Report

Receptor-Mediated Activation of Heterotrimeric G-Proteins in Living Cells

See allHide authors and affiliations

Science  23 Mar 2001:
Vol. 291, Issue 5512, pp. 2408-2411
DOI: 10.1126/science.1055835

Abstract

Receptor-mediated activation of heterotrimeric GTP–binding proteins (G-proteins) was visualized in living Dictyostelium discoideum cells by monitoring fluorescence resonance energy transfer (FRET) between α- and β- subunits fused to cyan and yellow fluorescent proteins. The G-protein heterotrimer rapidly dissociated and reassociated upon addition and removal of chemoattractant. During continuous stimulation, G-protein activation reached a dose-dependent steady-state level. Even though physiological responses subsided, the activation did not decline. Thus, adaptation occurs at another point in the signaling pathway, and occupied receptors, whether or not they are phosphorylated, catalyze the G-protein cycle. Construction of similar energy-transfer pairs of mammalian G-proteins should enable direct in situ mechanistic studies and applications such as drug screening and identifying ligands of newly found G-protein–coupled receptors.

Chemoattractants, hormones, neurotransmitters, odorants, and other sensory stimuli exert their effects on cells through a vast family of serpentine G-protein–coupled receptors (GPCRs). Excited receptors catalyze the exchange of guanosine triphosphate (GTP) for guanosine diphosphate (GDP) and the dissociation of the G-protein heterotrimer, allowing both the GTP-bound α subunit and free βγ complexes to signal to downstream effectors. The intrinsic guanosine triphosphatase (GTPase) activity of the α-subunit hydrolyzes the bound GTP and the heterotrimer reassociates, completing the cycle (1,2). These mechanisms have been derived from studies using isolated membranes and purified proteins and have not yet been directly observed in living cells.

In shallow gradients of chemoattractants, cells such as D. discoideum amoebae and leukocytes can activate physiological responses selectively at the cell's leading edge even when the chemoattractant receptors and G-proteins are distributed uniformly on the plasma membrane (3–6). The tightly restricted localization of the physiological responses requires a global adaptation process that also causes the responses to rapidly subside during persistent stimulation. Does global adaptation occur at the level of the G-protein cycle or at a subsequent step? To determine the kinetics and to localize sites of G-protein activation in cells and tissues, we developed a method to directly visualize G-protein disassociation. We reasoned that fluorescence resonance energy transfer (FRET) between probes attached to the α and βγ subunits would allow real-time measurements and imaging of G-protein signaling (7). To test this strategy, we tagged Gα2 and Gβ of D. discoideum with cyan and yellow fluorescent proteins and used FRET to observe the state of the G-protein heterotrimer in living cells (8).

We assessed the activities of the fusion proteins by phenotypic rescue of Gα2 and Gβ null cell lines. Fusion of enhanced yellow fluorescent protein (YFP) to the NH2-terminus of Gβ did not impair the capacity of Gβ to rescue the chemotactic defects of Gβ null mutants (gβ cells) (6,9). Guided by the crystal structures of several mammalian heterotrimers (10, 11), we inserted the enhanced cyan fluorescent protein (CFP) into the first loop (between αA and αB) of the helical domain of Gα2. Mutants that do not express Gα2 (gα2 cells) are completely deficient in chemoattractant-induced responses and therefore are unable to aggregate and differentiate (12). Stable expression of Gα2-CFP rescued the chemotactic and developmental defects of the gα2 cells (13). Chemoattractants triggered in vivo actin polymerization, and chemoattractant-filled micropipettes induced chemotactic responses. Similarly, co-transformation with Gα2-CFP and Gβ-YFP rescued the gα2 cells (Fig. 1) (14).

Figure 1

Functional interaction of Gβ-YFP and Gα2-CFP in living cells. (A) CFP was inserted into the helical domain of Gα2 in a location optimal for FRET with YFP fused to the NH2-terminus of Gβ (36). (B) Lysates of gα2 (lane 1), wild-type (lane 2), Gα2-CFP/gα2 (lane 3), α2-CFP:β-YFP cells (lane 4), gβ cells (lane 5), and Gβ-YFP/gβcell lines (lane 6) were subjected to immunoblot analyses. Gα2 antisera detect an appropriate band in wild-type cells (not shown) and a band at 72 kD in the Gα2-CFP/gα2 and α2-CFP:β-YFP cells. Antisera against GFP (Clontech, Palo Alto, CA) recognize the Gα2-CFP band at 72 kD in the Gα2-CFP/gα2 cells, a doublet of the 72-kD band and a 70-kD band of Gβ-YFP in the α2-CFP:β-YFP cells, and the 70-kD band in the Gβ-YFP/gβ cells. (C) In vivo actin polymerization assays of gα2 (light blue), wild-type (dark blue), and α2-CFP:β-YFP (green) cells were carried out as previously described (37). Cells were stimulated with 100 nM cAMP at t = 0 and fixed at times shown. Mean values for an experiment done in duplicate are shown. Two other independent experiments were performed and yielded similar results. (D) Cells were differentiated with repeated cAMP stimuli for 6 hours and were then examined for chemotactic response in a micropipette assay. At t = 0, a micropipette was filled with 1 μM cAMP and placed on the surface of the coverglass. The positions at time 0 and 10 min are shown. Bar, 10 μm.

Studies of cell lines expressing fluorescent subunits showed direct, specific transfer of resonance energy from Gα2-CFP to Gβ-YFP when the two proteins were co-expressed (Fig. 2A). We excited living, differentiated cells at 440 nm and recorded the emission spectrum between 460 and 600 nm. The co-transformed cells showed an extra emission peak near 527 nm, corresponding to the FRET fluorescence (15). Cell lines expressing Gα2-CFP alone or mixtures of cells containing either Gα2-CFP or Gβ-YFP alone did not display this additional peak. Co-transformation of Gα2-CFP and Gβ-YFP into gβcells gave similar results; here, we focused on the co-transformed gα2 cells (designated α2-CFP:β-YFP cells). The FRET fluorescence was observed in both membrane and supernatant fractions prepared from α2-CFP:β-YFP cells, suggesting heterotrimers exist in the cytoplasm as well as on the plasma membrane. The FRET fluorescence disappeared after treatment of membranes with Mg2+ plus GTP-γ-S, but not with GTP-γ-S alone (16).

Figure 2

Fluorescence spectra of cell lines. (A) Gα2-CFP/gα2 cells (light blue), α2-CFP:β-YFP cells (dark blue), and mixture of equal number of Gα2-CFP/gα2 and Gβ-YFP/gβ cells (green) were excited in a Spex Fluoromax-2 fluorimeter at 440 nm and emission spectra were collected and processed (15). (B) Emission spectra from α2-CFP:β-YFP cells before (dark blue) and after (light blue) treatment with 100 μM cAMP. (C) Kinetics of transient loss of FRET fluorescence after addition of 100 μM cAMP to α2-CFP:β-YFP cells (38). (D) Cells were observed with excitation and emission bandpass filters of 420 nm to 450 nm (Chroma Exciter 436/20) and 520 nm to 550 nm (Chroma Emitter 535/30), respectively. Mean intensities were calculated in IPLab Spectrum by manually circumscribing the membrane of multiple cells. The three curves represent individual cells from three different video sequences. Arrows indicate the frame at which the stimulus was added. (E) Difference images of consecutive frames captured at 3 s intervals were calculated in IPLab Spectrum. Stimulus was added between frames labeled 0 and 3 s; small cell indicated by the arrow was dislodged by the disturbance. Bar, 10 μm.

Addition of the chemoattractant cyclic adenosine 3′,5′ monophosphate (cAMP) to α2-CFP:β-YFP cells triggered a rapid, substantial loss of FRET fluorescence reflecting receptor-mediated activation and dissociation of the G-protein heterotrimer (Fig. 2B). There was a decrease in the YFP emission signal near 527 nm and a parallel increase in the CFP emission signals near 475 and 501 nm. The 490:527 ratio increased by 32%; the FRET fluorescence decreased by about 70% (17). Only α2-CFP:β-YFP cells showed a change in fluorescence intensity when stimulated (18). The large decrease suggests that the heterotrimer may dissociate rather than merely change conformation. We measured the kinetics of loss and restoration of the FRET fluorescence upon addition and removal of the stimulus (Fig. 2C). The response was maximal within 10 s of stimulation, the earliest time point taken. In other experiments, the response reached 90% of the maximum within a few seconds (16). As the cAMP was removed by the endogenous phosphodiesterase, the FRET fluorescence returned to its maximal level within 2 min. Activation is at least as fast as the most rapid receptor-mediated physiological responses such as actin polymerization and Pleckstrin homology (PH)–domain translocation.

We attempted to directly visualize the change in the FRET fluorescence by conventional fluorescence microscopy. When α2-CFP:β-YFP cells were excited with blue light (420 to 460 nM) and observed in the yellow range (525 to 540 nM), there was a fluorescent signal on the membrane as well as within the cell perimeter (Fig. 2, D and E). These fluorescent signals were lower in cells expressing α2-CFP or β-YFP alone, even though these cells were highly fluorescent when examined with the appropriate cyan or yellow filter sets. When a saturating dose of cAMP was added to the α2-CFP:β-YFP cells, the membrane signal decreased by about 15% within 6 s and did not reappear during the 1-min period of observation. The signal within the perimeter of the cell also displayed a slight decrease. We visualized the response by subtracting consecutive frames from one another (Fig. 2E). These microscopy results extend and confirm the fluorometric observations.

Treatment with increasing concentrations of cAMP and two of its analogs, 2′ deoxyadenosine-3′,5′ monophosphate (2′-dcAMP) and 8-bromoadenosine-3′,5′ monophosphate (8-Br-cAMP) decreased the FRET fluorescence in a dose-dependent manner (Fig. 3). To quantitate the response for each analog, we subtracted the spectrum for the highest concentration from that of each of the lower concentrations. The negative values from 475 to 520 nm reflect the gain in cyan fluorescence, and the positive values from 520 to 550 nm reflect the loss in yellow fluorescence (Fig. 3A; see also Fig. 2B). Processing of this data yielded EC50s for cAMP, 2′-dcAMP, and 8-Br-cAMP of about 10 nM, 20 nM, and 2 μM, respectively, which were shifted to the left of their reported binding affinities (average KDs ≈ 180 nM, 1μM, and 32 μM, respectively) (19). This suggests that the steady-state level of G-protein activation saturates before all receptors are fully occupied and favors the possibility that the receptors act catalytically (20). Addition of 1 mM 5′-AMP, which does not bind to the cAMP receptor (cAR1), had no effect on the FRET fluorescence. These data validate the hypothesis, based on previous genetic studies (12), that the heterotrimeric G-protein formed from Gα2 [1 of 11 known α-subunits in D. discoideum (21, 22)] and the unique βγ-complex are directly linked to the cAMP chemoattractant receptors.

Figure 3

Dose-response curves for cAMP, 2′-dcAMP, 8-Br cAMP and 5′ AMP. (A) Difference fluorescence spectrum of α2-CFP:β-YFP cells treated with increasing concentrations of cAMP. Cells were treated with caffeine and stimulated with 0 (dark blue), 1nM (magenta), 10 nM (orange), 100nM (light green), 1 μM (violet), 10 μM (brown), and 100 μM cAMP (dashed line) in the presence of 10 mM dithiothreitol, and analyzed in the fluorimeter 15 s after mixing. (B) Dose-response curves for cAMP (dark blue), 2′-dcAMP (light blue), and 8-Br cAMP (green), and 5′ AMP (orange). Data from two independent experiments were averaged for each curve. Data was processed as described (39).

Prolonged stimulation of D. discoideum cells with cAMP induces adaptation of a number of physiological responses, such as actin polymerization and PH-domain recruitment to the plasma membrane (23). Yet previous studies have suggested that receptors remain coupled to G-proteins in adapted cells (24). To examine adaptation of the G-protein heterotrimer, we persistently stimulated α2-CFP:β-YFP cells with 100 nM cAMP for 18 min. Half the cells were left to recover in the absence of cAMP and half were exposed to 10 μM cAMP for an additional 10 min. During the first 18 min of treatment, the FRET fluorescence in living cells decreased and remained near 20% of the initial value (Fig. 4A) (25). Cells that were no longer exposed to cAMP after 18 min showed a gradual increase in FRET fluorescence, reflecting the gradual degradation of cAMP by endogenous phosphodiesterase. Cells treated with 10 μM cAMP showed a further decrease in FRET fluorescence. In lysates from naı̈ve α2-CFP:β-YFP cells, GTP-γ-S stimulated the binding of PH-domains to membranes (3,26). This response was lost in cells pretreated with cAMP (Fig. 4B), indicating that these cells adapt normally. Because persistent stimulation does not return the G-proteins to the heterotrimeric or “off” position, adaptation of the physiological responses must occur at another point in the signaling pathway.

Figure 4

Continuous activation of the G-protein cycle. (A) α2-CFP:β-YFP cells were treated with 3 mM caffeine and were maintained at 100 nM cAMP by adding fresh cAMP every minute for 18 min and then divided into two sets. One set received no further cAMP, whereas the second was increased to 10 μM cAMP. Spectra were processed as described (39). (B) Cells were treated with cAMP for 0 or 15 min, lysed, and then incubated in the presence or absence of GTP-γ-S with a supernatant from cells expressing PHCRAC-GFP (3). After 2 min, membranes were collected and binding of the PHCRAC-GFP was assessed by immunoblot analysis using antibodies to GFP. (C) Immunoblot analysis shows separation of the cAR1 doublet into its lower (phosphorylated) and higher (nonphosphorylated) mobility forms (32). (D) Identical sets of cells received buffer (control), one addition of 100 μM cAMP for 15 s (15 sec), or 15 additions, once per minute (15 min). All sets were then washed at 0°. Bars show the fraction (±SD) of FRET fluorescence before (–) and after (+) addition of a second cAMP stimulus (10 μM). Immunoblot analyses of cAR1 in samples taken after washing and before restimulation show control (lanes 1 and 4), 15 s treated (lanes 2 and 5), and 15 min treated (lanes 3 and 6). The average fraction of unphosphorylated receptor remaining after 0, 15 s, and 15 min were 100, 89, and 9%, respectively. Duplicate experiments performed on different days were averaged.

The persistent activation of the G-protein cycle is surprising because agonist-induced phosphorylation of the receptors reaches a plateau within a few minutes (27). Immunoblots of selected time points showed the expected steady-state level of receptor phosphorylation (Fig. 4C). Next, sets of cells were stimulated with cAMP for 0, 15 s, or 15 min and then washed at 0°, a condition known to prevent receptor dephosphorylation (Fig. 4D, inset). After washing, the FRET fluorescence of the three sets of cells was nearly identical, indicating that the G-protein heterotrimer reassociated even though receptors remained phosphorylated. Subsequent stimulus addition caused a parallel loss of FRET fluorescence in each set of cells (Fig. 4D). Agonist-induced phosphorylation of other GPCRs facilitates a series of downstream events such as arrestin binding that uncouple excited receptors from G-proteins (28–31). Phosphorylation of cAR1 decreases its affinity for cAMP (32); however, the low affinity receptors continue to trigger the G-protein cycle.

The evidence presented here is consistent with the model of the steady-state G-protein cycle derived from in vitro measurements, although a mechanism of adaptation is absent (1). The rapid kinetics, reversibility, and dose-dependence suggest that occupied receptors repeatedly elicit G-protein activation. It is expected that regulators of G-protein signaling (RGS proteins) will influence the steady-state ratio of active and inactive G-proteins (33,34) and not shape the time courses of the physiological responses because adaptation does not directly involve the G-protein cycle. The persistent kinetics of G-protein activation suggests that its spatial pattern likely reflects the shallow external gradient of receptor occupancy rather than the tightly restricted localization of the physiological responses. Thus, restriction of the response to the cell's leading edge may involve pathways beyond or independent from the G-protein cycle.

  • * Present address: Department of Cell Biology and Anatomy, Johns Hopkins Medical Institutions, Baltimore, MD 21205, USA.

REFERENCES AND NOTES

View Abstract

Navigate This Article