|Home | About | Journals | Submit | Contact Us | Français|
Characterization of G protein βγ dimer isoform expression in different cellular contexts has been impeded by low levels of protein expression, broad isoform heterogeneity, and antibodies of limited specificity, sensitivity or availability. As a new approach, we used quantitative mass spectrometry to characterize native βγ dimers associated with adenosine A1:αi1 and adenosine A2A:αS receptor fusion proteins expressed in HEK-293 cells. Cells expressing A1:αi1 were cultured in media containing [13C6] Arg and [13C6] Lys, and βγ labeled with heavy isotopes purified. Heavy βγ was combined with either recombinant βγ purified from Sf9 cells, βγ purified from the A2A:αS expressed in HEK-293 cells cultured in standard media, or an enriched βγ fraction from HEK-293 cells. Samples were separated by SDS-PAGE, and protein bands containing β and γ were excised, digested with trypsin, separated by HPLC and isotope ratios analyzed by mass spectrometry. Three β isoforms, β1, β2 and β4, and seven γ isoforms, γ2, γ4, γ5, γ7, γ10, γ11 and γ12 were identified in the analysis. β1 and γ5 were most abundant in the enriched βγ fraction, and this βγ profile was generally mirrored in the fusion proteins. However, both A2A:αS and A1:αi1 bound more β4 and γ5 compared to the enriched βγ fraction; also, more β4 was associated with A2A:αS than A1:αi1. Both fusion proteins also contained less γ2, γ10 and γ12 than the enriched βγ fraction. These results suggest that preferences for particular βγ isoforms may be driven in part by structural motifs common to adenosine receptor family members.
The G protein1 βγ dimer participates in the initiation of signaling cascades by coupling Gα subunits to G protein coupled receptors (1), and once activated, βγ dimers can interact with and regulate a multitude of signaling proteins (2). Function of the Gα isoforms has been well established with respect to specific receptor coupling and downstream signaling pathways. However, the five β and 12 γ isoforms form a diverse constellation of βγ dimers (3-5), the functional significance of which is only beginning to be appreciated. A number of powerful genetic approaches, including homologous recombination (6;7) and RNA interference (8;9) have emerged to allow deletion or attenuation of β or γ genes of interest. Results of these studies revealed that regulation of specific β and γ isoforms is tightly integrated to many elements of G protein coupled receptor signaling pathways. Furthermore, the advent of real time PCR has enabled the analysis of transcriptional regulation with great precision (9). In contrast, characterization of β and γ isoforms at the protein level has relied predominantly on antibodies; limitations in this approach, such as cross reactivity and poor sensitivity, make quantitative characterization of this family of highly related proteins fraught with difficulty.
One advance in proteomics has been the development of SILAC (Stable Isotope Labeling with Amino Acids in Cell Culture) for the quantitation of proteins by mass spectrometry (10). An advantage of SILAC is that protein standards can be combined with samples and treated identically during the sample preparation steps necessary for mass spectrometry, thus allowing protein quantitation with great precision. This study describes a general procedure for purifying endogenous βγ dimers from cells by expressing either an epitope tagged adenosine A1 receptor:αi1 (A1:αi1) fusion protein or adenosine A2A receptor:αS (A2A:αS) fusion protein. After immobilization as an R:G complex on affinity beads, the receptor fusion protein can release native βγ when activated with AlF−4. When used in conjunction with SILAC and LC MS/MS mass spectrometry, this technique demonstrates femtomole sensitivity, the capability to identify and quantify individual β and γ isoforms in a mixed βγ population, and the precision to discern differences in βγ composition among different adenosine receptor G protein complexes and the overall βγ profile in a cell. These attributes combine to provide a powerful approach that can be used to characterize G protein β and γ isoforms under a variety of experimental conditions.
The baculovirus expressing the human γ5 subunit was engineered by digestion of the pcDNA3.1+ plasmid containing γ5 (Missouri S&T cDNA Resource Center) with PmeI in order to generate a blunt end γ5 insert. The baculovirus expression vector pFastBac1™ (Invitrogen) was digested with StuI to generate linear DNA with blunt ends; the γ5 insert was ligated into the vector, and cDNAs from positive clones were screened for correct orientation of insert, and verified for correct sequence. DH10bac™ cells (Invitrogen) were transformed with the pFastBac1™ vector containing γ5, and screening of bacmid DNA from positive clones for correct transposition was achieved by PCR. The origin of the baculoviruses encoding 6HIS-Gi1 α, β1, β2, β4, γ2, γ7, γ10, γ11, and γ12 has been published elsewhere (11;12).
Recombinant baculoviruses encoding the desired combination of 6HIS-Gi1 α subunit and βγ dimer were used to infect Sf9 cells at an MOI of three, which were then harvested and used to purify recombinant βγ dimers as described (11). This purification scheme yields a highly pure preparation of recombinant βγ dimer of defined composition (Fig. 1).
PCR primers for individual β and γ isoforms were designed using Beacon Designer™ software and tested using end point PCR to verify a single amplicon product. Amplicons were sequenced to verify the fidelity of the primer target interaction; validated primer sets are listed in Table 1.
Total cell RNA was extracted using the RNeasy Minikit (Qiagen); cDNAs were created with 1 μg RNA using the iScript cDNA Synthesis Kit (BioRad) and quantitative real time PCR was performed using the iQ SYBER Green Supermix (BioRad) in an iCycler PCR machine (BioRad). The ribosomal protein 13A was used as an internal control reference gene. Normalization of the target gene was accomplished by using the formula 2(Et-Rt), where Et and Rt are the threshold cycles for the experimental and reference genes, respectively (13).
A plasmid encoding a fusion protein of the human adenosine A1 receptor and rat Gi1α was kindly provided by Dr. Graeme Milligan (University of Glasgow, Scotland, UK). Restriction enzymes BamH1 and EcoRV were used to generate an insert consisting of the 3′ end of the A1 receptor and the entire Gi1 α subunit. The same restriction enzymes were used to digest the vector pDoubleTrouble containing the adenosine A1 receptor (14), and the fusion protein insert was ligated into the purified linearized pDoubleTrouble vector containing the HIS and FLAG epitope tags and the 5′ end of the A1 receptor.
Vector pcDNA3.1+ containing the gene for the human GS α short (Missouri S&T cDNA Resource Center), was modified by PCR mutagenesis in order to facilitate the fusion of human adenosine A2A receptor to GS α. Two endogenous SmaI sites were changed in order to eliminate the restriction site: a C to G mutation in the non-coding backbone region of the plasmid DNA; the other site was internal to the GS α cDNA in which nucleotide 963(G) was mutagenized to a (C), resulting in a silent mutation at residue R321. During the same multi-mutagenesis reaction (Stratagene, La Jolla, CA) a SmaI restriction site was incorporated at the 5′end of the GS α cDNA. Nucleotides 1A, 2T, 3G and 6C of the GS α cDNA were changed to CCC and G, respectively, resulting in a SmaI site. Construction of the A2A:GS α fusion was completed utilizing standard PCR techniques to amplify the wild-type A2A gene using modified primers encoding exogenous restriction sites KpnI at the 5′, TTA AAC TTA AGC TTG GTA CCA TGC CCA TCA TGG GCT CCT and NcoI at the 3′, CCC GAG GCA GCC CAT GGA CAC TCC TGC TCC ATC CT, termini. The PCR product was digested with NcoI and filled-in using Klenow to generate a blunt end. Following subsequent digestion with KpnI, the product was subcloned by ligation into the modified pcDNA3.1+ GS α vector that had been digested with KpnI and SmaI. The A2A- GS α fusion protein construct was subcloned into the vector pDoubleTrouble by digestion of the pcDNA3.1+ vector containing A2A:GS α with BstEII and PmeI to produce the fusion protein insert with a blunt 3′ end. A pDoubleTrouble vector containing the A2A receptor (14) was digested with BstEII and EcoRV to produce a linearized empty vector with a blunt end at the EcoRV site; the fusion protein insert was than ligated into the vector, resulting in a pDoubleTrouble vector containing an A2A:GS α fusion protein with a HIS/FLAG tag at the N-terminus.
Human HEK-293 cells were cultured in Dulbecco's Modified Eagle Medium (DMEM) with 10% fetal bovine serum. Stable cell lines expressing the adenosine A1 or A2A receptor fusion proteins were generated by supplementing the media with G418 (500 μg/ml, final concentration). For SILAC conditions used to generate heavy βγ dimers associated with the A1:αi1 fusion protein, DMEM Flex media (Invitrogen) was supplemented with 200 g/L glucose, 200 mM L-glutamine, 10 g/L phenol red, 10% dialyzed FBS (Invitrogen) (10,000 mw cutoff), 179.6 mg/L L-lysine: 2HCl [13C6] and 86.2 mg/L L- arginine:HCl [13C6] (Cambridge Isotope Laboratories). In order to fully incorporate the heavy amino acids, cells were cultured in SILAC media for five doubling times. Cells cultured in either light or heavy media were harvested by trituration with PBS containing 5 mM EDTA, washed with PBS, collected by centrifugation, and resuspended in buffer containing 20 mM HEPES, pH 7.4, 1 mM EGTA, and 100 μg/ml Pefabloc SC Plus, 2 μg/ml pepstatin, leupeptin, aprotinin and 20 μg/ml benzamidine before flash freezing and storage at −80 °C.
Membranes from HEK-293 cells were collected after lysis with a 21 g needle and centrifugation at 53,000 RPM in a 90 Ti rotor for 45 minutes at 4 °C. Buffer containing 20 mM Tris, pH 8.0, 1 mM EDTA, 1% cholate, 1 mM DTT, 5 μM GDP, 100 μg/ml Pefabloc SC Plus, 2 μg/ml pepstatin, leupeptin, aprotinin and 20 μg/ml benzamidine was used to extract βγ dimers, either alone or in heterotrimeric form. The cholate extract was subjected to DEAE chromatography as described in Graber et al. (15). Fractions containing β-common immunoreactivity determined by SDS-PAGE and western blotting were pooled, concentrated, and resolved by Superose 6 size exclusion chromatography as described in McIntire et al. (11). Fractions from the Superose 6 separation containing β-common immunoreactivity were pooled, concentrated and frozen at −80 °C.
Cell pellets from approximately 30 15 cm plates of HEK-293 cells expressing the adenosine A1:αi1 or A2A:αS fusion protein were lysed by nitrogen cavitation, and membranes collected by centrifugation at 35,000 RPM in a 45 Ti rotor for 45 minutes at 4 °C. Membranes were washed with HNG buffer (20 mM HEPES, pH 7.4, 20 mM NaCl and 10% glycerol) containing 100 μg/ml Pefabloc SC Plus, 2 μg/ml pepstatin, leupeptin, aprotinin and 20 μg/ml benzamidine, 5 μM GDP, and resuspended to a volume of 10 mg/ml protein with HNG buffer containing 1 mM EDTA, 1% n-dodecyl β-D-maltoside (DDM), 0.02% cholesteryl hemisuccinate, 100 μM adenosine, 100 μg/ml Pefabloc SC Plus, 2 μg/ml pepstatin, leupeptin, aprotinin and 20 μg/ml benzamidine. All buffers used in the purification procedure were 0.22 μm filtered, and all steps were performed at 4 °C unless otherwise noted (see Fig. 1 for flow chart of purification). After stirring for two hours, the DDM extract containing the receptor fusion protein was clarified by centrifugation as described above, and diluted to approximately 0.5% DDM with HNG buffer containing 1 mM EDTA, 100 μM adenosine, 100 μg/ml Pefabloc SC Plus, 2 μg/ml pepstatin, leupeptin, aprotinin and 20 μg/ml benzamidine. The diluted extract was allowed to incubate with 200 μl of FLAG M2 affinity resin for one hour, rocking end over end. FLAG beads were collected with a 5 ml centrifuge column, washed with 5 one ml volumes of HGN buffer containing 1 mM EDTA (HNGE buffer), 0.1 % DDM and 100 μM adenosine. The column was then washed with two 1 ml volumes of HNGE buffer containing 1% cholate and 100 μM adenosine. HNGE buffer containing 100 μM adenosine and 1% cholate was supplemented with AlF−4(Activation Buffer) for elution of βγ dimers associated with the receptor fusion protein; 200 μl of activation buffer warmed to room temperature was added to the column and collected. The column was then capped, and a second 200 μl was added and allowed to incubate for 15 minutes at room temperature in order to facilitate dissociation of βγ from receptor fusion protein. After the incubation, the second volume was collected, along with 4 more 200 μl volumes, and fractions containing βγ dimer were pooled, concentrated in an amicon concentrator, and exchanged twice with buffer containing 20 mM HEPES, pH 7.4, 20 mM NaCl, 1 mM EDTA, 0.1 % CHAPS and 1 mM DTT. The column bound receptor fusion protein and FLAG elution buffer (HGNE buffer containing 0.1 % DDM, 100 μM adenosine and 0.5 mg/ml FLAG peptide) were warmed to room temperature, and 200 μl of the elution buffer was applied to the column and collected. The column was then capped, and a second 200 μl elution volume was applied and allowed to incubate for 15 minutes. The column was then uncapped, and a total of five more 200 μl elution volumes were collected to recover the receptor fusion protein.
Samples were prepared based on the thin layer method described by Cadene et al. (16). Briefly, a thin layer matrix solution was prepared by diluting a saturated solution of cyanohydroxycinnamic acid in a 1:2 mixture of water:acetonitrile four fold with 2-propanol. A sample matrix solution was prepared by sonicating a cyanohydroxycinnamic acid saturated solution in a 3:1:2 mixture of formic acid:water:acetonitrile for 10 minutes, followed by centrifugation. The thin layer matrix was prepared by applying 10-20 μl of thin layer matrix solution on a plate and allowing it to spread. When only traces of solvent were remaining, the plate was gently wiped to leave only a thin film of matrix. The sample matrix solution was used to dilute the purified βγ samples 20 fold; within 10 minutes, 0.5 μl was loaded on the plate containing the thin layer, allowed to dry and washed with 2 μl of 0.1% TFA. Samples were then analyzed on a Bruker MicroFlex MALDI mass spectrometer in linear mode using the manufacturer's standard settings and collecting 200 shots.
Prior to gel electrophoresis, samples were incubated with 6x sample buffer at room temperature for one hour without boiling. Proteins were separated using 12% polyacrylamide gels, and visualized by staining with silver or Coomassie blue; alternatively, gels were transferred to nitrocellulose for western blotting with a β-common (sc-378, Santa Cruz) or α-common (NEI-800, DuPont NEN) antibody. Polyacrylamide gels used for generation of samples for mass spectrometric analysis were prepared by 0.22 m filtration of the separating and stacking solutions, as well as the running buffer; this step is important for removal of common protein contaminants, such as keratin, that can obscure the detection of sample proteins. Gels were stained in a 0.1% Coomassie Brilliant blue solution of 45:45:10 methanol:water:acetic acid, followed by destaining in a 45:45:10 methanol:water:acetic acid solution. Once protein bands were adequately visualized, gels were stored in a 10% acetic acid solution. The β protein, which separates from γ during SDS-PAGE, has an electrophoretic mobility of approximately 36 kDa, while the γ protein is present at the dye front. These portions of the gel were excised in order to recover protein for mass spectrometric analysis.
Gel pieces were transferred to siliconized tubes and washed and destained in 200 μl 50% methanol overnight. The gel pieces were dehydrated in acetonitrile, rehydrated in 30 μL of 10 mM dithiolthreitol in 0.1 M ammonium bicarbonate and reduced at room temperature for 0.5 h. The DTT solution was removed and the sample alkylated in 30 μL 50 mM iodoacetamide in 0.1 M ammonium bicarbonate at room temperature for 0.5 h. The reagent was removed and the gel pieces dehydrated in 100 μl acetonitrile. The acetonitrile was removed and the gel pieces rehydrated in 100 μl 0.1 M ammonium bicarbonate. The pieces were dehydrated in 100 μl acetonitrile, the acetonitrile removed and the pieces completely dried by vacuum centrifugation. The gel pieces were rehydrated in 20 ng/μl trypsin in 50 mM ammonium bicarbonate on ice for 10 min. Any excess enzyme solution was removed and 20 μL 50 mM ammonium bicarbonate added. The sample was digested overnight at 37 °C and the peptides formed extracted from the polyacrylamide in two 30 μl aliquots of 50% acetonitrile/5% formic acid. These extracts were combined and evaporated to 15 μl for MS analysis.
The LC-MS system consisted of a Thermo Electron LTQ Orbitrap XL mass spectrometer system with a Protana nanospray ion source interfaced to a self-packed 8 cm × 75 um id Phenomenex Jupiter 10 um C18 reversed-phase capillary column. 7.5 μL volumes of the extract were injected and the peptides eluted from the column by an acetonitrile/0.1 M acetic acid gradient at a flow rate of 0.4 μl/min over 1 hour. The nanospray ion source was operated at 2.5 kV. The digest was analyzed by acquiring a full scan mass spectrum using Fourier-transform ion cyclotron resonance at 100k resolving power to determine peptide molecular weights followed by 10 product ion spectra in the ion trap to determine amino acid sequence in sequential scans. This mode of analysis produces approximately 10,000 ms/ms spectra of ions ranging in abundance over several orders of magnitude. The data were analyzed by database searching using the Sequest algorithm against Human International Protein Index (v3.66).
Peptide spectra putatively identified by Sequest as belonging to G protein γ or β isoforms were manually verified. H:L peptide ratios were calculated using areas obtained from QualBrowser (Xcalibur 2.1) for the monoisotopic m/z (+/− 0.01Da) for the heavy and light forms. The H:L ratio for each protein was obtained by taking the average peptide ratio for all peptides observed for a particular isoform.
For A1:αi1 fusion protein βγ dimers purified from HEK-293 cells cultured in media containing [13C6] Arg and [13C6] Lys, protein concentration was determined by western blotting with the sc-378 β-common antibody (Santa Cruz), against a standard curve of purified recombinant βγ dimer from Sf9 cells. For β isoform quantitation by mass spectrometry, purified recombinant β1γ2, β2γ2 and β4γ2 dimers from Sf9 cells were each added to the A1:αi1 fusion protein βγ dimers at a 1:10 molar ratio. For γ isoform quantitation, purified recombinant β1γ2, β1γ5, β1γ7, β1γ10, β1γ11 and β1γ12 dimers from Sf9 cells were also each added at a 1:10 ratio. The combined samples were separated by SDS-PAGE (Fig. 1) and processed according to the procedure outlined above in SDS-PAGE and Western Blotting and Tryptic Digestion of Gel Bands and LC/MS/MS Analysis.
Typically, several peptides from each β or γ isoform (see Table 3) produced ion pairs that were used to determine an average peak area ratio between heavy and light ion traces (H:L). Expression of β and γ isoforms as μg/μl of sample was determined by multiplying the average (H:L) ratio for each isoform by the γ of each standard isoform added. The amount of Sf9 γ isoform added in a standard was determined indirectly as a function of β concentration, assuming a 1:1 β:γ ratio. In experiments comparing βγ dimers between fusion proteins, or between the A1:αi1 fusion protein and the enriched HEK-293 βγ fraction, the known heavy β and γ concentrations from the A1:αi1 fusion protein were used along with the H:L ratios to determine the relative amounts of light β and γ isoforms associated with the A2A:αS fusion protein, or present in the enriched βγ fraction from HEK-293 cells. In order to normalize the relative levels of β and γ isoforms present, each β and γ isoform was expressed as a percent of the total β and γ protein quantified, respectively, for each sample.
H:L ratios were first analyzed for variability, and ratios that were greater than two standard deviations from the mean of data sets containing at least five values were excluded from further analysis. N values higher than the number of peptides observed (Table 3, A1:αi1 vs enriched HEK-293 βγ fraction) occur when several charge states of the same peptide generate unique H:L ratios. Prior to pooling data from separate experiments, H:L ratios were converted into moles of β or γ, and then expressed as percent of total moles β or γ detected. In order to determine statistical significance, data sets were compared using the unpaired t test in GraphPad Prism® 5 to calculate two-tailed p-values.
Reagents for Sf9 cell culture and purification of βγ dimers have been described previously (17-20). GDP, CHS, adenosine, HEPES and anti-FLAG M2 agarose from Sigma; DDM from MP Biochemicals; FLAG peptide was synthesized at the University of Virginia Biomolecular Research Facility; CHAPS from Roche Molecular Biochemicals; 10% Genapol C-100 from CalBiochem; Ni2+-NTA Superflow resin from Qiagen; centricon 30 concentrators from Millipore; Superose™ 6 HR 10/30 column from Pharmacia; all other materials were of the highest available purity. Mass spectrometric analysis of peptides was performed at the W.M. Keck Biomedical Mass Spectrometry Laboratory.
A graphical overview of the process for the isolation and biochemical characterization of native βγ dimers from cells is provided in figure 1. The scheme for βγ purification from a receptor:α:βγ complex after expression of epitope tagged adenosine receptor:α fusion protein in cells cultured in “heavy” SILAC media to introduce [13C] labeled Lys and Arg into cellular proteins is presented in figure 1A. This general scheme could be used for any combination of receptor and α engineered in a fusion protein construct. Figure 1B illustrates recombinant βγ dimers purified from Sf9 cells cultured in normal or “light” media. The βγ dimers containing different β isoforms (stained with Coomassie blue, Fig. 1B, above) and different γ isoforms (stained with silver, Fig. 1B, below) were used to quantify β and γ isoforms isolated from HEK-293 cells in figure 1A. A known amount of light βγ standard is combined with the native mixture of heavy βγ derived from the receptor fusion protein for quantitative analysis by mass spectrometry. Figure 1C illustrates the purity of the combined βγ dimers from both sources after separation by SDS-PAGE and staining with Coomassie blue. The β and γ proteins from the gel in figure 1C (see boxes) are excised, digested with trypsin and analyzed by LC/MS/MS as described in Experimental Procedures.
Although the main focus of this study was to characterize the β and γ isoforms at the protein level, it was also important to correlate β and γ protein with mRNA levels. Validated human PCR primers designed to target specific β and γ isoforms are listed in Table 1; all primers were validated by sequence verification of the amplicon, with the exception of the γ2 primers, which specifically recognized a plasmid containing the target sequence. Since this study examined the βγ dimers from HEK-293 cells, the β3 and β5 primers, as well as the γ1, γ3, γ8, γ9 and γ13 primers were excluded from this analysis, as these isoforms were not observed experimentally at the protein level (21;22), and thus do not appear to have a prominent role in HEK-293 cell γ protein signaling. Of the β isoforms, β2 was the most abundant, with transcript levels over tenfold higher than β1 or β4 in wild type cells (Fig. 2, white bars). The γ4, γ5 and γ12 isoforms were all transcribed at high levels in wild type cells; however, transcripts for γ2, γ7, γ10 and γ11 were also detected (Fig. 2, white bars). β and γ mRNA levels were also compared to HEK-293 cell lines expressing the adenosine A1:αi1 fusion protein (Fig. 2, black bars) or the adenosine A2A:αS fusion protein (Fig. 2, hatched bars) to control for possible changes in transcription caused by fusion protein expression. There were no significant differences in β or γ transcript levels between the adenosine A1 and A2A receptor fusion protein stable cell lines; in addition, there were no differences in β or γ mRNA between the wild type HEK-293 cells and the cell line expressing the A1:αi1 fusion protein. However, the HEK-293 cell line expressing the A2A:αS fusion protein displayed an approximately eight-fold increase in β1 mRNA, increasing from 0.03% in wild type cells to 0.20% in the A2A:αS stable cell line (Fig. 2). Further, there was a three-fold increase in β4 mRNA, from 0.4% in wild type cells to 1.25% in the A2A:αS stable cell line (Fig. 2).
The high affinity agonist binding state of a receptor occurs when it is bound to heterotrimeric γ protein, which makes agonist binding a useful parameter for measuring the interaction between receptor and γ protein. The fusion of αi1 onto the C-terminus of the adenosine A1 receptor did not interfere with the ability of the receptor to interact with the complete heterotrimeric γ protein, as measured by high affinity agonist binding (23). Thus the fact that the A1:αi1 fusion protein was functionally similar to the analogous non-fused proteins (23) made it a natural initial choice for the examination of interactions between the adenosine A1 receptor, αi1 and β and γ isoforms. Figure 3 illustrates the purification of native HEK-293 βγ dimers from the 6HIS/FLAG tagged A1:αi1 fusion protein; the β subunit at 36 kDa can clearly be seen in both the silver stained gel and the β-common western blot (AlF4− elutions 2 and 3, Fig. 3). Figure 1C demonstrates the purity of native HEK-293 βγ dimers combined with Sf9 purified βγ dimers after staining with Coomassie blue. The A1:αi1 fusion protein can also be recovered by elution with FLAG peptide; the expected electrophoretic mobility of the receptor fusion protein is approximately 80 kDa. In agreement with this, FLAG immunoreactivity was observed in the FLAG elution fractions of the western blot between the 75 and 100 kDa molecular weight standards (Fig. 3, FLAG blot, elutions 2-4). The faint visualization of the receptor fusion protein in the silver stained gel of figure 3 may be due to glycosylation induced band broadening of the fusion protein, or differential protein staining with silver. Similar results were observed for the purification of βγ from the A2A:αS fusion protein (data not shown).
Figure 4A displays a mass spectrum of purified β1γ5 from Sf9 cells using MALDI mass spectrometry, a technique which is able to ionize intact γ subunits in the [M+H]+ charge state. The largest peak at approximately 7160 m/z (Fig. 4A, left dashed line) corresponds to the predicted mass of the γ5 subunit which undergoes the conventional posttranslational processing of prenylation, cleavage of the C-terminal three amino acids, and methylation of the prenylated cysteine (5). At the higher m/z of 7501.9 (Fig. 4A, middle dashed line), a smaller peak corresponds to the predicted mass (7501.3) of the γ5 lacking proteolytic cleavage of the three C-terminal SFL residues (22). In contrast, the largest peak at m/z of 7501.7 in the mass spectrum for βγ purified from the A1:αi1 fusion protein expressed in HEK-293 cells (Fig. 4B, middle dashed line) corresponds to a geranylgeranylated γ5 lacking proteolytic cleavage of the three C-terminal SFL residues with a predicted mass of 7501.3. Similar results were observed for the βγ purified from the A2A:αS fusion protein (Fig. 4C, middle dashed line), with the experimental m/z of 7504.0 in good agreement with the predicted mass of 7501.3. The mass spectrum for the “heavy” βγ purified from the A1:αi1 fusion protein expressed in HEK-293 cells under SILAC conditions was observed to have a peak (Fig. 4D, right dashed line) that was approximately 45 Daltons higher than the major peak in figure 4B. This peak with an m/z of 7546.4 was in agreement with the predicted mass of 7549.3 for a geranylgeranylated γ5 subunit lacking proteolytic cleavage of the three C-terminal SFL residues, in which all of the arginines and lysines have been replaced with [13C6] Arg and [13C6] Lys, respectively. The experimental implications of differential posttranslational modification (24) are illustrated in the silver stained gel of purified γ subunits in figure 1, where significant heterogeneity in electrophoretic mobility of different γ isoforms was observed under the separating conditions (12% SDS-PAGE).
Differential N-terminal processing was also observed in the MS/MS analysis of peptides from γ isoforms. Table 2 lists all of the γ isoforms characterized in this study, with the N-terminal sequence translated from the open reading frame of each γ gene, and the N-terminal structure for each γ isoform purified from the A1:αi1 and A2A:αS fusion proteins as determined by mass spectrometry. Although levels of the γ4 isoform were not quantified, data from the mass spectrometric analysis still allowed qualitative characterization of post-translational processing of this isoform. With the exception of γ4, all the γ isoforms for which N-terminal peptides were observed (γ2, γ5, γ7 and γ10) had undergone cleavage of the N-terminal methionine, followed by acetylation (Table 2). The presence of the N-terminal methionine in γ4 can be accounted for by the lysine at position 2 (Table 2), which has been reported to provide a poor binding environment for methionine aminopeptidase (25). N-terminal acetylation of proteins containing a Met-Lys at positions 1 and 2 has been reported to be variable in humans (26). A study of bovine brain derived γ3, which also contains an N-terminal Met-Lys motif, and is thus expected to retain the N-terminal methionine, found more than half of the protein to be acetylated at the N-terminal methionine (24). Although the characterization of N-terminal processing of γ isoforms in this study is not quantitative, only unacetylated N-terminal peptides were detected for γ4 (Table 2). The absence of N-terminal peptides for γ11 was probably due to the low abundance of this isoform, and the Lys at position 4 of γ12 (Table 2) likely resulted in a tryptic N-terminal peptide that was too small for successful analysis. Other studies, however, have reported that the γ12 isoform undergoes cleavage of the N-terminal methionine, followed by acetylation of the resulting N-terminal serine (24). This differential processing of the N-termini of γ isoforms implies a point of functional regulation in the γ subunit, which is discussed below.
The use of SILAC allows the simultaneous biochemical processing of chemically identical heavy and light proteins and peptides, which can be differentiated and measured in a mass spectrometer as heavy and light ion pairs. The ratio of the signal intensity of the ion pairs can thus be translated into quantitative information about the proteins in the sample. Ion pairs from the LC MS/MS analysis of heavy βγ tryptic peptides from SILAC treated Adenosine A1:αi1 fusion protein and light βγ tryptic peptides from Sf9 purified βγ were identified and quantified by mass spectrometry.
The use of SILAC technology enables the characterization of heavy and light ion pairs by mass spectrometry over a wide dynamic range. Figure 5A illustrates the ion pair for the peptide KVVQQLR from the abundant γ5 isoform. Since the peptide has both a lysine and an arginine, the net mass difference between the heavy and light peptides is 12 Daltons; however, since the ions have a charge of 2+, the m/z difference is only 6. The inset (Fig. 5A) illustrates the relationship between the retention time by HPLC and the ion traces for the monoisotopic heavy and light ions indicated by arrows (Fig. 5). Since peptides with the heavy isotopes are chemically identical to their light counterparts, all ion pairs will have identical retention times and thus will be affected equally by any ionization influencing artifacts introduced by the sample. The ratio of the signal intensities of the ion current peaks (Fig. 5A, inset) for the heavy and light peptides in the ion pair is used to quantify protein levels. An example of an ion pair from the less abundant γ11 comes from the peptide SGEDPLVK (Fig. 5B); in the large spectrum, only the light ion is visible. However, when the part of the x-axis containing the heavy ion is magnified 100x (Fig. 5B, grey box in inset), the heavy ion becomes visible. The SGEDPLVK ion is also [M+H]2+, however, since there is only one amino acid that can be exchanged for a heavy isotope, the net mass change is only 6 Daltons and the m/z difference is 3. Table 3 contains a complete list of all the peptides that produced ion pairs used to quantify protein levels of γ isoforms associated with the A1:αi1 fusion protein.
Ion pairs for β isoforms were also examined, and figure 6A illustrates an example from the ELAGHTGYLSCCR peptide from the most abundant β1 isoform. A peptide from the least abundant β4, TFVSGACDASSK, yields an ion pair that is illustrated in figure 6B with a 2+ charge state. All of the peptides from which ion pairs were observed for the quantification of β isoforms associated with the A1:αi1 fusion protein are listed in Table 3.
The protein concentration from the known Sf9 βγ standards and the ratios of the heavy and light ion pairs were used to calculate the moles of each β and γ isoform purified from the A1:αi1 fusion protein. After expressing each β and γ isoform as a percent of the total β and γ protein observed, respectively, levels of each β isoform (Fig. 7A) and γ isoform (Fig. 7B) purified from the A1:αi1 fusion protein were compared to every other β and γ isoform member (respectively) for differences in the percentage levels. All of the β isoforms were different from each other at the p < 0.001 level (Fig. 7A), with β1 over 12-fold more abundant than β4 (see Fig. 7C for bar graph expression of data in Fig. 7A). Significant differences were observed among many of the γ isoforms; notably, γ5 was estimated to be 78% of total γ isoforms, while γ11 was only 0.03% of total γ isoforms (Fig. 7B). Although γ5 was the most abundant γ isoform, γ2, γ7, γ10 and γ12 all presented between 2-12% of total γ isoforms (Fig. 7C). Data in figure 7B are expressed in bar graph format in figure 7D.
One important question arising from quantitation of the β and γ isoform composition of the A1:αi1 fusion protein is the relationship of the A1:αi1 βγ profile to that of another receptor, or to the βγ profile in the whole cell. Enrichment of the βγ fraction in HEK-293 cells was necessary to reduce background protein signal and increase the strength of β and γ peptide signals in the mass spectrometric analysis. Protein concentration of the enriched βγ fraction from HEK-293 cells based on quantitative western blotting was 7.9 ng β/μg protein (sc-378, Santa Cruz) and 3.1 ng α/μg protein (NEI-800, DuPont NEN); thus, the enriched βγ fraction likely contained both free βγ and heterotrimeric γ protein. The A2A:αS fusion protein was chosen for comparison purposes as a member of the same receptor family with distinctly different Gα coupling preferences. For the experimental comparisons, heavy βγ purified from the A1:αi1 fusion protein was added to a similar amount of light βγ purified from the A2A:αS fusion protein, and to the enriched βγ fraction from HEK-293 cells. The samples were then separated by SDS-PAGE, stained with Coomassie blue, and the gel bands containing the heavy and light β isoforms, and the heavy and light γ isoforms were excised and analyzed as described in Materials and Methods.
Using the heavy βγ dimers purified from A1:αi1 as standards, β and γ isoform levels in both the whole cell and specifically associated with A2A:αS were calculated (See Table 3 for list of peptides used to determine H:L ratios). Figure 8 illustrates that the A2A:αS fusion protein bound ~30% more β4 (6.8% of total β) compared to the A1:αi1 fusion protein (5.3% of total β). Further, the A1:αi1 fusion protein contained ~40% higher levels of β4 than the enriched βγ fraction from HEK-293 cells (Fig. 8: 5.3% vs 3.7% of total β), and the A2A:αS fusion protein contained ~80% higher levels of β4 than the enriched βγ fraction (Fig. 8: 6.8% vs 3.7% of total β). Differences in β1 and β2 levels were not significantly different among A1:αi1, A2A:αS, or total HEK-293 βγ; however, levels of β2 were trending lower in both A1:αi1 and A2A:αS compared to total HEK-293 βγ (Fig. 8), offsetting the higher levels of β4 in both fusion proteins. No differences were observed in types of γ isoforms between A1:αi1 and A2A:αS; however, the A1:αi1 and A2A:αS fusion proteins contained 25% and 21% higher levels of γ5, respectively, than total HEK-293 cell βγ. These higher levels of γ5 were offset by lower levels of γ2 (A1:αi1: 41% Δ; A2A:αS: 40% Δ) γ10 (A1:αi1: 61% Δ; A2A:αS: 54% Δ) and γ12 (A1:αi1: 36% Δ; A2A:αS: 29% Δ) compared to total HEK-293 cell βγ (Fig. 9). Though not significantly different, the γ7 isoform appeared to also trend lower in A1:αi1 and A2A:αS compared to total HEK-293 cell βγ (Fig. 9).
Numerical values (+/− SEM) for β and γ protein levels associated with the A2A:αS the fusion protein and the enriched βγ fraction from HEK-293 cells are reported in Table 4. Interestingly, levels of mRNA detected for β isoforms (Fig. 2) did not correlate with β protein levels in the HEK-293 enriched βγ fraction (Fig. 8). Whereas β2 mRNA was ten-fold higher than β1 mRNA, β1 protein was actually two-fold higher than β2 protein, suggesting that the β1 protein is relatively long lived in the cell. In contrast to the β isoform, protein levels for γ isoforms (Fig. 9) correlated roughly with the mRNA detected by QPCR (Fig. 2), suggesting that the γ4 isoform (which was not quantified) is moderately expressed in HEK-293 cells. These discrepancies suggest that QPCR data should be interpreted with caution, and if at all possible, verified with quantitative data at the protein level.
Quantitation of β and γ isoforms allowed examination of the ratio between β and γ subunits in dimers purified from the A1:αi1 and A2A:αS receptor fusion proteins, as well as in the enriched βγ fraction from HEK-293 cells. This was done by dividing the total moles of γ isoforms in a sample by the total moles of β isoforms in a sample. In theory there is a 1:1 ratio of β:γ subunits in a given sample of purified βγ dimer. Using quantitative values from this study, there were 0.69 and 0.72 moles of γ for every mole of β in the βγ dimers purified from the A1:αi1 and A2A:αS receptor fusion proteins, respectively; similarly, the ratio of γ to β in the enriched HEK-293 cell fraction was 0.71. This was somewhat expected, as γ4 was not included in the total estimates of γ protein for each receptor fusion protein. Furthermore, reports of instability of βγ dimers containing γ11 (11;27) suggest that γ11 levels in the analysis may be underestimates of the actual level of γ11 present prior to the steps used to either purify the receptor-G protein complex or enrich the fraction of βγ from HEK-293 cells. The theoretical βγ ratio of 1:1 is an important issue in the choice of Sf9 βγ standards, as the γ concentration was calculated indirectly from β. Most cases of unstable βγ dimer combinations involve β2, β3, β4 or β5 (see review (5)); for this reason, β1 was expressed with different γ isoforms for the generation of recombinant βγ standards used to calculate γ protein levels.
The preference of adenosine A1 and A2A receptors for β4 and γ5 isoforms is in agreement with previous reconstitution studies that demonstrated that β2γ2 and β4γ2 were more efficient than β1γ2 at coupling GS α to the adenosine A2A receptor (28). This study refines that work by identifying the β4 isoform as the preferred binding partner of the two fusion proteins. The previous work is also expanded through the demonstration that the adenosine A1 and A2A receptors share a preference for the γ5 isoform. Together, these studies suggest that in the case of adenosine receptors, binding specific βγ dimer combinations is more strongly determined by the receptor family than the identity of Gα subunit that binds a particular receptor isoform.
G protein βγ dimers exist at the beginning of a signaling event as part of a receptor:Gαβγ ternary complex (29), with βγ binding to both receptor and Gα subunit (11). The crystal structure of a heterotrimeric G protein provided evidence that binding sites for βγ on Gα family members are highly conserved (30), and few accounts of specificity, such as between Gqα and β5 (17), have been reported. Broad diversity in both potential βγ dimer combinations (11) and G protein coupled receptor isoforms (31) suggest that receptor, or possibly both receptor and Gα, influence the composition of β and γ isoforms in a receptor-G protein complex. Understanding preferences of specific receptors for particular combinations of α, β and γ will help to elucidate the structural determinants that favor these combinations.
Two limitations had to be overcome to successfully quantify β and γ dimers associated with specific receptors. The first limitation concerned the isolation of βγ in sufficient quantity and purity for biochemical analysis. Receptors have been precipitated with associated G proteins, however, affinity and stoichiometry between G protein and receptor can be variable (32), and often not adequate for biochemical analysis (unpublished observation). Thus, a receptor fusion protein strategy was employed in order to preserve the interactions among βγ, receptor and Gα during purification of the complex (Fig. 1). This approach allows a more consistent purification, and would likely be applicable to any receptor-α combination.
The second limitation involves the availability of immunological reagents for characterizing β and γ isoforms. Many of the antibodies available do not have sufficient sensitivity or specificity to distinguish between isoforms, and thus are not quantitative. Mass spectrometry in conjunction with SILAC was chosen as an innovative and powerful approach to quantify β and γ isoforms for several reasons: 1) Obviation of the need for specific antibodies; 2) Femtomole sensitivity; 3) Linearity of heavy:light ion ratios at all signal strengths; 4) Absolute specificity with respect to protein isoform and species; 5) Ability to characterize covalent modifications of protein isoforms.
Posttranslational modification of proteins is critical to understand because it can have the effect of increasing the functional heterogeneity of a protein. Covalent modifications of γ isoforms were probed by both MALDI and ESI mass spectrometry. MALDI mass spectrometry was able to initially characterize the modification state of the γ5 subunits from βγ populations purified from both adenosine receptor fusion proteins in HEK-293 cells. In contrast to the γ5 protein observed from the β1γ5 dimer purified from Sf9 cells (Fig. 4A), and other mammalian γ isoforms which appear to exist predominantly in the prenylated and C-terminally processed state (24), the γ5 protein associated with both A1:αi1 and A2A:αS receptor fusion proteins was mostly prenylated without C-terminal proteolytic processing. This pattern of processing for γ5 is in agreement with the results reported by Kilpatrick et al. (22), which the authors suggest may be related to protein-protein interactions.
ESI-MS/MS analysis was also able to reveal the N-terminal modification state of many of the γ isoform derived peptides identified in this study. The significance of differential N-terminal acetylation of γ isoforms was emphasized by a recent study in Saccharomyces cerevisiae that suggested N-terminal acetylation is a degradation signal in the N-end rule pathway (33). In the study, the Doa10 ubiquitin ligase preferentially recognized N-acetylated proteins, which targets the protein for ubiquitylation, resulting in shorter half lives from increased degradation. This likely has relevance for G protein stability, as the γ2 isoform has been shown to be a substrate for ubiquitylation (34). Proteins with a lysine at position 2, such as γ4 (Table 2), regardless of N-terminal acetylation status, were found to bind poorly to the Doa10 ubiquitin ligase (33); interestingly, the only other γ isoform with a lysine at position 2 is γ3. Although the N-terminus of γ11 was not characterized in this study, the proline at position 2 (Table 2) suggests that it is a poor substrate for N-terminal acetyltransferase (26), and thus a poor target for Doa10 ligase; only one other γ isoform, γ1, contains a proline at position 2. Taken together, the lack of or limited acetylation in γ1, γ3, γ4 and γ11, in addition to the lysine at position 2 in γ3 and γ4, suggest that these isoforms have the capacity for metabolic stability, and may mark a functional divide in the γ isoform family. Extrapolating the effects of acetylation on a physiological system, the degree of acetylation of βγ dimers contained in a receptor:G protein complex may affect the duration of βγ signaling through regulation of its half life.
A previously published report demonstrated that purified β2γ2 and β4γ2 were more efficient than β1γ2 at coupling GS α to the adenosine A2A receptor in a reconstitution assay (28). A distinction should be made that the present study measures interactions of receptors with the endogenous pool of βγ dimers in a cell, allowing for differences in both stoichiometry and subcellular localization to influence formation of an R:G complex. One interpretation of the two studies is that βγ dimers containing either β2 or β4 are able to couple GS α to the A2A receptor with high efficiency; however, in the context of the HEK-293 cell, both adenosine A1 and A2A receptors have a preference for βγ dimers containing the β4 isoform, likely β4γ5. It should also be noted that the adenosine A1 receptor has been documented to have a preference for the Gi3 α subunit over Gi1 α, Gi2 α or GO α (35). Although little specificity between Gα and βγ has been reported (21), it is possible that Gα isoforms modulate the specificity of receptor βγ interactions, and this may be reflected in the differences in affinity observed between A1:αi1 and A2A:αS fusion proteins for dimers containing β4. The variability in Gα isoform may also account for the preference of the adenosine A2A receptor for β2 and γ7 in striatum, where A2A receptor mediated elevation of cAMP occurs primarily via Golf α instead of GS α (36). Cell type and differences in transcription may also contribute to the identity of a heterotrimeric G protein that can interact with a receptor. Thus, the increase in β4 mRNA levels resulting from A2A:αS fusion protein expression in this study (Fig. 2) may be related to the higher levels of β4 protein observed with A2A:αS (Fig. 9). Interestingly, this correlates with another study that reported a decrease in β4 mRNA levels after ablation of GS α expression using RNAi (9). This suggests that signaling components within a transduction cascade can be regulated in concert beginning at the level of transcription, and thus there are likely many points of control that determine the final makeup of a receptor:G protein complex.
In this analysis, β and γ isoform are expressed as a percent of the total quantified, which is essentially a zero sum situation where increases in one isoform must be offset by decreases in others. This can be explained by figure 9, which illustrates a ~15 percentage point increase in γ5 associated with A1:αi1 over HEK-293 βγ. The combined percentage point decrease in γ2, γ7, γ10 and γ12 associated with A1:αi1 relative to HEK-293 βγ agrees very closely with this value (Figs (Figs77 and and9,9, and Table 4). In contrast, the increase in β4 levels associated with A1:αi1 compared to HEK-293 βγ, although significant, is less than two percentage points (Fig. 8). This increase is likely offset by the slight trend lower of β2 associated with A1:αi1 compared to HEK-293 βγ (Fig. 8). One interpretation of these differences is that increases in γ5 levels in A1:αi1 over HEK-293 βγ can not be explained by increases in preference for the β4γ5 dimer alone. Thus, A1:αi1 (and A2A:αS to a lesser extent) likely also has preferences for β1γ5 and/or β2γ5 dimers.
This specificity for γ5 may be related to the physiological properties of adenosine receptors. For instance, activation of the adenosine A2A receptor has been shown to attenuate the inflammatory effects of Helicobactor pylori induced gastritis (37). It has also been shown that H. pylori infection up-regulates γ5 mRNA levels in a human gastric cancer cell line (38); increased transcription of γ5 may be related to a mechanism by which adenosine receptors interact with specific G protein combinations to signal and counter the effects of inflammation. Indeed, there is mounting evidence that β and γ levels are dynamic and respond to extracellular cues. For instance, the γ3 transcript was up-regulated in rat hippocampus following oxidative stress (39), and in activated CD4 + T-cells (40). LPS stimulation of the microglial cell line BV-2 resulted in a transient increase in γ12 levels (41). Levels of β4 mRNA and protein increased in human microvascular endothelial cells in response to IL-1 and TNF-α (42), and IFN-β was shown to increase β4, γ2 and γ11 transcripts in Ubp43−/− bone marrow derived macrophages (43). These examples suggest that extracellular stimuli may prime a cell to express a particular profile of β and γ isoforms; it is of critical importance to determine if specific receptors within the cellular context have inherent preferences for the resulting βγ dimers.
As an example of how dynamic regulation of β and γ transcription may influence signaling, the IFN-β mediated increase in β4 mentioned above may be examined in the context of adenosine receptor signaling. Increased β4 expression could facilitate the population of adenosine A2A receptor complexes containing β4γ dimers, which compared to A2A receptor complexes containing β1γ dimers, have the ability to shift the equilibrium of the A2A receptor population toward more high affinity agonist binding sites (44). This would have the effect of lowering the concentration of adenosine required for activation of the A2A receptor. Interestingly, a similar mechanism for increased adenosine receptor signaling was proposed after the discovery that IFN-β induced the expression of CD73, an ecto-5′-nucleotidase that increases adenosine production (45); the authors proposed that increased adenosine receptor signaling may be one way that IFN-β ameliorates the progression of multiple sclerosis.
The question of how β and γ subunit diversity translates into signaling specificity has been enigmatic. It is possible that multiple mechanisms exist for heterogeneity of β and γ isoforms to influence cellular signaling. Regardless, this innovative approach will allow the question to be fully addressed through the quantitative measurement of changes in both β and γ isoforms with high precision under a variety of experimental conditions.
We would like to thank Dr. James Garrison for support during the course of these studies, and helpful discussions during the preparation of this manuscript. Thanks go to Dr. Joel Linden for the generous gift of the pDoubleTrouble vectors containing adenosine A1 and A2A receptors. We also thank Dr. Thurl Harris, Jessica Ng and Jiping Zhou for expert technical assistance. We acknowledge the University of Virginia Pratt Committee for its generous support of the Biomedical Research Facility.
†The work was supported by the American Heart Association National Scientist Development Grant 0535350N, a Pilot/Feasibility Award from the UVa Silvio O. Conte Digestive Disease Research Center and the National Institutes of Health Grant R01-DK-19952.
1The abbreviations used are: G protein, guanine nucleotide-binding regulatory protein; SILAC, stable isotope labeling with amino acids in cell culture; Sf9 cells, Spondoptera frugiperda cells; HEK cells, human embryonic kidney cells; DMEM, Dulbecco's Modified Eagle Medium; FBS, fetal bovine serum; DTT, dithiothreitol; CHAPS, 3-[(3-cholamidopropyl) dimethylammonio]-1-propanesulfonate; EDTA, ethylenediaminetetraacetic acid; HEPES, N-(2-hydroxyethyl)piperazine-N′-2-ethane-sulfonic acid; BSA, bovine serum albumin; DDM, n-dodecyl beta-D-maltoside; CHS, cholesteryl hemisuccinate; SDS, sodium dodecyl sulfate; MALDI, matrix-assisted laser desorption/ionization.