Modulation of Inhibitory Glycine Receptors by Phosphorylation by Protein Kinase C and cAZMP-dependent Protein Kinase*

Recent evidence has suggested a role for phospho- rylation in the regulation of ligand-gated ion channels. We shown that the inhibitory glycine receptor (GlyR) a subunit is phosphorylated in vitro by protein kinase C (PKC). In this report we further show that a subunits of the GlyR can also be phosphorylated by CAMP-depend-ent protein kinase (PKA) in an in vitro assay. Moreover, incubation of intact rat spinal cord neurons with specific PKC or PKA activators leads to increased phos- phorylation of the GlyR a subunits, strongly suggesting a physiological role in its functional modulation. The role of protein phosphorylation in modulating GlyR channels was explored in Xenopus oocytes iqjected with poly (A)’ mRNA isolated from nervous tissue. The treatment of oocytes with phorbol esters or dibutyryl CAMP resulted in a decrease or an enhancement, respectively, of glycine-evoked currents. Our results show

The abbreviations used are: GlyR, glycine receptor; nAChR, Ncotinic acetylcholine receptor; PAGE, polyacrylamide gel electrophoresis; PKA, CAMP-dependent protein kinase; PKC, protein kinase C; TPA, the phorbol ester 12-l)-tretadecanoylphorbol-13-acetate; GABA, y-aminobutyric acid. differing developmental and regional expression (Malosio et al., 1991a, and references therein). Such diversity has been increased by the demonstration of alternative splicing of the a1 subunit mRNA, leading to the expression of a variant (alina) that contains an 8-amino acid insert in a putative cytoplasmic domain (Malosio et al., 1991b). Biochemical and pharmacological studies have demonstrated that a subunits carry the binding sites for agonists and for strychnine (Ruiz-Gbrnez, et al., 1989, 1990Schimieden et al., 1989;Kuhse et al., 1990aKuhse et al., , 1990b. These proteins form homomeric channels in heterologus systems which are pharmacologically and functionally similar to the native GlyRs (Schmieden et al., 1989;Sontheimer et al., 1989;Grenningloh et al., 1990aGrenningloh et al., , 1990bPribilla et al., 1992). In contrast, the /3 subunit seems to be expressed in a single form throughout the brain at all developmental stages and is not required for ligand binding (Malosio et al., 1990b;Grenningloh et al., 1990b), although recent data indicate that it may confer resistance to blockers . GlyR a and /3 subunits are structurally homologus to those of central and peripheral nicotinic acetylcholine receptors (nAChR), GABAA receptors, and some subtypes of glutamate receptors (Betz, 1990). Subunits of this superfamily of ligand-gated ion channels share a common structural motif of four transmembrane domains, a large NH2-terminal extracellular domain and a large intracellular loop between the third and fourth transmembrane domains.
Recent evidence has suggested that a variety of neurotransmitters that regulate intracellular second messenger levels may affect the efficacy of synaptic transmission by modulating the phosphorylation of ion channels (Huganir and Greengard, 1990;Swope et al., 1992;Raymond et al., 1993a). In this regard, most of the subunits of the ligand-gated receptor channel superfamily present consensus sites for protein phosphorylation within the putative major intracellular loop. The nAChR has been reported to be phosphorylated and modulated by a least three different protein kinases: cAMP-dependent protein kinase (PKA), protein kinase C (PKC), and an endogenous tyrosine kinase (reviewed in Wagner et al., 1991). PKA and PKC have also been shown to phosphorylate /3 and y subunits of the GABAA receptor in vitro (Swope et al., 1992;Moss et al., 1992a). Our laboratory has recently reported that the a subunit of the GlyR is phosphorylated in vitro by PKC in a serine residue close to the fourth transmembrane domain (Ruiz-Mmez et al., 1991) and that glycinergic ligands modulate the rate of phosphorylation by this kinase (Vaello et al., 1992).
To clarify the possible physiological role of the phosphorylation of ligand-gated ion channels, two key questions have to be addressed: whether receptor phosphorylation occurs in intact nervous cells in response to specific activators of protein kinases, and which (if any) are the effects of such a covalent modification in receptor function. It is known that phosphorylation of nAChR regulates the rate of desensitization of the receptor (reviewed in Wagner et al., 1991;Swope et al., 1992;Raymond et al., 1993a), and recent reports using recombinant GABAA receptor subunits or GluR6 receptors expressed in human embryonic kidney cells have demonstrated direct phosphorylation of receptor subunits by PKA leading to either decreased or increased responses to receptor agonists (Moss et al., 1992b;Raymond et al., 1993b;Wang et al., 1993). In line with these reports, we report herein that the (x subunit of the GlyR is phosphorylated in intact spinal cord cells in response to the presence of either PKC or PKA activators. Moreover, glycineinduced currents in Xenopus oocytes injected with nervous tissue poly(A)+ mRNA were decreased after phorbol ester treatment of the oocytes and increased after incubation with dibutyryl-CAMP. These results strongly indicate a role for protein phosphorylation in modulating GlyR function in neuronal cells.

Glycine Receptor Phosphorylation in Spinal
Cord Cells-Spinal cord cells were isolated from adult male Wistar rats as previously described (Johnston and Geiger, 1989). Briefly, the spinal cord and brain stem were rapidly removed from the animal, the meninges eliminated and the tissue minced in 0.28 M sucrose at 4 "C. This medium was replaced with Krebs-HEPES medium (125 m~ NaCl, 5 m~ KC1,2.7 m~ CaCl,, 1.3 m~ MgSO,, 10 m~ glucose, 25 m~ HEPES-Tris buffer, pH 7.4) oxygenated with 95% O2, 5% COz at 37 "C. In order to dissociate the cells, the medium was supplemented with 0.6 pg/ml collagenase (Boehringer Mannheim), 1% bovine serum albumin, and 50 pg/ml soybean trypsin inhibitor. ARer 15 min at 37 "C with gentle shaking, the suspensions were mechanically dissociated with glass pipettes as reported (Song and Huang, 1990) and sequencially filtered through two types of nylon mesh screens (200 and 80 p of pore size, respectively). The cells were washed twice with Krebs-HEPES medium and incubated at 37 "C in the same oxygenated buffer for 1 h in the presence of [32Plorthophosphate (0.6-1 mCi/ml). After metabolic labeling, the cells were treated (in the presence of 1 p~ okadaic acid) with vehicle, phorbol esters, or forskolin at the desired concentrations and for the time indicated in the figure legends. Following treatment, cells were rapidly pelleted and homogenized in 10 volumes of ice-cold buffer A (25 m~ potassium phosphate, pH 7.4,0.6 M sucrose, 50 m~ NaF, 10 m~ sodium pyrophosphate, 0.2 m~ EDTA, 2.5 mg/ml bovine serum albumin and 100 p~ phenylmethylsulfonyl fluoride, 100 p benzetonium chloride, 2.5 m~ iodoacetamide, 16 milliunitdml aprotinin, 0.24 mg/ml bacitracin, 12 pg/ml soybean trypsin inhibitor (protease inhibitors mixture)). The homogenate was centrifuged at 20,000 x g for 1 h in order to obtain a membrane pellet free of myelin (Garcia-Calvo et al., 1989). Membrane proteins were subsequently solubilized by incubating 1 h at 4 "C with gentle shaking in buffer B (25 m~ potassium phosphate, pH 7.4, 1 M KCl, 5 m~ EDTA, 5 m~ EGTA, 5 m~ dithiothreitol, 50 m~ NaF, 10 m~ sodium pyrophosphate, either 1.5% Triton X-100 or 1% sodium cholate and the protease inhibitors mixture), followed by centrifugation at 150,000 x g for 1 h. The GlyR complexes were immunoprecipitated from the supernatant by using a specific polyclonal antibody (Ab 384) raised against a synthetic peptide corresponding to residues 384-392 of the rat al subunit (Ruiz-G6mez et al., 1991) or a polyclonal antibody raised against purified GlyR preparations (Ruiz-G6mez et al., 1990) obtained from rat spinal cord (Ab R, to be described elsewhere)., Immunoprecipitations were performed overnight at 4 "C essentially as previously reported Nit6rica et al., 1988), and immune complexes were collected using protein-agarose beads. After extensive washing, the phosphorylation of the GlyR was analyzed by autoradiography after resolving the immunoprecipitated proteins by SDS-PAGE (10% gels). A nonspecific immunoprecipitation control using a preimmune serum exactly in the same experimental conditions was routinely included in all experiments. Specific [3Hlstrychnine binding in immunoprecipitated pellets and resulting supernatants was measured as previously described by our laboratory (Ruiz-G6mez et al., 1990).
In other set of experiments, in vivo phosphorylation of GlyR was assessed in primary neuronal spinal cord cell cultures prepared from 14-to 17-day-old rat embryos as described (Ransom et al., 1977). Cultured cells were grown in Dulbecco's modified Eagle's minimal essential medium supplemented as described by Brewer and Cotman (1989), and maintained at 37 "C in 10% COP for 10-15 days. For metabolic labeling, M.-L. Vaello and F. Mayor, Jr., manuscript in preparation. the culture medium was replaced with phosphate-free Dulbecco's modified Eagle's minimal essential medium and [32Plorthophosphate (0.5 mCi/ml). After 18 h, the cultured cells were treated for 15 min as described above for dissociated adult rat spinal cord cells. The incubation was stopped by aspiring the medium and rapid and repeated washing with ice-cold phosphate-buffered saline. Cells were then lysed in buffer B with 1.5% Triton X-100 and GlyR complexes immunoprecipitated exactly as described above.
When desired, relative band phosphorylation was quantitated by scanning the autoradiograms using a Molecular Dynamics 300 A computing densitometer. In some experiments, phosphorylated proteins bands of interest were excised from the gel, washed, and subjected to partial hydrolysis with N-chlorosuccinimide or complete digestion with endoproteinase lysine C exactly as previously described by our laboratory (Ruiz-G6mez et al., 1991). The products of cleavage were resolved by SDS-PAGE using either 12-20% linear gradient gels or 20% polyacrylamide-glycerol gels and phosphopeptide maps visualized by autoradiography. The M, values of the labeled fragments were compared with those of the peptides that could theoretically be generated as a result of the specific cleavage of the rat (11 sequence (Grenningloh et al., 1987), and with the phosphopeptide maps generated by the same reagents using as a substrate the ( I subunit of the GlyR phosphorylated in vitro by PKC (Ruiz-G6mez et al., 1991).
In Vitro Phosphorylation Studies-GlyR was affinity-purified from adult rat spinal cord and phosphorylated (-1 pmol of [3Hlstrychninebinding sitedassay) by PKC (5 x units) exactly as previously reported by our laboratory (Ruiz-G6mez et al., 1990(Ruiz-G6mez et al., , 1991. The reaction was stopped by addition of SDS sample buffer followed by SDS-PAGE (10% gels). In order to detect GlyR phosphorylation by PKA, an alternative experimental strategy previously reported by Rossie and Catterall (1987) and Lai et al. (1990) was used. Such strategy has been termed "back-phosphorylation" and is based on the incubation of receptor immunoprecipitates with purified protein kinases. In our experiments, GlyR was solubilized from rat spinal cord membranes with Triton X-100 as described (Pfeiffer et al., 1982;Ruiz-G6mez et al., 1990) and immunoprecipitated with the specific anti-peptide antibodyAb384 as detailed above. In other experiments, forskolin-treated or control isolated spinal cord cells were used as a source of GlyR in order to estimate the occurrence of PKA-mediated phosphorylation in vivo. The immunoprecipitated complexes attached to protein A-agarose are first washed with solubilization buffer and then washed three more times at 4 "C with PKA phosphorylation buffer (50 m~ Tris-HC1, pH 7.4, 5 m~ MgC12, 0.1 m~ EDTA, 2 m~ dithiothreitol, 0.05% Triton X-100, 12 m~ NaF, 30 pg/ml soybean trypsin inhibitor 600 pg/ml bacitracin, 200 p~ ATP). The phosphorylation assay is performed by resuspending the washed pellet in 100 pl of PKA phosphorylation buffer containing lo-* units of purified PKA catalytic subunit (generous gift of Dr. Lisardo Boscl, Universidad Complutense, Madrid) and [y-32P]ATP (3-5 countdmidfmol). After incubation for 20 min at 30 "C, the reaction was terminated by rapid sedimentation of the protein A-agarose-attached proteins in a microfuge. The resulting precipitates were washed twice by resuspensiod sedimentation in stop buffer (50 m~ Tris-HC1, pH 7.4, 10 m~ EDTA, 2 m~ dithiothreitol, 1% Triton X-100, 0.2 M NaCl, 20 m~ NaF, 10 m~ sodium pyrophosphate, 600 pg/ml bacitracin, 10 pdml soybean trypsin inhibitor, 1 m~ benzamidine, 17 milliunitdml aprotinin, 2.5 m~ iodoacetamide, 0.1 m~ phenylmethylsulfonyl fluoride, 0.1 m~ benzetonium chloride). The final pellet was resuspended in SDS sample buffer, boiled for 5 min, pelleted again, and the supernatant subjected to SDS-PAGE (10% gels) and autoradiography. A control with preinmune serum was performed in all the experiments.

Expression in Xenopus
Oaytes and Electrophysiology-Poly(A)+ mRNA was isolated from the brain of 14-day-old rats using a established procedure (Fast Track, Invitrogen). Injection of RNA into Xenopus oocytes and voltage-clamp recordings were performed as previously described (Lerma et al., 1989). Briefly, mature oocytes were defolliculated and injected with 50 ng of poly(A)+ mRNA dissolved in water. Oocytes were transferred to Leibovitz L-15 culture medium (0.7 strength, Sigma) supplemented with 5 m~ HEPES, pH 7.6,lOO unitdml penicillin and 1 mg/ml streptomycin. After 5-9 days, oocytes were placed in a -100 p1 recording chamber and perfused with Ringer solution (10 m~ HEPES, pH 7.2, 116 m~ NaCl, 2 m~ KCl, 0.5 m~ CaCl,). The compounds to be assayed (glycine, kainate, GABA, strychnine, phorbol esters, or dibutyryl CAMP) were dissolved in this buffer medium, bath-applied for the time desired, and then washed out. All experiments were done at room temperature (20 "C). Whole-cell current measurements were performed under two-electrode voltage-clamp, at a holding potential of -60 mV.

PKC Activators Promote GlyR Phosphoryation in Intact Spinal Cord
Cells-Previous data obtained in our laboratory indicated that the a subunit of the GlyR purified from adult rat spinal cord was rapidly and stoichiometrically phosphorylated by PKC (Ruiz-G6mez et al., 1991). In order to test whether the receptor was phosphorylated in situ in response to the presence of messengers leading to PKC activation, spinal cord cells isolated from adult rat tissue were labeled with [32P]orthophosphate and treated for 15 min with 0.5 p~ of the phorbol ester 12-O-tetradecanoylphorbol-13-acetate, TPA (Fig. IA, lane 2 ) or vehicle alone (Fig. IA, lane 1 ). Membrane proteins were solubilized with cholate and immunoprecipitated with a specific polyclonal antipeptide antibody raised against the GlyR (see "Experimental Procedures" for details). In the TPA-treated cells, the GlyR antibody immunoprecipitates a phosphorylated band of -48 kDa, which corresponds to the reported molecular mass of the a subunit of the GlyR. In a different set of experiments, extracts of TPA-treated cells (obtained in this case with 1.5% Triton X-100 as detergent) were immunoprecipitated either with preimmune serum (Fig. Ill, P ) or AB-384 GlyR antibody (Fig. 1B, I ) . This result indicates that immunoprecipitation of the labeled 48 kDa band is specific, since it is not observed with the preimmune serum. As will be discussed later, additional, higher M, phosphorylated proteins seem to be immunoprecipitated with the GlyR antibody, specially when membrane proteins are extracted with "iton X-100. Phosphorylation of the 48-kDa polypeptide is promoted by TPA concentrations ranging from 0.1 to 10 p~ and is not observed in the presence of inactive TPA analogs (data not shown). Maximal phosphorylation in response to TPA is attained around 15 min of treatment and decreases thereafter (data not shown). A variable, small level of basal phosphorylation was also observed in different experiments (see Fig. 3). Fig. 2A shows that the immunoprecipitated 48-kDa phosphoprotein comigrates in SDS-PAGE gels with the a subunit of purified GlyR phosphorylated in vitro by PKC. Further, a phos- phorylated 48-kDa protein can be purified from TPA-treated cells by using NH2-strychnine affinity columns (data not shown). In addition to its specific recognition by GlyR antibodies, these results indicate that the 48 kDa band corresponds to the a subunit of the GlyR. Taken together, our results indicate that the a subunit of the GlyR is phosphorylated in intact spinal cord neurons upon activation of endogenous PKC.
The low abundance of GlyR makes direct sequencing and identification of phosphorylation sites not possible. Therefore, we have used sequence-specific proteolytic agents to cleave the 48-kDa protein and generate phosphopeptide maps. Such phosphopeptide maps were then compared with those previously obtained in our laboratory with the a subunit of the GlyR phosphorylated in vitro by PKC and from the theoretical products of specific cleavage in the rat a1 sequence (Ruiz-G6mez et al., 1991). When the 48-kDa phosphopeptide was subjected to digestion with N-chlorosuccinimide (cleaving a t tryptophan residues), a lower molecular mass phosphopeptide of 3 14 kDa was produced together with other labeled bands generated as a result of partial cleavage (Fig. 2 B ) . The size of the lower molecular weight phosphopeptide and the overall pattern of cleavage are similar to those obtained with the in vitro phosphorylated a subunit (Ruiz-G6mezet al., 1991). Complete digestion of the immunoprecipitated GlyR subunit with endopr0teina.w lysine C generated a single labeled peptide of -2.9 kDa (Fig. 2C ). Such fragment has a size close to the theoretical cleavage product 390-411 of the al sequence and is again analogous to that found using the same proteolytic agent with the in vitro phosphorylated GlyR a subunit (Ruiz-G6mez et al., 1991). Taken together, our results suggest that the same residue (serine 391) is preferentially phosphorylated by PKC both in vitro and in vivo.
Cyclic AMP-dependent Phosphorylation of the Glycine Receptor-In parallel experiments we investigated whether activation of endogenous PKA could also lead to GlyR phosphorylation. Fig. 3A shows the phosphorylation level of the immunoprecipitated GlyR a subunit when isolated adult rat spinal cells are incubated for 10 min with vehicle alone (lane 1 ), the adenylyl cyclase activator forskolin, which is known to raise intracellular CAMP levels leading to PKA activation (lane 2 ) and TPA (lane 3). In addition to the above described effect of PKC activators, a n increase over basal phosphorylation is also noted in response to PKA activators. This effect was unexpected, since a subunits of the GlyR do not show consensus phosphorylation sites for PKA (Grenningloh et al., 1987). However, a recent report by Betz and colleagues had shown that an alternative splicing form of the a1 subunit, alins, contained an 8-amino acid insert in the intracellular domain including a serine residue that might serve as a potential phosphorylation site for cyclic nucleotide-dependent protein kinases (Malosio et al., 1991b). Northern analysis has indicated that the alins form represents -30% of the total a1 subunit mRNA, which is expressed predominantly in the adult rat spinal cord (Malosio et al., 1991a(Malosio et al., , 1991b. In this regard, most of our in vivo phosphorylation experiments have been performed in spinal cord neuronal cells isolated from fresh adult rat tissue instead of in primary neuronal cultures from embryonic rat spinal cord. The rationale for this choice (besides the higher amounts of GIyRbinding sites available) was the fact that different a subunit isoforms are expressed in the adult animal ( a l ) and the primary cultures, which preferentially express the neonatal a2 isoform (Malosio et al., 1991a;Hoch et al., 19891, and the fact that our previous in vitro PKC phosphorylation studies were performed with adult tissue. In this regard, it is worth noting that when experiment analogous to that shown in Fig. 3A were done in primary neuronal spinal cord cultures (Fig. 3B) the forskolin effect is not apparent (lane 2 versus control in lane 1 ), whereas a stimulation of GlyR a subunit phosphorylation similar to that found in adult tissue is noted in cultured cells treated with phorbol esters (lune 3). Fig 3B also illustrates the previously mentioned finding that some GlyR-associated phosphorylated proteins are also immunoprecipitated with the antiserum. The faint band below the 48-kDa phosphopeptide probably represents a proteolytic product of the a subunit, whereas the heavily phosphorylated protein of -95 kDa may represent gephyrin, a tubulin-binding protein which is peripherally associated to cytoplasmic domains of the GlyR (Prior et al., 1992). Gephyrin has been shown to be phosphorylated by endogenous protein kinases (Langosh et al., 1992) and copurifies with the GlyR specially when solubilized with Triton X-100 (Pfeiffer et al., 1982;Garcia-Calvo et al., 1989;Prior et al., 1992) as is the case in the immunoprecipitations were this band is more clearly observed.
To evaluate whether the observed effect of forskolin on the in vivo phosphorylation of the a subunit of the GlyR was a direct or an indirect effect, we tried to demonstrate that purified PKA can directly phosphorylate the GlyR. Since previous attempts with the purified GlyR had proven unsuccessful (Ruiz-G6mez et al., 1991;Langosh et al., 19921, we used an alternative experimental design known as back-phosphorylation. In such experiments, adult rat spinal cord membranes were solubilized and the extracts immunoprecipitated with preinmune serum (Fig.   3C, lane 1 ) or specific anti-GlyR antibodies (Fig. 3C, lane 2 ), washed, and incubated in the presence of [y-32P]ATF' and purified PKA catalytic subunit. Fig. 3C shows that, under such experimental conditions, the a subunit of the GlyR is clearly phosphorylated by PKAin a dose-dependent way (not shown) to a estimated stoichiometry of -0.2 pmol of phosphate/pmol of [3Hlstrychnine-binding sites, thus suggesting that only a subpopulation of GlyR a subunits is being phosphorylated. It is worth noting that the extent of PKA back-phosphorylation attained in glycine receptors immunoprecipitated from forskolintreated spinal cord cells is diminished with respect to nontreated cells (compare lane 2 versus 1 in Fig. 301, thus suggesting that a fraction of glycine receptors are phosphorylated by PKA in vivo, so subsequent in vitro phosphorylation by this kinase is reduced.

Functional Consequences of GlyR Phosphorylation in Re-
sponse to PKA and PKC Activators-To examine the possible role of the observed GlyR phosphorylation in receptor function, we have used the Xenopus oocytes heterologous expression system. f i r injection with rat brain poly(A)+ mRNA, oocytes responded to the agonists GABA, kainate, and glycine under voltage-clamp conditions (Fig. 4A, control recordings, dotted  lines). Glycine-induced currents were specifically antagonized by strychnine (data not shown). Fig. 4, A and B, show that a brief (1-2 min) perfusion of the oocytes with TPA (0.1-1 J~M) promotes a time-dependent decrease in the glycine and GABAelicited currents, whereas the response to kainate is not significantly affected. Similar incubations with the inactive TPA analog 4-a-phorbol 12,13-didecanoate (2.5 p~) did not induce any change in response to the different agonists ( n = 2, data not shown). A similar effect of phorbol esters, which would lead to the activation of the oocyte PKC, either endogenous or directed by the injected poly(A)+ mRNA, on GABAA channels has been previously described in Xenopus oocytes injected with chicken brain mRNA (Sigel and Baur, 1988). The lack of effect of PKC activators on kainate currents has also been reported in the Xenopus oocyte system (Lotan et al., 1990). Thus, our results indicate that GlyR function is negatively modulated by PKC activators. On the contrary, a brief incubation of the oocytes with dibutyryl CAMP, which leads to endogenous PKA activation, results in a moderate but significant increase in glycineinduced currents without affecting the response to kainate (Fig. 5, A and B ). These results tie in nicely with recent reports showing that similar concentrations of dibutyryl CAMP promote an enhancement of glycine-activated currents in isolated spinal trigeminal neurons (Song and Huang, 1990). It is worth noting that the phosphatase inhibitor okadaic acid (1 p~) applied in the bath for 2 min led to a selective increase (27% after 60 min) in glycine-induced currents, with no change in kainate or GABA-induced responses (not shown). This result would suggest that a certain level of PKA-mediated GlyR phosphorylation is taken place in the oocyte under basal conditions. Interestingly, the subsequent incubation of a dibutyryl CAMPtreated oocyte (Fig. 5B) with TPA produced a decrease on glycine-induced responses (Fig. 5C) which was similar to that found in untreated oocytes (Fig. 41, thus suggesting independent sites for PKA and PKC phosphorylation and modulation of GlyR. Overall, our results strongly suggest that a dual, opposite modulation of GlyR function can be observed as a consequence of PKAor PKC-mediated protein phosphorylation.

DISCUSSION
The phosphorylation of ligand-gated receptor channels by second messenger-regulated protein kinases such as PKA and PKC seems to emerge as a major mechanism for the modulation of cross-communication between transduction pathways (Huganir and Greengard, 1990;Swope et al., 1992;Raymond et al., 1993a). A better understanding of such mechanisms require the identification of phosphorylation targets in these receptors both in vitro and in vivo and the characterization of the functional consequences of receptor phosphorylation. In this study, we show that the a subunit of the glycine receptor (GlyR) can be phosphorylated in intact spinal cord neurons in response to the presence of either PKC or PKA activators. Moreover, using the Xenopus oocyte heterologous expression system, we demonstrate that activators of these two different kinases promote opposite functional effects on the GlyR. In situ phosphorylation of the a subunit of the GlyR by PKC is consistent with our previous report showing that this kinase phosphorylates the a subunit of the receptor in vitro, in a cytoplasmic serine residue (Ser-391) close to the putative fourth transmembrane domain (Ruiz-Gmez et al., 1991). The time course of phosphorylation and in vivo experimental conditions are similar to those reported for PKC-mediated phosphorylation of the nAChR and other membrane proteins (Ross et al., 1988;Piller et al., 1989;Clark et al., 19911, thus supporting a direct effect. Phosphopeptide maps generated from the immunoprecipitated 48-kDa phosphoprotein are similar to those obtained after in vitro phosphorylation by PKC of the GlyR a subunit (Ruiz-Gmez et al., 1991). This result suggests that the same serine residue of the a subunit phosphorylated in vitro  is also phosphorylated in vivo upon activation of endogenous PKC, although this has to be confirmed by using recombinant a subunits and site-directed mutagenesis. Phorbol esters clearly increase GlyR phosphorylation in both spinal cord cells isolated form adult rat spinal cord, where a1 is the predominantly expressed subunit isoform (Malosio et al., 1991a) and in embryonic primary neuronal cultures, which express the neonatal form of the a subunits, a2 (Hoch et al., 1989). It is worth noting that these isoforms show high sequence homology in the proposed domain of phosphorylation.
Our data also indicate that the a subunit of the GlyR is phosphorylated in spinal cord neurons isolated form adult rat tissue in response to the presence of 10 p~ forskolin. This adenylyl cyclase activator has been used at the same concentration and in similar experimental conditions to assess PKAmediated phosphorylation of nAChR and GABAA receptors in both primary cultures and heterologus expression systems (Miles et al., 1987; Moss et al., 1992b; Swope et al., 1992).
Although modulation of GlyR function by PKA activators has been recently described in spinal trigeminal neurons (Song and Huang, 1990), the target of forskolin stimulation was somewhat surprising, since a subunits do not display a clear consensus sequence for PKA phosphorylation (Grenningloh et al., 1987). However, as mentioned under "Results," Betz and colleagues have reported the occurrence of an alternative splicing form of the a1 subunit which bears a potential PKA phosphorylation site (Malosio et al., 1991b). Such variant may represent -30% of the total a subunits in the adult spinal cord (Malosio et al., 1991a). In line with this observation, we show in this paper that the catalytic subunit of PKA can phosphorylate in vitro a subunits of the GlyR immunoprecipitated from adult rat spinal cord extracts (Fig. 3C). On such basis, it is tempting to speculate that PKA would phosphorylate a specific subset of a1 subunits both in vitro and in situ. This situation would be reminiscent to that observed in the case of the homologous GABAA receptor, where two alternative splicing forms of the 72 subunit have been shown to differ by the presence of an 8-amino acid insert (as the alins of the GlyR), bearing an additional site of phosphorylation by PKC (Moss et al., 1992a). However, the determination of the actual site of phosphorylation of the GlyR a subunit by PKA in vitro and in vivo remains a subject for future investigation. In this regard, it also would be of interest to confirm that alins (or an hypothetical aZins homolog) is not expressed in cultured spinal cord neurons. This would explain the lack of effect of forskolin in such an experimental system, although other explanations cannot be ruled out.
The fact that GlyR a subunits are phosphorylated both in vitro and in vivo in response to PKA or PKC activators disclose interesting possibilities of functional modulation of physiological interest, given the variety of neurotransmitters, neuropeptides, and growth factors known to regulate the intracellular activity of these protein kinases. As a first step for investigating functional consequences, we have tested the effect of PKA and PKC activators on the functionality of GlyRs expressed in Xenopus oocytes. This experimental system has been previously employed for assessing the modulation of receptors and channels by protein kinases (Sigel and Baur, 1988;Lerma et al., 1989;Lotan et al., 1990;Swope et al., 1990). Our results indicate that agents that activate PKC decrease glycine-induced currents, while PKA modulators increase the response to glycine. Although such data do not provide evidence for the actual phosphorylation of GlyR channels in the oocyte, the results obtained in intact spinal cord cells are consistent with a direct effect of phosphorylation in receptor function. Since GlyR a subunits have been shown to be critical for both ligand binding and channel function (Betz, 19901, it is reasonable to assume that the observed functional effects are related to a subunit phosphorylation. However, a role for phosphorylation of / 3 subunits or other GlyR-associated proteins in regulating GlyR function cannot completely be ruled out. The observed enhancement of glycine responses upon treatment with PKA activators is in agreement with a recent report showing positive modulation of GlyR by PKA in isolated spinal trigeminal neurons, by a mechanism involving increased probability of channel opening (Song and Huang, 1990). We are not aware of reports describing functional effects of PKC activators on GlyR function. Phosphorylation of muscle and neuronal nAChR by PKC has been shown to increase the rate of desensitization to agonists of these receptors. Activation of PKA in muscle cells also leads to a n increase in the rate of the rapid phase of desensitization. On the contrary, CAMP analogs seem to enhance the intensity of neuronal nAChR-mediated responses (Wagner et al., 1991;Swope et al., 1992). Very recent evidence indicates that PKA increases the amplitude of responses to glutamate of GluR6 homomeric receptors (Raymond et  CAMP-dependent phosphorylation decreases the amplitude of GABA response and slows the rate of desensitization of some receptor subtypes (Moss et al., 1992b). PKC phosphorylation has been suggested to decrease GABAA receptor function (Sigel and Bour, 1988). However, results on the physiological effects of phosphorylation of GABAA receptors have been complex and sometimes contradictory (discussed in Raymond et al., 1993a;Moss et al., 1992b). This fact may be due, a t least in part, to the potentially different regulation by phosphorylation of the different subunits isoforms (Moss et al., 1992a(Moss et al., , 1992b. In the case of the GlyR, if our suggestion regarding alins as the specific target for PKA phosphorylation is confirmed, only receptor subtypes containing this a subunit variant would have the ability to be modulated by CAMP-dependent protein phosphorylation. Thus, changes in the developmental andor cellular expression of particular subunit isoforms would lead to the assembly of receptor subtypes with different subunit composition and with variable patterns of phosphorylation and modulation, further contributing to generate receptor diversity and functional heterogeneity within the nervous tissue. In conclusion, our results indicate that the GlyR can be phosphorylated in intact spinal cord neurons in response to the presence of either PKC or PKA activators and that the stimulation of these transduction pathways leads to a decrease or an enhancement of GlyR functionality, respectively (see Fig. 6). Such dual mechanism of regulation would provide a very sensitive instrument for the integrated modulation of the GlyR by other extracellular messengers acting in the same neuron. The previous or simultaneous activation of a variety of G proteincoupled receptors modulating CAMP or diacylglycerol levels would lead to changes in the subsequent neuronal responses to glycine and in the transmission at medulla and spinal levels (Song and Huang, 1990). Such cross-talking mechanisms would help to integrate different extracellular signals at the neuronal membrane level, and account for transient changes in the eficacy of synaptic transmission, thus contributing to synaptic plasticity.