G protein-coupled receptor kinase 4γ interacts with inactive Gα_s and Gα₁₃

Abstract

G protein-coupled receptors (GPCRs) are regulated by multiple families of kinases including GPCR kinases (GRKs). GRK4 is constitutively active towards GPCRs, and polymorphisms of GRK4γ are linked to hypertension. We examined, through co-immunoprecipitation, the interactions between GRK4γ and the Gα and Gβ subunits of heterotrimeric G proteins. Because GRK4 has been shown to inhibit Gα_s-coupled GPCR signaling and lacks a PH domain, we hypothesized that GRK4γ would interact with active Gα_s, but not Gβ. Surprisingly, GRK4γ preferentially interacts with inactive Gα_s and Gβ to a greater extent than active Gα_s. GRK4γ also interacts with inactive Gα₁₃ and Gβ. Functional studies demonstrate that wild-type GRK4γ, but not kinase-dead GRK4γ, ablates isoproterenol-mediated cAMP production indicating that the kinase domain is responsible for GPCR regulation. This evidence suggests that binding to inactive Gα_s and Gβ may explain the constitutive activity of GRK4γ towards Gα_s coupled receptors.

G protein-coupled receptor kinases (GRKs) comprise a family of seven proteins that possess an amino terminal regulator of G protein signaling homology (RH) domain and a serine/threonine kinase domain similar to protein kinase C [1]. GRKs initiate homologous desensitization through phosphorylation of the third cytoplasmic loop or carboxy terminus of ligand bound/activated target receptors [1]. This desensitization of the activated receptor is crucial in regulating signal transduction as well as ligand-mediated receptor endocytosis. The GRK1-like family, GRK1 and GRK7, are found in rod and cone cells and are associated with opsins [2]; the GRK2-like family, GRK2 and GRK3 (also known as β-adrenergic receptor kinase 1 and 2, respectively), are ubiquitously expressed and associate with Gα_q/11-coupled receptors [1]; and finally, the GRK4-like family comprises GRK4, GRK5, and GRK6, which compared to the other GRKs little is known about their biology [3]. GRK5 and GRK6 are expressed in multiple tissues [1], whereas GRK4 is expressed in the testis [4], kidney tubular cells [5], brain [6], and myometrium [7]. In this study we focused on GRK4.

GRK4 was first identified during positional cloning of the Huntington’s disease locus [8]. Early in vitro studies demonstrated that GRK4 phosphorylates the β₂-adrenergic receptor (β₂AR) in a PIP₂-dependent manner [4]. Additionally, both GRK4 and GRK2, significantly decrease luteinizing hormone/chorionic gonadotropin receptor-mediated cAMP production [4]. These data suggest that GRK4 is similar to GRK2 in its action; however, further studies demonstrate that in a comparable system GRK4, but not GRK2, phosphorylates the β₂AR in an agonist-independent manner resulting in increased recycling of the β₂AR [9]. Additionally, GRK4, and particularly GRK4γ, is the only GRK that appears to be constitutively active [9]. In support of this, recent studies demonstrate that GRK4α phosphorylates the D1 dopamine receptor (D1R) independently of D1R agonists [10].

There are four published splice variants of GRK4 [4]. GRK4α is the longest transcript with 16 exons encoding a protein of 578 amino acids. GRK4β is missing exon 2 and produces a protein of 546 amino acids; the 32 missing amino acids occur after Gln¹⁷ of GRK4α, which, based upon the GRK6 crystal structure [11], correspond to the first two alpha helixes α0 and α1 of the RH domain. GRK4γ, the first GRK4 reported [8], lacks exon 15 and is 532 amino acids in length; the 46 missing amino acids occur after Glu⁵¹⁵ and correspond to less than half of alpha helix α10 and all of α11 of the RH domain as well as a portion of the unstructured region following the RH domain [11]. GRK4δ, a 500 amino acid protein, is missing both exons 2 and 15. None of the alternative splicing events eliminate GRK4 activity [4]; however, most studies of GRK4 are conducted with GRK4α. Polymorphisms in GRK4 (numbering based on GRK4α: R65L, A142V, and A486V) are associated with hypertension [5; 12; 13]. Two of the polymorphisms, R65L and A142V, lie within the RH domain and transgenic mice carrying the polymorphisms of GRK4γ are hypertensive compared to mice carrying human wild-type GRK4γ [5] (personal communication with P.A. Jose). The mechanism of action of GRK4γ, with regard to hypertension, is proposed to be hyper-phosphorylation/desensitization of the dopamine D1R [5]. This desensitization prevents excretion of salt and relieves the renal dopaminergic-mediated inhibition of the renin-angiotensin system [14].

We examined which of the Gα subunits interact with GRK4γ in an effort to understand the biochemistry of GRK4γ and eventually examine the polymorphisms in the RH domain associated with hypertension. We hypothesized that GRK4γ would interact with active/GTP-bound Gα_s but not inactive/GDP-bound Gα_s, and also possibly interact with other G proteins in a GTP-dependent manner. Counter-intuitively, the data indicate that GRK4γ interacts preferentially with inactive/GDP-bound Gα_s.

Materials and Methods

Materials

All agonists and drugs were obtained from Sigma-Aldrich (St. Louis, MO). The AU1 antibody was obtained from Covance (Berkley, CA). All other antibodies and materials for immunoprecipitation were obtained from Santa Cruz (Santa Cruz, CA). The following lysis buffers were used: TX-100 based buffer (1% Triton X-100, 100mM NaCl, 20mM Tris pH 7.5, 2mM EDTA, 10mM MgCl₂, 10mM NaF, 40mM β-glycerol phosphate, and 1mM PMSF), RIPA buffer (TX-100 based buffer plus 0.1% SDS and 0.5% deoxycholate), and detergent-free buffer (TX-100 buffer without Triton X-100) with cells passed through a 27 gauge needle 10 times. The cAMP EIA Kit was obtained from Cayman Chemical (Ann Arbor, MI) and used per the manufactures specifications.

cDNA Constructs

GRK2 cDNA was purchased from ATCC (Manassas, VA). Human wild-type GRK4γ and GRK2 were subcloned into the expression vector pcDNA4/TO (Invitrogen). Kinase dead (K216M, K217M) GRK4γ [15] was created with the Stratagene (La Jolla, CA) quick change mutagenesis kit. The wild-type and constitutively active QL mutant G proteins were gifts of Dr. J.S. Gutkind. Wild-type G proteins were subcloned into pEF-AU1, a derivative of pCEFL, obtained from Dr. J.S. Gutkind.

Cell Culture and Transfection

HEK293T cells were grown in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% (v/v) FBS and 100 units/ml of penicillin/streptomycin (Invitrogen, Carlsbad, CA) in 60mm culture dishes coated with poly-D-Lysine (Sigma). HEK293T cells were transiently transfected using Qiagen’s PolyFect Transfection Reagent (Germantown, MD). Functional assays were conducted using immortalized renal proximal tubule cells (RPTCs) isolated from Wistar-Kyoto (WKY) rats (a gift from Dr. U. Hopfer). PTCs were cultured in DMEM/F12 containing 5% (v/v) FBS, 10 ng/ml epithelial growth factor (EGF), 4μg/ml dexamethasone, ITS+ diluted at 4 ml per bottle of media, and 100 units/ml of penicillin/streptomycin. EGF and ITS+ premix were obtained from BD Biosciences (Bedford, MA). PTCs were grown in 60mm culture dishes for the cAMP assay and transfected utilizing the FuGene 6 protocol (Roche, Indianapolis, IN).

Preparation of Cellular Extracts and Co-IP

Transfected HEK293T cells were serum starved overnight (at least 8 hours) prior to all experiments. For experiments with the AU1-tagged G proteins, all samples and buffers were supplemented with 10μM GDP. To activate Gα proteins, samples were treated with 100μM AlCl₃ and 10mM NaF for 15 minutes at 37°C to generate AlF₄⁻,. All buffers used with the AlF₄⁻ treated samples contained 100μM AlCl₃ and 10mM NaF. For experiments with the wild-type and QL mutant Gα subunits, GDP and AlF₄⁻ were not used. Medium was removed and cells were lysed in one of the three lysis buffers, and clarified by centrifugation for 5 minutes at 4°C. An aliquot of the supernatant was reserved for analysis of the total cell lysate and the remaining sample was used for co-immunoprecipitation with an antibody directed against the AU1 tag, Gα subunit, or Gβ common for one hour at 4°C. Protein A/G PLUS-Agarose beads (Santa Cruz) were added for 30 additional minutes. Immune complexes were washed with PBS containing 1% NP-40 and 1mM PMSF, followed by SDS-PAGE on a 9% gel and western blot analysis.

Results and Discussion

Examination of common Gα subunits

Before conducting any co-immunoprecipitation (co-IP) studies, the levels of endogenous and transfected GRK4γ were examined in 293T cells. No endogenous GRK4γ was detected by western blot in 293T cells. Initial co-IP studies were conducted with a panel of the commonly used heterotrimeric G protein alpha subunits: s, i1-3, q, 11, 12, and 13 (Fig. 1). The Gα subunits were subcloned into pEF-AU1 in order to conduct the analyses in parallel using a single antibody for detection. Each Gα subunit was examined in the active, AlF₄⁻ treated, conformation (Fig. 1a and 1b) and inactive conformation (Fig. 1c and 1d). After stimulation with AlF₄⁻, GRK4γ co-immunoprecipitates with AU1-Gα_s and to a lesser extent AU1-Gα₁₃ (Fig. 1a). Surprisingly, when the experiment is conducted without AlF₄⁻, representing the inactive Gα subunits, GRK4γ also co-immunoprecipitates with AU1-Gα_s and to a lesser extent AU1-Gα₁₃ (Fig. 1c). These results were opposite to our initial hypothesis.

Figure 1.

GRK4γ interacts with G-proteins in the active and inactive state. 293T cells were transfected with an AU1-tagged Gα subunit, and GRK4γ (a & c) or GRK2 (b & d). The G-proteins were stimulated as indicated and immunoprecipitated with AU1 antibody. The first of the three blots in each set is the co-immunoprecipitation, and the lower two blots are the controls to indicate that the GRK and Gα subunits were expressed. Note, Gα_q and Gα₁₁ are not well expressed, which is why the IP blot was used instead of total cell lysate (TCL) in panels a, b & d. Each experiment was conducted a minimum of two times.

To confirm that the conditions represent active and inactive Gα subunits, we conducted identical studies with GRK2 (Fig. 1b and d). As shown in Figure 1b GRK2 interacts with AlF₄⁻ treated (active) AU1-Gα_q. Moreover, in samples lacking AlF₄⁻ (inactive) GRK2 is not co-immunoprecipitated with any of the AU1-Gα subunits (Fig 1d). These data partially fit what is known about GRK2 and G protein interactions[16; 17]; however, GRK2 should also interact with Gα₁₁. Most likely this interaction is not observed due to low expression of AU1-Gα₁₁. The expression levels of the various G proteins are significantly different, and Gα_q and Gα₁₁ are expressed the least (Figure 1). Therefore, we utilized wild-type Gα_q and Gα₁₁ and constitutively active Q209L Gα_q and Q209L Gα₁₁ in co-IP experiments with GRK2. As previously reported [16; 17], GRK2 interacted with the active form of Gα_q and Gα₁₁ (data not shown). Therefore, IP is an accurate measure of GRK4γ interactions with active and inactive G proteins.

Interaction of active and inactive Gα_s with GRK4γ

Due to the experimental setup of the Gα subunit panel, direct comparison of the inactive and AlF₄⁻ activated blots is not appropriate. To directly compare each Gα subunit in the active and inactive conformation, two series of experiments were conducted. First, experiments similar to those shown in figure 1 were conducted; each Gα subunit in the active and inactive conformation was run on the same gel for direct comparison of their interaction with GRK4γ. As shown in figure 2a, GRK4γ co-immunoprecipitates with inactive AU1-Gα_s to a greater extent than with AlF₄⁻ activated AU1-Gα_s. Second, in order to avoid artifacts that may arise from using tagged constructs and AlF₄⁻, untagged wild-type and constitutively active Q213L Gα_s subunits were expressed to obtain the inactive and active state of the G protein, respectively. GRK4γ robustly co-immunoprecipitated with inactive Gα_s, but little interaction was found with Q213L Gα_s (Fig. 2b). These data confirm the AU1-tagged/AlF₄⁻ treatment results. Thus indicating that the observed interaction between GRK4γ and inactive Gα_s is not an artifact of the tagged Gα_s, AlF₄⁻, or the antibodies used. However, in all experiments presented thus far, the co-IP was conducted in Triton X-100 based lysis buffer. To determine if the interaction of inactive Gα_s with GRK4γ is an artifact of the lysis buffer, we repeated the studies shown previously with three different lysis buffers (Fig. 2c). GRK4γ interacts with wild-type Gα_s to a much greater extent than Q213L Gα_s in all detergents examined, including detergent-free lysis. Overexposure of the IP blot indicates that native Gα_s will also pull down the overexpressed GRK4γ. Similar experiments were conducted with AU1-Gα_s with and without AlF₄⁻, and showed GRK4γ interaction with inactive Gα_s (data not shown). Therefore, the data are not an artifact of the lysis buffer. GRK4γ preferentially interacts with inactive Gα_s.

Figure 2.

GRKγ interacts preferentially with inactive Gα_s and Gβ. 293T cells were transfected with GRK4γ and either AU1-tagged Gα_s (a), or wild-type and constitutively active Gα_s (b & c). AU1 tagged Gα_s was stimulated with AlF₄⁻. 293T cells were lysed with either a triton X-100 (TX-100) based lysis buffer, a RIPA buffer, or a detergent-free buffer to examine the effects of the lysis buffer (c). Endogenous Gβ is co-immunoprecipitated with GRK4γ by Gα_s (d), conversely endogenous Gβ co-immunoprecipitates GRK4γ (e). In all panels the upper blots are the co-immunoprecipitation and the lower blots are the control total cell lysates (TCL). Each experiment was conducted a minimum of two times

Interaction of an RH-domain containing protein with an inactive Gα subunit has never been described. As stated, these results run counter to our hypothesis. Given that this is an overexpression system, and 293T cells are known to generate copious amounts of protein, we investigated the interaction of GRK4γ, Gα_s, and Gβ subunits. Gβ subunits are bound to inactive Gα subunits. If GRK4γ displaces Gβ from Gα_s, then it is likely that the binding is an artifact of overexpression. Using endogenous Gβ, overexpressed GRK4γ, and overexpressed wild-type Gα_s, we conducted co-IP studies with antibodies to Gα_s/olf (Fig. 2d) and Gβ common (Fig. 2e). In both cases GRK4γ was pulled down only in the presence of overexpressed Gα_s. Gβ was pulled down with the antibody to Gα_s/olf (Fig. 2d). Similarly, Gα_s was pulled down with the antibody to Gβ common, and GRK4γ appears to increase the pull down of Gα_s (Fig. 2e). Because there is no apparent displacement of Gβ by GRK4γ, the interaction is most likely specific. The Gβ data are difficult to interpret because the antibody is interacting with four of the five Gβ variants, which will pull down all the Gα subunits. Alternatively, it is unknown if Gβ binds to GRK4γ. Gβ binds to the plextrin homology (PH) domain of GRK2 [17], however the GRK4 family members do not contain a PH domain [11]. Although it is unlikely that Gβ directly binds to GRK4γ, the experiments conducted here cannot discriminate between direct binding and indirect binding through Gα_s or Gα₁₃. However, it is clear that GRK4γ, Gα_s, and Gβ are bound together in one complex.

Interaction of active and inactive Gα₁₃ with GRK4γ

Experiments similar to those presented in figures 2a and 2b were conducted with all Gα subunits examined in figure 1. Initially weak interactions between GRK4γ and many of the Gα subunits were observed, but only the AU1-Gα₁₃ interaction with GRK4γ proved to be specific when using the wild-type and QL mutants. This supports previous findings that GRK4 does not interact with Gα_q or Gα₁₁ [18]. In accordance with figure 1, GRK4γ is co-immunoprecipitated with inactive AU1-Gα₁₃ and to a lesser extent when treated with AlF₄⁻ (Fig. 3a). Repetition with wild-type and constitutively active Q226L Gα₁₃ indicated that inactive Gα₁₃ binds to GRK4γ significantly more than active Gα₁₃ (Fig. 3b). To confirm these results, the wild-type and Q226L Gα₁₃ co-IP was repeated under detergent-free conditions. As shown in figure 3c, inactive Gα₁₃ co-immunoprecipitates with GRK4γ, whereas constitutively active Gα₁₃ does not interact with GRK4γ. Furthermore, GRK4γ is co-immunoprecipitated with Gβ only in the presence of overexpressed Gα₁₃ (Fig. 3d) similar to what is seen with Gα_s (Fig 2e). However, Gβ pulled down an equal amount of Gα₁₃ in all conditions suggesting that the system was saturated. As noted previously, the binding order cannot be determined in such an assay; however, the data indicate that GRK4γ, Gα₁₃, and Gβ are in a complex. However, the interaction of GRK4γ with Gα₁₃ is weaker than that of GRK4γ with Gα_s (Fig. 1a and 1c).

Figure 3.

GRK4γ interacts with inactive Gα₁₃. 293T cells were transfected with GRK4γ and either AU1 tagged Gα₁₃ (a), or wild-type and constitutively active Gα₁₃ (b & c). The AU1 tagged Gα₁₃ was stimulated with AlF₄⁻. To test if the lysis buffer had any effects, 293T cells were lysed in detergent-free buffer (c). Endogenous Gβ co-immunoprecipitates GRK4γ and Gα₁₃ (d). In all panels the upper blots are the co-immunoprecipitation and the lower blots are the control total cell lysates. Each experiment was conducted twice.

Function of GRK4γ

Recent structural studies indicate that the interaction of Gα_q with the RH domain of GRK2 occurs in a manner distinct from Gα_q/RGS interactions in that the interaction is more like an interaction with an effector not a RGS [17]. GRK2 does not increase the intrinsic catalytic rate of Gα_q, which is proposed to be a general method of action of GRKs. The RH domain of GRK6, a member of the GRK4 subfamily, structurally resembles GRK2 suggesting that the mechanisms of action and areas of Gα subunit binding are similar [11]. However, the area of GRK2 that interacts with Gα_q, α5-α6 helixes of the bundle sub-domain of the RH domain, is highly diverged in GRK6 (3.1% identity and 34.4% similarity). Furthermore, the predicted α5-α6 helixes of GRK4 has a 15.5% identity and 28.1% similarity with GRK2, and GRK4 and GRK6 only share 25% identity and 59.4% similarity over the same sequence. Although the α5-α6 region of GRK4 and GRK6 may not be the region that interacts with Gα subunits, the absence of significant similarity suggests that the interaction of the Gα subunit with its respective GRK does not necessarily have one unified mechanism.

In order to examine the physiological consequence of the interactions with inactive Gα subunits, functional assays were conducted to test the effects of GRK4γ and kinase dead K216M, K217M GRK4γ (KD GRK4γ) on Gα_s- and Gα₁₃-mediated cellular signaling. We generated KD GRK4γ as previously reported [15] (Fig 4a). As a control, we examined the interaction of KD GRK4γ with wild-type Gα_s (Fig. 4b). KD GRK4γ interacts identically to wild-type GRK4γ (Fig. 2d). Similar studies were conducted with Q213L Gα_s, wild-type Gα₁₃, and Q226L Gα₁₃ that demonstrated KD GRK4γ interacts with the inactive Gα subunits (data not shown). Thus, the dual lysine to methionine mutation can be used to discriminate between kinase activity driven effects of GRK4γ and interactions with Gα subunit driven events.

Figure 4.

Functional properties of GRK4γ. Kinase dead (KD) GRK4γ was generated through oligonucleotide-directed mutagenesis (a). This mutation does not abrogate binding to inactive Gα_s (b). cAMP production was measured in WKY RPTCs transfected with cDNA expression vectors, indicated at the base of the histogram, and stimulated for 30 min with Isoproterenol (n = 4). Bars with different letters are significantly different (p < 0.05) via ANOVA with Tukey-Kramer multiple-comparison post hoc test.

To examine Gα_s-mediated signaling, we utilized WKY RPTCs; one of the few cells to express endogenous GRK4 [4–7]. WKY RPTCs were transfected with empty vector, GRK4γ, KD GRK4γ, or GRK2 and then stimulated with vehicle, or 30 nM or 10 μM Isoproterenol for 30 min in the presence of 500 μM IBMX, a phosphodiesterase inhibitor. Intracellular cAMP measurements indicated that 30 nM Isoproterenol is saturating, thus, only this data is shown (Fig. 4c). All data were normalized to the control empty vector transfection. GRK2 served as a positive control. No effects were observed in unstimulated RPTCs. Stimulation doubled cAMP levels in KD GRK4γ and vector transfected RPTCs, but had no significant effect in GRK4γ and GRK2 transfected RPTCs. An independently conducted experiment utilizing 1 μM fenoldopam, a partial D1-like receptor agonist [19], and Isoproterenol confirmed GRK4γ’s inhibitory action, and revealed that KD GRK4γ can enhance fenoldopam-mediated cAMP accumulation (data not shown). Fenoldopam had a marginal effect on the RPTCs except when KD GRK4γ was present, suggesting that KD GRK4γ acts in a classic dominant negative fashion. Experiments were conducted to examine Gα₁₃-mediated events; however, these studies were inconclusive.

Summary and Conclusion

A caveat of these studies is that they are conducted with GRK4γ. The gamma isoform is missing a portion of the carboxy terminus of full-length GRK4. Based on the co-crystallization of GRK2, Gα_q, and Gβ, the missing portion of GRK4γ is most likely not directly involved in the interaction with the Gα subunit [17]. However, this does not preclude a variation in protein folding that alters GRK4γ-Gα interactions. Therefore, these studies should not be extended to all GRK4 variants or the other GRK4 family members.

GRK4 is the most active GRK under basal conditions, and GRK4γ appears to have the lowest level of receptor-mediated activity [9]. This may, in part, be explained by our results indicating that GRK4γ binds preferentially to inactive Gα_s. Binding of GRK4γ to the inactive receptor-Gα complex would allow GRK4γ greater access to the G protein-coupled receptor, and thus could result in increased phosphorylation of the inactive receptor. Moreover, the apparent lower affinity for active Gα_s may also explain the aforementioned low level of receptor-mediated activity. GRK4γ and GRK2 functionally inhibit Gα_s-mediated accumulation of cAMP (Fig. 4). Thus, in RPTCs, GRK4γ functions as hypothesized, but through a novel mechanism. This may be a cell type specific response since similar studies show little effect of wild-type GRK4γ in regulating dopamine-mediated cAMP production in CHO cells [5].

In conclusion, GRK4γ binds to inactive Gα_s > inactive Gα₁₃ ≫ active Gα_s ≫ active Gα₁₃, yet still appears to function normally by inhibiting receptor-mediated signaling in a physiologically relevant cell type. These phenomena may account for the previously reported increased basal activity of GRK4γ. The molecular mechanism mediating this interaction requires further elucidation. Once the mechanism of wild-type GRK4γ binding to Gα_s is determined, the effects of the polymorphisms in the RH domain associated with hypertension can then be clearly examined.