Cross-talk and co-trafficking between rho1/GABA receptors and ATP-gated channels.

Gamma-aminobutyric-acid (GABA) and ATP ionotropic receptors represent two structurally and functionally different classes of neurotransmitter-gated channels involved in fast synaptic transmission. We demonstrate here that, when the inhibitory rho1/GABA and the excitatory P2X2 receptor channels are co-expressed in Xenopus oocytes, activation of one channel reduces the currents mediated by the other one. This reciprocal inhibitory cross-talk is a receptor-mediated phenomenon independent of agonist cross-modulation, membrane potential, direction of ionic flux, or channel densities. Functional interaction is disrupted when the cytoplasmic C-terminal domain of P2X2 is deleted or in competition experiments with minigenes coding for the C-terminal domain of P2X2 or the main intracellular loop of rho1 subunits. We also show a physical interaction between P2X2 and rho1 receptors expressed in oocytes and the co-clustering of these receptors in transfected hippocampal neurons. Co-expression with P2X2 induces retargeting and recruitment of mainly intracellular rho1/GABA receptors to surface clusters. Therefore, molecular and functional cross-talk between inhibitory and excitatory ligand-gated channels may regulate synaptic strength both by activity-dependent current occlusion and synaptic receptors co-trafficking.

Neuronal activity is regulated by a number of transmitters acting on different receptor types (1). Fast neurotransmission is achieved through different classes of transmitter-gated channels, including the P2X and nicotinic receptor superfamilies (1,2). The family of P2X ATP-gated cation channels is composed of seven genes coding for subunits with two transmembrane domains, intracellular N and C termini, and a large extracellular loop (2). The nicotinic superfamily includes the GABA-gated 1 channels along with the acetylcholine, 5-HT 3 , and glycine receptors that share several structural features, including a large extracellular N-terminal domain, four hydrophobic transmembrane domains (M1-M4), and a long cytoplasmic loop connecting M3 and M4 (1). GABA receptor channels have been classified into two subtypes based on their pharmacological properties. GABA A receptors are inhibited by bicuculline, whereas GABA C receptors are insensitive to this antagonist (1). Diversity of GABA A receptors is achieved by pentameric assembly of multiple subunits, including ␣1-6, ␤1-3, ␥1-3, ␦, , and ⑀. GABA C receptors are composed of 1-3 subunits that can assemble into homo-oligomers (3,4). Recent data suggest that a 1 subunit could co-assemble with a GABA A subunit (5,6). Neuronal ATP and GABA C ionotropic receptors are involved in fast excitatory and inhibitory synaptic transmission, respectively, and display overlapping distribution in many regions of the nervous system, including DRG, dorsal horn of the spinal cord (7,8), retina (9 -13), hippocampus (14 -15), cerebellum (14,16), and anterior pituitary (17,18).
Recently, Jo and co-workers described ATP and GABA corelease from the same axon terminals into the dorsal horn of the spinal cord and into hypothalamic neurons where they activate P2X and GABA A receptors, suggesting that a single synapse can be excitatory or inhibitory (19,20). Functional cross-inhibition between receptors activated by ATP and either acetylcholine or GABA has been reported lately in neurons or transfected cells (21)(22)(23)(24)(25), and we recently demonstrated that a physical interaction between P2X and 5-HT 3 receptors leads to an activity-dependent cross inhibition between the two cationic channels (26). We decided to explore the possibility that ligandgated chloride channels may also interact with P2X ATP-gated channels through similar mechanisms.
Because P2X and 1/GABA receptors are co-expressed in the central nervous system (7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18), we investigated potential molecular and functional interactions between these two types of transmitter-gated channels in heterologous expression systems. We report here an intracellular inhibitory cross-talk between P2X 2 and GABA C 1 receptors. Moreover, we also show a link between the physical coupling of these channels and their subcellular localizations in mammalian neurons.

EXPERIMENTAL PROCEDURES
Plasmid Constructions-Wild-type rat P2X 2 , a C-terminal truncated form of P2X 2 (P2X 2 TR), and wild-type rat 1 clones were available from previous work (18,27). Carboxyl-terminal epitope-tagged P2X 2 subunits with either hexahistidine (His 6 )-tagged or enhanced YFP-tagged sequences and a C-terminal His 6 -tagged 1 subunit (1-HIS) were obtained as described previously (28). N-terminal GFP-tagged 1 subunit (1-GFP) was generated by insertion of two enhanced GFPs in tandem into a unique BamHI restriction site. This artificial BamHI site was created at amino acids positions Gly 49 -Ala 50 by site-directed mutagenesis (QuikChange, Stratagene). The intracellular C-terminal domain of P2X 2 (P2X2-CT) and the main intracellular loop (IL2) of the 1 subunit were amplified by PCR using 5Ј-and 3Ј-primers incorporating an initiation methionine and a stop codon, respectively. All constructs were subcloned into pcDNA3 (Invitrogen) and verified by automatic dideoxy DNA sequencing.
Neuronal cultures were performed according to a modification of the procedure described by Goslin et al. (29). Briefly, hippocampi of rat embryos were dissected at day 18. Dissociation was achieved with a Pasteur pipette after trypsinization. Cells were plated on poly-D-lysinecoated coverslips (Electron Microscopy Sciences) in complete Neurobasal medium supplemented with B27 (Invitrogen) containing 1 mM L-glutamine, penicillin G (10 units/ml), and streptomycin (10 mg/ml). Four hours after plating, the coverslips were transferred to dishes containing conditioned medium obtained by incubating the complete medium described above on glial cultures (70 -80% confluency) for 24 h. The medium was partially changed every 3-4 days.
Hippocampal neurons were transfected at 7-8 days in vitro as follows. For each coverslip, a total of 1.5 g of plasmid DNA was mixed with 50 l of Neurobasal medium without B27 supplement. For cotransfection experiments, the proportions of each plasmid were first adjusted to reach similar levels of expression, and the plasmids were thoroughly mixed prior to the addition of 15 l of Plus™ packaging reagent (Invitrogen). After 15 min of incubation, 1.5 l of LipofectAMINE 2000 (Invitrogen) in 50 l of Neurobasal medium was added, and incubation was continued for another 30 min. After the addition of 150 l of complete Neurobasal medium containing B27 supplement, the mix was applied on the neuronal culture for 3 h at 37°C. Expression was then conducted for 48 h in the original medium that was added back to the neurons. Under these conditions, 85-90% of transfected neurons were cotransfected with two plasmids.
Electrophysiology and Data Analysis-Two-electrode voltage clamp recordings were performed using glass pipettes (1-2 megaohms) filled with 3 M KCl solution. Oocytes were perfused at a flow rate of 10 -12 ml/min with Ringer's solution (pH 7.4) containing 115 mM NaCl, 5 mM NaOH, 2.5 mM KCl, 1.8 mM CaCl 2 or BaCl 2 , and 10 mM HEPES. Membrane currents were recorded through an OC-725B amplifier (Warner Instruments) and digitized at 500 Hz. All drugs (purchased from Sigma) were dissolved in the perfusion solution and applied using a computer-driven valve system (BPS8, Ala Scientific). We compared the peak of actual responses to the peak of predicted additive responses obtained with Axograph software (Axon Instruments). All recordings were made at room temperature. Statistical differences between means were assessed using Student's t test. Dose-response curves were fitted to the Hill sigmoidal equation, and EC 50 values were determined by a non-linear regression analysis test using Prism 2.0 (Graphpad, San Diego, CA).
Co-purification and Western Blotting-Following the measurement of 1-GFP alone, P2X 2 -YFP alone, P2X2-YFP ϩ 1-HIS, or P2X 2 -HIS ϩ 1-GFP expression levels by electrophysiological recordings, batches of 15 oocytes were homogenized in 10 mM HEPES and 0.3 M sucrose and solubilized in 0.8% Triton X-100 and protease inhibitors (Sigma) at 4°C for 2 h. Total proteins were incubated overnight with equilibrated nickel resin (Qiagen) under agitation at 4°C. After several washes with 20 mM imidazole, bound His 6 -tagged proteins were eluted with 0.5 M imidazole solution and then loaded onto a 8% SDS-PAGE and transferred to a polyvinylidene difluoride membrane. Immunostainings of associated tagged receptors were performed with anti-GFP antibodies (1:2500; Molecular Probes) at room temperature for 2 h followed by 1 h of incubation with anti-rabbit peroxidase-labeled secondary antibodies (1:2000, Jackson ImmunoResearch, West Groove, PA) for visualization by enhanced chemiluminescence (ECL, Amersham Biosciences).
Immunocytochemistry and Confocal Microscopy-Immunofluorescence was performed 9 -10 days after plating and 2 days after transfection. Cells were washed with phosphate-buffered saline containing 0.1 mM CaCl 2 and 0.1 mM MgCl 2 (PBSϩ) at 37°C and then fixed with paraformaldehyde (4%) containing 4% sucrose at 37°C in PBSϩ. Incubation with the primary antibodies was performed in 2% bovine serum albumin, 3% normal goat serum, 3% normal donkey serum, and 0.1% Triton X-100 in PBS for 2 h at room temperature. As the intensity of anti-1 labeling was dependent on Triton X-100 concentration, the detergent could not be completely omitted. However, we found that 0.002% Triton X-100 significantly enhanced anti-1 immunostaining without labeling intracellular organelles. For external labeling, therefore, immunocytochemistry was performed on non-permeabilized cultures using 0.002% Triton X-100. However, because the intensity of anti-1 immunolabeling decreased from 0.1 to 0.002% Triton X-100, the proportion of surface versus total receptors could not be quantified. Labeling was performed with affinity-purified anti-1 (1:200) directed against the extracellular N-terminal sequence (16) and revealed with Cy3-conjugated donkey anti-rabbit for 1 h at room temperature (1:1000, Jackson ImmunoResearch). Alexa Fluor® 488-phalloidin (1:3000 to 1:10,000 dilutions, Molecular Probes) staining was occasionally performed together with the secondary antibody. Immunofluorescence images were generated using a Leica laser-scanning confocal microscope. Contrast and brightness were chosen to ensure that all relevant pixels were within linear range. . When GABA was applied during application of ATP, and, conversely, when ATP was applied during GABA application, significant inhibition (p Ͻ 0.0005) of the responses was also observed (Fig. 1B). Amplitude of I ATP and GABA was 73 Ϯ 4.5% and I GABA and ATP was 77 Ϯ 4% of the predicted current (n ϭ 16).

Reciprocal Cross-inhibition between GABA C and P2X 2 Receptors-We
We showed previously, by recordings of oocytes expressing 1 GABA C receptors and 5-HT 3A receptors, that co-activation by simultaneous application of GABA (10 M) and 5-HT (100 M) evoked responses (I GABA ϩ 5-HT ϭ Ϫ2.7 Ϯ 0.6 A) corresponding to the sum of I GABA ϭ Ϫ1.2 Ϯ 0.4 A and I 5 Ϫ HT ϭ Ϫ1.5 Ϯ 0.2 A, demonstrating that responses mediated by these two independent receptors are additive under similar experimental conditions (26). These results showed that 1 and P2X 2 channels do not function independently during co-activation and that the subunit-specific cross-talk induces rapid and reciprocal current occlusion.
Current Occlusion Was Observed at Low Receptor Densities and Was Not Due to Cross-activation-To check if the crossinhibition was dependent on the density of receptors, the expression level of both receptors was decreased while the ratio of P2X 2 /1 subunits was kept constant. Co-application of ATP (100 M) and GABA (10 M) evoked currents with an amplitude (I ATP ϩ GABA ϭ Ϫ0.4 Ϯ 0.1 A) that was significantly lower than the expected sum of currents evoked separately by ATP (I ATP ϭ Ϫ0.3 Ϯ 0.1 A) and by GABA (I GABA ϭ Ϫ0.3 Ϯ 0.1 A; Fig. 2A). The amplitude of actual ATP ϩ GABA responses represented 61.5 Ϯ 2% of the predicted current (n ϭ 8), indicating that current occlusion during co-activation of GABA C and P2X 2 receptors was not related to artificially high receptor densities.
Oocytes injected with P2X 2 RNA only did not respond to 10 -100 M application of GABA, and responses to 100 M ATP (I ATP ϭ Ϫ4.6 Ϯ 0.1 A, n ϭ 6) were not modulated by coapplication of GABA (I ATP ϩ GABA ϭ Ϫ4.65 Ϯ 0.1A, n ϭ 6) (Fig.  2B). Similarly, in oocytes expressing 1 GABA C receptors only, the application of saturating concentration of ATP did not activate GABA C channels, and co-application of ATP ϩ GABAinduced currents (I ATP ϩ GABA ϭ Ϫ1.9 Ϯ 0.4 A, n ϭ 5) were identical in kinetics and amplitude to GABA responses (I GABA ϭ Ϫ1.9 Ϯ 0.4 A, n ϭ 5; Fig. 2C). The ATP concentrationresponse curve of the P2X 2 channel was not modified by coexpression with 1 receptors (EC 50 for ATP was 2.6 Ϯ 1.2 and 4.3 Ϯ 1.3 M, n ϭ 7, respectively, for oocytes expressing P2X 2 alone and the P2X 2 ϩ 1 subunit; Fig. 2D). Normalized ATP (100 M) responses obtained from oocytes expressing P2X 2 alone or 1 ϩ P2X 2 indicated that P2X 2 channel kinetics were not affected by the presence of 1 receptors (Fig. 2D). Conversely, GABA concentration-response curves determined from oocytes expressing either 1 receptor alone or 1 ϩ P2X 2 (Fig.  2E) had similar EC 50 (2.5 Ϯ 1.9 and 1.7 Ϯ 1.5 M, respectively, n ϭ 5). Kinetics of GABA C receptors were identical in the absence or presence of P2X 2 receptors (Fig. 2E). These results showed that the cross-inhibition between GABA C and P2X 2 receptor is not mediated by agonist cross-modulation or by a shift in agonist sensitivity and further indicated that functional cross-talk occurs only when both channels are co-activated.
Voltage, Ionic Flux Direction, and Ca 2ϩ Independence-To investigate a potential voltage-dependence of the cross-talk, we recorded responses obtained from the same oocyte expressing 1 and P2X 2 channels after successive applications of ATP (100 M), GABA (10 M), and a mixture of both agonists at different holding potentials. Currents induced by co-application of ATP ϩ GABA (actual) were significantly lower than the expected sum of individual ATP and GABA responses (prediction) at holding potential ranging from Ϫ80 to 20 mV (Fig. 3, A and B). Amplitude of I ATP ϩ GABA was 65 Ϯ 6% at Ϫ80 mV, 70 Ϯ 5% at Ϫ60 mV, 75 Ϯ 4% at Ϫ40 and Ϫ20 mV, 60.3 Ϯ 9% at 0 mV, and 52 Ϯ 10% at 20 mV of the corresponding predicted current (n ϭ 7). P2X 2 channels are non-selective cationic channels and display a current-voltage relationship with marked inward rectification and a reversal potential close to 0 mV (30), whereas 1 channels are mainly permeable to Cl Ϫ and display a linear I-V relationship with a reversal potential close to Ϫ20 mV from oocytes recorded in normal ringer solution (31). In oocytes co-expressing GABA C and P2X 2 channels, the experimental current-voltage relationship for P2X 2 was inwardly rectifying, whereas that for 1 channels was linear with reversal potential of ϩ3.4 Ϯ 1.3 mV for ATP and Ϫ19 Ϯ 2 mV for GABA respectively (Fig. 3, A and C). Current-voltage relationship for actual and predicted ATP ϩ GABA currents were linear with a slight inward rectification and displayed no difference in reversal potential (Ϫ6.7 Ϯ 6 mV for actual and Ϫ7.6 Ϯ 4 mV for predicted). These data indicated that both channels participated to inhibited ATP ϩ GABA currents induced by co-application of ATP and GABA.
It has been proposed that cross-inhibition between P2X and GABA A receptors expressed in DRG neurons was mainly due to action of Cl Ϫ efflux (generated by GABA inward currents) on P2X channels and by Ca 2ϩ entry on GABA receptors (25). We expressed P2X 2 and 1 in order to obtain outward currents with similar amplitudes. In these conditions, at 30-mV holding potential, we obtained similar responses to each agonist (I ATP ϭ 0.3 Ϯ 0.1 A and I GABA ϭ 0.4 Ϯ 0.1 A, n ϭ 6). ATP ϩ GABA induced responses with amplitudes (I ATP ϩ GABA ϭ 0.4 Ϯ 0.1, n ϭ 6) significantly smaller (p Ͻ 0.0005) than the arithmetic sum of I ATP and I GABA (Fig. 3D). The ATP ϩ GABA current represents only 51 Ϯ 4% of the prediction, showing that current occlusion occurred during Cl Ϫ entry. P2X 2 and 1 GABA receptors also elicited non-additive responses when calcium ions (Ca 2ϩ ) were removed (or substituted with Ba 2ϩ , not shown) from extracellular Ringer solution (Fig. 3E). ATP ϩ GABA responses represent 56.7 Ϯ 7% of the prediction (n ϭ 6). Taken together, these results indicated that cross-inhibition is voltage-independent and did not arise from Cl Ϫ efflux via 1 channels or Ca 2ϩ entry via P2X 2 channels.
Intracellular Domains of Both Receptors Are Involved in Non-additivity-To determine whether intracellular domains of receptors were involved in cross-inhibition between P2X 2 and GABA C receptors, we first co-expressed 1 subunits with P2X 2 TR, a C-terminal truncated form of P2X 2 (Fig. 4A). Activation of P2X 2 TR with 100 M ATP evoked inward responses with slow to fast desensitizing kinetics, depending on oocytes as shown previously (27). As for wild-type P2X 2 , P2X 2 TR receptors expressed alone in oocytes were not gated or modulated by the application of GABA (Fig. 4B). Co-activation of P2X 2 TR and 1 channels evoked currents (I ATP ϩ GABA ϭ Ϫ0.8 Ϯ 0.2 A), not significantly different (p Ͼ 0.5, n ϭ 11) from the sum of the individual responses (I ATP ϭ Ϫ0.3 Ϯ 0.1 A and I GABA ϭ Ϫ0.4 Ϯ 0.1 A). I ATP ϩ GABA represented 107 Ϯ 3.5% of the prediction (Fig. 4C). Similarly, additive responses were recorded when GABA was applied during application of ATP and when ATP was applied during application of GABA (Fig. 4D), accounting for 97.2 Ϯ 1 and 98.5 Ϯ 0.6% of the predicted currents respectively (n ϭ 5). These results showed that P2X 2 TR and 1 receptors were independent, indicating a prominent role of the cytoplasmic domain of these channels in the cross-talk.
To determine whether both intracellular domains were involved in cross-inhibition between P2X 2 and 1 channels, we examined the ability of minigenes coding for P2X2-CT, the C-terminal domain of P2X 2 receptors (amino acids 374 -472; Fig. 5A), and minigenes encoding the intracellular loop between the M3 and M4 transmembrane domains of 1 subunits (1-IL2, amino acids 366 -457; Fig. 5A) to disrupt the functional cross-talk by competition. As illustrated in Fig. 5, B and C, over-expression of P2X2-CT (Fig. 5B) or 1-IL2 (Fig. 5C) significantly blocked the cross-inhibition observed between P2X 2 and GABA C receptors. Co-application of ATP ϩ GABA elicited currents that were not significantly different (p Ͼ 0.5) from the arithmetic sum of individual ATP and GABA applications. I ATP ϩ GABA , recorded in the presence of either P2X2-CT or 1-IL2, represented 101.4 Ϯ 6% (n ϭ 10) and 100.4 Ϯ 3% (n ϭ 7), respectively, of the predicted sum. These data confirm that interactions between cytoplasmic domains are critical for the reciprocal current occlusion between P2X 2 and GABA C receptors.
Physical Coupling and Co-trafficking of 1 and P2X 2 Receptors-To assess the existence of complexes containing P2X 2 and GABA C receptors by using affinity purification and Western blot, we first verified cross-inhibition in oocytes expressing P2X 2 -HIS and 1-GFP channels (Fig. 6A1) or, conversely, expressing 1-HIS and P2X 2 -YFP channels (Fig.  6B1). Currents recorded by co-application of ATP ϩ GABA represented 69.7 Ϯ 6.7% (Fig. 6, A1) and 77.6 Ϯ 4% (Fig. 6, B1) of the prediction (p Ͻ 0.05, n ϭ 5), showing that the addition of tags did not affect the functional cross-talk between P2X 2 and GABA C receptors. After affinity purification of P2X 2 -HIS in total protein extracts from oocytes on nickel resin and labeling using polyclonal anti-GFP antibodies, we detected a band corresponding to the expected size of 1-GFP (Fig. 6A2). Reciprocally, after purification of 1-HIS, a band corresponding to P2X 2 -YFP was revealed by anti-GFP anti-FIG. 2. Non-additivity at low channel densities and no crossmodulation or shift in agonist sensitivity in oocytes co-expressing P2X 2 and GABA C channels. A, co-application of ATP ϩ GABA induced currents (Actual) significantly smaller than the arithmetic sum (Predicted) of the individual ATP (100 M) and GABA (10 M) responses in oocytes expressing low levels of each receptor. B and C, superimposed current traces obtained with 10 M of GABA, 100 M of ATP, or a mixture of ATP and GABA from oocytes expressing P2X 2 receptors alone (B) and homomeric 1 receptors (C). D, ATP dose-response curves from oocytes expressing P2X 2 alone (filled squares) or with 1 (filled circles). Normalized ATP responses obtained from oocytes expressing P2X 2 only and P2X 2 ϩ 1 are presented in the inset. Duration of application was 5 s. E, GABA dose-response curves from oocytes expressing 1 alone (filled squares) and with P2X 2 (filled circles). Normalized GABA (10 M) responses obtained from oocytes expressing 1 only and P2X 2 ϩ 1 are presented in the inset. Duration of application was 5 s. In panels D and E, mean peak currents were normalized to the maximal response (mean Ϯ S.E.) from 5-7 oocytes. Vh ϭ Ϫ60 mV. bodies (Fig. 6B2), suggesting that GABA C receptors form a stable complex with P2X 2 receptors.
To challenge the hypothesis that the cross-inhibition implicated receptor-receptor interactions and, consequently, subunit colocalization, we visualized GABA C and P2X 2 receptors in transfected hippocampal neurons. Wild-type and tagged 1 and P2X 2 receptors were first expressed in hippocampal neurons at 7 days in vitro to compare their subcellular distribution. No significant differences in addressing were noticed between wild-type and tagged receptors, indicating that the GFP/YFP tags did not affect receptor trafficking (not shown). The homomeric GABA C receptors exhibited a punctate distribution restricted to the close vicinity around cell bodies and proximal dendrites in agreement with an earlier report (32). Confocal microscopy, associated with Alexa Fluor 488-phalloidin staining to label the F-actin cytoskeleton and reveal the outline of the cell, showed that most puncta were intracellular (Fig. 7A). This observation was confirmed by surface labeling of the extracellular 1 epitope. At lower concentrations of Triton X-100 (0.002%), the extracellular component of the 1 immunolabeling could be visualized (Fig. 7B). In this condition, very low levels of uniform immunoreactivity were detected on cell bodies and proximal dendrites, suggesting that only a minor fraction of 1 subunits were located at the cell surface. Thus, GABA C receptors expressed in hippocampal neurons exhibit a distribution compatible with a predominant intracellular localization.
In contrast to GABA C receptors, we found P2X 2 -YFP receptors uniformly distributed throughout the neurons, with a predominant surface localization consistent with previous reports (33). YFP fluorescence was not restricted to proximal dendrites but extended to the very extremities of neuronal processes (Fig.  7C, middle). This localization was divided into diffuse and clustered distribution, with a typical topology suggestive of P2X receptors at the plasma membrane, occasionally associated in dendritic clusters (Fig. 7D, middle).
In 40 -50% of neurons that were cotransfected with 1 and P2X 2 -YFP, significant differences in GABA C receptor distribution were noticed (Fig. 7C) when compared with neurons transfected with 1 alone (Fig. 7A). In conditions of co-expression, whereas P2X 2 -YFP receptor topology seemed unaffected by the presence of the 1-subunit, GABA C receptors were translocated to the cell surface and colocalized with P2X 2 -YFP clusters (Fig.  7, C and D). Co-clustering of 1 and P2X 2 -YFP subunits was particularly obvious on proximal dendrites (Fig. 7D). The external immunolabeling of 1 subunits appeared to overlap with P2X 2 -YFP fluorescence, indicating that the addressing of GABA C receptors was changed by the presence of P2X 2 -YFP receptors and that the two receptors were largely colocalized in surface clusters (Fig. 7E). Furthermore, 1 immunofluorescence was found more distally to the cell body, indicating that the targeting of GABA C receptors was also following the one exhibited by P2X 2 -YFP subunits. The specificity of this effect was verified by transfecting neurons with 1 GABA C and 5-HT 3A receptors; no changes in subcellular labeling were ob- , or a mixture of both agonists recorded from oocytes co-expressing P2X 2 and GABA C receptors (n ϭ 6). Non-additivity of ATP and GABA responses were observed at 30 mV holding membrane potential (D) and in calcium-free extracellular solution (E); ***, p Ͻ 0.0005. served following co-transfection (not shown). It can be concluded from these results that, in neurons expressing both receptors, P2X 2 receptors regulate the addressing of GABA C receptors in a dominant way by inducing their translocation from an internal vesicular compartment to surface clusters common to both receptors. DISCUSSION No functional or biochemical interaction was previously described between identified subtypes of excitatory and inhibitory ligand-gated channels. We demonstrate here a functional cross-talk and a physical coupling between P2X 2 and 1 GABA C receptors. A reciprocal current inhibition (25-50%) was recorded from oocytes co-expressing homomeric P2X 2 and 1 channels when they were simultaneously activated, showing that these two unrelated subtypes of receptor channels are not functionally independent. The percentage of current occlusion is identical whatever the sequence of agonist application (coapplication, GABA during application of ATP, or vice versa), indicating that each type of receptor has an equal propensity to inhibit the other one. Similarities in kinetics (i.e. activation and desensitization) between recorded and predicted currents (see Fig. 1A) induced by the co-application ATP ϩ GABA suggested that the cross-talk closed an important proportion of both receptors. Application of ATP did not gate or modulate GABA C receptors, and GABA did not activate or modulate P2X 2 receptors, showing that nonspecific cross-activation or cross-desensitization between one type of transmitter and the other type of receptor could not explain the cross-inhibition observed during co-application. A similar reciprocity of cross-inhibition was observed between P2X 2 and nicotinic or 5-HT 3 receptors (24,26). Interestingly, an inhibitory cross-talk between P2X and GABA A channels reported in dorsal root ganglion neurons indicates a more pronounced inhibitory effect of GABA receptors on P2X channels than the reverse (25). It was proposed that the chloride efflux generated by GABA-gated channels could inhibit P2X receptors and, inversely, that the Ca 2ϩ influx mediated by P2X channels could inhibit the GABA-gated channels (25). It is unlikely that these mechanisms play a major role in the functional interaction between the P2X 2 and GABA C receptors studied here, because we observed a reciprocal current occlusion when the oocytes co-expressing both channels are clamped between Ϫ80 to 30 mV (above the reversal potential of showing co-expression of P2X 2 TR and 1 channels. C, simultaneous application of ATP ϩ GABA (Actual) elicited responses that were not significantly different (ns; p Ͼ 0. 5, n ϭ 11) than the arithmetic sum of individual ATP and GABA responses (Predicted). D, additives responses were also recorded when GABA was applied during ATP application (ATP and GABA) and when ATP was applied during GABA responses (GABA and ATP). Amplitudes of responses normalized to the predicted response from each oocyte. Vh ϭ Ϫ60 mV. chloride ions) and in the absence of extracellular calcium ions. The cross-talk was also recorded at low channel densities, giving rise to small current amplitudes at saturating concentrations of agonists and suggesting a significant affinity of coupling between P2X 2 and GABA C receptors.
Co-expression of 1 subunits with a functional truncated version of the P2X 2 subunit lacking most of its C-terminal domain resulted in receptor independence, as indicated by current additivity. Moreover, an excess of minigene coding for the C-terminal domain of P2X 2 or for the main intracellular loop (IL2) of 1, co-expressed with P2X 2 and GABA C receptors, resulted also in the loss of functional cross-talk despite normal individual current phenotypes. The results of these mutagenesis and competition experiments underlined the importance of the intracellular domains of both receptor subunits in this cross-inhibition. The crucial role of cytoplasmic domains was also demonstrated in the functional coupling between P2X and 5-HT 3 receptor channels. However, in competition experiments these cytoplasmic domains were not able to disrupt physical association between 5-HT 3 and P2X receptors (26). We also observed co-purification of GABA C and P2X 2 receptors expressed in oocytes. No clear sequence homology exists between the sequences of the large intracellular loops of members of the nicotinic receptor superfamily and between the C-terminal domains of the different P2X family members. Taken together, this led us to formulate the hypothesis of the existence of two distinct mechanisms involved in the physical and functional coupling between these two families of ligand-gated channels.
To identify putative sequences required in both mechanisms, the propensity of other members of P2X family to interact with GABA C or different subunit combinations of GABA A receptors should now be investigated.
Our immunolocalization experiments in the primary culture of neurons revealed that expression of the 1 subunits was mainly intracellular and compatible with a prominent vesicular localization. Whether these vesicular compartments are part of the routing mechanism involved in 1 subunit trafficking, an early endosomal compartment involved in subunit recycling, or a lysosomal network remains to be determined. Co-expression with P2X 2 resulted in a dramatic change in the distribution of GABA C receptors that were translocated to the cell surface and colocalized with P2X 2 clusters. When co-expressed with P2X 2 , the 1 receptor was also found more distally to the cell body, indicating that the radial topology of GABA C was driven by P2X 2 receptors. The co-targeting in transfected neurons and the co-purification of P2X 2 and 1 receptor channels expressed in oocytes demonstrate the existence of physical complexes formed by the constitutive association of P2X 2 and GABA C receptor channels in a different expression system.
Our findings suggest also that, besides the regulation of neuronal activity by modulation of current amplitude during co-activation, physical association between receptors could regulate the location and the number of surface receptors, which is an important means for synaptic neuronal regulation. Recent studies showed that P2X 1 and P2X 4 receptors undergo rapid agonist-dependent internalization and recycling, contrary to P2X 2 , which appeared to be more stable at the cell surface (33)(34)(35)(36). The molecular mechanism of P2X receptor trafficking has yet to be elucidated despite identification of some associated intracellular proteins (37,38). Dynamic regulation of GABA A and GABA C receptor numbers at the membrane is regulated by interactions with several proteins. For example, gephyrin and dystrophin have been shown to play a role in the anchoring of GABA A receptors, although no physical interaction with gephyrin has been demonstrated in vitro (39,40). GABARAP, a microtubuleassociated protein, and Plic-1, an ubiquitin-related protein, interact with GABA A receptor subunits and are involved in receptor trafficking (41,42). MAP1B is a large protein that binds actin and tubulin and is the only GABA C receptor binding partner identified so far. Whereas expression of the 1 subunit in COS cells produced diffuse surface and intracellular labeling, coexpression with MAP1B resulted into a punctuate intracellular localization of 1 receptor (43). Proteins associated with P2X or GABA receptors bind directly to the same cytoplasmic domains critical for their cross-inhibition, suggesting that these proteins and/or cytoskeleton could also participate in the functional coupling.
The P2X 2 receptor subunit is widely expressed in the nervous system, whereas the 1 subunit shows a restricted expression profile. Both of these subunits have been detected in the same regions, suggesting that they are co-localized in the same neurons (7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18). Lately, the 1 subunit was immunolocalized in DRG neurons and in lamina I and II of the spinal superficial dorsal horn, and 1-containing receptors have been shown to modulate nociceptive transmission (8). However, it is still unclear if 1 forms homooligomeric receptors or if it associates with GABA A subunits to form a hybrid receptor/channel (8). P2X 2 receptors are also expressed in these neuronal populations and have also been shown to be involved in modulation of pain/sensory pathways in spinal cord (44). Physical association between 1 and P2X 2 receptors could ensure a co-clustering on the soma of DRG neurons, on primary nociceptive afferent fibers, or in interneurons constituting lamina II. The resulting cross-inhibition could modify the balance between ATP and GABA signaling and, consequently, could contribute to normal or pathological pain/sensory transduction.
Accumulating data suggest that, besides modulation through intracellular second messenger cascades, direct interactions may regulate receptor signaling. This has been illustrated by interactions between different G-protein-coupled receptors (45), between GPCRs and ligand-gated channels (46,47), and between different ligand-gated channels (Refs. 26 and 48 and this work). Moreover, our findings show also that, in addition to a functional role in modulation of receptor activity, receptor-receptor interaction may regulate receptor trafficking and targeting. Thus, interactions between distinct neuronal ligand-gated channels, excitatory or inhibitory, appear to be a widespread mechanism for the fine regulation of fast transmission in peripheral and central synapses.