivity for GRK5 (IC50 ~50 nM) > GRK6 GRK2 (IC50 ~2 µM)
GRK1.
Calmodulin inhibition of GRK5 is mediated via a reduced ability of the
kinase to bind to both receptor and phospholipid. Interestingly,
calmodulin also activates autophosphorylation of GRK5 at sites distinct
from the two major autophosphorylation sites on GRK5. Moreover,
calmodulin-stimulated autophosphorylation directly inhibits GRK5
interaction with receptor even in the absence of calmodulin. Using
glutathione S-transferase-GRK5 fusion proteins either to
inhibit calmodulin-stimulated autophosphorylation or to bind directly
to calmodulin, we determined that an amino-terminal domain of GRK5
(amino acids 20-39) is sufficient for calmodulin binding. This domain
is abundant in basic and hydrophobic residues, characteristics typical
of calmodulin binding sites, and is highly conserved in GRK4, GRK5, and
GRK6. These studies suggest that calmodulin may serve a general role in
mediating calcium-dependent regulation of GRK activity.
G protein-coupled receptor kinases
(GRKs)1 form a family of serine/threonine
protein kinases with the unique ability to recognize specifically the
agonist-activated state of G protein-coupled receptors (1, 2).
GRK-mediated phosphorylation promotes the binding of an arrestin
protein, thereby uncoupling the receptor from G protein and terminating
receptor signaling. Six members of the GRK family have been identified,
and based on their sequence homology they have been divided into three
subfamilies (2). GRK1 (rhodopsin kinase) forms one group; GRK2
(-adrenergic receptor kinase) and GRK3 a second; and GRK4, GRK5, and
GRK6 combine into a third subfamily.
All GRKs share a similar structural organization with a poorly
conserved amino-terminal domain of ~185 residues, a conserved protein
kinase catalytic domain of ~270 residues, and a variable length
carboxyl-terminal domain of 105-230 residues (3). However, although
all GRKs have a similar overall structure and function, various
subfamily members also have certain unique features. For example,
various GRKs utilize different mechanisms to promote membrane
association, an event critical for receptor interaction. GRK1 is
farnesylated (4), GRK2 and 3 interact with phospholipids and G protein
subunits via pleckstrin homology domains (5-8), GRK4 (9) and
GRK6 (10) are palmitoylated, and GRK5 binds to phospholipids via
polybasic regions in the amino- and carboxyl-terminal domains
(11, 12).
Another characteristic that appears specific for the GRK subtype involves regulation of kinase activity. For example, in the visual system, GRK1 has been shown to be inhibited by the Ca2+-binding protein recoverin (13). Calcium binding to recoverin promotes its association with GRK1, inactivating the kinase and thereby reducing its ability to phosphorylate rhodopsin. Since calcium levels are decreased upon light activation of rod cells (14), recoverin binding to rhodopsin kinase might provide a mechanism for adaptation of the system to ambient light. Because recoverin has no effect on GRK2, regulation by recoverin may be specific for GRK1. Recent studies also have demonstrated that GRK2 and GRK5 are subject to regulatory phosphorylation via protein kinase C (PKC), a Ca2+/phospholipid-dependent kinase. GRK2 phosphorylation by PKC leads to an ~2-3-fold activation of the kinase, possibly via an increased ability of GRK2 to bind to membranes (15, 16). In contrast, GRK5 is inhibited severalfold when phosphorylated by PKC due to both a decreased activity and affinity for receptor (17).
These examples illustrate how GRKs can be regulated via changes in
intracellular Ca2+ concentrations. Another universal
mediator of calcium signaling is calmodulin. Calmodulin is a
ubiquitously expressed Ca2+-binding protein that functions
as a Ca2+-dependent regulator of multiple
pathways including cyclic nucleotide metabolism, ion transport, protein
phosphorylation-dephosphorylation cascades, cytoskeletal function, and
cell proliferation (18, 19). In the present study we evaluated whether
calmodulin can regulate GRK activity. We show that calmodulin inhibits
GRK activity with a specificity of GRK5 (IC50 ~50
nM) > GRK6 GRK2 (IC50 ~2 µM)
GRK1. The calmodulin binding domain of GRK5 was
localized within the amino-terminal domain (residues 20-39). These
findings suggest that calmodulin may play an important role in
regulating GRK function in a subtype-specific manner.
Restriction endonucleases, Vent DNA polymerase,
and other molecular biology reagents were purchased from New England
Biolabs or Boehringer Mannheim. SP Sepharose was obtained from
Pharmacia Biotech Inc.. Calmodulin (bovine brain, >98% pure),
calmodulin-agarose, and phosphatidylcholine (soybean type II-S) were
from Sigma. Phosphatidylserine (bovine brain, 99% pure) was from
Avanti Polar Lipids, Inc. Rat PKC- and bovine GRK1, overexpressed
and purified from Sf9 cells, were generous gifts from Dr. C. Stubbs and
Drs. R. J. Lefkowitz and J. A. Pitcher, respectively. All other
materials were from sources previously described (17).
Expression plasmids for GRKs were constructed by cloning the coding sequences of bovine GRK2 (20) and human GRK5 and GRK6 in the vector pBC12BI (21). COS-1 cells were grown to ~80-90% confluence in 60-mm dishes at 37 °C in a humidified atmosphere containing 5% CO2, 95% air in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 100 units/ml penicillin, and 100 µg/ml streptomycin. Cells were transfected with 4 µg of DNA/dish using LipofectAMINE following the manufacturer's instructions (Life Technologies, Inc.). Forty-eight h after transfection, cells were harvested and lysed by scraping into 1 ml of ice-cold 20 mM Tris-HCl, pH 8.0, 2 mM EDTA, 200 mM NaCl, 1% Triton X-100 with protease inhibitors (5 µM aprotinin, 5 mM benzamidine, 20 µM leupeptin, 2 µM pepstatin A, 1 mM phenylmethylsulfonyl fluoride) and supernatants were prepared by centrifugation for 7 min at 100,000 × g (4 °C). GRKs were then partially purified by chromatography on SP Sepharose as described (17).
Four-µl aliquots of the partially purified GRK were then assayed by
incubating with rod outer segment (ROS) membranes (100 pmol of
rhodopsin) in 20 µl of 20 mM Tris-HCl, pH 8.0, 4 mM MgCl2, 0.1 mM CaCl2,
0.1 mM [-32P]ATP (1,000 cpm/pmol) in the
presence of the indicated concentration of calmodulin for 6 min at
30 °C in room light. The reactions were stopped with 200 µl of
ice-cold buffer (20 mM Tris-HCl, pH 8.0, 10 mM
EDTA, 100 mM NaCl) and centrifuged for 10 min at 100,000 rpm (4 °C). Pellets containing phosphorylated rhodopsin were
dissolved in SDS loading buffer, and the samples were then
electrophoresed on a 10% SDS-polyacrylamide gel (22). Gels were
stained with Coomassie Blue, dried, and autoradiographed, and the
32P-labeled proteins were excised and counted to determine
the pmol of phosphate transferred. Urea-treated ROS membranes
containing rhodopsin were prepared from bovine retinas as described
previously (23).
Bovine GRK2 and human GRK5 were overexpressed and purified
from Sf9 cells as described (7, 24). GRK-mediated phosphorylation was
assayed by incubating 0.8 pmol of GRK with either ROS membranes (80 pmol of rhodopsin), casein (10 µg), or phosvitin (10 µg) in 20 µl
of 20 mM Tris-HCl, pH 8.0, 4 mM
MgCl2, 0.1 mM CaCl2 (or 2 mM EGTA), 0.1 mM [-32P]ATP
(1,000 cpm/pmol) in the presence of the indicated concentrations of
calmodulin for 6 min at 30 °C in room light. The reactions were
stopped with 5 µl of SDS sample buffer, and the samples were electrophoresed on a 10% SDS-polyacrylamide gel. Gels were stained with Coomassie Blue, dried, and autoradiographed, and the
32P-labeled proteins were excised and counted.
The autophosphorylation-defective
mutant GRK5-DD (Ser484 and Thr485 mutated to
Asp) was overexpressed and purified from Sf9 cells as described (17).
Autophosphorylation reactions contained 4 pmol (0.27 µg) of either
wild type GRK5 or GRK5-DD in 20 µl of 20 mM Tris-HCl, pH
8.0, 4 mM MgCl2, 0.1 mM
CaCl2, 0.1 mM [-32P]ATP (5,000 cpm/pmol), 0.1 mg/ml ovalbumin, and either 0.85 mg/ml phospholipid
vesicles or the indicated concentration of calmodulin. Reactions were
incubated at 30 °C for 10 min and stopped with 5 µl of SDS sample
buffer. To assess the rate of calmodulin-stimulated autophosphorylation, 100 pmol of rhodopsin was phosphorylated with 2 pmol (0.1 µM) of GRK5 in the absence or presence of 0.5 µM calmodulin. Reactions were incubated at 30 °C and
at the indicated times were stopped with SDS sample buffer. Samples
were electrophoresed, and the 32P-labeled proteins were
excised and counted as described above.
To determine the kinetics for ATP, GRK5-DD (16 pmol) was
autophosphorylated in 20 mM Tris-HCl, pH 8.0, 4 mM MgCl2, 0.1 mM CaCl2,
0.1 mg/ml ovalbumin, and 2-100 µM
[-32P]ATP (10,000 cpm/pmol) in the absence or presence
of 0.8 µM calmodulin. Km and
Vmax values were derived from double-reciprocal plots of the data.
Sixty pmol of GRK5 was autophosphorylated in a 40-µl reaction at 30 °C for 15 min in the presence or absence of either 3 µM calmodulin or 0.07 µM PKC, 1 µM phorbol 12-myristate 13-acetate, and 0.85 mg/ml phospholipid vesicles as described above and then purified by batchwise chromatography on SP Sepharose. Briefly, phosphorylation reactions were stopped on ice, mixed with an equal volume of 20 mM Tris-HCl, pH 8.0, 2 mM EDTA, 2 mM EGTA, 100 mM NaCl, 0.4% Triton X-100, and then incubated for 10 min with 50 µl of a 50% suspension of SP Sepharose in buffer A (20 mM Tris-HCl, pH 8.0, 2 mM EDTA, 1 mM EGTA, 50 mM NaCl, 0.02% Triton X-100). The resin was pelleted, washed two or three times with 1 ml of buffer A, and the bound kinase was eluted with two 75-µl aliquots of 20 mM Tris-HCl, 1 mM EDTA, 600 mM NaCl, 0.02% Triton X-100. The supernatants were combined, diluted with 150 µl of 20 mM Tris-HCl, pH 8.0, 4 mM MgCl2, 1 mM EDTA, and then used for further analysis. Aliquots of the phosphorylated kinase before and after SP Sepharose purification were electrophoresed on an SDS-polyacrylamide gel to enable assessment of autophosphorylation, phosphorylation by PKC, and recovery from SP Sepharose. Typically, 70-80% of the GRK5 was recovered by this procedure, whereas PKC and calmodulin did not bind to SP Sepharose. Aliquots (~20 ng) of the SP Sepharose-purified GRK5 were also electrophoresed and subjected to Western blot analysis using antibodies raised against either amino acids 556-571 or 489-590 of human GRK5 as described (17).
Four-µl aliquots (~0.6 pmol) of the SP Sepharose-purified GRK5 were
assayed by incubating with either ROS membranes (60 pmol of rhodopsin)
or phosvitin (10 µg) in 20 µl of 20 mM Tris-HCl, pH
8.0, 4 mM MgCl2, 1 mM EDTA, 1 mM EGTA, 0.1 mM [-32P]ATP
(1,000 cpm/pmol) for 6 min at 30 °C in room light. When the effect
of calmodulin was tested the reactions also included the indicated
concentration of calmodulin and 0.2 mM CaCl2
(with no EGTA) and were incubated for 2 min at 30 °C. Reactions were stopped with 5 µl of SDS sample buffer, the samples were
electrophoresed on a 10% SDS-polyacrylamide gel, gels were stained
with Coomassie Blue, dried and autoradiographed, and the
32P-labeled proteins were excised and counted. To assess
the kinetics of receptor phosphorylation, 25-660 pmol of rhodopsin was
phosphorylated with GRK5 autophosphorylated in the presence or absence
of calmodulin in 20 mM Tris-HCl, pH 8.0, 4 mM
MgCl2, 1 mM EDTA, 0.1 mM
[
-32P]ATP (6,000 cpm/pmol). Km and
Vmax values were derived from double-reciprocal
plots of the data.
The ability of GRK5 to associate with either receptor or phospholipid was analyzed by incubating 8-µl aliquots (~1.2 pmol) of SP Sepharose-purified 32P-labeled autophosphorylated GRK5 in the presence or absence of the indicated concentration of phospholipid vesicles or ROS membranes (250 pmol of rhodopsin) in 60 µl of 20 mM Tris-HCl, pH 8.0, 2 mM MgCl2, 0.1 mM CaCl2, 80 mM NaCl, 0.1 mg/ml ovalbumin, and the indicated concentration of calmodulin at 30 °C for 5 min in room light. The samples were centrifuged at 100,000 rpm for 6 min, the pellets were resuspended in 60 µl of reaction buffer, and equal aliquots of the supernatant and pellet fractions were electrophoresed on a 10% SDS-polyacrylamide gel. The gels were dried, autoradiographed, and the 32P-labeled proteins were excised and counted. Pelleted GRK5 was expressed as a percentage of the total after subtracting the amount of GRK5 pelleted in the absence of phospholipids or ROS (~10-15%).
Phospholipid vesicles were prepared by sonicating 76 mg of phosphatidylcholine and 9 mg of phosphatidylserine in 5 ml of 10 mM Tris-HCl, pH 8.0, 100 mM NaCl, 0.1 mM EDTA on ice four times for 20 s.
Expression and Purification of Glutathione S-Transferase (GST) Fusion ProteinsDNA sequences coding for various regions of GRK5 were generated using the polymerase chain reaction and then used to replace a BamHI/SalI fragment in the vector pGEX-4T-2 (Pharmacia). The polymerase chain reaction-derived portions of the constructs were sequenced in their entirety using the dideoxy chain termination method. The GST-GRK5 fusion proteins were expressed in Escherichia coli and purified over glutathione-agarose using standard procedures (25). The purity of the proteins was >95% as determined by Coomassie Blue staining. Protein concentrations were determined by dye binding assay (Bio-Rad) using bovine serum albumin as a standard.
Interaction of GST Fusion Proteins with CalmodulinTo assess the ability of GST fusion proteins to block calmodulin-mediated activation of GRK5, 4 pmol of GRK5-DD was autophosphorylated in the presence of a 2 µM concentration of the indicated fusion protein and in the absence or presence of 0.1 µM calmodulin. Reactions were processed by gel electrophoresis as described above, and the level of autophosphorylation was determined by excising and counting the 32P-labeled bands. None of the GST fusion proteins significantly affected the basal autophosphorylation of GRK5-DD.
The binding of GRK5 and GST-GRK5 fusion proteins to calmodulin-agarose was performed in buffer B (20 mM Tris-HCl, pH 8.0, 100 mM NaCl, 1% Triton X-100, 0.01% SDS, 0.1 mg/ml ovalbumin, and either 0.1 mM CaCl2 or 2 mM EGTA). Ten pmol of GRK5 or fusion protein was incubated with 20 µl of calmodulin-agarose beads for 20 min in a total volume of 0.2 ml at 4 °C. The resin was pelleted, washed with 0.5 ml of buffer B, and bound proteins were eluted with two 100-µl aliquots of 20 mM Tris-HCl, 1 mM EDTA, 100 mM NaCl, 1% Triton X-100, 0.01% SDS, 0.1 mg/ml ovalbumin, 10 mM EGTA. The supernatants were combined, and the amount of bound and eluted protein was determined by immunoblotting using a rabbit polyclonal antibody generated against a GST fusion protein containing amino acids 98-136 of human GRK5 (17). This antiserum recognizes both GRK5 and all GST proteins. Approximately 50-60% of GRK5 was bound to calmodulin-agarose in the presence of Ca2+ and could be eluted with EGTA, whereas no binding was detected in the presence of EGTA.
Analysis of GRK Binding to Calmodulin on BIAcoreDirect binding of GRK to calmodulin was also assessed by perfusing solutions of either GRK5 or GST-GRK5 fusions over the surface of a BIAcore sensor chip containing calmodulin. To immobilize calmodulin, it was first biotinylated either at lysine residues using NHS-LC-Biotin (Pierce) or at a unique cysteine residue using Iodoacetyl-LC-biotin (Pierce). The calmodulin was desalted on Sephadex G-15 to remove free biotinylation reagent and then trapped on the surface of a sensor chip containing covalently attached streptavidin (Sensor chip SA5, BIACORE, Inc). This yielded ~2,000 relative units of calmodulin on the streptavidin chip which retained its activity for several days.
For analysis of calmodulin/GRK5 interaction, solutions of the kinase
were injected across chip surfaces containing either calmodulin,
streptavidin only, or another calcium-binding protein, recoverin. The
running buffer contained 20 mM Hepes, pH 7.4, 200 mM NaCl, 0.02% Surfactant P20 (BIACORE, Inc.), 0.01 mg/ml
bovine serum albumin, 0.1 mM -mercaptoethanol, 1 mM CaCl2. GRK5 and the GST-GRK5 fusions were
diluted in this buffer to a final concentration of 200 nM
prior to injection. The volume of injected sample was 40 µl, and the
flow rate was 10 µl/min. Experiments were performed on a BIAcore 2000 instrument with SPR data points collected at 1 Hz and the data analyzed
using BIAEvaluation 2.1 software (BIACORE, Inc.).
In an effort to
elucidate further the potential role of calcium in regulating GRK
function, we tested whether calmodulin could modulate the activity of
various GRKs. Our initial studies compared the effect of calmodulin on
COS-1 cell overexpressed preparations of GRK2, GRK5, and GRK6 to
phosphorylate light-activated rhodopsin. Protein extracts from control
COS-1 cells displayed very low rhodopsin phosphorylation activity,
whereas cells transfected with GRK2, GRK5, or GRK6 expression
constructs had a much higher level of phosphorylation (Fig.
1A). In the presence of calmodulin the
phosphorylation of rhodopsin by GRK5 and GRK6 was significantly
inhibited with IC50 values of ~0.25 µM for
GRK5 and ~0.7 µM for GRK6 (Fig. 1B). In
contrast, GRK2 was inhibited only at the highest concentration of
calmodulin tested (IC50 > 3 µM). Thus,
although all three GRKs tested were inhibited by calmodulin, GRK2 was
much less sensitive than GRK5 and GRK6.
Because the COS-1 cell extracts contain many other proteins that could
potentially influence the assay, we also studied the effect of
calmodulin on purified GRKs. Calmodulin effectively inhibited the
ability of GRK5 to phosphorylate rhodopsin with an IC50
~50 nM (Fig. 2). Calmodulin also inhibited
the activity of GRK2, although much less effectively (IC50
~2 µM) than GRK5, whereas the activity of GRK1 was only
modestly inhibited even at 10 µM calmodulin. The effect
of calmodulin on GRK5 was also completely dependent on the presence of
Ca2+ (data not shown). The higher IC50 values
observed for the COS-expressed GRKs versus the purified GRKs
might be because of the presence of additional calmodulin-binding
proteins in the cruder preparations which could bind calmodulin and
reduce its effective concentration.
The high sensitivity of GRK5 to inhibition by both
Ca2+/calmodulin and PKC (17) strongly suggests that GRK5
will not be involved in regulating receptors coupled to
Gq/11 and phospholipase C since these receptors promote
increased free calcium levels when activated, presumably leading to
inhibition of GRK5. Thus, even if GRK5 can phosphorylate such receptors
in vitro, it is unlikely that this would occur in intact
cells. This may explain why coexpression of
1b-adrenergic receptors with GRK5 results in enhanced
basal phosphorylation but no significant agonist-induced
phosphorylation of the receptor (26). Similarly, recent studies have
demonstrated that although GRK5 can phosphorylate myocardial type 1A
angiotensin II receptors in vitro (27), desensitization of
this receptor in transgenic mice overexpressing GRK5 was not affected
(28). In contrast, the lower affinity of calmodulin for GRK2 suggests that it would not be regulated by calmodulin in most cells, although calmodulin levels in brain are high (1-10 µM) (18).
Since GRK5 was more sensitive to inhibition by calmodulin than the other GRKs, we focused the remainder of the study on the GRK5/calmodulin interaction. To assess further the effect of calmodulin on the activity of GRK5, we utilized soluble substrates such as casein and phosvitin. Although phosphorylation of casein by GRK5 was not altered by calmodulin (data not shown), GRK5 phosphorylation of phosvitin was inhibited with an IC50 ~0.6 µM (Fig. 2C). The inhibition of phosvitin phosphorylation suggests that calmodulin interacts with regions of GRK5 which are likely involved in substrate binding. However, the ~10-fold reduced sensitivity of inhibition of phosvitin phosphorylation by calmodulin relative to rhodopsin phosphorylation implies that calmodulin may either more effectively inhibit GRK5 binding to receptor substrates, and/or it may also inhibit GRK5 binding to phospholipid.
Effect of Calmodulin on GRK5 Binding to MembranesUnlike the
other GRKs that utilize either covalent lipid modifications (GRK1, 4, and 6) or interaction with G protein subunits (GRK2 and 3) to
enhance binding to phospholipid membranes, GRK5 appears to interact
directly with phospholipids via regions rich in basic amino acids. GRK5
displays significant association with either phospholipid vesicles or
with rhodopsin-containing ROS membranes (11, 17, 29). When tested in a
direct binding assay, calmodulin was found to inhibit GRK5 binding to
ROS membranes significantly with an IC50 ~0.3-0.4
µM (Fig. 3, A and
B). However, this IC50 was some 6-8-fold higher
than the IC50 for inhibition of rhodopsin phosphorylation.
Indeed, at the highest calmodulin concentration tested, ~20% of the
kinase remained bound to the ROS membranes even though rhodopsin
phosphorylation was reduced >99% (compare Figs. 2B and
3B). These results suggest that calmodulin can directly
inhibit GRK5 interaction with receptor. The binding of GRK5 to
phospholipid vesicles was also inhibited by calmodulin (Fig. 3,
C and D). However, although this inhibition was
substantial at relatively low lipid concentrations (0.017 mg/ml), it
could be largely overcome at higher phospholipid (0.85 mg/ml). These results imply a competitive type of inhibition and taken together with
the rhodopsin studies suggest that calmodulin can directly compete for
both the lipid and receptor binding sites of GRK5.
Calmodulin Activates GRK5 Autophosphorylation
GRK5 appears to
be activated via a rapid phospholipid-stimulated autophosphorylation at
residues Ser484 and Thr485 (11, 29). To our
surprise calmodulin significantly enhanced the autophosphorylation of
GRK5 (Fig. 4A). In an attempt to further characterize this finding we studied the effect of calmodulin on the
autophosphorylation-defective mutant GRK5-DD, which has both
Ser484 and Thr485 mutated to aspartate (17).
Although autophosphorylation of GRK5-DD was not enhanced by
phospholipids, calmodulin still significantly enhanced the
autophosphorylation with an overall increase comparable to that seen
for wild type GRK5. These data indicate that interaction with
calmodulin results in increased autophosphorylation of GRK5 at sites
distinct from Ser484 and Thr485. Interestingly,
calmodulin also significantly activates autophosphorylation of GRK6,
but it has no effect on the autophosphorylation of GRK1 and GRK2 (data
not shown).
The EC50 for calmodulin activation of GRK5-DD autophosphorylation was ~40 nM (Fig. 4B), very similar to the IC50 for calmodulin inhibition of rhodopsin phosphorylation. Kinetic studies reveal that calmodulin increases the Vmax of autophosphorylation (by ~8-fold), and the affinity for ATP (Km is reduced from ~17 to ~5 µM) (Fig. 4C). To characterize initially the potential role of calmodulin-stimulated autophosphorylation in inhibition of GRK5 we performed time course experiments. These studies reveal that after 1 min, rhodopsin phosphorylation was inhibited 95-96% (~20-fold), whereas the stoichiometry of GRK5 autophosphorylation was only 0.15 mol/mol (Fig. 4D). Moreover, as GRK5 autophosphorylation increased from 0.15 to 0.72 mol/mol, the fold inhibition of rhodopsin phosphorylation remained unchanged. These results suggest that in the presence of calmodulin, autophosphorylation does not appear to play a major role in inhibition of GRK5 activity.
To characterize directly the effect of calmodulin-stimulated
autophosphorylation on GRK5 activity, GRK5 was incubated with ATP in
the presence or absence of calmodulin or PKC, purified by SP Sepharose
chromatography, and then assayed. Whereas phosphorylation of phosvitin
by PKC-phosphorylated GRK5 was inhibited 3-4-fold compared with wild
type GRK5, phosvitin phosphorylation by GRK5 autophosphorylated in the
presence of calmodulin was slightly increased (27 ± 5%) (Fig.
5A). In contrast, GRK5 phosphorylation of
rhodopsin was inhibited ~6-7-fold by either PKC phosphorylation or
calmodulin-stimulated autophosphorylation. To determine whether the
reduced receptor phosphorylation by autophosphorylated GRK5 was due to
a loss in either phospholipid or receptor binding, direct binding to
phospholipid vesicles and ROS was tested. GRK5 association with
phospholipids was not affected by calmodulin-stimulated autophosphorylation (57% bound with wild type GRK5 versus
59% with autophosphorylated GRK5; compare Figs. 3D and
5B). In contrast, calmodulin-stimulated autophosphorylation
dramatically reduced GRK5 binding to ROS (from 57 to 12%, compare
Figs. 3A and 5B). In addition, kinetic analysis
revealed that autophosphorylated GRK5 had a ~6-fold increased
Km and a ~2-fold reduced Vmax for rhodopsin phosphorylation compared with
wild type GRK5 (data not shown). Taken together, these data suggest
that calmodulin-stimulated autophosphorylation predominantly affects
GRK5 interaction with receptor without disrupting the catalytic
activity or association of GRK5 with phospholipids.
Our studies have revealed some similarities in the effects of calmodulin-stimulated autophosphorylation and PKC phosphorylation of GRK5 (17). In both cases phosphorylation inhibits GRK5 binding to ROS and phosphorylation of rhodopsin. However, several lines of evidence suggest that the site(s) phosphorylated by PKC and autophosphorylated in the presence of calmodulin are distinct. First, PKC phosphorylation inhibits phosvitin phosphorylation by GRK5, whereas calmodulin-stimulated autophosphorylation slightly activates phosphorylation of phosvitin. In addition, although PKC phosphorylation of GRK5 blocks binding of an antibody generated against residues 556-571 of GRK5 (17), calmodulin-stimulated autophosphorylation had no effect on antibody binding to GRK5 (data not shown). Identification of the calmodulin-stimulated autophosphorylation site is currently under investigation.
Whereas calmodulin-stimulated autophosphorylation directly inhibits rhodopsin phosphorylation by GRK5, the addition of calmodulin further inhibits GRK5 (Fig. 5, C and D). This demonstrates that calmodulin is still able to bind to autophosphorylated GRK5. However, the effect of calmodulin appeared to be neither additive nor synergistic with autophosphorylation since both forms of GRK5 (wild type and calmodulin-stimulated autophosphorylated) were inhibited comparably by 150 or 500 nM calmodulin.
Taken together, our data suggest that calmodulin binding to GRK5 can directly inhibit GRK5 activity and that this inhibition is not mediated primarily via kinase autophosphorylation. First, calmodulin inhibits GRK5 binding to phospholipid and ROS membranes even in the absence of ATP (Fig. 3), whereas the addition of ATP has no effect on binding. Second, calmodulin-stimulated autophosphorylation increases slightly the phosphorylation of phosvitin by GRK5 (Fig. 5A), whereas phosvitin phosphorylation is inhibited in the presence of calmodulin (Fig. 2C). Finally, whereas calmodulin-stimulated autophosphorylation inhibits GRK5 activity 6-7-fold (Fig. 5D), high concentrations of calmodulin (>1 µM) are able to inhibit the kinase activity >100-fold (Fig. 2B). Thus, although calmodulin-stimulated autophosphorylation may contribute to a higher sensitivity of GRK5 to calmodulin, the direct binding of calmodulin to GRK5 likely plays the major role in inhibiting GRK5 activity.
Based on our data we propose the following model for GRK5 regulation by calmodulin. At resting calcium concentrations GRK5 is active and able to phosphorylate agonist-occupied receptors. When a cell is stimulated and intracellular calcium levels rise, calmodulin binds to GRK5 and inhibits directly receptor phosphorylation. However, since calmodulin-stimulated autophosphorylation also inhibits GRK5 activity, the kinase should remain inhibited even when calcium levels go down and calmodulin dissociates from the enzyme. Presumably, GRK5 will eventually be dephosphorylated and return to its basal level of activity. Thus, calmodulin-stimulated autophosphorylation may prolong the inhibitory effect of a transient increase in intracellular calcium levels on GRK5. A similar regulatory cycle has been demonstrated for the calmodulin-dependent protein kinase, CaM-kinase II, where calmodulin binding and calmodulin-stimulated autophosphorylation activate rather than inhibit the kinase (18). GRK5 is the first example of an enzyme negatively regulated by calmodulin in this manner.
Identification of the Calmodulin Binding Site in GRK5Since
calmodulin interaction with GRK5 reduces the binding of GRK5 to both
phospholipid and receptor, we next focused on identifying the region of
GRK5 which interacts with calmodulin. Analysis of the GRK5 amino acid
sequence reveals that it contains several regions (in amino- and
carboxyl-terminal domains) with features typical for a calmodulin
binding site (i.e. mainly basic and hydrophobic residues
that form an -helix) (30). To identify the calmodulin binding domain
in GRK5 we generated several GST-fusion proteins containing various
regions of GRK5 and then assessed the ability of these proteins to
inhibit calmodulin-stimulated autophosphorylation of GRK5-DD. Neither
GST alone nor a GST fusion protein containing the carboxyl-terminal 102 amino acids of GRK5 had an effect on GRK5-DD autophosphorylation (Fig.
6A). However, a fusion protein containing
residues 1-200 of GRK5 blocked almost completely the stimulation of
autophosphorylation by calmodulin. Thus, GST-GRK5-(1-200) appears to
bind calmodulin thereby effectively sequestering it from GRK5-DD and
inhibiting calmodulin-stimulated autophosphorylation.
Since the amino-terminal domain of GRK5 contains four or five potential calmodulin binding domains, several additional constructs from this region were tested. A fusion protein containing the first 98 residues of GRK5 still effectively inhibited calmodulin-stimulated GRK5 autophosphorylation, whereas GST-GRK5-(50-200) had no effect. Constructs containing residues 20-49 or 20-39 of GRK5 also effectively blocked calmodulin activation of GRK5-DD autophosphorylation. As a further test of calmodulin binding, we measured direct binding of several GST-GRK5 fusion proteins to calmodulin-agarose. GST-GRK5-(1-200) and GST-GRK5-(20-49) bound to calmodulin-agarose in the presence of Ca2+ and could be eluted with EGTA, whereas GST-GRK5-(50-200) did not bind to calmodulin-agarose (data not shown).
The ability of GRK5 and several of the GST-GRK5 fusion proteins to bind calmodulin was also tested using SPR technology on a BIAcore instrument. This method detects directly the interaction among biological macromolecules because of an increase in the mass of the protein complex and a corresponding change in the refractive index of the solution close to the surface of the instrument's sensor chip (31). GRK5 bound to immobilized calmodulin in a calcium-dependent manner (Fig. 6B). Additional studies suggested a Kd of ~10 nM for this interaction, a value in good agreement with the EC50 for calmodulin-dependent inhibition of rhodopsin phosphorylation. In addition, the interaction of GRK5 with calmodulin was specific since no GRK5 binding was observed when another calcium-binding protein, recoverin, was tested. In contrast, under similar conditions, GRK1 does display specific calcium-dependent binding to recoverin.2 As expected, the 1-200- and 20-39-containing GST-GRK5 fusion proteins bound to calmodulin, whereas the 98-136 and 50-200 fusions did not.
Taken together, these data clearly indicate that the major calmodulin
binding domain is within residues 20-39 of GRK5. Although calmodulin
binding domains have no obvious consensus sequence, most of them adopt
a basic amphiphilic -helical structure that contains a large number
of positively charged residues as well as hydrophobic residues that
repeat with a three to four period (30). The region of GRK5 which binds
calmodulin has similar characteristics with a total of 9 basic and 5 hydrophobic residues (Fig. 7A). This portion
of GRK5 also has significant homology with the myristoylated
alanine-rich PKC substrate (MARCKS), a well characterized
calmodulin-binding protein (19). Helical wheel projection of this
region of GRK5 shows the segregation of basic and hydrophobic residues
to opposite sides of the helix, thereby making them available for
interaction with acidic and hydrophobic patches of calmodulin
(Fig. 7B).
Although the 20-39 amino acid region of GRK5 is substantially conserved within the GRK4 subfamily (GRK4, 5, and 6), it differs significantly from the corresponding regions of GRK1, 2, and 3 (Fig. 7A). These differences may account for the much higher affinity of GRK5 for calmodulin compared with GRK1 and GRK2. Similarly, the apparent higher affinity of GRK5 for calmodulin compared with GRK6 may indicate that the three nonconserved amino acid differences among these kinases within this region are involved in calmodulin binding. Alternatively, other protein determinants may also play a role in calmodulin interaction. For example, since GRK6 is palmitoylated it might bind more tightly to phospholipid membranes than GRK5 and thus might be inhibited less effectively by calmodulin. Recently residues 22-29 of GRK5 were suggested to be involved in the binding of phosphatidylinositol 4,5-bisphosphate (12). Since this region overlaps with the calmodulin binding domain of GRK5 it is possible that calmodulin can compete with phosphatidylinositol 4,5-bisphosphate for binding to GRK5. This is supported by our observation that calmodulin inhibits directly the GRK5 binding to phospholipid vesicles. Indeed, it has been shown for a number of proteins, such as GAP-43 (32), MARCKS (33), and nitric oxide synthase (34), that the calmodulin binding site is involved directly in phospholipid interaction. However, since the carboxyl-terminal region of GRK5 has also been implicated in phospholipid binding (11), this may explain why calmodulin does not inhibit completely the membrane binding of GRK5 (Fig. 3B).
GRK4 is the only member of the GRK family demonstrated to undergo alternative splicing (9, 35). Exons in both the amino-terminal and carboxyl-terminal domains of GRK4 can be alternatively spliced, resulting in a total of four different protein variants of GRK4, each of which appears capable of promoting receptor phosphorylation and desensitization (9). Interestingly, the region homologous to the calmodulin binding domain of GRK5 (residues 20-39) lies entirely within the first alternatively spliced exon of GRK4 (residues 18-49). This raises the intriguing possibility that different splice variants of GRK4 could be differentially regulated by calmodulin (i.e. variants lacking this exon would not be inhibited by calmodulin). In addition, although there is no similarity in the exon-intron organization of the GRK2 and GRK4 genes (9, 36), organization of the GRK4, 5, and 6 genes appears to be highly conserved.3 This creates the possibility of calmodulin-insensitive splice variants of GRK5 which are expressed in certain cell types or during some stage of development.
While this manuscript was in preparation, Chuang et al. (37)
also reported that calmodulin inhibits GRK-mediated phosphorylation of
rhodopsin with IC50 values virtually identical to those
observed in our experiments. However, one important difference between the two studies was that Chuang et al. included G protein
subunits in their assays to activate GRK2, whereas our studies
were performed in the absence of G
. Since
G
can interact directly with calmodulin (38), these
authors speculated that calmodulin inhibition of GRK2 could be mediated
via sequestering G
from GRK2. However, since
the IC50 values for calmodulin inhibition of GRK2
activity were identical in both studies, this suggests that the
inhibition is the result of a direct interaction between calmodulin and
GRK2. Moreover, our preliminary studies also demonstrate that GRK2
binds directly to calmodulin as detected by SPR (data not shown).
In conclusion, our studies demonstrate that calmodulin is a potent
inhibitor of GRK activity with a selectivity for GRK5 (IC50 ~50 nM) > GRK6 GRK2 (IC50 ~2
µM)
GRK1. Calmodulin inhibition of GRK5-mediated
receptor phosphorylation is caused by inhibition of kinase interaction
with both receptor and phospholipid, and the major region of calmodulin
interaction lies within amino acids 20-39 of GRK5, a region that is
highly conserved in GRK4, 5, and 6. These studies also further
establish the role of Ca2+ in the regulation of GRK
activity. While GRK1 is inhibited by Ca2+/recoverin and
GRK5 and GRK6 (and likely GRK4) are potently inhibited by
Ca2+/calmodulin, the effect of Ca2+ on GRK2
(and likely GRK3) will depend on the relative contribution of
PKC-mediated activation and calmodulin-mediated inhibition of these
kinases.
We thank Dr. Daniel Ladant for initial insight into the effect of calmodulin on GRK2.