Differential Targeting of Gßγ-Subunit Signaling with Small Molecules

See allHide authors and affiliations

Science  21 Apr 2006:
Vol. 312, Issue 5772, pp. 443-446
DOI: 10.1126/science.1120378


G protein βγ subunits have potential as a target for therapeutic treatment of a number of diseases. We performed virtual docking of a small-molecule library to a site on Gβγ subunits that mediates protein interactions. We hypothesized that differential targeting of this surface could allow for selective modulation of Gβγ subunit functions. Several compounds bound to Gβγ subunits with affinities from 0.1 to 60 μM and selectively modulated functional Gβγ-protein-protein interactions in vitro, chemotactic peptide signaling pathways in HL-60 leukocytes, and opioid receptor–dependent analgesia in vivo. These data demonstrate an approach for modulation of G protein–coupled receptor signaling that may represent an important therapeutic strategy.

The βγ subunits of heterotrimeric guanine nucleotide binding proteins (G proteins) are released upon ligand activation of G protein–coupled receptors (GPCRs). Free Gβγ subunits bind and regulate multiple target proteins within the cell—including phospholipase C (PLC) β2 and PLC β3, phosphoinositide 3 kinase (PI3K) γ, adenylyl cyclase, N-type Ca2+ channels, and inwardly rectifying K+ channels—and mediate physiological processes such as neutrophil chemotaxis, vascular cell proliferation, and cardiac chronotropy (1, 2).

To investigate the molecular nature of Gβγ-target recognition, we screened random-peptide phage display libraries for binding to Gβ1γ2 and identified a series of peptides that bound to a single preferred protein-protein interaction surface (“hotspot”) on Gβ (3). One of the peptides (SIRK) blocked Gβγ-dependent regulation of PLCβ2 and PI3Kγ but not regulation of type I adenylyl cyclase or N-type Ca2+ channels, thus demonstrating the potential for selective targeting of Gβγ signaling. The crystal structure of Gβ1γ2 in a complex with a SIRK peptide derivative (SIGK) reveals the preferred interaction surface as a region overlapping the Gα-switch II domain binding surface on top of the Gβ propeller (46). Subclasses of peptides appeared to bind distinct subsurfaces of the hotspot and differentially affect G protein subunit interactions. Many Gβγ effectors also use diverse mechanisms for binding within this surface (7, 8), and because peptides showed some selectivity, we hypothesized that small organic molecules also might selectively modulate Gβγ-target interactions.

We used FlexX virtual screening software (9) in the Sybyl molecular modeling package to dock 1990 compounds in the chemical diversity set from the National Cancer Institute (NCI) to the interaction hotspot of Gβ1γ2. The diversity library was designed to contain chemical core structures representative of the larger 250,251-compound library from NCI. The docked models were ranked using five scoring functions in C-Score: D-score, G-score, F-score, Chemscore, and PMF-score (10). Two consensus scores that equally weight the five scoring functions were also used to rank the compounds. The top 1% of compounds from each scoring function (85 compounds total) were tested for their ability to compete with phage-displaying SIGK for binding to biotinylated Gβ1γ2 (bGβ1γ2) in an enzyme-linked immunosorbent assay (ELISA) (3). Nine of the 85 compounds inhibited SIGK binding with median inhibitory concentration (IC50) values ranging from 100 nM to 60 μM (Fig. 1B and table S1).

Fig. 1.

Small molecule binding to the hotspot on Gβγ. (A) Structure of SIGK bound to Gβγ at the hotspot. Amino acids within 6.5 Å of SIGK were targeted with the FlexX module of Sybyl virtual docking software. (B) Competition ELISA data for three of the compounds identified in the virtual screen. Compounds were tested for their ability to inhibit binding of a phage displaying the peptide SIGKAFKILGYPDYD (3, 4). M119 (NSC119910) (squares); M109 (NSC109208) (triangles); M117 (NSC117079) (circles). Data for all the binding compounds are summarized in table S1. (C) Structures of representative βγ-binding compounds. (D) Competition of M119 for interactions between Gαi1 and Gβ1γ2. F-αi1 and M119 were simultaneously added to bGβ1γ2 immobilized on streptavidin beads. The amount of bead-based fluorescence was assessed by flow cytometry as described (11, 12).

One compound, M119, with a high apparent affinity for bGβ1γ2 (ELISA IC50 = 200 nM) (Fig. 1, B and C, and table S1) was selected as a lead to define structure-activity requirements (SAR) for binding to Gβ1γ2. Twenty-one compounds from the NCI library with similar structures to M119 were tested for relative Gβ1γ2 binding affinities (Fig. 1C and table S2). For example, the apparent affinity of M119B for bGβ1γ2 is Embedded Image that of M119, with the key chemical difference being the loss of two hydroxyl groups. Thus, specific chemical characteristics may be required for interactions with the Gβγ hot-spot. We next tested whether M119 could disrupt protein interactions with a bona fide Gβγ binding partner, Gαi1. The hotspot for protein interaction overlaps with the Gα switch II binding surface on Gβγ. The overall Gαi1-βγ interaction surface spans 1800 Å2 (5, 6), and the dissociation constant (Kd) for Gαi1 binding to Gβγ is ∼1 nM (11). M119 competed with fluorescein isothiocyanate (FITC)–αi1 (Fαi) for binding to bGβ1γ2 with an IC50 value of 400 nM (Fig. 1D). However, unlike SIGK and related peptides that bind to this surface (12), M119 did not promote dissociation of Gαi from Gβγ (fig. S1).

FlexX docking software predicted that compounds M201 and M119 (Fig. 1C) bound to distinct subsurfaces in the hotspot, but M201 did not compete for SIGK binding. Nevertheless, we tested M119 and M201 in in vitro reconstitution assays of Gβγ-dependent activation of PLCβ2, PLCβ3, and PI3Kγ and binding to GRK2. M119 attenuated Gβ1γ2-dependent activation of PLCβ2 (IC50 value of 3 μM), PLCβ3, and PI3Kγ (Fig. 2, A to C, left panels). M201, on the other hand, did not affect PLCβ2 activation by Gβ1γ2 but potentiated Gβ1γ2-dependent activation of both PLCβ3 and PI3Kγ (Fig. 2, A to C, right panels). M119 also inhibited direct binding of bGβ1γ2 to PLCβ2 and PLCβ3, whereas M201 did not block binding of PLCβ2 and enhanced binding of PLCβ3 to Gβ1γ2 (fig. S2). M119 and M201 both inhibited GRK2 binding to bGβ1γ2 with similar IC50 values of approximately 5 μM (Fig. 2D). A weakly binding compound M119B (Fig. 1C and table S2) did not have effects in these assays (fig. S3). These data suggest that both M201 and M119 bind to Gβγ but differentially modulate Gβγ interactions with effectors. These are only two of multiple diverse compounds identified, which suggests the potential for multiple modes of Gβγ-dependent target modulation by these small molecules.

Fig. 2.

Differential effects of M119 and M201 on βγ-dependent regulation of downstream targets. (A) Effects of M119 and M201 (NSC201400) on Gβγ-activation of PLCβ. Purified PLCβ2 (0.25 ng) was assayed in the presence (triangles) or absence (squares) of 100 nM purified Gβ1γ2. (B) Effects of M119 and M201 effects on purified PLCβ3 (0.5 ng) activity in the presence (triangles) or absence (squares) of 100 nM purified Gβ1γ2. (C) Effects of M119 and M201 on activation of PI3Kγ by Gβ1γ2. Assays contained 10 ng of purified p101/p110 PI3Kγ heterodimer with or without 100 nM purified Gβ1γ2. Left,: (triangles) 100 nM Gβ1γ2 or (squares) no βγ. (D) Effects of M119 and M201 on Gβγ-GRK2 interactions. M119 or M201 and 25 nM purified GRK2 were added simultaneously to 250 nM bGβ1γ2 immobilized on streptavidin agarose. Associated GRK2 was detected with antibody to GRK2. Data are representative experiments [mean ± SEM, except (D)] and were repeated at least three times each. The concentrations of compound required for these assays are apparently higher than predicted by the phage ELISA assays, but may reflect the different assay conditions and protein concentrations required for each assay.

To test the effects of differentially targeting Gβγ on GPCR signaling in intact cells, M119 (and a similar compound, M158C) (Fig. 1C and table S2) and M201 were tested for their ability to modulate fMLP receptor–dependent signaling in differentiated HL-60 leukocytes. The fMLP receptor couples to Gi in these cells and activates PLCβ2 (PLCβ3 is a minor isoform in these cells), PI3Kγ, and ERK through Gβγ signaling (13, 14). Pretreatment of differentiated HL-60 cells with M119 and M158C (Fig. 3A and fig. S4A), but not M201 (Fig. 3B and fig. S4B), attenuated fMLP-induced Ca2+ increases. M119 had no effect on carbachol-dependent increases in Ca2+ in HEK293 cells stably expressing the Gq-linked M3-muscarinic receptor, confirming a specific effect of M119 on Gβγ-dependent Ca2+ mobilization (fig. S4C). fMLP-dependent GRK2 translocation to the membrane fraction of HL-60 cells, on the other hand, was substantially inhibited by incubation with either M119 or M201 (Fig. 3C). Thus, M119 and M201 differentially modulate PLCβ2 regulation by Gβγ, yet both inhibit GRK2 binding in intact cells.

Fig. 3.

Effects of M119 and related compounds on Gβγ signaling in dimethyl sulfoxide (DMSO)–differentiated HL-60 cells. (A) M119 and related compounds block fMLP-dependent Ca2+ release in differentiated HL-60 cells, loaded with 1 μM Fura2-AM. Cells were pretreated with DMSO or 10 μM compounds for 5 min before stimulation with 250 nM fMLP. The change in fluorescence was monitored at 340/380 nm. Data are representative of five independent experiments. Compounds are significantly different than DMSO control, P < 0.01. See fig. S4A for dose dependence. (B) Same as A except 10 μM M201 was tested. Data are representative of four independent experiments. See fig. S4B for pooled data. (C) M119 and M201 inhibition of GRK2 translocation. Differentiated HL-60 cells were treated with 10 μM compound before stimulation with 250 nM fMLP. Translocation of endogenous GRK2 was determined by Western blotting with a GRK2 antibody and quantitative chemiluminescence. Data are mean ± SEM from five experiments, *P < 0.05, **P < 0.01 analysis of variance (ANOVA). (D) M119 and M158C (NSC158110) inhibition of GFP-PHAkt translocation. Differentiated HL-60 cells stably overexpressing GFP-PHAkt were treated with 10 μM compound before stimulation with 100 nM fMLP. Translocation of GFP-PHAkt to the membrane was determined by Western blotting with antibody to GFP and quantitative chemiluminescence. Data are mean ± SEM from four experiments. ***P < 0.001 ANOVA. (E) Lack of effect of M119, M158C, and M201 on fMLP-induced ERK1 and ERK2 activation. Differentiated HL-60 cells were pretreated with 10 μM of compound prior to stimulation with 1 μM fMLP for 5 min. Levels of phosphorylated and total ERK were determined by Western blotting. This experiment was repeated three times with similar results.

To assess the ability of M119 and M158C to inhibit fMLP receptor–dependent regulation of PI3Kγ activation, we stimulated HL-60 cells stably overexpressing a green fluorescent protein (GFP)–tagged pleckstrin homology (PH) domain from Akt (15) with fMLP and assessed translocation of the GFP-PHAkt to the membrane by subcellular fractionation and Western blotting. The PH domain from Akt binds to PIP3 produced by PI3K activity at the membrane. Pretreatment of the cells with 10 μM of M119 or M158C inhibited fMLP-dependent translocation of GFP-PHAkt to the membrane (Fig. 3D), consistent with the ability of these compounds to inhibit activation of PI3Kγ by Gβγ.

Stimulation of differentiated HL-60 cells with fMLP also results in pertussis toxin–sensitive activation of various MAP kinases, including ERK1 and ERK2, p38, and JNK (16). However, pretreatment of HL-60 cells with M119, M158C, and M201 did not block fMLP-induced activation of ERK1 and ERK2 (Fig. 3E). Ca2+ mobilization occurred at low fMLP (100 nM) concentrations relative to that required to observe ERK1 and ERK2 (1 μM) activation, indicating that the selective lack of effect of M119 on ERK activation is not due to excess G protein activation.

We tested the efficacy of our compounds in an in vivo model. Compared with wild-type animals, PLCβ3–/– mice are 10 times as sensitive to the antinociceptive effects of the μ-agonist morphine (17). Because M119 blocks Gβγ-dependent activation of PLCβ3, we tested whether coadministration of M119 with morphine would also increase morphine-induced antinociception. Coadministration of M119 with morphine intra cerebroventricularly resulted in an 11-fold increase in the analgesic potency of morphine (Fig. 4A) [ED50 values and 95% confidence limits: 0.069 (0.023 to 0.201) nmol and 0.743 (0.341 to 1.62) nmol, respectively], whereas administration of 100 nmol M119 alone had no effect on baseline antinociception (table S3). M119 also had no effect on morphine-dependent antinociception in PLCβ3–/– mice (Fig. 4B). These data highlight the specificity of M119 actions and the selective nature of M119 both in vitro and in vivo. Gβγ subunits regulate many aspects of signaling critical for the actions of opioid agonists (18). If M119 were globally blocking Gβγ subunit functions, we expect that morphine-induced antinociception would have been attenuated rather than potentiated with M119 coadministration.

Fig. 4.

M119 effects on morphine-induced antinociception in (A) wild-type and (B) PLCβ3–/– mice. Increasing concentrations of morphine were administered intra-cerebroventricularly to mice with (squares) or without (triangles) concomitant administration of 100 nmol M119. Antinociception was measured 20 min after injection using the 55°C tail-flick test. Data are mean ± SEM from 7 to 10 animals at each point.

Protein interaction interfaces present difficult drug targets because of the generally large interaction surface area and flat topology of the interaction surfaces. Hotspots at protein interfaces comprise a small fraction of the overall interaction surface, yet are responsible for most of the energetics of binding, and it has been proposed that targeting such surfaces could successfully disrupt protein-protein binding (19). As a result of extensive screening, some protein interaction interfaces have been targeted with small molecules (1921). Our screen of only 1990 molecules identified multiple compounds with apparent affinities in the high nM to low μM range. These molecules demonstrate that multiple small molecule binders of Gβγ could be developed that differentially modulate functions downstream of GPCRs. Numerous studies in animal models have implicated Gβγ-subunit targeting as a therapeutic strategy in diseases such as heart failure (22), prostate cancer (23), vascular disease (24), and inflammatory disease (13). Thus, more extensive screening for molecules that bind to the Gβγ hotspot will yield a repertoire of potentially therapeutically useful small molecules.

Supporting Online Material

Materials and Methods

Tables S1 to S3

Figs. S1 to S4


References and Notes

Stay Connected to Science

Navigate This Article