Regulator of G protein signaling (RGS) proteins are considered key modulators of G protein-coupled receptor (GPCR)-mediated signal transduction. These proteins act directly on Gα subunits in vitro to increase their intrinsic rate of GTP hydrolysis; this activity is central to the prevailing view of RGS proteins as negative regulators of agonist-initiated GPCR signaling. However, the specificities of action of particular RGS proteins toward specific GPCRs in an integrated cellular context remain unclear. Here, we developed a medium-throughput assay to address this question in a wholly endogenous context using RNA interference. We performed medium-throughput calcium mobilization assays of agonist-stimulated muscarinic acetylcholine and protease-activated receptors in human embryonic kidney 293 (HEK293) cells transfected with individual members of a “pooled duplex” short interfering RNA library targeting all conventional human RGS transcripts. Only knockdown of RGS11 increased both carbachol-mediated calcium mobilization and inositol phosphate accumulation. Surprisingly, we found that knockdown of RGS8 and RGS9, but not other conventional RGS proteins, significantly decreased carbachol-mediated calcium mobilization, whereas only RGS8 knockdown decreased protease-activated receptor-1 (PAR-1)-mediated calcium mobilization. Loss of responsiveness toward carbachol and PAR-1 agonist peptide upon RGS8 knockdown appears due, at least in part, to a loss in respective receptor cell surface expression, although this is not the case for RGS9 knockdown. Our data suggest a cellular role for RGS8 in the stable surface expression of M3 muscarinic acetylcholine receptor and PAR-1, as well as a specific and opposing set of functions for RGS9 and RGS11 in modulating carbachol responsiveness similar to that seen in Caenorhabditis elegans.
- calcium flux
- muscarinic acetylcholine receptor
- protease-activated receptor-1
- regulators of G protein signaling
- short interfering RNA duplexes
g protein-coupled receptors (GPCRs) constitute the largest class of drug targets (20), and there is considerable interest in elucidating the varied cellular controls over GPCR signal transduction. Agonist binding promotes activation of heterotrimeric G proteins, which in turn modulates various downstream effector proteins to trigger cell-specific signaling events (18). Cellular signaling by GPCRs is regulated through various mechanisms (e.g., Refs. 8 and 24), including “regulator of G protein signaling” (RGS) proteins (42). RGS proteins are GTPase-accelerating proteins (GAPs) of heterotrimeric G protein Gα subunits (4). Given their ability to accelerate the GTP hydrolysis rate of Gα subunits, RGS proteins are considered key negative regulators of receptor-promoted cellular signaling that oppose the action of extracellular agonists (26).
Thirty-seven human RGS proteins are grouped into eight families based on sequence, structural, and functional similarities (42). Conventional RGS proteins (RGS1 through RGS22) share a central function as Gαi/o/q-directed GAPs and span four families and one outlier (Fig. 1A). RZ subfamily members possess an NH2-terminal poly-cysteine region, the R4 subfamily includes the largest number of members (some possessing an NH2-terminal amphipathic α-helix), R7 subfamily members bind Gβ5 via their GGL domain, and R12 subfamily members variably contain multiple distinct domains including PSD-95/Discs large/Zonula occludens-1 (PDZ), phosphotyrosine binding (PTB), and Ras-binding (RBD) domains along with the GoLoco motif (reviewed in Ref. 42). The precise molecular determinants underlying any GPCR selectivity to RGS protein action are not well characterized. However, the existence of multiple additional domains outside the hallmark RGS domain increases the range of possible mechanisms linking RGS protein activity to particular receptor(s).
One major question regarding GPCR regulation by RGS proteins is which superfamily member(s) regulates a given receptor in an integrated cellular context. To date, a relatively small number of RGS proteins have been identified compared with the number of verified human GPCRs, which consists of at least 799 members (20, 21). RGS-insensitive Gα mutants have provided convincing evidence that endogenous RGS proteins control the kinetics, duration, and amplitude of cellular GPCR signal transduction in both ex vivo and whole organism contexts (e.g., Refs. 15 and 16); however, this approach does not identify the particular RGS protein(s) operative on select GPCR responses. Yet, evidence is accumulating that specific RGS proteins exhibit receptor-selective effects. For example, RGS1 is 1,000-fold more potent as an inhibitor of the M3 muscarinic acetylcholine receptor (M3 mAChR) than the cholecystokinin receptor in pancreatic acinar cells, whereas RGS2 is equipotent in inhibiting the action of both receptors (44). Ribozyme-mediated depletion of RGS3 levels in A-10 rat aortic smooth muscle cells selectively enhances M3 mAChR-promoted signaling, whereas depletion of RGS5 potentiates angiotensin II signaling through the AT1a receptor (38). As a potential molecular mechanism for the receptor specificity found in these cellular contexts, some RGS proteins have been shown to bind the third intracellular loops (3iLs) of certain GPCRs. RGS2 is reported to bind directly to the 3iL of Gq/11-coupled M1 mAChR and not to the Gi/o-coupled M2 mAChR (5); the same group has shown that RGS2 binds selectively to the 3iL of Gq/11-coupled α1A-adrenergic receptor but not to α1B- or α1D-ARs (11). These reports suggest that direct interactions between specific RGS proteins and certain receptors may provide a mechanism for receptor-selective regulation.
To begin to delineate RGS protein specificity for particular GPCRs in an integrated cellular context, here we describe our efforts in developing a medium throughput, quantitative assay for GPCR signal transduction allowing for a survey of the effects of individual RGS protein knockdown across the superfamily using RNA interference. The knockdown screening strategy chosen was one of using pools of chemically synthesized, short interfering RNA (siRNA) duplexes, with each pool containing four different duplexes targeting the same RGS gene transcript. This pooled siRNA screening strategy has been found to reduce off-target effects while providing a high frequency of effective knockdown and has previously been successful in whole human genome screens, which have identified a variety of genetic modifiers of cellular physiological events (e.g., Refs. 10 and 40). We performed GPCR agonist-evoked calcium flux assays in cells transfected with siRNA pools targeting each conventional RGS protein to identify any changes in agonist potency or efficacy upon RGS protein knockdown. We examined responses from the muscarinic acetylcholine receptors (mAChRs) and the protease-activated receptor-1 (PAR-1) that are endogenously expressed and provide robust calcium mobilization upon agonist exposure (i.e., carbachol and the PAR-1-selective agonist peptide TFLLRNPNDK, respectively) (34). RGS11 knockdown increased carbachol efficacy (consistent with negative regulator function via GAP activity); yet, in an opposite manner, knockdown of the related RGS9 decreased carbachol efficacy and potency without affecting cell surface receptor expression. Moreover, RGS8 knockdown decreased the efficacy of both carbachol and PAR-1 agonist peptide signaling to calcium mobilization; this effect is related to a potential new function for an RGS protein in modulating receptor cell surface expression.
MATERIALS AND METHODS
Agonists, antibodies, and other reagents.
Carbachol was from Sigma and the PAR-1-specific agonist peptide TFLLRNPNDK was synthesized with a carboxyl amide and purified by reverse-phase high-pressure liquid chromatography (UNC Peptide Facility, Chapel Hill, NC). Sheep anti-RGS9 polyclonal antibody was a generous gift from Dr. Kirill A. Martemyanov (University of Minnesota). The rabbit polyclonal anti-RGS11 (Rb417/419) was previously described (35), the horseradish peroxidase (HRP)-conjugated anti-hemagglutinin (HA) monoclonal antibody (clone 3F10) was obtained from Roche Diagnostics, and the anti-β-actin antibody AC-74 was purchased from Sigma. The anti-PAR1 polyclonal antibody has previously been described (27). HRP-conjugated goat anti-mouse and goat anti-rabbit antibodies were from GE Healthcare (Piscataway, NJ). Expression plasmids for HA-tagged mAChRs (M1-M5) and RGS8 were purchased from The Missouri S&T cDNA Resource Center (Rolla, MO).
All siRNA used in this study were chemically synthesized by Dharmacon RNAi Technologies (Lafayette, CO). siRNA directed toward human RGS transcripts (RGS1 to -22) and M1, M3, and M4mAChR mRNAs were purchased as siGENOME SMARTpools containing four distinct siRNA oligonucleotide duplexes; for those RGS transcripts examined after the primary SMARTpool screen, the four individual siRNA duplexes of the pool were obtained separately from Dharmacon. The siRNA duplex targeting a highly conserved region of sequence shared between human Gαq and Gα11 mRNAs (5′-AAGATGTTCGTGGACCTGAAC-3′) was previously described (2). The siRNA sequence used for targeting PAR-1 mRNA was 5′-AGAUUAGUCUCCAUCAAUA-3′. An additional siRNA duplex was designed for RGS9 (designated “duplex 5”: 5′-GGAACAAAGCAGACAGAUAUU-3′). The nonspecific (“ns”) siRNA 5′-CTACGTCCAGGAGCGCACC-3′ was used as a negative control for all siRNA experiments as previously described (27).
Cell culture and transfection.
Human embryonic kidney 293 (HEK293) cells were obtained from the American Type Culture Collection and were maintained in DMEM (Invitrogen, Carlsbad, CA) supplemented with 10% fetal bovine serum (Cellgro, Manassas, VA) at 37°C in a humidified atmosphere containing 5% CO2. Transient transfections of HEK293 cell monolayers grown to 75–90% confluence were performed using Lipofectamine 2000 (Invitrogen, Carisbad, CA) for siRNA (150 nmol) and using FUGENE 6 (Roche, Indianapolis, IN) for DNA plasmid in Opti-MEM, each according to the particular manufacturers′ instructions. Stable HEK293 cell lines were generated by transfecting HEK293 cells with indicated constructs and culturing in DMEM supplemented with 10% fetal bovine serum and 1 μg/ml of G418 (Invitrogen).
Fluorescence imaging plate reader calcium flux assays.
One day before the assay, HEK293 cells previously transfected with siRNA were seeded at 100,000 cells per well of a 96-well plate precoated with poly-d-lysine. Calcium flux assays were performed with no-wash dye [fluorescence imaging plate reader (FLIPR) calcium assay kit, Molecular Device, Sunnyvale, CA] as previously described (37). Fluorescence responses of cells were measured with a FLIPRTETRA device upon the addition of variable concentrations of agonist, or vehicle, in the presence of assay buffer [20 mM HEPES, pH 7.4, 1× Hanks balanced salt (Invitrogen) and 2.5 mM probenecid]. Peak responses (“maximal amplitude” values) were normalized to those of cells transfected with the nonspecific (ns)-siRNA control and therefore expressed as percentages of the evoked response from ns-siRNA treated cells. Sigmoidal dose-response curve fitting, error-bar calculations, and pair-wise “comparison of fits” tests for significant changes to Rmax and/or EC50 values (using the extra sum-of-squares F test) were performed using Prism v. 5.0c (GraphPad Software, La Jolla, CA) as described (37).
To analyze the effects of siRNA-mediated knockdown on target proteins, siRNA-transfected HEK293 cells in individual wells of a 12-well plate were washed twice with ice-cold phosphate-buffered saline and then lysed with buffer [20 mM HEPES, pH 7.5, 10 mM EDTA, 150 mM NaCl, 1% Triton X-100, and complete protease inhibitors (Roche, Indianapolis, IN)] at 4°C on a rocker for 30 min. Lysates were centrifuged at 4°C at 30,000 rpm for 30 min and quantified by BCA protein content measurement (Pierce). Equivalent amounts of total protein across a series of siRNA duplex transfections were resolved on 4–12% precast SDS polyacrylamide gels (Novex/Invitrogen), transferred to nitrocellulose, and immunoblotted using anti-RGS9 (1:100), anti-RGS11 (1:200), or anti-HA tag (1:1,000) primary antibodies, followed by an incubation with relevant HRP-labeled secondary antibodies and visualization by chemiluminescence (ECL+, GE Healthcare). Blots were stripped and reprobed using anti-actin (1:1,000) for protein loading standardization.
Standard and quantitative RT-PCR protocols.
First-strand cDNA was generated by reverse transcription on total RNA isolated from HEK293 cells using M-MLV Reverse Transcriptase and oligo(dT)12–18 (Invitrogen) according to the manufacturer's instructions. Negative control samples were generated in parallel by performing “-RT” reactions in the absence of reverse transcriptase. Polymerase chain reaction (PCR) was conducted on first-strand cDNA using primers specific for RGS1 to -22 and GAPDH (the latter as a positive control). PCR amplicons were resolved by 1.2% (wt/vol) agarose gel electrophoresis and detected with ethidium bromide staining. Sequence and predicted PCR amplicon sizes are shown in Supplementary Table S.1 (see AJP-Cell website for supplementary material).
For qRT-PCR experiments to validate the efficacy of siRNA-mediated knockdown of RGS transcripts, HEK293 cells transfected with siRNA duplexes were washed once with PBS (48 h posttransfection) and then scraped and resuspended in 500 μl of PBS. Total RNA extraction and subsequent qRT-PCR was performed exactly as previously described (19) using gene-specific primers and 6-carboxyfluorescein (FAM) and 6-carboxytetramethylrhodamine (TAMRA) dual-labeled probes. Primer and probe sequences were as follows. Human β-actin: forward, 5′-GGTCATCACCATTGGCAATG-3′; reverse, 5′-TAGTTTCGTGGATGCCACAG-3′; probe 5′-FAM-CAGCCTTCCTTCCTGGGCATGGA-TAMRA-3′; Human RGS8: forward, 5′-CCCATAGGATCTTTGAGGAG-3′; reverse, 5′-TCGGGTCTGGAAGTCAATGT-3′; probe, 5′-FAM-TCCCGTGGAGCCTGCACATCCAC-TAMRA -3′. The number of cycles until threshold (Ct) was determined using an ABI Prism 7700 Sequence Detector System (Applied Biosystems, Foster City, CA). To normalize for variation in the total number of cells and the efficiency of the mRNA extraction, the Ct value for β-actin was subtracted from Ct values for RGS8. The change in RGS8 expression was then determined relative to cells treated with nonspecific siRNA using the 2−(ΔΔCt) method as previously described (41).
Inositol phosphate assay.
Inositol phosphate (IP) accumulation assays were performed as previously detailed (37). In brief, HEK293 cells were seeded at a density of 1.2 × 105 cells per well of 12-well plates. On the next day, cells were transfected with siRNA using Lipofectamine 2000 according to the manufacturer's instructions. The following day, cells were metabolically labeled for 18 h with myo-[3H]inositol at 4 μCi/ml in inositol-free DMEM (MP Biomedicals, Solon, OH) containing 0.5% BSA and 20 mM HEPES, pH 7.5. On the day of the experiment, cells were washed once with phosphate-buffered saline (PBS) and then incubated in prewarmed DMEM (without inositol) containing 0.5% BSA, 20 mM HEPES, pH 7.5, and 20 mM LiCl for 10 min. Thereafter, cells were stimulated with 25 μM carbachol for 30 min. After stimulation, the medium was removed and the reactions were terminated by addition of 150 μl of 50 mM formic acid and incubation for 1 h at room temperature. Then, 50 μl of formic acid supernatatant were mixed with 75 μl of 2.67 mg/ml of SPA beads diluted in ice-cold water (RNA binding YSi beads, GE Healthcare) in a 96-well/plate and agitated at 4°C for 30 min. The radioactivity was counted with a Wallac MicroBeta luminescence counter (PerkinElmer), and statistical significance of observed differences between siRNA treatments were obtained by unpaired Student's t-test with Welch's correction (GraphPad Prism).
Receptor expression measurements.
To measure total cellular levels of endogenous M3 muscarinic receptors, competition binding assays were performed using 75 μg of total cell membrane preparations from HEK293 cell monolayers previously transfected with siRNA as previously detailed (1). Briefly, membranes were incubated with 0.5 nM of the tritium-labeled, nonsubtype-specific, muscarinic orthosteric antagonist 3-quinuclydinyl benzylate ([3H]QNB) in Tris buffer (50 mM Tris·HCl, pH 7.7) at room temperature for 1 h with varied concentrations of unlabeled atropine (0 to 10 μM) before assay termination by rapid filtration using a Brandel harvester onto GF/B grade filter paper (Whatman, Maidstone, UK) presoaked with 0.3% polyethyleneimine. Filters were then washed with three 2-ml aliquots of ice-cold washing buffer (50 mM Tris·HCl, pH 7.4). Nonspecific binding was defined in the presence of 100 μM atropine, and radioactivity was determined by liquid scintillation counting.
ELISA assays were used to quantify cell-surface receptor expression of ectopically expressed, HA-tagged M3 mAChR and endogenous PAR-1 levels. Briefly, 2.0 × 105 HEK293 cells (either the parental line or an in-house derivative stably expressing HA-tagged M3 mAChR) were grown overnight in 12-well plates before transient transfection with 150 nmoles of specific or ns-siRNA. After an additional 24 h in culture, 1.5 × 105 cells were transferred to 24-well plates precoated with 0.1 mg/ml poly-d-lysine (BD BioCoat, Franklin Lakes, NJ) and cultured for an additional 24 h. Cells were then washed once with PBS and fixed with 4% paraformaldehyde/PBS (USB, Cleveland, OH) for 5 min at room temperature. Cells were then washed three additional times with PBS and nonspecific sites blocked with PBS containing 1% BSA for 45 min. HA tag-specific monoclonal antibody, PAR-1-specific polyclonal antibody, or preimmune serum was then added at a dilution of 1:1,000, 1:200, 1:100, respectively, in 1% BSA-PBS for 60 min. After incubation with primary antibody, cells were washed three times with PBS and blocked again with 1% BSA-TBS for 15 min. Incubation with HRP-conjugated sheep anti-mouse or donkey anti-rabbit antibody diluted 1:1,000 in 1% BSA-PBS was carried out for 60 min. The cells were washed three times with PBS and incubated with 300 μl of a preformulated ABTS substrate [one-step 2,2′-azino-bis-3-ethylbenzthiazoline-6-sulfonic acid (ABTS), Thermo Scientific, Rockford, IL]. Plates were then incubated at room temperature for color development. Absorbance of a 100-μl aliquot of the colorimetric reaction was read at 405 nm using a multimode plate reader POLARstar Omega (BMG Labtech). Cells not transfected with HA-tagged M3 mAChR or incubated with preimmune serum were assayed in parallel to determine background signal levels. ELISA measurements were normalized by total protein quantitation of whole cell lysates, as performed using the BCA assay (Pierce), to control for any cell death/cell toxicity effects of siRNA transfection. Statistical significance of observed differences between siRNA treatments were obtained by one-way ANOVA with Dunnett's posttest comparison (Prism version 5).
siRNA-based screen of RGS protein effects on mAChR and PAR-1 signaling to calcium mobilization.
To conduct an unbiased screen of RGS protein superfamily members for individual effects on GPCR signaling, we chose to use the human embryonic kidney-derived cell line HEK293 that expresses endogenous muscarinic acetylcholine receptors and PAR-1 and exhibits excellent efficiency in transient transfection. A profiling of the expression pattern of RGS transcripts in HEK293 cells was performed using RT-PCR with primer pairs (sequences and amplicon sizes in Supplementary Table S.1) directed to mRNA transcripts of all conventional (i.e., GAP-competent, noneffector function) RGS proteins. mRNA transcripts for the majority of conventional RGS proteins were detectable by RT-PCR from HEK293 total RNA (Fig. 1B), except those of RGS1, -6, -18, and -21; this expression pattern is consistent with the suspected neuronal origin of the HEK293 cell line (33), given that several of the detected transcripts encode RGS proteins considered neuronal specific in expression (e.g., R7 subfamily members RGS7, -9, and -11 that bind the neuronal-specific Gβ5 subunit; reviewed in Ref 42).
In the primary screen (see scheme in Fig. 2), we measured intracellular calcium release in siRNA-treated HEK293 cells following independent application of varied concentrations of two different agonists: carbachol, a nonselective mAChR agonist, and the peptide TFLLRNPNDK, a PAR-1-selective agonist. A variety of additional GPCR agonists were initially tested on HEK293 cells, including bradykinin, lysophosphatidic acid, UTP, and endothelin-1; however, only carbachol and PAR-1 agonist peptide demonstrated significant calcium mobilization. As shown in Fig. 2, A and B, robust calcium flux signals were obtained upon application of each agonist to cells transiently transfected with a ns-siRNA duplex. As a positive control for siRNA transfection and knockdown efficiency, we routinely performed parallel transfections of positive control siRNA duplexes either targeting both Gαq and Gα11 transcripts (for carbachol/mAChR signaling studies) or targeting the PAR-1 mRNA directly. These positive control siRNA transfections were routinely observed to diminish agonist-evoked calcium flux responses (e.g., Fig. 2, A and B). As an additional measure to control for any cell-death or cell toxicity effects during the siRNA screen, thapsigargin injection was performed following initial agonist application and intracellular calcium recapture (i.e., 300 s after carbachol application; see Fig. 2A). Inhibiting sarco- and endoplasmic reticulum calcium-activated (SERCA) pump activity in this way leads to a generalized increase in intracellular calcium via passive diffusion from the endoplasmic reticulum (23, 28); equivalent increases in calcium-based fluorescence signals from control and siRNA-treated cells (e.g., Fig. 2A) allowed for the elimination of false positives owing to cell death or toxicity (i.e., RGS16; see Table S.2).
Representative dose-response calcium mobilization data are presented in Fig. 3 from the primary screen using Dharmacon siRNA SMARTpools across the RGS family from RGS1 to RGS22. With the use of the agonist carbachol, knockdown of RGS2, -4, and -6 (Fig. 3A) and most other RGS transcripts (data not shown) had no significant effect on calcium mobilization (summarized in Supplementary Table S.2). In this carbachol screen, knockdown of three RGS family members gave significant changes in responses. Knockdown of RGS11 increased the calcium response at each carbachol concentration tested, leading to a significant increase in carbachol efficacy over that of control siRNA-transfected cells (Table S.2) but no significant change in carbachol potency (EC50); conversely, knockdown of either RGS8 or RGS9 yielded significant decreases in carbachol efficacy, with the latter gene knockdown also decreasing carbachol potency (EC50 of 1.98 μM vs. 1.05 μM on control cells; Table S.2). With the use of the PAR-1 agonist peptide, knockdown of RGS4, -6, -9, and -11 (Fig. 3B) and most other RGS transcripts (data not shown) had no effect on dose-dependent calcium mobilization (Table S.2). However, RGS2 knockdown led to a significant increase of PAR-1 agonist peptide efficacy (Fig. 3B, top left). Interestingly, RGS8 knockdown decreased agonist efficacy, similar to the effect observed in the carbachol screen. Unlike RGS8, the effects of RGS2, RGS9, and RGS11 knockdown were agonist/receptor-system selective in nature.
Validation of hit identification with individual siRNA duplexes.
To validate the results of the primary screen, we separately tested each of the four individual siRNA duplexes included in the SMARTpool for each positive RGS gene target. For RGS8, only transfection of the second or third constituent siRNA duplexes led to significant decreases in efficacy of carbachol and PAR-1 agonist peptide (Fig. 4, A and B; Table S.3). For RGS9, only the first constituent duplex gave the same result as that observed with the siRNA SMARTpool (i.e., reduced carbachol efficacy and potency). We therefore designed another siRNA duplex (“duplex 5”) directed against a unique sequence within the RGS9 mRNA to confirm the primary screen result. RGS9 siRNA duplex 5, like duplex 1, led to decreases in carbachol potency and efficacy (Fig. 4C), with the latter being statistically significant (Table S.3). For RGS11, transient transfection of each of the four constituent duplexes led to significant increases in carbachol efficacy (Fig. 4D; Table S.3). Upon observing similar effects from at least two different siRNA duplexes targeting different sequences within the same RGS transcript, we concluded that these three RGS family members were worthy of more detailed evaluation, given the unlikely outcome that two different siRNA duplexes elicit the same off-target effects (13).
RGS transcript knockdown was subsequently verified for these three hits by detection of RGS protein levels and, in the case of RGS8, also with quantitative RT-PCR given that endogenous RGS8 protein levels were not detectable neither with commercial antibodies nor antiserum provided by Dr. O. Saitoh (Nagahama, Japan). RGS8-directed siRNA duplexes 2 and 3 elicited a dramatic knockdown of HA-epitope tagged RGS8 protein levels in dually transfected HEK293 cells (as compared with nonspecific siRNA control), with duplex 1 also causing RGS8 reduction albeit to a lesser degree (Fig. 5A). Quantitative RT-PCR results confirmed the knockdown efficiency of RGS8 siRNA duplexes 2 and 3 on endogenous RGS8 transcript levels (Fig. 5B). Endogenous RGS9 protein levels were observed to decrease significantly only in cells with RGS9 siRNA duplexes 1 and 5 (Fig. 5C); only the long splice isoform of RGS9 was detected in these cells (i.e., RGS9L a.k.a. RGS9–2 of 77 kDa; Swiss-Prot O75916). For RGS11, all siRNA duplexes tested decreased endogenous RGS11 protein levels. The pattern of responses from the functional calcium mobilization assays (Fig. 4) generally matched the results from individual duplex-mediated RGS protein level knockdown assessed by immunoblotting and densitometry quantification (Fig. 5).
Characterization of functional mAChR subtypes and confirmation of carbachol signaling results.
Calcium mobilization following carbachol activation in HEK293 cells has previously been attributed to either the M1 mAChR subtype (25) or the M3 mAChR subtype (22). From this discord, we sought to determine which mAChR subtype(s) are functioning in the HEK293 cells used in our study. We first determined by RT-PCR which mAChR mRNAs could be detected within total RNA extracted from our HEK293 cells; Fig. 6A demonstrates that PCR amplicons were detected for the M1, M3, and M4mAChR transcripts.
To further investigate the subtype(s) of mAChR that are functional in HEK293 cells, we analyzed calcium flux responses following carbachol activation in cells treated with siRNA SMARTpools against the three receptor subtypes identified by RT-PCR. We first verified the specificity of these siRNA SMARTpools on HEK293 cell monolayers previously transfected with HA-tagged mAChR expression vectors. Subsequent immunoblotting demonstrated that each siRNA SMARTpool was specific and effective in decreasing protein levels of its respective target mAChR subtype (Fig. 6B). Using these selective siRNA, we observed that the knockdown of M3 mAChR dramatically attenuated calcium release following carbachol treatment (Fig. 6C), suggesting that the carbachol responses by our HEK293 cells is driven mainly by the Gq-coupled M3 mAChR subtype. Consistent with this result, pertussis toxin treatment (which selectively abrogates Gi/o-linked GPCR signaling) was not observed to affect carbachol-induced calcium mobilization in the HEK293 cell line used in our studies (Fig. S.1).
To confirm the carbachol signaling results obtained using fluorescence-based detection of intracellular calcium flux, we examined another signaling output from mAChR activation, namely IP accumulation measured after 30 min of carbachol exposure. A positive control transfection of M3mAChR-directed siRNA dramatically decreased IP production after carbachol treatment (Fig. 7). Consistent with the previous calcium mobilization results (Fig. 4), the two RGS8 siRNA duplexes (2 and 3) and the two RGS9 siRNA duplexes (1 and 5) each decreased IP production (Fig. 7). Finally, consistent with previous measurements, transient transfection with each of the four RGS11 siRNA duplexes increased carbachol-induced IP production, with the first three duplexes giving statistically significant results (Fig. 7).
RGS8 regulates M3 mAChR and PAR-1 cell surface expression.
One possible explanation for the changes observed in carbachol-evoked IP production and calcium mobilization from cells transfected with RGS-selective siRNA duplexes is a reduction of total and/or cell-surface receptor expression levels. We therefore examined whether changes in mAChR expression levels could account for the reduced signaling observed upon siRNA treatment. Competitive radioligand binding was first performed on total membrane extracts of HEK293 cells using the tritium-labeled, nonsubtype-specific mAChR antagonist 3-quinuclydinyl benzylate ([3H]QNB) and unlabeled atropine to assess nonspecific binding. Application of M3mAChR-directed siRNA was observed to greatly decrease specific QNB binding (Fig. 8A), consistent with its effects in reducing M3 mAChR expression levels. No significant changes were seen in specific QNB binding upon transfection of RGS8, -9, or -11 siRNA versus nonspecific siRNA, suggesting that observed changes in carbachol responses upon RGS8, RGS9, or RGS11 knockdown were not due to altered expression levels of total mAChRs.
Cell-surface expression assays were next performed by ELISA using HEK293 cells stably expressing the HA-tagged M3 mAChR. As a positive control for reduction of cell surface expression, we again used M3mAChR-directed siRNA and, as expected, we observed significant cell surface receptor loss (Fig. 8A). Transfection of RGS8 siRNA duplex 2 or duplex 3 resulted in statistically significant reductions in M3 mAChR cell surface expression, whereas no significant changes were observed upon transfection of RGS9 or RGS11 siRNA (Fig. 8A).
We also evaluated whether siRNA-mediated knockdown of RGS8 affected endogenous PAR-1 cell surface expression levels. We took advantage of an in-house, specific anti-PAR1 antibody, directed against the extracellular part of the receptor (27), to perform an ELISA for endogenous PAR-1 cell surface levels. A positive control transfection of the PAR-1-directed siRNA duplex completely abolished detection of cell surface-bound PAR-1 protein (Fig. 8B). Transfection of either RGS8-directed siRNA duplex led to a significant decrease of PAR-1 levels as well (Fig. 8B). Total protein quantification of whole cell lysates from both of these ELISA experiments indicated that siRNA transfection did not affect the total protein content (data not shown), thereby excluding toxicity or cell death as a trivial explanation for PAR-1 or M3 mAChR ELISA signal loss.
In this study, we employed a protein family-wide siRNA collection, coupled with a medium-throughput functional analysis of GPCR signaling, to address whether there is demonstrable specificity of action of endogenous RGS proteins toward specific endogenous GPCRs. Previous in vitro biochemical and cellular overexpression studies have provided considerable evidence of which Gα subunits can serve as substrates for RGS domain GAP activity and which GPCR signaling pathways can be blunted by such activity (reviewed in Refs. 26 and 42); however, acute perturbation of endogenous RGS protein expression should be considered a more robust approach for identifying physiological pairings of RGS proteins with GPCR circuitry, without the complications of long-term compensation (as with genomic knockouts) or disturbances in component stoichiometries (as with overexpression approaches). Our present data highlighting differential roles for RGS8, RGS9, and RGS11 in mAChR and PAR-1 signal regulation suggest that this screening approach will be equally valuable to the panoply of GPCRs for which signaling output upon agonist activation can be quantified using medium- to high-throughput cellular methods.
Here, we have provided experimental evidence that the knockdown of endogenous RGS11 levels in HEK293 cells increases the efficacy of carbachol to stimulate mAChR signaling to IP production and intracellular calcium mobilization; such increased signal output is consistent with the prevailing notion that RGS proteins are negative regulators of GPCR signaling by virtue of their GAP activity on activated Gα subunits. Surprisingly, in an opposing direction, RGS9 knockdown decreases both the potency and efficacy of carbachol in this setting. These two members of the R7-subfamily appear to act on M3 mAChR signaling in HEK293 cells without affecting its cell surface expression nor affecting the signaling of PAR-1 in the same cell line.
R7 subfamily members (RGS6, -7, -9 and -11) are characterized by their ability to bind Gβ5 via a central GGL domain (35). This unique Gβ5 pairing, coupled with the structural mimicry between GGL domains and Gγ subunits (7), presents the possibility that R7-RGS proteins not only serve as GAPs for activated Gα subunits, but also serve to couple inactive Gα subunits to GPCRs, akin to the known function of conventional Gβγ subunits. Recombinant Gβ5/R7-RGS protein dimers have been shown in vitro to stimulate the steady-state GTPase activity of Gα subunits from the Gαi/o subfamily, but not of Gαq nor Gα11, when added to proteoliposomes containing heterotrimer-coupled mAChRs (14). However, fluorescence resonance energy transfer (FRET)-based studies in intact mammalian cells have suggested an interaction between Gβ5/RGS7 dimers and Gαq (43). Given the possibility of direct Gαq interactions and our observations that three experimental treatments (RGS9 SMARTpool, RGS9 duplexes 1 and 5) give similar effects as knockdown of Gαq/11 subunits (decreased carbachol potency and efficacy), one could speculate that endogenous RGS9 may play a role in the normal coupling of Gαq/11 subunits to muscarinic acetylcholine receptors in this cell system.
In the nematode worm Caenorhabditis elegans, the two orthologs of Gβ5/RGS protein dimers (GPB-2/EGL-10 and GPB-2/EAT-16) have been implicated in regulating both Gαo and Gαq-mediated signaling that underlies body-bending and egg-laying behaviors (6). Indeed, the Gαo and Gαq pathways function antagonistically in C. elegans, with Gαo negatively regulating the Gαq pathway, possibly via EAT-16 (12). In the same way, our present study in the human cell line HEK293 has uncovered that two members of the R7 subfamily, RGS9 and RGS11, regulate signaling by Gαq-coupled mAChRs in opposing directions. These opposing effects of RGS9 and RGS11 are apparently not common to all Gαq-coupled receptors given that no effect of RGS9 or RGS11 knockdown was observed on PAR-1 signaling in the same system. Studies by Slepak and colleagues (31, 32) involving the overexpression of RGS7, Gβ5, and M3 mAChR proteins in CHO-K1 cells have led these authors to conclude that “open” and “closed” configurations of the RGS7/Gβ5 dimer exist with differential abilities to engage the M3 mAChR third intracellular loop and COOH-terminus; such receptor interaction dynamics, if they also exist for the related RGS9/Gβ5 and RGS11/Gβ5 complexes, could help explain the receptor-selective and opposing functional effects of these latter dimers.
As a potential molecular mechanism for receptor-selective actions of RGS proteins, it has also been demonstrated that some R4 subfamily RGS proteins can directly interact with the third intracellular loops of certain GPCRs to regulate their function (5, 11, 17). These studies have also included RGS8 in their analyses. Saitoh and colleagues (30) have shown that RGS8 overexpression decreases M1 mAChR signaling but does not inhibit signaling from the related M3 mAChR, even though both are Gαq-coupled receptors. The action of RGS8 on mAChR signaling in this context may not necessarily arise from RGS domain GAP activity. Saitoh's group has reported that recombinant RGS8 binds in vitro specifically to Gαo and Gαi3 subunits but not to Gαq from brain extracts (29); moreover, overexpressed RGS8 is attracted to the plasma membrane by coexpression of activated Gαo(QL), but not by Gαq(QL) coexpression (30). In contrast, our more recent comprehensive study of Gα selectivity by R4-subfamily members using surface plasmon resonance demonstrated strong binding of RGS8 to both Gαq and Gαi1 subunits (36). Nevertheless, studies by several groups suggest that RGS8, like other R4-subfamily members, can interact with GPCRs such as the mAChRs either directly (3, 5, 17) or indirectly via the scaffold protein spinophilin (9, 39), thereby providing a possible explanation for our observation that RGS8 knockdown leads to decreased M3 mAChR and PAR-1 surface expression and thus reduced signaling to second messenger production. Decreased receptor surface expression could reflect a role for RGS8 in receptor biosynthesis and plasma membrane trafficking, stable coupling to heterotrimeric G proteins (3), constitutive endocytosis, or receptor recycling. Future studies will thus need to focus on elucidating the mechanism by which RGS8 functions to stabilize receptor cell-surface levels and the molecular determinants underlying this underappreciated aspect of the overall role for RGS proteins in regulating GPCR signaling.
This study was supported by National Institute of General Medical Sciences Grant R01 GM062338 (to D. P. Siderovski), postdoctoral fellowships from the Heart & Stroke Foundation of Canada (to G. Laroche and P. M. Giguère), and infrastructure support from the NIMH Psychoactive Drug Screening Program (to B. L. Roth).
No conflicts of interest, financial or otherwise, are declared by the author(s).
We thank Dr. Kirill Martemyanov (Univ. of Minnesota) for RGS9 antiserum, Dr. Liu Hao (National University of Singapore) for early design considerations for the human RGS gene transcript RT-PCR primers, and Dr. Angelique Whitehurst (UNC-Chapel Hill) for critical discussion and advice in conducting siRNA-based screens.
- Copyright © 2010 the American Physiological Society