点击显示 收起
【关键词】 Analysis
The specificity of G protein signaling demonstrated by in vivo knockouts is greater than expected based on in vitro assays of function. In this study, we investigated the basis for this discrepancy by comparing the abilities of seven 1 complexes containing 1, 2, 5, 7, 10, 11, or 12 to interact with s and of these subunits to compete for interaction with 1 in live human embryonic kidney (HEK) 293 cells. complexes were imaged using bimolecular fluorescence complementation, in which fluorescence is produced by two nonfluorescent fragments (N and C) of cyan fluorescent protein (CFP) or yellow fluorescent protein (YFP) when brought together by proteins fused to each fragment. Plasma membrane targeting of s-CFP varied inversely with its expression level, and the abilities of YFP-N-1YFP-C- complexes to increase this targeting varied by 2-fold or less. However, there were larger differences in the abilities of the CFP-N- subunits to compete for association with CFP-C-1. When the intensities of coexpressed CFP-C-1CFP-N- (cyan) and CFP-C-1YFP-N-2 (yellow) complexes were compared under conditions in which CFP-C-1 was limiting, the CFP-N- subunits exhibited a 4.5-fold range in their abilities to compete with YFP-N-2 for association with CFP-C-1. CFP-N-12 and CFP-N-1 were the strongest and weakest competitors, respectively. Taken together with previous demonstrations of a role for in the specificity of receptor signaling, these results suggest that differences in the association preferences of coexpressed and subunits for each other can determine which complexes predominate and participate in signaling pathways in intact cells.Cells integrate multiple receptor-G protein pathways to respond to stimulation by hormones and neurotransmitters. Given the numerous mammalian G protein isoforms (23 subunits, 5 subunits, and 12 subunits), maintenance of signaling specificity is clearly a vital cellular function. subunits have been thought to play the most important role in specificity because they exhibit greater diversity in their interactions with receptors and effectors than do the different complexes when tested in vitro (Clapham and Neer, 1997; Robishaw and Berlot, 2004). However, in vivo knockout experiments have demonstrated unique functions for particular complexes. For instance, mice containing targeted deletions of either 7 (Schwindinger et al., 2003) or 3 (Schwindinger et al., 2004) exhibit distinct phenotypes. Moreover, ribozyme-mediated depletion of 7 in HEK-293 cells results in a corresponding decrease in the level of 1 (Wang et al., 1999) and 7-knockout mice express reduced levels of olf (Schwindinger et al., 2003), suggesting that associations of particular subunits may be obligatory for their stability. The basis for this discrepancy in specificity in vivo versus in vitro is not well understood.
G protein signaling specificity is most likely maintained at multiple levels, including cell-type-specific expression, subcellular localization, and protein-protein interactions (Robishaw and Berlot, 2004). Only a subset of the genes encoding G protein-coupled receptors and G protein subunits is expressed in individual cells, and their expression levels can change during development with functional consequences (Iiri et al., 1995). Subcellular compartmentalization of G proteins and receptors may facilitate or impair interactions between coexpressed proteins (Ostrom, 2002). At the molecular level, reconstitution experiments have indicated differences in the combinations that are preferred by particular receptors (Figler et al., 1997; Richardson and Robishaw, 1999; Hou et al., 2000; Lim et al., 2001; McIntire et al., 2001).
Herein, we focus on subcellular regulation of signaling specificity. Among the potential regulatory mechanisms are preferential associations of and subunits, of subunits with complexes, and of combinations with G protein-coupled receptors. Based on studies using the yeast two-hybrid system (Yan et al., 1996) and reticulocyte lysates (Schmidt et al., 1992; Ray et al., 1995; Dingus et al., 2005), most and subunits can form complexes. However, because the cellular environment can affect assembly (Clapham and Neer, 1997; Lukov et al., 2005; Li et al., 2006), it would be optimal to study dimerization in vivo. In addition, because cells express multiple isoforms of and subunits, determining the relative preferences of the subunits for each other in intact cells would help to predict which complexes are most likely to form in vivo.
To study assembly in live cells, we have applied the strategy of BiFC (Hu et al., 2002) to localize and compare the relative amounts of different dimers (Hynes et al., 2004b). This approach involves the production of a fluorescent signal by two nonfluorescent fragments of CFP or YFP (N and C) when they are brought together by interactions between proteins fused to each fragment. This makes it possible to image complexes exclusively, rather than individual and subunits. By imaging the fluorescent signals formed by pair-wise combinations of 1, 2, and 5 with 1, 2, and 7 (Hynes et al., 2004b), we found that can direct trafficking of . In addition we found that upon stimulation of the 2-adrenergic receptor, both s-CFP and YFP-N-1YFP-C-7 internalize from the plasma membrane to the cytoplasm and colocalize on vesicles (Hynes et al., 2004a).
In this report, we compare the abilities of seven different 1 complexes to interact with s and of these subunits to compete for interaction with 1 in live HEK-293 cells. Interaction of s with was measured using a plasma membrane targeting assay. Plasma membrane targeting of s-CFP varied inversely with its expression level, and the abilities of YFP-N-1YFP-C- complexes to increase this targeting varied by 2-fold or less. To compare the interactions of different subunits with 1, we produced CFP-C-1, YFP-N-2, and CFP-N- subunits. When the intensities of coexpressed CFP-C-1CFP-N- (cyan) and CFP-C-1YFP-N-2 (yellow) complexes were compared under conditions in which CFP-C-1 was limiting, the CFP-N- subunits exhibited a 4.5-fold range in their abilities to compete with YFP-N-2 for association with CFP-C-1. Taken together with previous demonstrations of a role for in the specificity of receptor signaling, these results suggest that differences in association preferences of coexpressed and subunits may determine which complexes predominate and participate in cellular signaling pathways.
Production of Fluorescent Fusion Proteins. YFP(1-158)1, YFP(159-238)1, YFP(159-238)2, and YFP(159-238)7 were produced as described previously (Hynes et al., 2004b). To produce YFP(159-238)5, YFP(159-238)10, YFP(159-238)11, and YFP(159-238)12, the bovine 5 cDNA, the human 10 cDNA, the bovine 11 cDNA, and the human 12 cDNA (obtained from Janet Robishaw, Weis Center for Research, Danville, PA) were each amplified by a PCR that added a BamHI site and a linker sequence (Arg-Ser) to the 5' end and a BglII site to the 3' end, digested with BamHI and BglII, and subcloned into the BglII site of YFP(159-238)pcDNAI/Amp, as described previously (Hynes et al., 2004b), so that YFP(159-238) was fused to the amino terminus of each of the subunit cDNAs.
To produce CFP(159-238)1 in pcDNAI/Amp, CFP(159-238) was amplified by a PCR from ECFP (Clontech, Mountain View, CA) containing a substitution of His for Asn-164. The PCR introduced a BamHI site at the 5' end of CFP(159-238) and a BglII site at the 3' end. A BglII site was introduced into the polylinker of pcDNAI/Amp (Invitrogen, Carlsbad, CA) 3' to the BamHI site and CFP(159-238) was subcloned into these sites to produce CFP(159-238)pcDNAI/Amp. A BglII site in the human 1 cDNA (obtained from Janet Robishaw) was removed using the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA), and the cDNA was amplified by a PCR that added a linker sequence (Arg-Ser-Ile-Ala-Thr) containing a BamHI site on the 5' end and a BglII site on the 3' end. This PCR product was digested with BamHI and BglII and subcloned into the BglII site of CFP(159-238)pcDNAI/Amp so that CFP(159-238) was fused to the amino terminus of 1.
To produce CFP(1-158) constructs in pcDNAI/Amp, CFP(1-158) was amplified by a PCR from ECFP (Clontech) that introduced a BamHI site at the 5' end and a BglII site at the 3' end and subcloned into the BamHI and BglII sites of the modified pcDNAI/Amp vector described above to produce CFP(1-158)pcDNAI/Amp. Each of the subunit cDNAs was amplified by a PCR, as described above for the YFP(159-238) constructs, digested with BamHI and BglII, and subcloned into the BglII site of CFP(1-158)pcDNAI/Amp so that CFP(1-158) was fused to the amino terminus of each of the subunits. Cer(1-158) constructs in pcDNAI/Amp were produced in the same way, using monomeric Cerulean (Rizzo et al., 2004) (obtained from David Piston, Vanderbilt University, Nashville, TN), which contains S72A, Y145A, H148D, and A206K substitutions in ECFP, as the PCR template.
mRFP-Mem was constructed by a fusion PCR using pEYFP-Mem (Clontech) and mRFP1/pcDNA3 (obtained from Roger Tsien, University of California, San Diego) as templates. The PCR product contained the first 20 residues of neuromodulin fused to the amino terminus of mRFP and had a unique 5' BglII site and 3' NotI site. This PCR product was digested with BglII and NotI and subcloned into BglII/NotI digested pEYFP-Mem.
Imaging of Transfected Cells. HEK-293 cells (American Type Culture Collection, Manassas, VA) were plated at a density of 2 x 105 cells per well on four-well chambered coverslips (Lab-Tek II; Nalge Nunc, Naperville, IL). On the following day, the cells were transiently transfected using 0.25 µl of LipofectAMINE 2000 Reagent (Invitrogen). Plasmids were transfected using the following amounts: s-CFP, 0.15 µg or as described in the legend to Fig. 1; YFP-N-1, YFP-C- subunits, CFP-C-1, CFP-N- subunits, YFP-N-2, CFP-N, and CFP-C, 0.075 µg; and YFP and mRFP-Mem, 0.0025 µg. The cells were imaged 2 days after transfection at 63x using a Zeiss Axiovert 200 fluorescence microscope equipped with computer-controlled filter wheels, shutters, xyz stage (Ludl, Hawthorne, NY), and ORCA-ER camera (Hamamatsu Corporation, Bridgewater, NJ), under the control of IPLab software (Scanalytics, Inc., Fairfax, VA). A single triple-pass dichroic mirror (86006bs; Chroma, Brattleboro, VT) was used to ensure image registration. Excitation and emission band-pass filters for CFP (430/25, 465/30), YFP (495/20, 535/25), and mRFP1 (565/25, 630/60) were obtained from Chroma. One hour before imaging the culture medium was replaced with 20 mM HEPES-buffered minimal essential medium with Earle's salts without bicarbonate. During imaging, the cells were maintained at 37°C using a CSMI stage incubator (Harvard Apparatus, Holliston, MA).
Fig. 1. Targeting of s-CFP to the plasma membrane is plasmid dose-dependent and can be increased by co-expression with YFP-N-1YFP-C-7. A, left, representative images of s-CFP in HEK-293 cells transfected with the indicated amounts of s-CFP-expressing plasmid. The cells were also transfected with mRFP-Mem, a plasma membrane marker, and either YFP, a cytoplasm marker, or YFP-N-1 and YFP-C-7, as indicated. The plasma membrane fractions (determined as described under Materials and Methods) of s-CFP for each cell and for YFP-N-1YFP-C-7 in the bottom cell are listed. The black line and black square on the mRFP-Mem image indicate the pixels used in the measurements of average membrane and cytoplasm intensity. YN indicates YFP-N and YC indicates YFP-C. Scale bar, 5 µm. Right, graphs show the intensities of each of the three signals along the white line drawn on the mRFP-Mem image from the outside of the cell into the cytoplasm. The location of the plasma membrane is highlighted in gray. The intensity values for each color were scaled to a value of one in the cytoplasm region. B, plasma membrane fractions of s-CFP in cells transfected with the indicated amounts of s-CFP plasmid. Values represent the mean ± S.E. The number of cells analyzed for each condition was as follows: 0.017 µg of plasmid, 61; 0.05 µg of plasmid, 64; 0.1 µg of plasmid, 38; 0.15 µg of plasmid, 60. C, plasma membrane fractions of s-CFP (first and second bars) and YFP-N-1YFP-C-7 (third bar) in cells transfected with 0.15 µg of s-CFP plasmid in the absence (185 cells) or presence (144 cells) of 0.075 µg each of plasmids expressing YFP-N-1 and YFP-C-7. Values represent the mean ± S.E.
Image Analysis. The membrane marker mRFP-Mem was included in all transfections and was used as the primary criterion for selecting cells for imaging and analysis. Cells having a clear plasma membrane border and adjacent region of cytoplasm were identified. If the cell also had detectable intensity of the CFP and/or YFP labeled proteins that were also transfected, then the cell image was recorded. Exposure times for each color varied depending on cell intensity, and the following corrections were made so that the intensity values correspond to a 1-s exposure with image background subtracted. The background intensity for each image was determined by averaging the intensity of a region of pixels outside the cell. The image background was subtracted from each image, and the remaining intensity was scaled to correspond to an exposure time of 1 s. To eliminate the effect of day-to-day variation in lamp intensity, an additional correction was applied based on images of a slide containing a mixture of blue-green, yellow-green, and red Fluo-Sphere 1-µm beads (Invitrogen) that were taken each day. A small amount of bleed-through of CFP intensity into the YFP images (0.6%) was then subtracted. All image processing was performed using IPlab software.
The average fluorescence intensity of each cell was determined by tracing a border around the edge of the cell in the mRFP-Mem image using a Cintiq pen-based display screen (Wacom, Vancouver, WA). The average pixel intensity of the entire cell including the border was calculated for each color.
The plasma membrane fraction is a measurement of the distribution of a labeled protein between the plasma membrane and cytoplasmic compartments. The plots in Fig. 1A illustrate the pixel intensity values for the membrane marker (mRFP-Mem) and cytoplasm marker (YFP) along the white line drawn perpendicular to the plasma membrane in the mRFP-Mem image. Plasma membrane pixels (black line) corresponding to a length of plasma membrane were identified and marked using the mRFP-Mem image. Cytoplasm pixels (black box) were marked with a 12 x 12 pixel box adjacent to the plasma membrane pixels in a region that was devoid of labeled intracellular vesicles or membranes. The plasma membrane to cytoplasm ratios of labeled G protein subunits (ProteinRatio), plasma membrane marker (MemRatio), and cytoplasm marker (CytoRatio) were calculated by dividing the average intensity of each label in the marked plasma membrane region by the average intensity in the marked cytoplasm region. Based on images of cells transfected with YFP, mRFP-Mem, and s-CFP (Fig. 1A), the CytoRatio was very consistent from cell to cell, with a mean value of 0.62 (S.E. = 0.01, n = 223). Because most of the experiments described here required both the CFP and YFP color channels for labeled G protein subunits, the average CytoRatio of 0.62 was used in the calculations of plasma membrane fraction. The plasma membrane fraction of a labeled protein is defined as the plasma membrane to cytoplasm ratio of the protein relative to that of the plasma membrane and cytoplasm markers, and was calculated using the following equation: Plasma Membrane Fraction(Protein) = (ProteinRatio-CytoRatio)/(MemRatio-CytoRatio). A value of zero corresponds to a completely cytoplasmic distribution, and a value of one corresponds to a completely plasma membrane distribution.
To determine the average cellular intensity of s-CFP at which its plasma membrane fraction was half-maximal in the absence or presence of a YFP-N-1YFP-C- complex (Fig. 2A, Table 1), the plasma membrane fraction of s-CFP was fitted to the equation Y = b + (a - b)/[1 + (X/c)d], where X is the average intensity of s-CFP, Y is the observed plasma membrane fraction of s-CFP, a is the maximum plasma membrane fraction of s-CFP, b is the minimum plasma membrane fraction of s-CFP, c is the average intensity of s-CFP at which its plasma membrane fraction is half-maximal, and d is the slope factor. S.E. of the fits were determined by Kaleidograph (Synergy Software, Reading, PA).
Fig. 2. Targeting of s-CFP by YFP-N-1YFP-C- complexes. HEK-293 cells were transfected with s-CFP, YFP-N-1, and a YFP-C- subunit as described under Materials and Methods. A, plot of plasma membrane fraction of s-CFP as a function of its intensity in the absence or presence of the indicated YFP-N-1YFP-C- complexes. B, average intensities of each of the YFP-N-1YFP-C- complexes coexpressed with s-CFP. YN indicates YFP-N and YC indicates YFP-C. C, plot of plasma membrane fraction of s-CFP as a function of the ratio of intensities of YFP-N-1YFP-C- complexes to that of s-CFP. D, plasma membrane fractions of each of the YFP-N-1YFP-C- complexes coexpressed with s-CFP. All data were derived from the same cell images for each YFP-N-1YFP-C- combination. The numbers of cells analyzed were as follows: no (181), 11 (111), 12 (132), 15 (168), 17 (142), 110 (120), 111 (145), and 112 (155). In A and C, measurements from individual cells were sorted into equal size bins along the logarithmic x-axis and averaged. Bins containing fewer than three measurements were omitted. All values represent the means ± S.E.
TABLE 1 s-CFP intensities and YN-1YC-/s-CFP intensity ratios at which plasma membrane fraction of s-CFP is half-maximal (calculated from the data in Fig. 2)
To determine the YFP-N-1YFP-C-/s-CFP intensity ratio at which the plasma membrane fraction of s-CFP was half-maximal (Fig. 2B, Table 1), the plasma membrane fraction of s-CFP was fitted to the same equation, Y = b + (a - b)/[1 + (X/c)d], where X is the YFP-N-1YFP-C-/s-CFP intensity ratio, Y is the observed plasma membrane fraction of s-CFP, a = 0, b = 1, c is the YFP-N-1YFP-C-/s-CFP intensity ratio at which the plasma membrane targeting of s-CFP is half-maximal, and d is the slope factor. S.E. of the fits were determined by Kaleidograph.
Measurement of Fluorescence in Cell Populations. HEK-293 cells (1.6 x 106 per 60-mm dish) were transfected with plasmids as described in the legends to Figs. 5 and 6 using 6 µl of Lipofectamine 2000 Reagent (Invitrogen) according to the manufacturer's instructions. Two days later, cells were washed once in 4 ml of HEPES-buffered salt solution + CaCl2 media (20 mM HEPES, pH 7.2, 118 mM NaCl, 4.6 mM KCl, 10 mM D-glucose, and 1 mM CaCl2). HEPES-buffered salt solution + EDTA media (2 ml; 20 mM HEPES, pH 7.2, 118 mM NaCl, 4.6 mM KCl, 10 mM D-glucose, and 0.5 mM EDTA) were then added to the dish, cells were scraped off the dishes with a rubber policeman, triturated with a pipet to break up clumps, and suspended in a 1-cm square glass cuvette with a magnetic stir bar.
Fig. 5. Quantification of the fluorescence intensities of CFP-C-1Cer-N- complexes in cell populations. A, intensities of CFP-C-1Cer-N- complexes when CFP-C-1 is not limiting. HEK-293 cells (1.6 x 106 per 60-mm dish) were transfected with 2.4 µg of plasmid expressing CFP-C-1 and 0.0033, 0.01, or 0.03 µg of each Cer-N- subunit plasmid as indicated. The total amount of plasmid in each transfection was maintained at 2.43 µg by making up the difference with pcDNAI/Amp. Values represent the means ± S.E from three experiments performed in duplicate, and data were fit by linear regressions. CC indicates CFP-C. B, immunoblot showing expression levels of Cer-N- subunits. HEK-293 cells were transfected as in A, except that 100-mm dishes were used, and the number of cells and amounts of plasmids and Lipofectamine 2000 reagent were scaled up accordingly by a factor of 2.78. Membranes were prepared, resolved by SDS-polyacrylamide electrophoresis, and immunoblotted with a polyclonal antibody to GFP peptides as described under Materials and Methods. Lane 1 shows membranes from cells transfected with vector (pcDNAI/Amp). Lanes 2, 3, and 4 show membranes from cells transfected with plasmid expressing 2.4 µg of CFP-C-1 and 0.0033, 0.01, or 0.03 µg (per 1.6 x 106 cells) of Cer-N-2, respectively. Membranes in lanes 5 to 10 are from cells transfected with 2.4 µg of CFP-C-1 and 0.03 µg (per 1.6 x 106 cells) of the indicated Cer-N- subunit. Similar results were obtained in two additional experiments. C, the expression level of Cer-N-2 exhibited a linear relationship to the amount of transfected plasmid. The relative amounts of Cer-N-2 protein in lanes 2 to 4 of the immunoblot in B were quantified as described under Materials and Methods. Similar results were obtained in two additional experiments. D, quantification of expression levels of Cer-N- subunits obtained from immunoblots. The relative expression levels of Cer-N- subunits in cells transfected with 2.4 µg of plasmid expressing CFP-C-1 and 0.03 µg of each Cer-N- subunit plasmid (per 1.6 x 106 cells) were quantified as described under Materials and Methods and are expressed as percentage of Cer-N-2 expression. Values represent the means ± S.E from three experiments. E, normalization of CFP-C-1Cer-N- intensities to Cer-N- expression levels. The mean slopes of CFP-C-1Cer-N- intensity versus Cer-N- plasmid dose were divided by the relative expression levels of the corresponding to Cer-N- subunits, expressed as percentage of Cer-N-2 expression (each from three experiments). The S.E. resulting from division of mean CFP-C-1Cer-N- intensity ± S.E. by Cer-N- expression level ± S.E. was calculated using an equation for error propagation (http://www.rit.edu/~uphysics/uncertainties/Uncertaintiespart2.html).
Fig. 6. Competition between Cer-N- and YFP-N-2 subunits for limiting amounts of CFP-C-1 in cell populations. The intensity of CFP-C-1YFP-N-2 was measured in the presence of a competing Cer-N- subunit or empty vector. HEK-293 cells were transfected with 0.6 µg each of plasmids expressing CFP-C-1 and YFP-N-2, and 0, 0.01, 0.03, 0.09, 0.27, 0.81, or 2.43 µg of each Cer-N- subunit. The total amount of plasmid in each transfection was maintained at 3.63 µg by making up the difference with pcDNAI/Amp. YN indicates YFP-N, and CC indicates CFP-C. A, CFP-C-1YFP-N-2 intensity expressed as a function of µg of cotransfected Cer-N- subunit plasmid. Values represent the means ± S.E. from three experiments performed in duplicate. B, CFP-C-1YFP-N-2 intensity expressed as a function of the relative amounts of coexpressed Cer-N- subunits. The plasmid amounts of each of the Cer-N- subunits used in A were multiplied by their relative expression levels (fraction of the Cer-N-2 level from Fig. 5D and Table 2).
Data were collected on a PC1 photon-counting spectrofluorometer (ISS, Champaign, IL) configured with motorized filter wheels on both the excitation path between the excitation monochrometer and the sample and on the emission path between the sample and the emission monochrometer. The slits on the excitation and emission monochrometers were set to 16 nm. To eliminate the light scattering signal from the cell suspensions during measurements of CFP and YFP intensity, band-pass filters were used in the excitation filter wheel in combination with long-pass filters in the emission filter wheel. For CFP measurements, the excitation monochrometer was set to 430 nm with a 430/25 band-pass filter, and the emission monochrometer was set to 480 nm with a 455 long-pass filter. For YFP measurements the excitation was set to 492 nm with a 492/18 band-pass filter, and emission was set to 530 nm with a 515 long-pass filter.
The cell density of each sample was determined from a light-scattering measurement at 650 nm. Excitation and emission monochrometers were set to 650 nm, and a 1.3 OD neutral density filter in combination with a long-pass filter at 590 nm was used in the excitation filter wheel. All filters were from Chroma. To subtract the signal that was attributable to autofluorescence, measurements of CFP, YFP, and light scattering were made each day on a dilution series of cells transfected with vector alone. The linear relationship between the autofluorescence signal in the CFP and YFP channels and the light-scattering signal was calculated. The light-scattering measurement of samples containing fluorescent proteins was used to calculate the amount of autofluorescence to subtract from the CFP and YFP measurements. Control of the monochrometers, motorized filter wheels, and data acquisition was done using the Vinci software program (ISS) and allowed the measurements of CFP, YFP, and light scattering to be made in quick succession without repositioning the sample.
In multicolor BiFC experiments, the IC50 for inhibition of association of YFP-N-2 with CFP-C-1 by Cer-N- subunits was defined as micrograms of Cer-N- subunit plasmid that produced a 50% decrease in the intensity of CFP-C-1YFP-N-2. To determine IC50 values, the data were fit to the equation Y = (a)/[1 + (X/b)c], where X is micrograms of transfected Cer-N- plasmid, Y is the amount of fluorescence produced by CFP-C-1YFP-N-2, a is the maximum amount of fluorescence produced by CFP-C-1YFP-N-2, b is the half-maximal inhibitory concentration (IC50) of the Cer-N- subunit, and c is the slope factor. In Table 2, these IC50 values were normalized by multiplying by the relative expression levels of the Cer-N- subunits (Fig. 5D). The S.E. resulting from the multiplication of IC50± S.E. by Cer-N- expression level ± S.E. was calculated using an equation for error propagation (http://www.rit.edu/~uphysics/uncertainties/Uncertaintiespart2.html).
TABLE 2 Competition of CerN- subunits with YN-2 for dimerization with CC-1 in live HEK-293 cells
Values represent the mean ± S.E. from three experiments.
Membrane Preparations and Immunoblots. HEK-293 cells were transfected as described above for measurement of fluorescence in cell populations, except that 100-mm dishes were used and the number of cells and amounts of plasmids and Lipofectamine 2000 reagent were scaled up accordingly by a factor of 2.78. Forty-eight hours after transfection, membranes were prepared as described previously (Hynes et al., 2004a). Fifty micrograms of membrane proteins were resolved by SDS-polyacrylamide electrophoresis (12.5%), transferred to nitrocellulose, and probed with a polyclonal antibody to GFP, Living Colors A.v. Peptide Antibody (Clontech, Mountain View, CA), which is directed against three peptides derived from GFP residues 100 to 238. The antigen-antibody complexes were detected according to the ECL Western blotting protocol (GE Healthcare, Little Chalfont, Buckinghamshire, UK). Chemiluminescence was imaged using a Lumi-Imager (Roche Applied Science, Indianapolis, IN). Bands in the images were quantified using IPLab software. The intensity of a nonspecific band obtained in membranes from cells transfected with vector alone (pcDNAI/Amp) that comigrated with Cer-N-1, Cer-N-2, Cer-N-11, and Cer-N-12, was subtracted from the intensities of these constructs.
Plasma Membrane Targeting of s-CFP by YFP-N-YFP-C- Complexes Can Be Used to Measure S- Interaction. One potential mechanism for maintaining G protein signaling specificity in vivo is by selectivity in the interactions between the G protein and subunits. To investigate whether there are differences among complexes in their abilities to interact with s, we took advantage of the fact that the localization of s-CFP in HEK-293 cells, when transiently expressed without exogenous , depended on the amount of transfected plasmid (Fig. 1, A and B). When cells were transfected with 0.017 µg of s-CFP-expressing plasmid, distinct plasma membrane labeling was observed (Fig. 1, A and B). However, transfection with increased amounts of s-CFP plasmid resulted in proportionately decreased distributions of the label to the plasma membrane (Fig. 1, A and B). Because interaction with is required for plasma membrane targeting of s (Evanko et al., 2000), the simplest explanation for this relationship between s-CFP expression level and localization was that insufficient endogenous limited plasma membrane localization of the higher levels of s-CFP. To test for this, we coexpressed s-CFP with 17, imaged using BiFC, which involves the production of a fluorescent signal by two nonfluorescent fragments of CFP or YFP (N and C) when brought together by interactions between proteins fused to each fragment (Hynes et al., 2004b). Expression of either YFP-N-1 or YFP-C-7 separately or YFP-N and YFP-C together does not produce fluorescence (Hynes et al., 2004b). In addition, fluorescence is not obtained upon coexpression of YFP-N-2 and YFP-C-1 (Hynes et al., 2004b), consistent with other studies indicating that 2 and 1 do not interact to form a functional dimer (Iniguez-Lluhi et al., 1992; Schmidt et al., 1992). In support of the idea that plasma membrane targeting of over-expressed s-CFP required exogenous , coexpression of YFP-N-1YFP-C-7 resulted in plasma membrane localization of s-CFP in cells transfected with 0.15 µg of s-CFP-expressing plasmid (Fig. 1, A and C), as observed previously (Hynes et al., 2004a). In the absence of coexpressed YFP-N-1YFP-C-7, the distribution of s-CFP in cells transfected with this amount of plasmid was similar to that of free YFP, used as a cytosolic marker (Fig. 1A).
Seven Different YFP-N-1YFP-C- Complexes Target s-CFP to the Plasma Membrane with Similar Efficacies. We used the above coexpression system to compare the abilities of YFP-N-1YFP-C- complexes containing 1, 2, 5, 7, 10, 11, or 12 to target s-CFP to the plasma membrane in HEK-293 cells. These complexes were selected for study because 1 is widely expressed and the subunits are highly conserved, with the exception of 5, whereas the subunits are more numerous and less well conserved. 1 (Wang et al., 1999) and each of these subunits, except 1 (Wang et al., 1997), have been detected at the protein level in HEK-293 cells. 1 and 11 are covalently modified by the 15-carbon isoprenoid, farnesyl, whereas the other subunits contain the more hydrophobic 20-carbon isoprenoid geranylgeranyl (Wedegaertner et al., 1995).
Targeting of s-CFP to the plasma membrane, in cells transfected with 0.15 µg of plasmid, was determined in the absence and presence of each of the YFP-N-1YFP-C- complexes (Fig. 2A). The s-CFP intensities in individual cells were distributed over a 100-fold range. Even in the presence of coexpressed YFP-N-1YFP-C- complexes, the fraction of s-CFP that associated with the plasma membrane decreased as its intensity increased (Fig. 2A). However, coexpression of each of the YFP-N-1YFP-C- complexes increased the s-CFP intensity at which its plasma membrane targeting was half-maximal (Fig. 2A). YFP-N-1YFP-C-1, the least effective complex, and YFP-N-1YFP-C-10, the most effective complex, produced half-maximal plasma membrane targeting of s-CFP at intensities that were 2- and 3-fold greater, respectively, then when it was expressed alone (Fig. 2A, Table 1). The decreased effectiveness of YFP-N-1YFP-C-1 was not due to a decreased amount of this complex, because its average intensity was similar to that of the other YFP-N-1YFP-C- complexes (Fig. 2B).
To normalize the s-CFP targeting abilities of the YFP-N-1YFP-C- complexes to their relative concentrations, the plasma membrane fraction of s-CFP was expressed as a function of the YFP-N-1YFP-C- to s-CFP intensity ratio in each cell (Fig. 2C). This ratio reflects the relative intracellular concentration of each of the different YFP-N-1YFP-C- complexes compared with that of s-CFP but is not a molar ratio of to s because the relative intensities of YFP produced by BiFC and of CFP have not been determined. Based on the YFP-N-1YFP-C- to s-CFP intensity ratio that resulted in half-maximal plasma membrane targeting of s-CFP, YFP-N-1YFP-C-1 and YFP-N-1YFP-C-10 were still the least and most effective, respectively, at s-CFP targeting, differing by 2-fold in their efficacies (Fig. 2C, Table 1). The decreased effectiveness of YFP-N-1YFP-C-1 may be due in part to the fact that it exhibited somewhat less plasma membrane targeting itself than did the other dimers (Fig. 2D). The relatively low plasma membrane fractions of both YFP-N-1YFP-C-1 and YFP-N-1YFP-C-11 (Fig. 2D) are consistent with the fact that 1 and 11 are farnesylated, rather than geranylgeranylated like the other subunits tested here. Overall, however, the s-CFP targeting abilities of the different YFP-N-1YFP-C- complexes were fairly similar, suggesting that the corresponding 1 complexes exhibit similar affinities for s.
Production of CFP-C-1CFP-N- Dimers for Use in Multicolor BiFC. Another potential source of functional diversity among G protein and subunits could be preferential formation of particular complexes. To investigate this possibility, we chose to compare the abilities of 1, 2, 5, 7, 10, 11, and 12 to compete for association with 1 using multicolor BiFC (Hu and Kerppola, 2003). This approach consists of simultaneous visualization of the two fluorescent complexes formed when proteins fused to YFP-N and CFP-N interact with a common binding partner fused to CFP-C. Complexes containing YFP-N and CFP-C fusion proteins are yellow, whereas those containing CFP-N and CFP-C fusion proteins are cyan, because the amino terminal fragment of the fluorescent protein determines the spectral properties of the complex (Hu and Kerppola, 2003). We have produced YFP-N-YFP-C- dimers previously (Hynes et al., 2004a,b) (Figs. 1 and 2). Now, to compare the interactions of different subunits with the same subunit, we fused a carboxylterminal CFP fragment (residues 159-238) to 1 to produce CFP-C-1, and amino-terminal CFP or YFP fragments (residues 1-158) to the subunits, producing CFP-N- and YFP-N- subunits (Fig. 3A).
Fig. 3. Production of fluorescent 1 complexes using BiFC for multicolor competition studies. A, model of CFP-C-CFP-N-. The C-terminal CFP fragment, CFP(159-238), referred to as CFP-C (magenta), is fused to G (red) and the N-terminal CFP fragment, CFP(1-158), referred to as CFP-N (green), which contains the chromophore, is fused to G (blue). Linker sequences between the CFP fragments and the G protein subunits are orange. The structures of CFP-N and CFP-C are adapted from the structure of GFP (Ormo et al., 1996). The structures of and are from the structure of an t/i1 chimera complexed with tt (Lambright et al., 1996). B, CFP-C-/CFP-N- dimers containing 1 complexed with 1, 2, 5, 7, 10, 11, and 12 exhibit similar plasma membrane localization patterns in HEK-293 cells. Cells were transfected with the indicated constructs and mRFP-Mem (membrane marker) as described under Materials and Methods. The control image is the CFP image of a cell expressing only mRFP-Mem. All images are 10-s exposures, and similar pixel ranges are displayed. CN indicates CFP-N and CC indicates CFP-C. Scale bar, 5 µm. C, average intensities of each of the CFP-C-1CFP-N- complexes. D, plasma membrane fractions of the CFP-C-1CFP-N- complexes. Values for C and D represent the means ± S.E of 147 to 243 measurements.
Coexpression of CFP-C-1 with each of the CFP-N- subunits produced a fluorescent signal in the plasma membrane of HEK-293 cells that was not seen when individual subunits were expressed alone or when CFP-N and CFP-C were expressed together (Fig. 3B). These results indicate that the BiFC method can be applied to imaging complexes both when the amino terminal fragment of the fluorescent protein is attached to and the carboxyl terminal fragment is attached to and vice versa. The intensities of the different complexes varied by 2-fold or less (Fig. 3C), indicating that CFP-C-1 interacts similarly with each of the CFP-N- subunits when coexpressed with one of them at a time. The plasma membrane fractions of each of the different CFP-C-1CFP-N- complexes, expressed in the absence of s-CFP, were similar (Fig. 3D) and were slightly less than those of YFP-N-1YFP-C- dimers coexpressed with s-CFP (Fig. 2D), suggesting that s plays a role in targeting , as observed previously (Takida and Wedegaertner, 2003). However, this effect seems to be much slighter than that of on s targeting.
Seven CFP-N- Subunits Vary in Their Abilities to Compete with YFP-N-2 for Association with CFP-C-1. Because it is likely that the predominance of particular dimers in individual cells is influenced by the relative association preferences of the expressed and subunits, we postulated that preferential interactions between 1 and particular subunits might be revealed when a limiting amount of CFP-C-1 was coexpressed with a CFP-N- and a YFP-N-. To test for this, we compared the abilities of each of the CFP-N- subunits to compete with YFP-N-2 for binding to CFP-C-1 by measuring the cyan fluorescence (from a CFP-C-1CFP-N- complex) and yellow fluorescence (from CFP-C-1YFP-N-2) obtained when CFP-C-1 was coexpressed with the CFP-N- subunit and YFP-N-2. Figure 4A shows plots of YFP versus CFP intensity for individual HEK-293 cells transfected with equal amounts of plasmid encoding either CFP-C-1 and YFP-N-2 (red circles), CFP-C-1 and a CFP-N- subunit (blue circles,), or CFP-C-1, YFP-N-2, and a CFP-N- subunit (green circles). Cells expressing only CFP-C-1 and YFP-N-2 (Fig. 4A, red circles) or only CFP-C-1 and a CFP-N- subunit (Fig. 4A, blue circles), exclusively exhibited yellow or cyan fluorescence, respectively, whereas cells coexpressing CFP-C-1, YFP-N-2, and a CFP-N- subunit (Fig. 4A, green circles), exhibited both yellow and cyan fluorescence. Under these transfection conditions, cells expressing CFP-C-1 and YFP-N-2 exhibited an average yellow fluorescence intensity of 8.74 (S.E. = 0.48, n = 182), whereas cells expressing CFP-C-1 and CFP-N-2 exhibited an average cyan fluorescence intensity of 9.53 (S.E. = 0.58, n = 178). Cells coexpressing CFP-C-1, YFP-N-2, and CFP-N-2 exhibited average yellow and cyan fluorescence intensities of 3.53 (S.E. = 0.21, n = 180) and 5.09 (S.E. = 0.25, n = 180), respectively, indicating that CFP-C-1 was limiting. Figure 4B shows representative cells expressing CFP-C-1YFP-N-2, CFP-C-1CFP-N-2, or both. Because complexes do not dissociate in the absence of denaturants (Clapham and Neer, 1997), except for 5 complexes (Jones et al., 2004), the relative abilities of CFP-N- and YFP-N- subunits to compete for interaction with CFP-C-1 presumably reflect their abilities to associate with this subunit. Differences in association could reflect variations in the affinities of the subunits for 1, but other factors such as differential targeting to distinct cellular compartments or differences in interactions with chaperonins or other associated proteins (Clapham and Neer, 1997; Lukov et al., 2005; Li et al., 2006) could also regulate complex formation in vivo.
Fig. 4. Competition between different subunits to form complexes with 1 using multicolor BiFC in individual microscopically imaged cells. HEK-293 cells were transfected as described under Materials and Methods. A, plots of YFP versus CFP intensity for individual cells transfected with plasmids encoding either CFP-C-1 and YFP-N-2 (red), CFP-C-1 and a CFP-N- subunit (blue), or CFP-C-1, YFP-N-2, and a CFP-N- subunit (green). One hundred seventy-eight to 310 cells were imaged for each condition. For clarity of presentation, only cells with intensities less than 25 are shown. The average autofluorescence signals in the YFP channel (1.36, S.E. = 0.03, n = 1390) and in the CFP channel (0.95, S.E. = 0.02, n = 1397), corresponding to the average intensity of cells transfected with CFP-C-1 and a CFP-N- subunit (blue) and CFP-C-1 and YFP-N-2 (red), respectively, were subtracted before plotting and curve fitting. The linear fit of the competition data (green), calculated using Igor Pro (WaveMetrics, Inc., Lake Oswego, OR), is shown as a solid line with 95% confidence intervals indicated by dashed lines. B, representative images of cells expressing CFP-C-1YFP-N-2 or CFP-C-1CFP-N-2 alone and of the YFP and CFP signals in a cell coexpressing CFP-C-1, YFP-N-2, and CFP-N-2. The average CFP and YFP intensity values of the displayed cells are listed below the images. CN indicates CFP-N, YN indicates YFP-N, and CC indicates CFP-C. Scale bar, 5 µm.
The fluorescence intensities of cells coexpressing CFP-C-1, YFP-N-2, and a CFP-N- exhibited a linear relationship between the CFP and YFP intensities (Fig. 4A, green circles). This linearity suggests that the cells expressed each of the transfected plasmids in the same proportion, regardless of overall expression level, and also indicates that the abilities of the CFP-N- subunits to compete with YFP-N-2 for association with CFP-C-1 were the same over the observed range of expression levels. The slopes of linear fits to these data indicate the relative abilities of the CFP-N- subunits to compete with YFP-N-2 for dimerization with CFP-C-1, with the relative efficacies varying inversely with the slopes. These slope values spanned a range of 5-fold (Fig. 4A). CFP-N-2 was the most effective competitor, followed by CFP-N-7. The other CFP-N- subunits were clearly less effective competitors.
To compare the abilities of the seven subunits to compete for 1 more precisely, we measured the fluorescence intensities of populations of HEK-293 cells in a spectrofluorometer. This approach was justified by the linear relationship between YFP and CFP intensities of individual cells over a wide range of intensities (Fig. 4A) and made it possible to pool results from a much larger number of cells than was feasible by imaging individual cells (on the order of 106 compared with 102). This method made it possible to determine the relative amounts of each competitor that decreased the intensity of CFP-C-1YFP-N-2 by 50%, enabling finer distinctions to be made among the set of subunits. For these experiments, we used a modified version of ECFP, referred to as Cerulean, which is 2.5-fold brighter than ECFP (Rizzo et al., 2004), to produce Cer-N- subunits. The fluorescence intensity of Cer-N-2CFP-C-1 was 3.96-fold greater than that of CFP-N-2CFP-C-1 (S.E. = 0.8, n = 3).
Cells were transfected with a range of amounts of Cer-N- subunit plasmids to enable determination of BiFC intensities and competitive abilities as a function of plasmid dose, keeping the total amount of plasmid constant with empty vector. In the presence of an excess of cotransfected CFP-C-1 plasmid, linear relationships between the amounts of transfected Cer-N- subunit plasmids and CFP-C-1Cer-N- intensities were obtained (Fig. 5A). The relative amounts of complexes formed between the different Cer-N- subunits and CFP-C-1, obtained from the slopes of linear fits to the data, varied by less than 3-fold, with CFP-C-1Cer-N-2 and CFP-C-1Cer-N-12 exhibiting the greatest and least intensities, respectively (Table 2). Using an anti-GFP antibody, the relative expression levels of the Cer-N- subunits were determined by immunoblotting membranes from cells transfected in the same way as when BiFC intensities were compared (Fig. 5B). The relationship between Cer-N- subunit expression level and amount of transfected plasmid was also linear (Fig. 5C). The Cer-N- subunit expressed at the lowest level, Cer-N-12, exhibited levels that were 21% of the one expressed at the highest level, Cer-N-2 (Fig. 5D, Table 2). The ratios of the CFP-C-1Cer-N- intensities (Fig. 5A, Table 2) to the expression levels of the corresponding Cer-N- subunits (Fig. 5D, Table 2) were used to normalize the 1-interacting abilities of the different subunits. These ratios were fairly similar, varying by 2-fold or less (Fig. 5E). These results, in agreement with the results from imaging individual cells (Figs. 2B and 3C), suggest that the abilities of different subunits to form complexes with 1, when tested one at a time, are similar.
As was seen when the fluorescence intensities of individual cells were measured (Fig. 4), there were differences in the abilities of the Cer-N- subunits to compete with YFP-N-2 for interaction with limiting amounts of CFP-C-1 in cell populations (Fig. 6). In agreement with the microscope studies, Cer-N-2 caused the largest decrease in intensity of CFP-C-1YFP-N-2, followed by Cer-N-7 (Fig. 6A, Table 2). The amount of Cer-N-2 plasmid that reduced CFP-C-1YFP-N-2 intensity by 50% was 15% of that of the weakest competitor, Cer-N-10 (Table 2). To control for Cer-N- expression levels, the plasmid amounts of these subunits were multiplied by their relative expression levels (Fig. 5D, Table 2). The most significant difference that resulted from this normalization was that Cer-N-12, which was expressed at the lowest level of the Cer-N- subunits, became the most effective competitor (Fig. 6B, Table 2). In addition, the lower level of Cer-N-10 expression compared with that of Cer-N-1 caused Cer-N-1 to become the weakest competitor (Fig. 6B, Table 2). Thus, 1 was the least effective subunit both at competing for association with 1 and, when complexed with 1, at targeting s to the plasma membrane (Fig. 2, A and C). When expression levels were controlled for, Cer-N-12 was 4.5 times more effective at competing for association with CFP-C-1 than was Cer-N-1 (Fig. 6B, Table 2).
Previous studies of the specificity of G protein complex formation relied on comparisons of interactions between single pairs of and subunits and generally were not conducted in intact mammalian cells. Given that cells coexpress multiple isoforms of and , we sought to determine whether the predominance of particular complexes might be determined by the relative association preferences of the subunits. We found that multicolor BiFC can be used to compare the abilities of different subunits to compete for limiting amounts of a shared subunit in live cells. This strategy demonstrated a 4.5-fold range in the association preferences of 1 for seven subunits. CFP-N-12 and CFP-N-1 were the strongest and weakest competitors, respectively. Differences of this magnitude were not seen using single-color BiFC (in which each subunit was individually coexpressed with an excess of 1), using the yeast two-hybrid system (Yan et al., 1996), immunoprecipitation of complexes from tissue extracts (Asano et al., 1999), or analysis of subunits synthesized in vitro using reticulocyte lysates (Schmidt et al., 1992; Ray et al., 1995; Dingus et al., 2005).
Preferential association of particular and subunits is likely to be of functional importance in regulating interactions between G protein-coupled receptors and G proteins. For instance, ribozyme-mediated suppression of 7 in HEK-293 cells specifically reduced expression of 1 and disrupted activation of Gs by -adrenergic and D1 dopamine receptors, but not by prostaglandin E1 and D5 dopamine receptors (Wang et al., 1997, 1999, 2001). In mice lacking 7, D1 dopamine receptor-mediated stimulation of adenylyl cyclase activity in the striatum was lost (Schwindinger et al., 2003). Moreover, an in vitro study comparing the abilities of i11 complexes to produce high affinity agonist binding to the 2a-adrenergic receptor showed that complexes containing 2, 3, 4, 7, and 11 were 3 times more effective than complexes containing 5 and 10, and 30 times more effective than i111 (Richardson and Robishaw, 1999). Another study comparing the abilities of i11 complexes containing 1, 2, 7, 10, or 11 to produce high affinity agonist binding to 2a-adrenergic, A1 adenosine, 5-hydroxytryptamine1A, and µ-opioid receptors found a 3- to 8-fold range in potencies; i1111 was the most effective for A1 adenosine and 5-hydroxytryptamine1A receptors and i117 was the most potent for the 2a-adrenergic and µ-opioid receptors (Lim et al., 2001). In addition, the magnitude of activation of i142 by endogenous 2-adrenergic receptors in HeLa cells was 3 times that of i122 as determined by FRET analysis (Gibson and Gilman, 2006) and 42 was 12 times more effective than 12 at coupling Gs to the A2a adenosine receptor in a reconstituted system (McIntire et al., 2001).
The role of in mediating specificity at the level of interactions with subunits and effectors, compared with receptors, may be more variable. Regarding subunit interaction, we did not observe large differences in the abilities of 1 complexes to target s-CFP to the plasma membrane. 11 was 2-fold less effective than 110, the most effective complex, but it also exhibited less plasma membrane targeting itself. It is possible that greater differences might have been seen among a wider selection of dimers in interactions with s as well as with other subunits. For instance, a previous study examined the ability of 1-5 in combination with 2 or 3 to promote plasma membrane targeting of -binding deficient mutants of s and q (Evanko et al., 2001) and found that whereas 13 was equally effective as 12 at targeting mutant s, 5 and 3 complexes were ineffective and weak, respectively, and 4 complexes effectively targeted mutant s but not mutant q. The preferences of subunits and complexes for each other in targeting assays could be due to differences in relative affinities, accessibilities, or other cellular factors. Direct comparisons of the affinities of and subunits for each other are more readily carried out in vitro. For instance, a comparison of the abilities of 1 complexes containing 1, 2, 7, 10, or 11 to compete for binding to fluorescein-labeled i1 in a flow cytometry assay demonstrated that 11 was 2- to 5-fold less potent than the other complexes (Lim et al., 2001). At the level of effector interaction, many combinations exhibit similar abilities to modulate effector proteins (Iniguez-Lluhi et al., 1992), although preferential interactions have been seen (McIntire et al., 2001). In addition, effector regulation by in vivo can be more selective than that in vitro (Diverse-Pierluissi et al., 2000). Moreover, the pronounced phenotypes obtained using targeted deletions of subunits (Schwindinger et al., 2003, 2004) suggest that differences in the interactions of subunits with receptors, subunits, and effectors, although not always robust when examined individually, may have additive effects that lead to a high degree of signaling specificity.
The ability of multicolor BiFC to determine preferences in the formation of particular complexes in live cells is advantageous because in vivo factors can affect assembly. For instance, the 90-kDa heat shock protein can be immunoprecipitated with nondimerized subunits but not complexes (Clapham and Neer, 1997); binding of phosducin-like protein to 1 is required for association with 2 in vivo (Lukov et al., 2005), and an immature form of 13 interacts with PDZ-containing proteins (Li et al., 2006). It is possible that coexpression of phosducin or another retina-specific chaperonin might cause 1, which interacts with 1 in the retina, to compete more effectively for 1 in HEK-293 cells. In addition, relationships between subcellular compartmentalization and association of particular and subunits can only be investigated using intact cells. 5 can associate with focal adhesions (Hansen et al., 1994), whereas 12 can associate with actin filaments (Ueda et al., 1997). Application of BiFC will enable determination of the subunits with which these subunits are associated and may demonstrate differential localization patterns depending on the associated subunit. For instance, we found previously that 5 complexes localize to the cytoplasm, either diffusely or on intracellular membranes, depending on the associated subunit, whereas the corresponding 1 complexes localize to the plasma membrane (Hynes et al., 2004b).
Comparing the abilities of additional and subunits to interact with each other using multicolor BiFC will provide a more comprehensive picture of which complexes are likely to predominate in particular cells. For instance, the presence of other subunits will affect which subunits interact preferentially with 1. Based on analysis of the subunit composition of complexes purified from bovine tissues, 5 and 12 seem to associate selectively with 4 (Asano et al., 1999). If 12 interacts preferentially with 4 rather than 1, this could explain why 12 could not prevent a decrease in the level of 1 upon ribozyme-mediated depletion of 7 in HEK-293 cells (Wang et al., 1997). It is not clear why 2, which also competed for 1 more effectively, could not substitute for 7, but it is possible that 2 is expressed at a lower level or interacts preferentially with a different subunit in HEK-293 cells. To address such questions, multicolor BiFC can be applied to additional analyses of subunits competing for subunits as well as of subunits competing for subunits.
Cell-type specific patterns of and expression, as well as their association preferences, will determine which complexes predominate in particular cells. For instance, 5 and 12 are the major subunits in HEK-293, HeLa, and BRL-3A cells, whereas 2 and 5 prevail in F9 and NG108-15 cells (Ueda et al., 1998). In addition, differences in expression levels have been observed during development. One study found that retinoic acid-induced differentiation of HL-60 cells into neutrophil-like cells involves induction of 2 expression and potentiation of fMLP stimulation of phospholipase C via Gi (Iiri et al., 1995). Another study found that 5 is the predominant subunit in undifferentiated HL-60 cells, and retinoic acid-induced differentiation induces expression of both 2 and 7, with 2 replacing most of the 5 (Ueda et al., 1998). Likewise, 2 replaces 5 during neuronal differentiation in rat brain (Morishita et al., 1999). Because interactions between specific G protein subunits seem to be important for mutual stabilization (Wang et al., 1999; Schwindinger et al., 2003), alterations in the expression of particular subunits may cause changes in the levels of their binding partners.
Taken together with demonstrations that targeted deletions of specific subunits can produce unique effects and that both and composition play a role in determining receptor specificity, our results showing preferences of 1 for particular subunits suggest that multiple levels of selective interactions contribute to G protein signaling specificity. Combining diverse approaches to compare the expression levels of the G protein subunits, their preferences for association, and the preferences of G protein-coupled receptors for particular heterotrimers will elucidate which complexes are most likely to form in a particular cellular environment and mediate specific signaling pathways.
Acknowledgements
We thank David Piston for the monomeric cerulean plasmid, Roger Tsien for the monomeric RFP1 plasmid, Janet Robishaw for plasmids expressing and subunits, and Gerda Breitwieser for helpful discussions and critical reading of the manuscript.
ABBREVIATIONS: HEK, human embryonic kidney; BiFC, bimolecular fluorescence complementation; YFP, yellow fluorescent protein; CFP, cyan fluorescent protein; ECFP, enhanced cyan fluorescent protein; mRFP, monomeric red fluorescent protein; Cer, monomeric cerulean protein.
【参考文献】
Asano T, Morishita R, Ueda H, and Kato K (1999) Selective association of G protein 4 with 5 and 12 subunits in bovine tissues. J Biol Chem 274: 21425-21429.[Abstract/Free Full Text]
Clapham DE and Neer EJ (1997) G protein beta gamma subunits. Annu Rev Pharmacol Toxicol 37: 167-203.
Dingus J, Wells CA, Campbell L, Cleator JH, Robinson K, and Hildebrandt JD (2005) G protein betagamma dimer formation: Gbeta and Ggamma differentially determine efficiency of in vitro dimer formation. Biochemistry 44: 11882-11890.
Diverse-Pierluissi M, McIntire WE, Myung CS, Lindorfer MA, Garrison JC, Goy MF, and Dunlap K (2000) Selective coupling of G protein complexes to inhibition of Ca2+ channels. J Biol Chem 275: 28380-28385.[Abstract/Free Full Text]
Evanko DS, Thiyagarajan MM, Siderovski DP, and Wedegaertner PB (2001) G isoforms selectively rescue plasma membrane localization and palmitoylation of mutant Gs and Gq. J Biol Chem 276: 23945-23953.[Abstract/Free Full Text]
Evanko DS, Thiyagarajan MM, and Wedegaertner PB (2000) Interaction with G is required for membrane targeting and palmitoylation of Gs and Gq. J Biol Chem 275: 1327-1336.[Abstract/Free Full Text]
Figler RA, Lindorfer MA, Graber SG, Garrison JC, and Linden J (1997) Reconstitution of bovine A1 adenosine receptors and G proteins in phospholipid vesicles: betagamma-subunit composition influences guanine nucleotide exchange and agonist binding. Biochemistry 36: 16288-16299.
Gibson SK and Gilman AG (2006) Gialpha and Gbeta subunits both define selectivity of G protein activation by alpha2-adrenergic receptors. Proc Natl Acad Sci USA 103: 212-217.[Abstract/Free Full Text]
Hansen CA, Schroering AG, Carey DJ, and Robishaw JD (1994) Localization of a heterotrimeric G protein gamma subunit to focal adhesions and associated stress fibers. J Cell Biol 126: 811-819.[Abstract/Free Full Text]
Hou Y, Azpiazu I, Smrcka A, and Gautam N (2000) Selective role of G protein subunits in receptor interaction. J Biol Chem 275: 38961-38964.[Abstract/Free Full Text]
Hu CD, Chinenov Y, and Kerppola TK (2002) Visualization of interactions among bZIP and Rel family proteins in living cells using bimolecular fluorescence complementation. Mol Cell 9: 789-798.
Hu CD and Kerppola TK (2003) Simultaneous visualization of multiple protein interactions in living cells using multicolor fluorescence complementation analysis. Nat Biotechnol 21: 539-545.
Hynes TR, Mervine SM, Yost EA, Sabo JL, and Berlot CH (2004a) Live cell imaging of Gs and the 2-adrenergic receptor demonstrates that both s and 17 internalize upon stimulation and exhibit similar trafficking patterns that differ from that of the 2-adrenergic receptor. J Biol Chem 279: 44101-44112.[Abstract/Free Full Text]
Hynes TR, Tang L, Mervine SM, Sabo JL, Yost EA, Devreotes PN, and Berlot CH (2004b) Visualization of G protein dimers using bimolecular fluorescence complementation demonstrates roles for both and in subcellular targeting. J Biol Chem 279: 30279-30286.[Abstract/Free Full Text]
Iiri T, Homma Y, Ohoka Y, Robishaw JD, Katada T, and Bourne HR (1995) Potentiation of Gi-mediated phospholipase C activation by retinoic acid in HL-60 cells. Possible role of G2. J Biol Chem 270: 5901-5908.[Abstract/Free Full Text]
Iniguez-Lluhi JA, Simon MI, Robishaw JD, and Gilman AG (1992) G protein subunits synthesized in Sf9 cells. Functional characterization and the significance of prenylation of . J Biol Chem 267: 23409-23417.[Abstract/Free Full Text]
Jones MB, Siderovski DP, and Hooks SB (2004) The G{beta}{gamma} dimer as a novel source of selectivity in G-protein signaling: GGL-ing at convention. Mol Interv 4: 200-214.[Abstract/Free Full Text]
Lambright DG, Sondek J, Bohm A, Skiba NP, Hamm HE, and Sigler PB (1996) The 2.0 ? crystal structure of a heterotrimeric G protein. Nature (Lond) 379: 311-319.
Li Z, Benard O, and Margolskee RF (2006) G13 interacts with PDZ domain-containing proteins. J Biol Chem 281: 11066-11073.[Abstract/Free Full Text]
Lim WK, Myung CS, Garrison JC, and Neubig RR (2001) Receptor-G protein gamma specificity: gamma11 shows unique potency for A(1) adenosine and 5-HT(1A) receptors. Biochemistry 40: 10532-10541.
Lukov GL, Hu T, McLaughlin JN, Hamm HE, and Willardson BM (2005) Phosducin-like protein acts as a molecular chaperone for G protein betagamma dimer assembly. EMBO (Eur Mol Biol Organ) J 24: 1965-1975.
McIntire WE, MacCleery G, and Garrison JC (2001) The G protein subunit is a determinant in the coupling of Gs to the 1-adrenergic and A2a adenosine receptors. J Biol Chem 276: 15801-15809.[Abstract/Free Full Text]
Morishita R, Shinohara H, Ueda H, Kato K, and Asano T (1999) High expression of the gamma5 isoform of G protein in neuroepithelial cells and its replacement of the gamma2 isoform during neuronal differentiation in the rat brain. J Neurochem 73: 2369-2374.
Ormo M, Cubbitt AB, Kallio K, Gross LA, Tsien RY, and Remington SJ (1996) Crystal structure of the Aequoria victoria green fluorescent protein. Science (Wash DC) 273: 1392-1395.
Ostrom RS (2002) New determinants of receptor-effector coupling: trafficking and compartmentation in membrane microdomains. Mol Pharmacol 61: 473-476.[Free Full Text]
Ray K, Kunsch C, Bonner LM, and Robishaw JD (1995) Isolation of cDNA clones encoding eight different human G protein subunits, including three novel forms designated the 4, 10 and 11 subunits. J Biol Chem 270: 21765-21771.[Abstract/Free Full Text]
Richardson M and Robishaw JD (1999) The 2A-adrenergic receptor discriminates between Gi heterotrimers of different subunit composition in Sf9 insect cell membranes. J Biol Chem 274: 13525-13533.[Abstract/Free Full Text]
Rizzo MA, Springer GH, Granada B, and Piston DW (2004) An improved cyan fluorescent protein variant useful for FRET. Nat Biotechnol 22: 445-449.
Robishaw JD and Berlot CH (2004) Translating G protein subunit diversity into functional specificity. Curr Opin Cell Biol 16: 206-209.
Schmidt CJ, Thomas TC, Levine MA, and Neer EJ (1992) Specificity of G protein and subunit interactions. J Biol Chem 267: 13807-13810.[Abstract/Free Full Text]
Schwindinger WF, Betz KS, Giger KE, Sabol A, Bronson SK, and Robishaw JD (2003) Loss of G protein 7 alters behavior and reduces striatal olf level and cAMP production. J Biol Chem 278: 6575-6579.[Abstract/Free Full Text]
Schwindinger WF, Giger KE, Betz KS, Stauffer AM, Sunderlin EM, Sim-Selley LJ, Selley DE, Bronson SK, and Robishaw JD (2004) Mice with deficiency of G protein gamma3 are lean and have seizures. Mol Cell Biol 24: 7758-7768.[Abstract/Free Full Text]
Takida S and Wedegaertner PB (2003) Heterotrimer formation, together with isoprenylation, is required for plasma membrane targeting of G. J Biol Chem 278: 17284-17290.[Abstract/Free Full Text]
Ueda H, Morishita R, Katoh-Semba R, Kato K, and Asano T (1998) G protein gamma subunits coimmunoprecipitated with antibodies against alpha subunits: identification of major isoforms in cultured cells by silver stain and immunoblotting with conventional transfer procedure. J Biochem (Tokyo) 124: 1033-1037.[Abstract/Free Full Text]
Ueda H, Saga S, Shinohara H, Morishita R, Kato K, and Asano T (1997) Association of the gamma12 subunit of G proteins with actin filaments. J Cell Sci 110: 1503-1511.
Wang Q, Jolly JP, Surmeier JD, Mullah BM, Lidow MS, Bergson CM, and Robishaw JD (2001) Differential dependence of the D1 and D5 dopamine receptors on the G protein 7 subunit for activation of adenylylcyclase. J Biol Chem 276: 39386-39393.[Abstract/Free Full Text]
Wang Q, Mullah B, Hansen C, Asundi J, and Robishaw JD (1997) Ribozyme-mediated suppression of the G protein 7 subunit suggests a role in hormone regulation of adenylylcyclase activity. J Biol Chem 272: 26040-26048.[Abstract/Free Full Text]
Wang Q, Mullah BK, and Robishaw JD (1999) Ribozyme approach identifies a functional association between the G protein 17 subunits in the -adrenergic receptor signaling pathway. J Biol Chem 274: 17365-17371.[Abstract/Free Full Text]
Wedegaertner PB, Wilson PT, and Bourne HR (1995) Lipid modifications of trimeric G proteins. J Biol Chem 270: 503-506.[Free Full Text]
Yan K, Kalyanaraman V, and Gautam N (1996) Differential ability to form the G protein complex among members of the and subunit families. J Biol Chem 271: 7141-7146.[Abstract/Free Full Text]
作者单位：Weis Center for Research, Geisinger Clinic, Danville, Pennsylvania