Dissecting G Protein-coupled Receptor Signaling Pathways with Membrane-permeable Blocking Peptides

To determine the intracellular signaling mechanism of the 5-HT2C receptor endogenously expressed in choroid plexus epithelial cells, we implemented a strategy of targeted disruption of protein-protein interactions. This strategy entails the delivery of conjugated membrane-permeable peptides that disrupt domain interaction at specific steps in the signaling cascade. As proof of concept, two peptides targeted against receptor-G protein interaction domains were examined. Only GqCT, which targets the receptor-Gq protein interacting domain, disrupted 5-HT2C receptor-mediated phosphatidylinositide hydrolysis. GsCT, targeting the receptor-Gs protein, disrupted β2 adrenergic receptor-mediated activation of cAMP but not 5-HT2C receptor-mediated phosphatidylinositide hydrolysis. The peptide MPS-PLCβ1M, mimicking the domain of phospholipase Cβ1 (PLCβ1) interacting with active Gαq, also blocked 5-HT2C receptor activation. In contrast, peptides PLCβ2M and Phos that bind to and sequester free Gβγ subunits were ineffective at blocking 5-HT2C receptor-mediated phosphoinositol turnover. However, both peptides disrupted Gβγ-mediated α2A adrenergic receptor activation of mitogen-activated protein kinase. These results provide the first direct demonstration that active Gαq subunits mediate endogenous 5-HT2C receptor activation of PLCβ and that Gβγ subunits released from Gαq heterotrimeric proteins are not involved. Comparable results were obtained with metabotropic glutamate receptor 5 expressed in astrocytes. Thus, conjugated, membrane-permeable peptides are effective tools for the dissection of intracellular signals.

To determine the intracellular signaling mechanism of the 5-HT 2C receptor endogenously expressed in choroid plexus epithelial cells, we implemented a strategy of targeted disruption of protein-protein interactions. This strategy entails the delivery of conjugated membrane-permeable peptides that disrupt domain interaction at specific steps in the signaling cascade. As proof of concept, two peptides targeted against receptor-G protein interaction domains were examined. Only G q CT, which targets the receptor-G q protein interacting domain, disrupted 5-HT 2C receptor-mediated phosphatidylinositide hydrolysis. G s CT, targeting the receptor-G s protein, disrupted ␤2 adrenergic receptor-mediated activation of cAMP but not 5-HT 2C receptor-mediated phosphatidylinositide hydrolysis. The peptide MPS-PLC␤1M, mimicking the domain of phospholipase C␤1 (PLC␤1) interacting with active G␣ q , also blocked 5-HT 2C receptor activation. In contrast, peptides PLC␤2M and Phos that bind to and sequester free G␤␥ subunits were ineffective at blocking 5-HT 2C receptormediated phosphoinositol turnover. However, both peptides disrupted G␤␥-mediated ␣ 2A adrenergic receptor activation of mitogen-activated protein kinase. These results provide the first direct demonstration that active G␣ q subunits mediate endogenous 5-HT 2C receptor activation of PLC␤ and that G␤␥ subunits released from G␣ q heterotrimeric proteins are not involved. Comparable results were obtained with metabotropic glutamate receptor 5 expressed in astrocytes. Thus, conjugated, membrane-permeable peptides are effective tools for the dissection of intracellular signals.
The 5-HT 2 receptor family consists of three members, 5-HT 2A , 5-HT 2B , and 5-HT 2C . All three receptors belong to the G protein-coupled serpentine receptor superfamily. Their pharmacological profiles are very similar, leading to difficulty in defining their functional roles. 5-HT 2 receptors have been implicated in behaviors such as sleep, feeding, aggression, pain, and anxiety and are thought to play a role in a number of central nervous system disorders including affective disease, schizophrenia, and epilepsy (1). In addition, 5-HT 2 receptors may play a major role in mediating the actions of hallucinogenic drugs (2) as well as antipsychotic drugs (3,4). Mice expressing nonfunctional 5-HT 2C receptors exhibit epileptic and obese phenotypes (5,6), suggesting that these receptors play a crucial role in moderating central nervous system function.
Expression of the 5-HT 2C receptor is exceptionally high in the choroid plexus (7,8), where it plays a role in the regulation of production and composition of cerebrospinal fluid (9 -11). Initial studies of 5-HT 2C receptor signaling showed that these receptors activate the downstream intracellular effector, phospholipase C␤ (PLC␤) 1 resulting in the hydrolysis of phosphatidylinositol-4,5-bisphosphate into inositol-1,4,5-triphosphate and diacylglycerol (12). In addition, activation of the 5-HT 2C receptor has been observed to release arachidonic acid (13), increase cyclic GMP (14), and regulate potassium channels and Ca 2ϩ -activated chloride channels (15)(16)(17)(18). These observations suggest that 5-HT 2C receptor activation results in the induction of multiple signaling pathways. However, it is unclear how each individual intracellular signaling pathway contributes to modulation of the cell as a whole. In this paper, we employed a novel strategy to dissect intracellular signaling pathways, which combines a newly developed peptide synthesis technology with the application of targeted disruption of protein-protein interactions. This strategy is applied to examine PI hydrolysis signaling, a well defined pathway associated with 5-HT 2C receptor activation.
The current model of 5-HT 2C receptor signaling suggests that G q/11 heterotrimers are the immediate G protein mediators of receptor signaling based on two indirect observations. First, activation of PLC␤ predominantly occurs through the G␣ q family. Second, 5-HT 2C receptor-mediated PI hydrolysis is largely pertussis toxin (PTX) -insensitive, which suggests that G i/o heterotrimers, which activate PLC via their ␤␥ subunits, are not involved (19). However, previous studies that directly examined the identity of the G protein-mediating 5-HT 2C receptor signaling were all conducted in artificial systems (20 -22) where the receptors may have promiscuous interactions with various heterotrimeric G proteins. To address this question in a native environment, we examined the G protein mediator of 5-HT 2C receptor signaling in primary cultures of choroid plexus epithelial cells and further assessed the role of G␣ and G␤␥ subunits.

EXPERIMENTAL PROCEDURES
Materials-Most peptides used were synthesized in our laboratory; G s CT, G q CT, G o CT, and PLC␤2M were also synthesized by Genosys (The Woodlands, Texas). Antibodies against PLC␤ isozymes and G i/o/z were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). G␣ q antibodies were a kind gift of Dr. Tom Martin (University of Wisconsin-Madison, WI).
Peptide Design-The sequence of the various peptides and their proposed targets are presented in Table I. The membrane-permeable sequence (MPS) peptide was based on a hydrophobic membrane-permeable sequence described previously (23)(24)(25). Lysine and serine residues, as a pseudo-dipeptide, were added at the carboxyl terminus to serve as a linker and a masked aldehyde to facilitate conjugation. The lysine was attached to the carboxyl terminus of the MPS sequence by its primary amine, and then serine was attached at the side chain amine (Fig. 1).
Phospholipase C␤1-mimicking peptide (PLC␤1M) was derived from amino acids 1053-1084 of the PLC␤1 enzyme. This design stems from the observation that loss of the last 10 kDa from the carboxyl terminus of PLC␤1 results in the loss of interaction with active G␣ q (26). Additional work, using a series of deletion mutants, defined region 1030 -1142 as the domain required for interaction with G␣ subunits (27). A specific segment within this region (amino acids 1053-1084) was observed to dose dependently inhibit GTP␥S-dependent activation of PLC using either purified PLC␤1 or a crude membrane assay.
Phospholipase C␤2 mimicking peptide (PLC␤2M) is based on residues 564 -583. The domain of PLC␤2 interacting with G␤␥ subunits has been determined utilizing a peptide fragment strategy. Two twenty amino acid segments of PLC␤2 (564 -583 and 574 -593) were defined as the domains binding to G␤␥ (28). The segment with the optimal interaction with G␤␥ subunits was observed to span amino acids 564 -583. Synthetic peptide of this region exhibited specific binding to G␤␥ subunits as well as specific inhibition of G␤␥-effector interactions.
Phosducin-like peptide (Phos) was derived from carboxyl-terminal residues 168 -195 of phosducin-like protein (PhLP). PhLP isolated from rat brain (29) was determined to be an ubiquitous inhibitor of G␤␥mediated signaling. The region of PhLP conferring interactions with G␤␥ subunits was delineated to be in the carboxyl-terminal domain (30). Expression of glutathione S-transferase fusion proteins including residues 168 -195 of PhLP had inhibitory effects on G o GTPase activity, demonstrating the ability to bind G␤␥. In addition, carboxyl-terminal peptides of PhLP (including amino acids 168 -195) inhibited G␤␥-enhanced rhodopsin phosphorylation by ␤ARK.
G␣ carboxyl-terminal peptides (G q CT and G s CT) were designed based on the last 10 amino acids of the carboxyl terminus, a region of G␣ subunits that has been recognized as a site of interaction between G proteins and receptors (31,32). Peptides corresponding to the last 11 residues of G i/o/s proteins have been demonstrated to be effective in the specific inhibition of receptor-G protein interaction (33). For the purpose of this paper, the corresponding carboxyl-terminal peptides were derived from the following residues: G s (amino acids 385-394) and G q/11 (amino acids 350 -359).
For chemical conjugation of the MPS and the cargo peptide, two requirements of these peptides were necessary. First, cargo peptide contained an amino-terminal cysteine residue (Fig. 1). Second, the carboxyl-terminal domain of the signal sequence (MPS) contained a masked aldehyde moiety. For this purpose, a pseudo-dipeptide Lys(Ser) was attached to the carboxyl-terminal of the MPS sequence (Fig. 1). The peptide is synthesized with a lysine carboxyl terminus, and then a serine residue was chemically added to the ⑀-amine moeity of the lysine residue. To make the MPS peptide chemically reactive, NaIO 4 treatment was used to oxidize the serine residue to an aldehyde moiety. The oxidized MPS was purified by preparative reverse phase-HPLC and the aldehyde product was again verified using matrix-assisted laser desorption ionization-mass spectrometry. Conjugation of the MPS with cargo peptide was accomplished by dissolving both peptides in 0.5 ml of dimethylformamide and by adding 0.5 ml of 0.2 M sodium acetate buffer, pH 5.4, to the mixture. After agitation for 16 -20 h, the conjugate, as a thiazolidine bond formed between the amino-terminal cysteine of the cargo peptide and the MPS aldehyde, was purified using preparative reverse phase-HPLC and the product characterized by matrix-assisted laser desorption ionization-mass spectrometry (Fig. 2). Several peptides had limited solubility (Table I), and these were used in our experiments at the maximum soluble concentration.
Primary cultures of astrocytes were prepared as described previously (35). Brains were removed from postnatal day 2-5 rats, the meninges were carefully removed, and the cerebral cortex was dissected using visual landmarks. Cells were dissociated in horse serum by mechanical trituration, collected by centrifugation, and resuspended in 1 ml of fetal bovine serum. Dissociated cells were cultured in DMEM containing 10% fetal bovine serum in 75-cm 2 culture flasks coated with poly-D-lysine. After 7 days, cells were shaken in an orbital shaker at 37°C overnight to remove nonastrocytic cells. The cells were then replated in 24-or 48-well plates for functional assays.
Phosphoinositide (PI) Hydrolysis Assay-CPE cells plated in 48-well a This lysine is attached to the carboxyl terminus of the serine by its side-chain group (see Figure 1). b The C-terminal cysteine moiety is added to facilitate chemical conjugation through thiazolidine formation.
plates were incubated for 16 -20 h with 2 Ci/ml myo-[ 3 H]inositol (20 -25 Ci/mmol, NEN Life Science Products) in serum-free, inositolfree DMEM to label phospholipid pools. Labeling medium was aspirated, and the cells washed twice with HBSS containing 1 mM Ca 2ϩ and 1 mM Mg 2ϩ . Cells were treated with peptides solubilized in HBSS (ϩCa 2ϩ /Mg 2ϩ ) at 37°C for 30 min. Subsequently, 10 mM lithium chloride and 10 M pargyline were added to the cells for 10 min prior to agonist activation for 30 min at 37°C. The reaction was stopped by aspirating the solution and fixing with 25 l of methanol/well. [ 3 H] Inositol monophosphates were isolated as described previously (36). HEK-␣ 2A cells were plated in 24-well plates for PI hydrolysis assay and assayed as described above for CPE cells, except pargyline was not added. Cells were incubated with thrombin receptor-activating peptide for 30 min.
Primary cultured astrocytes were plated in 48-well plates and assayed for PI hydrolysis as described for HEK-␣ 2A cells above. Metabo-tropic glutamate receptors were activated with 100 M (R,S)-3,5-dihydrophenylglycine.
ADP-Ribosylation Assay-ADP ribosylation was performed as described previously (37). Briefly, CPE cells were plated in 100-mm plates and cultured for 3-4 days in DMEM. Cells were incubated with 500 ng/ml PTX for 16 h in the absence of serum. Membranes, suspended in 50 mM Tris, pH 8.0, containing 5 mM MgCl 2 and 1 mM EDTA buffer, were subjected to ADP ribosylation at 30°C for 1 h in a 50-l reaction containing 100 g of membrane protein, 1 mM ATP, 20 mM arginine, 20 mM thymidine, 100 mM NaCl, 0.25% Lubrol, 5 mM dithiothreitol, 1 g/ml PTX, and 2.5 Ci of [ 32 P]nicotinamide adenine dinucleotide. The reaction was terminated by adding 1.2 ml of 20 mM HEPES, pH 8.0. Membranes were pelleted and separated by polyacrylamide gel electrophoresis, and ribosylated proteins were visualized using a Molecular Dynamics PhosphorImager system.
Western Blot-Protein extraction from cells and separation on acryl-

FIG. 2. Example of sample preparation.
A, products from the conjugation of PLC␤1M with MPS were separated by preparative HPLC. B, matrix-assisted laser desorption ionization-mass spectrometry was performed on the fractions collected in A. Results shown are from the fraction in A denoted by an asterisk. C, the fraction identified to be the MPS-PLC␤1M peptide (denoted by * in A) is checked with analytical HPLC to verify purity.
amide gel was done as described previously (38). The molecular masses were determined using Sigma high molecular weight markers.
For the MAP kinase assay, HEK-␣ 2A cells in 24-well plates, containing serum-free medium, were treated with appropriate peptides for 30 min at 37°C prior to activation with 100 M epinephrine for 2 min. Supernatant was aspirated, and cells were solubilized with 1X sample buffer (62.5 mM Tris, 2% SDS, pH 6.8, containing 10% (v/v) glycerol). Proteins were separated in 12% SDS-polyacrylamide gels. Active MAP kinase was detected using the Promega anti-phospho-MAP kinase antibody at 1:1000 dilution with an overnight incubation at 4°C. Total MAP kinase was detected using NEB total MAP kinase antibodies at 1:500 dilution with an overnight incubation at 4°C. Secondary peroxidase-conjugated donkey anti-rabbit antibodies were used at 1:2000 dilution with incubation at room temperature for 30 min. Immunoreactive protein bands were visualized by treatment of blots with NEB ECL reagent and subsequent exposure to Kodak Biomax film.
For the detection of PLC␤ isozymes and G protein isoforms, CPE were removed from Spargue-Dawley rats and solubilized in Tris buffer containing 10 mM CHAPS (39). The CHAPS-soluble fraction was fractionated in 7.5% SDS-polyacrylamide gels. Detection of PLC␤ isozyme was performed using PLC␤ isozyme-specific antibodies as per the manufacturer recommendations (Santa Cruz Biotechnology). G q/11 detection was achieved using affinity purified polyclonal antibodies provided by Dr. Tom Martin.
cAMP Assay-Primary cultured astrocytes plated in 48-well plates were labeled for 16 -20 h with 2 Ci/ml [ 3 H]adenosine in serum-free DMEM. Peptides solubilized in HBSS (ϩCa 2ϩ /Mg 2ϩ ) were added to cells and incubated at 37°C for 30 min prior to initiation of the assay. Agonist was added and the incubation continued for 30 min at 4°C in presence of 1 mM isobutylmethylxanthine. The reaction was stopped with 10% trichloroacetic acid containing 2 mM ATP and 2 mM cAMP. Accumulated [ 3 H]cAMP was separated on alumina columns as described previously (40).

PLC␤ Signaling Machinery in CPE Cells-Immunoblots
were utilized to evaluate signaling molecules including the PLC␤ isozymes, which have differential specificity for activation either by G␣ q or G␤␥ subunits. Three isoforms of PLC␤ were detected in the choroid plexus (Fig. 3A). Using anti-PLC␤1 antibodies two bands at approximately 140 and 100 kDa were detected; the latter is an expected degradation product of PLC␤1 (Santa Cruz antibody protocol). PLC␤2 and PLC␤3 were detected with apparent masses of approximately 100 and 140 kDa, respectively. However, PLC␤4 was not present at a detectable level. We also probed CPE extracts with anti-G i/o as well as anti-G␣ q antibodies to verify the potential for G␤␥-and G␣ q -mediated signaling (Fig. 3B). These results demonstrated that G i/o as well as G␣ q are expressed in CPE.
G Protein Mediators of Endogenous 5-HT 2C Receptor Signal in CPE Cells-Given that both PLC␤2 and G i/o exist in CPE cells, the possibility that activation of PLC␤ in CPE cells is mediated by G␤␥ subunits released from G i/o heterotrimers was examined using an indirect method based on PTX sensitivity. The 5-HT 2C receptor PI hydrolysis response in CPE cells is predominantly insensitive to PTX (Fig. 3C), which ADP ribosylates G i/o heterotrimers leading to their inactivation. As a control for PTX activity, CPE cells were pretreated overnight with PTX and then subjected to an in vitro ADP-ribosylation assay. Cells treated overnight with PTX were not ADP-ribosylated by PTX added in vitro (Fig. 3D), whereas non-PTXtreated cells were ADP-ribosylated. These results suggest that G i/o heterotrimers are predominantly not involved in the endogenous 5-HT 2C receptor PI signal.
To determine directly the heterotrimeric G protein mediator of endogenous 5-HT 2C receptor signaling, we introduced the membrane-permeable MPS-G q CT peptide to cultured CPE cells. This peptide is designed to disrupt receptor coupling to G q/11 heterotrimeric protein (Fig. 4A). MPS-G q CT, at 5 M, was able to block PI hydrolysis resulting from treatment of CPE cells with 100 nM serotonin (Fig. 4B). A peptide designed to disrupt receptor coupling to G s heterotrimer (MPS-G s CT) was ineffective in perturbing endogenous 5-HT 2C receptor-mediated PI hydrolysis. Signal sequence (MPS) peptide alone was also ineffective demonstrating that the domain conferring permeability does not attenuate the observed PI signal. Additionally, membrane-impermeable, nonconjugated G q CT peptide was unable to disrupt 5-HT 2C receptor signaling, which is consistent with the idea that without membrane permeability the peptide is not functional. This is also an indication that the effect of the G q CT peptide is not due to general toxicity during Extracts of rat choroid plexi were separated by 7.5% and 12.5% SDS-polyacrylamide gel electrophoresis for PLC␤ isozymes and G␣ proteins, respectively. C, effects of PTX on 5-HT 2C receptor-mediated PI hydrolysis stimulated by serotonin in CPE cells. Graph is representative of six independent experiments; each point represents an average of triplicate determinations. D, ADP-ribosylation assay of CPE cells. Lane 1, control-untreated CPE cells. Lane 2, CPE cells were pretreated overnight using 500 ng/ml PTX. the course of the experiment. The observation that neither MPS nor the G q CT peptide alone was an effective inhibitor of receptor-G q coupling validates the MPS-importing strategy.
As an additional proof-of-concept for the use of membranepermeable peptides designed from the carboxyl terminus of G␣ subunits, we examined the functional effect of MPS-G s CT on endogenous ␤2 adrenergic receptor signaling in cultured astrocytes. As seen in Fig. 4B, MPS-G s CT is effective in blocking, to almost basal levels, ␤2 adrenergic receptor-mediated activation of adenylate cyclase. In contrast, the peptide MPS-G q CT was not functionally disruptive in this system. Furthermore, at the same concentration, MPS-G s CT was ineffective in blocking 5-HT 2C receptor-mediated PI hydrolysis in CPE (Fig. 4B). These results contribute collectively to demonstrate peptide specificity and their lack of toxicity.

Demonstration of Function and Specificity of Membrane-permeable Peptides Targeting G Protein Subunits Using HEK-␣ 2A
Cells-To assess directly whether the activation of PLC␤ is mediated by active G␣ q subunits or free G␤␥ subunits, we designed the following peptides: MPS-PLC␤1M targeted against the disruption of G␣ q -PLC␤ interaction; and MPS-PLC␤2M and MPS-Phos, both designed to bind and sequester free G␤␥ subunits thereby preventing subsequent activation of PLC␤. Because CPE cells lack the appropriate receptor-signaling pathways to determine peptide function, specificity and toxicity, we exploited HEK cells stably expressing ␣ 2A -adrenergic receptors (HEK-␣ 2A ) for this purpose. HEK-␣ 2A cells endogenously express thrombin receptors as well as transfected ␣ 2A adrenergic receptors, which signal through the G i heterotrimeric proteins leading to G␤␥-mediated activation of MAP kinase (41); this serves as a suitable model to evaluate the effects of MPS-PLC␤2M and MPS-Phos peptides on free G␤␥mediated signaling (Fig. 5A). Thrombin receptors have been observed to activate a PTX-insensitive PI signal postulated to be through the G␣ q heterotrimeric proteins (Fig. 5B). These two receptor systems provide divergent signaling pathways to test the aforementioned functional peptides.
Using antibodies directed against the phosphorylated, active form of MAP kinase or against a region of MAP kinase away from the phosphorylation site, we could discern active versus inactive forms of MAP kinase and visualize total MAP kinase. Activation ␣ 2A -adrenergic receptors in HEK-␣ 2A cells with 100 M epinephrine (Fig. 5A, control lanes) results in an increase in the level of active MAP kinase as compared with the basal levels. Total MAP kinase labeling of the same blot indicates that the levels of MAP kinase are equal or even higher in basal versus control lanes. Pretreatment of cells with MPS-PLC␤2M and MPS-Phos peptides disrupted activation of MAP kinase by 100 M epinephrine through the ␣ 2A receptors (Fig. 5A). However, the nonconjugated, membrane impermeant forms of Phos or PLC␤2M did not inhibit MAP kinase activation. MPS alone had no functional effect on MAP kinase activation. In addition, PTX pretreatment abrogated subsequent MAP kinase activation confirming that MAP kinase activation is through G␤␥ subunits released from the G i/o protein. When tested in the thrombin receptor PI hydrolysis pathway, the MPS-Phos and MPS-PLC␤2M peptides were shown to have no inhibitory effects (Fig. 5B) demonstrating the specificity and lack of toxicity of these peptides. These results validate the interpretation that MPS-Phos and MPS-PLC␤2M are functional and specific to target the sequestration and disruption of signaling by free G␤␥ subunits.
Pretreatment of HEK-␣ 2A cells with 100 M MPS-PLC␤1M produced no disruptive effect on ␣ 2A receptor-mediated activation of MAP kinase, indicating that this peptide is apparently not toxic to the cells and does not nonspecifically disrupt G␤␥-mediated signaling (Fig. 5A). However, signaling of endogenous thrombin receptors in HEK-␣ 2A cells was disrupted by MPS-PLC␤1M (Fig. 5B), demonstrating that in the same cells PI hydrolysis blockade can be achieved.
Role of Active G␣ q and Free G␤␥ Subunits in Mediating Endogenous 5-HT 2C Receptor Signaling-To examine the direct contribution of active G␣ q subunits mediating 5-HT 2C receptor signals, MPS-PLC␤1M, designed to disrupt the G␣ q -PLC␤ interaction, was applied to cultured CPE cells. This peptide blocked, down to basal levels, serotonin-induced PI The graph illustrates data from two independent experiments with triplicate determinations. Individual responses were normalized to the average control value corresponding to that particular experiment and are plotted as mean Ϯ S.E. Statistical analyses were performed using one-way analysis of variance (ANOVA) with a nonparametric TUKEY test. hydrolysis (Fig. 6A). However, in the same assay, the G␤␥sequestering peptides, MPS-PLC␤2M and MPS-Phos, at levels that blocked MAP kinase activation, did not significantly inhibit 5-HT 2C receptor-mediated PI signaling. Because evidence suggests that the 5-HT 2C receptor in CPE is the sole mediator of serotonin stimulation (12), these results indicate that active G␣ q subunits are involved in 5-HT 2C receptor signaling. The effect of MPS-PLC␤1M was dose-dependent, as seen in Fig. 6B, with an IC 50 of 55 M. Additionally, concentration response studies showed that MPS-PLC␤1M decreased the maximal signal produced by 5-HT without altering the EC 50 (Fig. 6C). Although this type of effect may be attributed to general toxicity, results observed in earlier experiments do not support this conclusion.

Role of Membrane-permeable Peptides in Endogenous Receptors Expressed in Primary Cultures
Astrocytes-To further demonstrate that the effects of these peptides are not receptoror cell type-specific, we analyzed their effects on metabotropic glutamate receptor (mGluR) signaling in primary cultures of astrocytes. Type I mGluR, consisting of mGluR1 and mGluR5, activate PLC␤ (42). Overexpression of G␣ q augments PI hydrolysis of mGluR1a transfected into HEK-293 cells, suggesting that this receptor couples through G␣ q heterotrimeric protein (43). Furthermore, in astrocytes, the activation of mGluR5 signaling appears to be PTX-insensitive (44 -46). However, the exact identity of the G protein mediating endogenous mGluR5 signal remains unclear. We addressed this question by analyzing primary cultures of astrocytes, which express mGluR5 but not mGluR1 (47). (R,S)-3,5-Dihydrophenylglycine, a mGluR1and mGluR5-specific agonist, was used to activate mGluR5 in these cells. In the presence of MPS-PLC␤1M (300 M), mGluR5-mediated PI hydrolysis was significantly inhibited. In contrast, treatment with MPS-PLC␤2M (100 M) or MPS-Phos (100 M) peptides did not decrease signaling relative to controls (Fig. 7). These results suggest that G␤␥ subunits are not involved in mediating PLC␤ activation of mGluR5, affirming the observed lack of PTX sensitivity and presumed lack of involvement of G i/o heterotrimers. These results indicate that signaling of mGluR5 receptors in astrocytes is mediated by active G␣ q subunits, which is consistent with the current consensus of the involvement of G␣ q heterotrimers. DISCUSSION Most studies of receptor function involve the use of agonists and antagonists to modulate receptor activity in an attempt to understand how these receptors regulate cell physiology. Many G protein-coupled receptors activate multiple independent intracellular signaling cascades, and it is equally important to understand how these various signals contribute to the physiological actions of drugs and to whole cell function. Current tools available for studying multicomponent intracellular signaling pathways are limited in comparison to the plethora of available receptor ligands. Contemporary methods for assessing the contribution of specific signaling molecules generally involve molecular strategies such as antisense oligonucleotides to knockdown expression of a specific protein or transfection experiments to overexpress the protein of interest. However, these approaches disrupt the stoichiometry of the protein within the signal transduction system and are not always applicable to native cell systems. For signaling proteins with enzymatic activity, chemical inhibitors or toxins may be available to provide a more temporally controlled dissection of signaling. The drawbacks of these compounds are that many of them have specificity problems and may be toxic. Currently, there is no systematic broadly applicable approach to the dissection of intracellular signaling events in native cells or tissues.
In this paper, we describe a strategy that involves the use of functional peptides coupled to membrane-permeable peptides for dissecting intracellular signaling pathways. Signaling in many cases requires an activated protein to contact directly with its immediate downstream mediator. Thus, it is possible to disrupt protein-protein interactions of specific coupling domains by introducing into cells only the binding domain. Specificity is intrinsic to the amino acids encoding the peptide as well as their inherent binding properties. The blocking peptide strategy to disrupt protein-protein interactions has been applied sporadically by other investigators in signaling studies. Hamm and Rarick (31) as well as Taylor and Neubig (32) reviewed the use of peptides as probes for G protein signal transduction. Hamm and Rarick (31) analyzed the use of peptides to study receptor-G protein interaction and G proteineffector interaction and pointed out that, with a blocking peptide strategy, signaling pathways can be disrupted at any level, provided that protein-protein interactions exist. However, a limitation of this strategy lies in the intracellular import of peptides. Because most peptides do not readily penetrate the cell membrane, their use in disrupting intracellular signaling has been primarily limited to in vitro analyses, a major impediment to general use of this methodology. To alleviate this problem, we have adopted a noninvasive approach to introduce peptides into cells, the addition of a MPS to the blocking peptides. The feasibility of this approach has been demonstrated in experiments to prevent the inducible nuclear import of transcription factors in human monocytic, endothelial, and T lymphocyte cell lines (23,25). We have also implemented a recently developed approach of modular peptide synthesis, which involves separate synthesis of the MPS and functional peptides with subsequent chemical ligation of the two peptides under mild aqueous conditions (24). The biological activity of the resulting modular peptides has been extensively documented by comparing modular peptides with those synthesized conventionally (23,25,48,49). This modular approach not only provides versatility in preparing multiple MPS-coupled functional peptides for structure and function studies, but it also serves to increase yield and reduce the cost of synthesis.
The conjugation of MPS with blocking peptide has great potential as a versatile tool for studying intracellular signaling. This approach was validated in recombinant cell lines and then applied to cultured CPE cells for direct analysis of endogenous 5-HT 2C receptor signaling at the receptor-G protein level and at the G protein-effector level. A decapeptide mimicking the carboxyl terminus of G␣ q subunits (MPS-G q CT), a domain conferring specific interaction with receptors, profoundly dis-rupted 5-HT 2C receptor-mediated PI hydrolysis in CPE cells. A peptide targeting the disruption of receptor-G s interaction (MPS-G s CT) was ineffective, although the same G s peptide was able to block ␤2-AR-mediated cAMP formation in astrocytes. The MPS domain alone was inactive, suggesting that the carrier peptide made no significant contribution to the blockade of any signal cascades tested. Nonconjugated peptides also did not have a significant effect, consistent with the premise that, without attachment of the MPS sequences, the blocking peptides are unable to penetrate the plasma membrane. These results indicate that the G␣ q heterotrimer mediates 5-HT 2C receptor signaling in choroid plexus, consistent with previous observations in reconstituted systems (20), and also demon- strate the specificity of the G protein-targeted approach.
It is well documented that G␤␥ subunits released from G i/o heterotrimers have the ability to activate effectors, including PLC␤2 (50 -53). More recently, effector activation by G␤␥ released from G s heterotrimers has been reported (54). The possibility also exists for signaling mediated by G␤␥ subunits released from G q heterotrimers. For example, studies in Xenopus oocytyes have shown that the M3 muscarinic receptor, which is G q/11 -coupled, activates PLC␤ mainly through G␤␥ subunits (55). However, transfected (COS) cells expressing the G q/11 -coupled parathyroid hormone or calcitonin receptor failed to show augmented PI hydrolysis when cotransfected with G␤␥ subunits (56). Studies such as these in artificial systems may not represent the in vivo situation, therefore, we developed specific membrane-permeable peptides to evaluate the contribution of G protein subunits to the PI hydrolysis signal mediated by endogenous 5-HT 2C receptors. To elucidate the function and specificity of the newly developed conjugated G␣ q and G␤␥ peptides, we used an HEK-␣ 2A stable cell line expressing the cloned ␣ 2A adrenergic receptor as well as endogenous thrombin receptors. ␣ 2A adrenergic receptor activation of MAP kinase is mediated by G␤␥ subunits released by G i/o heterotrimer (41), whereas thrombin receptor signaling, in these cells, has been observed to be mediated by G␣ q protein (57). The addition of MPS-PLC␤1M peptide into the HEK-␣ 2A cell line blocked the subsequent activation of PLC␤ by thrombin receptor-activating peptide, consistent with the expected role for G␣ q . The specificity of MPS-PLC␤1M peptide was confirmed, because it had no effect on ␣ 2A -adrenergic receptor-mediated activation of MAP kinase, a G␤␥-dependent response. In contrast, G␤␥sequestering peptides, MPS-Phos and MPS-PLC␤2M, were both effective in disrupting ␣ 2A -adrenergic receptor activation of MAP kinase. The consistent results of both G␤␥-sequestering peptides, which have different size and amino acid composition, provide converging support for their functionality and specificity. When the G␤␥-sequestering peptides were tested on thrombin receptors in HEK-␣ 2A cells, they failed to block receptoractivated PI hydrolysis, providing evidence for the specificity of these constructs.
The current studies of 5-HT 2C receptors in CPE cells suggest that G␤␥ does not contribute to the G q/11 heterotrimer-mediated PI hydrolysis signal in this endogenous receptor system. This conclusion is based on results obtained with the specific peptides that block active G␣ q and free G␤␥ subunits released from activated G q/11 heterotrimers. MPS-PLC␤1M blocked 5-HT-mediated PI hydrolysis in CPE cells, whereas both G␤␥sequestering peptides, MPS-Phos and MPS-PLC␤2M, had no effect at concentrations that eliminated ␣ 2A -adrenergic receptor-mediated MAP kinase activation. Analysis of mGluR5 in astrocytes, another putative endogenously expressed G q/11 -coupled receptor, using the same membrane-permeable peptides showed similar results. MPS-PLC␤1M markedly decreased mGluR5-mediated PI hydrolysis, whereas both MPS-PLC␤2M and MPS-Phos did not attenuate signaling in this pathway. The consistency of these results obtained with two different endogenous receptors in different native environments suggests that, unlike in artificial systems (55), endogenous G␤␥ subunits released from G q/11 heterotrimers may not contribute to the activation of PLC␤ in native systems. However, these results do not rule out a role of G␤␥ subunits released from G q/11 heterotrimers in other receptor signaling cascades. For example, 5-HT 2C receptors also induce the release of arachidonic acid via phospholipase A 2 activation (58, 59) as well as an increase in the intracellular level of cGMP (14). 5-HT 2C receptor activation also modulates various potassium channels (17,18). Currently, it is not known which intracellular pathways are involved in the activation of these various effectors. The availability of cell-permeant peptides that block at the level of receptor-G protein and G protein-effector should allow a more precise dissection of these various downstream signals.
In conclusion, we have directly demonstrated that PLC␤ signaling of endogenously expressed 5-HT 2C receptors in CPE cells is mediated by G q/11 heterotrimeric protein and specifically by the G␣ q subunit. In addition, the G␣ q subunit is responsible for the metabotropic GluR5 receptor activation of PLC␤ in primary cultures of astrocytes. These studies serve as the first direct demonstration that active G␣ q subunit released from G q/11 heterotrimers mediates the downstream activation of PLC␤ in native systems. We have also provided evidence that subsequent to receptor-G q/11 activation, G␤␥ subunits released from G q/11 heterotrimers do not contribute to the activation of PI hydrolysis signal in natural systems where the stoichiometry of signaling molecules is undisturbed. These studies demonstrate that membrane-permeable peptides, generated by a chemical ligation strategy, are effective tools in the dissection of multiple intracellular signals of G protein-coupled receptors in their native environment.
Acknowledgments-We thank Dr. Tom Martin for providing antibodies against G␣ q , Dr. Lee Limbird for providing the HEK-␣ 2A cells, and Richard Peavy and Dr. Jeff Conn for their assistance in establishing primary cultured astrocytes. We also thank Dr. Yi-an Lu and Cheng-Wei Wu for their technical assistance and expertise in peptide synthesis and purification and Dr. Viet Nguyen (Mass Spectrometry Center, Vanderbilt University Medical Center) for his technical assistance with mass spectrometric analysis of peptides. The assistance of Antoinette Poindexter, Ray Price, and Dr. Paul Gresch in preparing cultured CPE cells is greatfully acknowledged. We also thank Dr. Jon Backstrom for his assistance with Western blots and constructive discussions.