Receptor-mediated Reversible Translocation of the G Protein βγ Complex from the Plasma Membrane to the Golgi Complex*[boxs]

Heterotrimeric G proteins have been thought to function on the plasma membrane after activation by transmembrane receptors. Here we show that, after activation by receptors, the G protein βγ complex selectively translocates to the Golgi. Receptor inactivation results in Gβγ translocating back to the plasma membrane. Both translocation processes occur rapidly within seconds. The efficiency of translocation is influenced by the type of γ subunit present in the G protein. Distinctly different receptor types are capable of inducing the translocation. Receptor-mediated translocation of Gβγ can spatially segregate G protein signaling activity.

Heterotrimeric G proteins have been thought to function on the plasma membrane after activation by transmembrane receptors. Here we show that, after activation by receptors, the G protein ␤␥ complex selectively translocates to the Golgi. Receptor inactivation results in G␤␥ translocating back to the plasma membrane. Both translocation processes occur rapidly within seconds. The efficiency of translocation is influenced by the type of ␥ subunit present in the G protein. Distinctly different receptor types are capable of inducing the translocation. Receptor-mediated translocation of G␤␥ can spatially segregate G protein signaling activity.
Heterotrimeric (␣␤␥) G proteins are localized to the plasma membrane of mammalian cells, facilitating interaction and activation by transmembrane G protein-coupled receptors (1)(2)(3). Extensive characterization of the effectors on which the G proteins act has suggested that the activated G protein ␣ and ␤␥ subunits function on the plasma membrane (3)(4)(5)(6). It has been thought that the post-translational addition of a lipid moiety to the ␣ subunit and the ␥ subunit aids in the localization of G␣ and G␤␥ complex to the plasma membrane, where they act on the effector molecules (7). However, there is little information about the properties of these proteins or the signaling they mediate in intact live mammalian cells, because studies attempting to observe G protein function in living mammalian cells have been limited.
To visualize the impact of receptor activation and inactivation on the spatial distribution of G protein subunits, we tagged G protein subunits with the yellow and cyan mutant forms of the green fluorescent protein, YFP 1 and CFP, respectively. The fusion proteins were expressed in mammalian cell lines and observed after activating overexpressed or endogenous receptors using fluorescence-based imaging methods. Although we have previously obtained evidence indicating the direct involvement of the G protein ␥ subunit in receptor interaction (8,9), we examined the effect of receptor activation on ␣o-CFP, ␤1, and different ␥ subunit types tagged with YFP in Chinese hamster ovary (CHO) cells overexpressing M2 muscarinic receptors. We discovered that ␥-YFP translocated from the plasma membrane to the cell interior on receptor activation and translocated back to the plasma membrane on inactivation. The ␤ subunit co-translocated with the ␥ subunit. The rapidity of the translocation process and proportion of ␤␥ complex that translocated were dependent on the ␥ subunit type present in the expressed G protein. Experiments using a marker for the Golgi complex and a Golgi disrupting agent, brefeldin A, indicated that the ␤␥ complex translocates to the Golgi complex. The translocation was sensitive to G i/o -and G q -coupled receptor stimulation. Endogenous receptors also stimulated G␤␥ translocation. The translocation of the ␤␥ complex is selective, because ␣o-CFP or a chimeric ␣o-q-CFP that couples to G q -coupling receptors do not translocate from the plasma membrane in response to receptor stimulation. The rapidity of the G␤␥ complex translocation process suggests that it is diffusion-mediated.
Imaging-Cells were cultured on glass coverslips and transiently transfected with appropriate combinations of different G protein subunits as described in the text and figure legends. The coverslips were washed with Hanks' buffer saline solution supplemented with 10 mM Hepes, pH 7.4, and 1 mg/ml glucose and mounted on an imaging chamber with an internal volume of 25 l (Warner Instruments). A fluid delivery system, including a programmable valve controller and Teflon valves (10 ms open/closure time), was used to deliver the buffer with or without agonist or antagonist through the chamber at a rate of 1 ml/min with a pressure-regulated flow controller (Automate Scientific). The cells were visualized with a Zeiss Axioscope fluorescent microscope using a 63ϫ oil immersion objective (1.4 numerical aperture) and 100-W mercury lamp with a Hamamatsu CCD Orca-ER camera. The shutter, emission, and excitation filter wheels were con-* This work was supported by National Institutes of Health Grants GM46963 and GM69027 and American Heart Association Postdoctoral Fellowship 0225378Z (to M. A.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. trolled by a Sutter Lambda 10 -2 optical filter changer (Sutter Instrument Company) run by MetaVue (Universal Imaging) software. The filter and beam splitter combinations (from Chroma Technology) were as follows: for CFP (CC), D436/20 excitation (x), D480/40 emission (m); for YFP (YY), D500/20 (x), D535/30 (m); a polychroic beam splitter (Chroma 86002BS); and 3 or 10% neutral density filters. In those cases where both CFP and YFP fusions were co-expressed in cells, the cells expressing relatively similar levels of the two fusions were selected based on emission intensities. Images were acquired at 20-s intervals. The exposure times were between 0.6 and 1.4 s. The cells were treated with pertussis toxin at a final concentration of 100 ng/ml 6 -15 h prior to the analysis. To disrupt Golgi in the cells, 10 M brefeldin A was used in the culture medium 6 -10 h before cell analysis.

RESULTS AND DISCUSSION
CHO cells stably expressing M2 muscarinic receptors were transiently transfected with cDNAs encoding G protein ␣o, ␤1 and ␥11 subunits tagged with YFP. ␥11 is a farnesylated ␥ subunit that is present in several mammalian tissues (16). YFP was attached to the N terminus of ␥11, because we had earlier discovered that the tag did not affect receptor activation of the G protein and it does not affect the farnesylation that occurs at the C terminus of the subunit (13). The cells were imaged as described above. Images of individual cells before and after sequential exposure to an agonist, carbachol, and an antagonist, atropine, are shown in Fig. 1A and Supplemental Fig. 1 (Movie 1). The emission signal from ␥11-YFP is initially localized to the plasma membrane. After M2 muscarinic receptor activation, the ␥11-YFP translocates to the cell interior. Subsequent inactivation of the M2 receptor leads to the translocation of ␥11-YFP back to the plasma membrane. Plots of the time courses of changes in the YFP emission intensity on the plasma membrane and the intracellular region where the translocated YFP signal localizes are shown in Fig. 1B (left and  right panels). The t for the translocation of ␥11-YFP protein in both directions is Ͻ20 s (Fig. 1B). When M2-CHO cells expressing ␣o-CFP, ␤1, and ␥11-YFP were exposed to different concentrations of agonist and the translocation response measured, the translocation of G␤␥ was exquisitely sensitive to the extent of receptor activation (Supplemental Fig. 1). To examine the impact of this translocation on the G protein ␣ subunit, we co-expressed ␣o-CFP with ␤1 and ␥11-YFP into the same cells. CFP was inserted downstream of Gly-92 of ␣o (13). ␣o-CFP did not affect the translocation properties of ␥11-YFP compared with native ␣o (Fig. 1C). As anticipated, based on our previous findings (13), the CFP emission increases with receptor activation and decreases with receptor inactivation due to the corresponding changes in the fluorescence resonance energy transfer occurring from CFP to YFP (Fig. 1C). This result indicates that the ␥ subunit translocation results from the activation of the ␣ subunit in the G protein heterotrimer. Consistent with this result, pertussis toxin-treated M2-CHO cells expressing ␤1 and ␥11-YFP do not show receptor-mediated translocation of ␥11-YFP, demonstrating a requirement for both the endogenous ␣i subunit and receptor activation of a G protein for translocation to occur (data not shown). Although it is required for translocation of ␥11-YFP, the ␣ subunit itself does not translocate in response to receptor activation (Supplemental Fig. 1 (Movie 2)). To examine whether this selective translocation of the ␥ subunit included the ␤ subunit, we co-expressed ␤1-YFP along with ␥11-CFP and observed the response of ␤1-YFP to receptor stimulation. Fig. 1D shows that the ␤ subunit co-translocates with the ␥ subunit consistent with extensive previous evidence that the ␤ subunit is in a tight complex with the ␥ subunit (1-3). When the ␤1-YFP protein was co-expressed with untagged ␥11, it also translocated, indicating that the fluorescent protein tag of ␥11-YFP did not affect translocation (Supplemental Fig. 1). In a control experiment, when ␣o-CFP and ␤1-YFP were expressed in M2-CHO cells without ␥11, very little receptor-dependent translocation of ␤1-YFP was detected, indicating that translocation was dependent on the presence of the ␥11 subunit (data not shown). It is likely that ␥11 subunit dependence of ␤ subunit translocation is linked to the role of the ␥ subunit in receptor interaction. We then examined whether a distinctly different ␥ subunit type, ␥5 (which has 33% identity at the primary structure level with ␥11), translocates similarly. M2-CHO cells were transiently transfected with ␣o-CFP, ␤1, and ␥5-YFP. Unlike ␥11, which is farnesylated, ␥5 is geranylgeranylated. The results of imaging the cells in the presence of agonist and antagonist are shown in Fig. 1E and Supplemental Fig. 1 (Movie 3). ␥5-YFP translocates in response to the receptor state, although the proportion of ␥5 translocated from the plasma membrane at steady state is distinctly less (Supplemental Fig. 1, compare Movie 2 with Movie 3), and the rate at which translocation occurs is also slower with a t of ϳ1 min.
To identify the organelle to which the G protein ␤␥ complex was translocated, we co-expressed a trans Golgi complex marker, galT tagged with YFP (17,18), with ␥11-CFP in M2-CHO cells and imaged the cells after M2 activation. Overlaying the images of CFP and YFP emission showed that ␥11-CFP was localized in the Golgi complex ( Fig. 2A). To confirm this localization, we treated the transfected M2-CHO cells with brefeldin A, which is known to disrupt the Golgi complex (19). When these cells were examined after agonist treatment, translocation of ␥11-YFP to the cell interior could no longer be detected (Fig. 2B). Control cells expressing galT-YFP did not show galT localization, indicating that the Golgi complex was indeed disrupted in these cells (Fig. 2B).
We then tested the generality of the receptor-mediated G␤␥ translocation process in the following systems. To eliminate the possibility that translocation was peculiar to M2 receptors, we tested the ability of a distinctly different isoform of the acetylcholine receptor, M3, to induce translocation of ␤␥11. M3 couples to the G q class of ␣ subunits, whereas M2 couples to the G i/o class (11). In CHO cells stably expressing M3 receptors, we co-expressed ␤1 and ␥11-YFP with a chimeric ␣o subunit containing the C-terminal tail of ␣q (Fig. 3A). This chimeric G␣ o-q is activated by the M3 receptor (but not M2) as we have shown before (13). The cells were treated with pertussis toxin to prevent cross-activation of G i proteins in CHO cells by M3 (CHO cells do not express G o (13)). Fig. 3A shows that ␥11-YFP translocated to the Golgi complex when M3 receptors were activated and translocated back to the plasma membrane when they were inactivated. To examine whether receptors other than muscarinic receptors can induce translocation, we performed similar experiments with CHO cells stably expressing 5HT1A receptors. ␥11-YFP translocated as in the case of M2 and M3, depending on the activation state of the 5HT1A receptor (Fig. 3B). Translocation was not dependent on overexpression of receptors because an endogenous receptor in CHO cells, 5HT1B also induced translocation (Fig. 3C). To test whether a cell line from a different species could induce translocation, we transfected a human lung fibrosarcoma cell line (HT1080) with cDNAs for M2 receptors, ␣o-CFP, ␤1, and ␥11-YFP and imaged the cells in the presence of agonist and antagonist as above. Translocation was induced, indicating that the process is not peculiar to CHO cells (Fig. 3D).
Based on a variety of in vitro experiments, it has been thought that G proteins, which are peripherally associated with the plasma membrane, are activated by transmembrane receptors, and after subunit dissociation, act on effector molecules on the plasma membrane (1)(2)(3)7). Consistent with this proposal, known effectors of G proteins (such as adenylyl cyclase and ion-conducting channels) are membrane proteins, and others (such as phospholipase C isozymes) act on substrates localized to the plasma membrane (3)(4)(5)(6). The rapid translocation of the G protein ␤␥ complex from plasma membrane to the Golgi complex in response to receptor stimulation is therefore very surprising. This hitherto unanticipated translocation suggests that the ␤␥ complex may act on an effector molecule located in the Golgi complex. G␣i3 has previously been shown to be present in the Golgi, although receptor-dependent effects on this localization were not identified (17). The   FIG. 2. Colocalization of ␥11 and a Golgi complex marker. A, M2-CHO cells expressing ␣o, ␤1, ␥11-CFP, and galT-YFP (a trans Golgi marker) were exposed to carbachol for 2 min, and images of ␥11-CFP and galT-YFP were captured, pseudocolored, and overlaid. Shown are ␥11-CFP (red) and galT-YFP (green). B, same cells treated with brefeldin A and exposed to carbachol. All are representative of n Ն 4. G protein ␤␥ complex has been shown earlier to act on protein kinase D in the Golgi complex to regulate Golgi disassembly (20). However, that work did not identify the signaling pathway or its mechanistic basis that allowed the G protein ␤␥ complex to act on the Golgi. The relationship between the translocation seen here and the previously reported action of G␤␥ on protein kinase D in Golgi is not clear. It is possible that the function of an unidentified effector in Golgi is modulated by receptor-regulated translocation of G␤␥. There is now evidence that, in addition to its role in the plasma membrane, Ras acts on effectors in the Golgi complex (21). A recent report shows that a protein resident in the Golgi complex binds MEK when signaling is activated by a tyrosine kinase receptor (22). These evidences for the Golgi complex acting as a distinct site for downstream elements of the tyrosine kinase signaling pathways makes the cytoplasmic surface of this organelle an attractive locale for integrating regulated cross-talk with G protein pathways through the translocation of the ␤␥ complex seen here. Overall, the ability of one signaling arm of the G protein to spatially isolate its activity from the plasma membrane to an internal organelle demonstrates a novel mode by which G protein-mediated signaling can be spatially segregated.
Another role for this translocation may be the rapid decrease in the concentration of the ␤␥ complex in the plasma membrane resulting in a dampening of signaling activity because of the reduced concentration of heterotrimer available for activation. This effect can be further modulated by differential magnitudes and rates of translocation as seen in the case of ␥11 and ␥5. The rate of translocation of G␤␥ back to the plasma membrane on receptor inactivation is as rapid as the translocation to the Golgi on activation of the receptor. Both processes occur within a time frame that suggests that translocation is diffusionmediated rather than actively controlled. The rapid reversibility of the translocation suggests that the receptor can exert tight control over the signaling effects of the spatial segregation of G␤␥.
The results here demonstrate that by imaging signaling ac-tivity in a live cell, previously unanticipated novel mechanisms underlying signaling can be identified. The ability to image the translocation of G␤␥ in a living cell in response to a receptor agonist, as well as an antagonist, can be of value in the rapid, non-invasive, high throughput and high content screening of potential therapeutic compounds directed at G protein-coupled receptors. These receptors form the most important target of commercially available therapeutic drugs at present.