Cyclic AMP and Synaptic Activity-Dependent Phosphorylation of AMPA-Preferring Glutamate Receptors

The Journal of Neuroscience, December 1994, 1412): 7585-7593

Craig Blackstone,lva Timothy H. Murphy,lrb Stephen J. Moss,~~~~~ Jay M. Baraban,lm* and Richard L. Huganir1v3 ‘Departments of Neuroscience, ‘Psychiatry and Behavioral Sciences, and the 3Howard Hughes Medical Institute, The Johns Hopkins University School of Medicine, Baltimore, Maryland 21205

Several studies have suggested that the function of gluta- protein kinase C. These results indicate that AMPA receptors mate receptor channels can be regulated by protein phos- containing the GIuRl subunit may be regulated by extra- phorylation. Furthermore, a basal level of phosphorylation cellular signals working through the CAMP second messen- may be necessary to maintain receptor function. Little is ger system as well as by synaptic activity, possibly acting known, however, about the phosphorylation state of gluta- through protein kinase C. Such regulation by protein phos- mate receptor channels in neurons and how it is regulated phorylation may be involved in short-term changes in syn- by synaptic activity. In this study, we have investigated the aptic efficacy thought to involve the functional modulation phosphorylation of the AMPA-preferring glutamate receptor of AMPA receptors. subunit GluRl in cortical neurons in primary culture. These [Key words: glutamate, AMPA receptor, phosphorylation, neurons elaborate extensive processes, form functional syn- CAMP, synaptic activity, long-term potentiation] apses, and exhibit spontaneous 4-8 set bursts of synaptic activity every 15-20 sec. In cultures in which this synaptic Glutamate receptors mediate excitatory neurotransmissionin activity was suppressed by tetrodotoxin and MK-801, the the CNS, playing critical roles during synaptogenesis,in mech- GIuRl protein was phosphorylated on serine residues within anismsof synaptic plasticity suchas long-term potentiation (LTP) a single tryptic phosphopeptide, as determined by phos- and long-term depression(LTD), and in someneuropathological phoamino acid analysis and phosphopeptide mapping. This conditions (Ito et al., 1989; Siegelbaumand Kandel, 1991; Bliss same peptide was basally phosphorylated in recombinant and Collingridge, 1993; Shaw, 1993; Linden, 1994; Lipton and GluRl receptors transiently expressed in human embryonal Rosenberg,1994). Thesereceptors can be divided into two broad kidney 293 cells. Treatment of these synaptically inactive categories:ionotropic receptors, oligomeric proteins that form cortical neurons with the adenylyl cyclase activator forskolin ligand-gated ion channels; and metabotropic receptors, mono- resulted in a robust increase in phosphorylation on serine mers that are coupled through guanine nucleotide-binding pro- residues on a phosphopeptide distinct from the basally teins to second messengercascades. The ionotropic receptors phosphorylated peptide. Again, this same phosphopeptide have classically been divided into a-amino-3-hydroxy-s-meth- was observed in recombinant GluRl receptors isolated from yl-4-isoxazolepropionate (AMPA), kainate, and N-methyl-D-as- 293 cells coexpressing the catalytic subunit of CAMP-de- partate (NMDA) receptorsbased on their preferred agonistsand pendent protein kinase. Spontaneous synaptic activity in electrophysiological properties (Monaghan et al., 1989). Over cultures of cortical neurons resulted in a consistent, rapid the past several years, molecular cloning studieshave identified (within 1 O-30 set) increase in phosphorylation on serine and many subunits of the ionotropic receptors. These have been threonine residues. Interestingly, these phosphopeptides classifiedon the basisof sequencesimilarity and pharmacology were also phosphorylated when neurons from inactive cul- as membersof the AMPA (GluR1-4) kainate (GluR5-7, KAl, tures were stimulated with phorbol esters, which activate 2), and NMDA (NMDAR 1, 2A-D) receptor complexes (Gasic and Hollmann, 1992; Nakanishi, 1992; Sommer and Seeburg, 1992; Seeburg, 1993; Hollmann and Heinemann, 1994). Two Received Oct. 11, 1993; revised May 20, 1994; accepted June 2, 1994. additional subunits related in sequence(6 1 and 62) remain “or- C.B. and T.H.M. contributed equally to this study. We are grateful to Drs. Michael Hollmann and Stephen Heinemann for the GluR 1 RoPcDNA and to Dr. phan” receptors (Yamazaki et al., 1992; Lomeli et al., 1993). Michael D. Uhler for the cDNA of the catalytic subunit of PKA (Ca). This work Further diversity is generatedthrough alternative RNA splicing was supported by an MSTP fellowship (C.B.), an NRSA fellowship (T.H.M.), USPHS Grants DA-00266 and MH-00926 (J.M.B.), and the Howard Hughes and posttranscriptional mRNA editing of many of these sub- Medical Institute (R.L.H.). units (Sommer et al., 1990, 1991; Gallo et al., 1992; Sugihara Correspondence should be addressed to Dr. Richard L. Huganir, Department et al., 1992; Seeburg, 1993).The predicted amino acid sequences of Neuroscience, Howard Hughes Medical Institute, The Johns Hopkins Univer- sity School of Medicine, 725 North Wolfe Street, PCTB 900, Baltimore, MD derived from thesecloning studieshave permitted the genera- 21205-2185. tion of subunit-specific antibodies againstthese proteins, facil- 1 Present address: Harvard-Longwood Neurology Program and Harvard Med- itating the study of such biochemical properties asreceptor sub- ical School, Boston, MA 02 115. b Present address: Kinsmen Laboratory for Neurological Research, Division of unit composition and modification by protein phosphorylation Neurological Sciences, Department of Psychiatry, University of British Columbia, and glycosylation (Rogerset al., 1991; Blackstoneet al., 1992b; Vancouver, BC V6T 123, Canada. c Present address: IRC Institute for Molecular Cell Biology and Department of Wenthold et al., 1992, 1994; Moss et al., 1993; Raymond et al., Pharmacology, University College, London WClE 6BT, UK. 1993b; Roche et al., 1994; Sheng et al., 1994). Copyright 0 1994 Society for Neuroscience 0270-6474/94/147585-09$05.00/O One posttranslational mechanismimplicated in the long-term 7586 Blackstone et al. - Glutamate Receptor Phosphorylation functional modulation of glutamate receptors is phosphoryla- rnM NaCl, 5.0 mM KCl, 2.5 mM CaCl,, 1.0 mM MgSO,, 10 mM sodium tion by intracellular protein kinases (Blackstone et al., 1992~; HEPES, 1 mM NaHCO,, and 22 mM glucose (pH 7.4) supplemented with l-2 mCi/ml 32P- orthophosphate. The labeling medium was then Swope et al., 1992; Raymond et al., 1993b). Application of removed and replaced with phosphate-free HBSS containing picrotoxin activators of CAMP-dependent protein kinase (PKA) as well as ( 10 PM) for synaptically active cultures, or picrotoxin ( 10 PM), TTX ( 1 purified PKA has been shown to potentiate the function of PM), and MK-80 1 (3 PM) for synaptically inactive cultures. In the pres- AMPA/kainate receptors in cultured rat hippocampal neurons ence of the GABA, receptorchannel antagonist picrotoxin, virtually all (Greengard et al., 1991; Wang et al., 1991) and teleost retinal cortical neurons within a culture exhibit highly synchronous spontaneous synaptic activity; a combination of TTX and MK-80 1 is used to horizontal cells (Liman et al., 1989). CAMP analogs, possibly block synaptic activity in these cultures (Murphy et al., 1992a, 1994). working through PKA, increase the current and calcium influx In experiments utilizing activators of intracellular protein kinases, the through recombinant GluRl/GluR3 AMPA receptor channels “synaptically inactive” culture solution was supplemented with either expressed in Xenopus oocytes (Keller et al., 1992). Similarly, forskolin (10 PM) and IBMX (75 PM), or PDA (5 PM) whereindicated. Cultures were incubated with these solutions for 20 min at room tem- homomeric GluR6 kainate receptors transiently expressed in perature unless otherwise indicated. human embryonal kidney 293 cells are directly phosphorylated We sought to determine the time course of GluRl phosphorylation and potentiated by PKA (Raymond et al., 1993a; Wang et al., after the initiation of synaptic activity. Since these cultures show prom- 1993). Another serine/threonine kinase, calcium/calmodulin- inent spontaneous firing and associated protein kinase activity (Murphy dependent protein kinase II (CaMKII), also enhances kainate- et al., 1994) we needed a method to suppress the endogenous synaptic activity that was readily reversible. Previous observations indicated that gated currents in cultured rat hippocampal neurons (McGlade- lowering extracellular calcium to 0.5 mM and raising magnesium to 2 McCulloh et al., 1993). These reports strongly support a key mM suppressed the amplitude and duration of intracellular calcium role for direct protein phosphorylation of AMPA/kainate glu- transients, without affecting resting intracellular calcium levels (Murphy tamate receptors in altering the excitability of neurons in the et al., 1992a, 1994). Cortical cultures were labeled with l-2 mCi/ml 32P-orthophosphate for 4 hr in a medium containing phosphate-free CNS. HBSS with low calcium (0.5 mM vs 2.25 mM), high magnesium (2 mM Previous studies have shown that the recombinant GluRl,,, vs 1 mM), and picrotoxin (10 PM). Restoration of normal calcium and (Hollmann et al., 1989) AMPA receptor protein expressed in magnesium concentrations (2.25 mM and 1 mM, respectively) by adding 293 cells is a substrate for both tyrosine and serine/threonine an equal volume of medium containing 4 mM calcium but no magne- kinases (Moss et al., 1993). In vitro phosphorylation studies sium resulted in a rapid increase in the frequency and duration of synaptic bursts. In initial studies, we observed that treating cultures with using purified recombinant GluRl (from Sl9 cells) as well as the medium containing 0.5 mM CaCl, and 2 mM MgSO, reduced GluRl GluRl isolated from rat brain synaptosomes and postsynaptic phosphorylation to levels found in TTX-treated cultures maintained in densities demonstrated phosphorylation by CaMKII and pro- the regular medium (2.25 mM calcium, 1 mM magnesium). To determine tein kinase C (PKC) but not by PKA (McGlade-McCulloh et the time course of GluRl phosphorylation, we restored calcium and magnesium concentrations to normal levels for different amounts of al., 1993). In the present study, we sought to determine whether time prior to harvesting the cultures. GluRl can also be phosphorylated in cultured neurons by the All reactions were terminated by aspirating the medium and adding activation of intracellular protein kinases through both synaptic an ice-cold lysis buffer consisting of 50 mM sodium phosphate (pH 7.5), activity and the extracellular application of membrane-per- 1.0% Triton X-100,0.5% deoxycholate, 0.2% SDS, 50 mM NaF, 10 mM meable protein kinase activators. Using metabolic labeling, im- sodium pyrophosphate, 5 mM EDTA, 5 mM EGTA, 1 mM sodium orthovanadate, 1 PM okadaic acid, and 1 PM microcystin-LR directly munoprecipitation, and phosphoamino acid and phosphopep- to the plates. The cells were scraped off of the plates, transferred to a tide map analyses, we report the regulation of GluRl AMPA microfuge tube, mixed by inversion, and then centrifuged (15,000 rpm, receptor phosphorylation by activators of PKA and PKC and 15 min. 4°C) in a TOMY MTX- 150 refrigerated microcentrifuge. The compare these effects with those resulting from brief bursts of supematant ‘was transferred to another microfuge tube, and antilGluR1 antiserum (final dilution 1:200) bound to urotein A-Sepharose CL-4B synaptic activity in cultured cortical neurons. (Pharmacia) that had been preblocked w;th 20 mg/mi bovine serum albumin was added. Following gentle agitation for 2 hr, the beads were Materials and Methods washed sequentially with the following solutions: lysis buffer alone (twice); Materials. 32P-orthophosphate(8500-S 120 Ci/mmol) was obtained from lysis buffer with 750 mM NaCl (three times); 10 mM phosphate buffer DuPont-New England Nuclear Research Products. TPCK-treated tryp- (pH 7.5) with 0.1% Triton X- 100,50 mM NaF, and 5 mM EDTA (twice). sin (EC 3.4.21.4; type XIII, from bovine pancreas), bovine serum al- Following removal of the final wash, SDS-PAGE sample buffer was bumin, ninhydtin, picrotoxin, and tetrodotoxin (TTX) were from Sig- added directly to the beads to elute antibody-bound GluR 1. ma. Aprotinin (Trasylol) was from Mobay Chemical. (+)-MK-801 was Cell transfection and metabolic labeling. Human embryonal kidney purchased from Research Biochemicals International. Okadaic acid was 293 cells (ATCC CRL 1573) were maintained as described previously obtained from LC Services. Forskolin (Coleus forskohlii), microcystin- (Moss et al., 1993). Approximately 12-24 hr prior to transfection, cells LR, phorbol 12,13-diacetate (PDA), phorbol 12-myristate, 13-acetate were passaged and plated at a density of 1O6 per plate (10 cm). Cells (TPA), and 3-isobutyl- 1-methylxanthine (IBMX) were obtained from were then transfected by calcium phosphate coprecipitation, with each Calbiochem. Anti-GluR 1,-GluR2/3, and-GluR4 antibodies used in this plate receiving 20 pg of plasmid DNA. The cDNAs for GluR 1Rop (Holl- study have been described previously (Blackstone et al., 1992a,b, Martin mann et al., 1989) and the catalytic subunit of PKA (Ca) (Uhler et al., et al., 1993). 1986) were each subcloned into eukaryotic expression vectors under Primary cortical neuron cultures and metabolic labeling. Cultures were control of the cytomegalovirus promoter. For expression of GluRl,,, prepared from cerebral cortex dissected from embryonic day 17 alone, 20 pg of the plasmid was used, while for cotransfection experi- Sprague-Dawley rats using a papain dissociation method as described ments 17 pg of GluR 1R,,P was cotransfected with 3 pg of PKA col DNA. previously (Murphy and Baraban, 1990; Murphy et al., 1992a). Cells Forty-eight hours after transfection, the cells were labeled in phosphate- were resuspended at 1.2 x lo6 cells/ml in minimal essential medium free media (GIBCO) for 4 hr with 2-4 mCi/ml 32P-orthophosphate. (MEM) supplemented with 5.5 gm/liter glucose, 2 mM glutamine, 10% Cells were then harvested and the GluRl protein was isolated by im- fetal calf serum, 5% heat-inactivated horse serum, 50 U/ml penicillin, munoprecipitation using anti-GluR 1 antibodies as described previously and 0.05 mg/ml streptomycin. They were then plated onto poly-L-ly- (Blackstone et al., 1992b Moss et al., 1993). sine-coated 12-well dishes (1 ml medium) and placed in a CO,-buffered Gel electrophoiesis and immunoblotting. Immunoprecipitates were 37°C incubator. Cells were fed as described previously (Murphy et al., heated at 100°C for 2 min. and then resolved bv SDS-PAGE in 8.0% 1992a). Cultures were allowed to mature for 3 weeks before use. acrylamide gels. The gels were fixed in 25% methanol/lO% acetic acid After 3 weeks in vitro, cultures were labeled for 4 hr at 37°C in a and dried, 3zP-labeled GluR 1 was revealed by autoradiography at - 70°C phosphate-free Hank’s balanced salt solution (HBSS) consisting of 137 using Kodak XAR-5 film with a DuPont Cronex intensifying screen. The Journal of Neuroscience, December 1994, 74(12) 7567

For immunoblot analysis of AMPA receptor subunits expressed by cortical neurons in culture, total homogenates (100 pg protein/lane) were Rl R2/3 R4 subjected to SDS-PAGE (8.0% gels). Proteins were then transferred to Immobilon P membrane (PVDF; Millipore) by electroblotting (30 V, overnight), blocked for 1 hr with 0.5% nonfat dry milk (Carnation)/ 205 - 0.1% Tween 20 in Tris-buffered saline (TBS: 50 mM Tris-HCl. DH 7.4. 150 mM NaCl), and probed with affinity-purified anti-GluRl ,%luR2; 3, or-GluR4 antibodies (0.5 &ml) for 1 hr at room temperature. Fol- lowing several washes with blocking buffer, horseradish peroxidase- conjugated donkey anti-rabbit secondary antiserum (1:5000; Amer- sham) was applied for 45 min. The blots were washed several times 116- with TBS, and specific immunodetection was revealed by enhanced chemiluminescence (ECL, Amersham). Phosphopeptide mapping andphosphoamino acid analysis. Slices containing 3ZP-labeled GluRl were excised from the gel, and the amount 97 - of radioactivity incorporated was determined by Cerenkov counting of the gel slice using a Beckman model LS 580 1 liquid scintillation counter. Next, the GluRl phosphoprotein was digested with trypsin (300 &ml, in 50 mM ammonium bicarbonate) overnight at 37°C. The samples were 66- dried in a Savant SpeedVac Concentrator, and then resuspended in 1 ml H,O and dried again.The resultingtryptic fragmentswere finally resuspended in 10 pl of H,O and spotted onto Kodak Chromagram thin-layer cellulose sheets along with the markers phenol red and basic fuchsin. Separation of the phosphopeptides in the first dimension was by electrophoresis (500 V) in acetic acid : pyridine : H,O (19: 1:89; pH 3.5) until the markers had migrated 6 cm. The cellulose plates were air dried, and ascendingchromatography was performedin the second 45 - dimension in pyridine : butanol : acetic acid : H,O (15: 10:3: 12) for 16 cm. Plates were air dried, and 32P-labeled phosphopeptides were revealed by autoradiography. For phosphoamino acid analysis, tryptic peptides prepared as above were resuspended in 6 M HCl. heated at 100°C for 1 hr. dried down in a SpeedVac,and finally resuspendedin 10~1 of H,O. Tne sampleswere spotted onto thin-layer cellulose plates along with the marker phenol red and IO pg eachof the phosphoaminoacid standards phosphoserine, Figure 1. AMPA receptors in cultured cerebral cortical neurons. Total phosphothreonine,and phosphotyrosine. The phosphoaminoacids were homogenates (100 pg protein/lane) from cultured neurons (3 weeks in separatedby electrophcresis(506 V), first in-for&c acid: aceticacid : culture) prepared from cerebral cortices dissected from embryonic day H,O (1:10~89: DH 1.9)for 5 cm. andthen in aceticacid : nvridine : H,O 17 Sprague-Dawleyrats were subjectedto SDS-PAGEand immuno- (1%1:‘89, pH j:5) for another8 cm. The celluloseplate was’ stained &h blotted with antibodies against GluRl , GluR2/3, or GluR4 as indicated. ninhydrin( 1.O%, in acetone)to revealthe phosphoamino acid standards. The sizes of the molecular mass standards (in kDa) are indicated to the 32P-labeled phosphoamino acids were revealed by autoradiography. left. Protein content determination. Protein concentrations were deter- minedby the bicinchoninicassay (Pierce) with bovine serumalbumin as the standard. abolic labeling with 32P-orthophosphateand immunoprecipitation, we assessedthe basal, activity-independent phosphory- Results lation of GluRl and phosphorylation promoted by activators The expressionof AMPA receptors in cortical neurons grown of PKA and PKC. Furthermore, we have compared the effects in culture for 3 weeks was evaluated using affinity-purified an- of theseactivators with those of endogenoussynaptic stimula- tibodies that specifically recognize the GluRl or GluR4 sub- tion. units, and an antibody (denoted GluR2/3) that detects both Since previous studies have suggestedthat a basal level of GluR2 and GluR3 (Fig. 1). Neurons in primary culture showed protein phosphorylation may be important in the maintenance a similar expressionof AMPA receptor subunitsto that of many of receptor function (Wang et al., 199 l), we first examined phos- cortical neurons in vivo (Petralia and Wenthold, 1992; Martin phorylation of the 106 kDa GluR 1 protein after the spontaneous et al., 1993). By immunoblot analysisthese antibodies detected synaptic activity of the cultured cortical neuronswas suppressed proteins ranging from 102 to 108 kDa, sizes identical to those with TTX and MK-801 (i.e., basal phosphorylation). Immu- reported for recombinant and native receptor subunits in pre- noprecipitation of GluR 1 from 3*P-orthophosphate-labeledcul- vious studies (Blackstone et al., 1992a,b; Martin et al., 1993). tures followed by SDS-PAGE analysis indicated that this re- Expression of GluRl-3 in neuronswas relatively high in these ceptor subunit was basally phosphorylated in the absenceof cultures; however, little or no GluR4 immunoreactivity was synaptic activity (Fig. 2). Phosphopeptide map analysis indi- detected. For each of thesepreparations, immunodetection was cated that this phosphorylation resided within a single tryptic completely abolishedby preadsorption of the antibodies to the phosphopeptide. To confirm that this phosphopeptide was in- synthetic peptide (50 &ml) to which they were raised (data not deed derived from GluRl and to compare basalphosphoryla- shown).Immunocytochemical studiesrevealed that virtually all tion sitesbetween native and recombinant receptors,the GluRl neuronsin the culture were immunostained by both the GluRl protein was transiently expressedin 293 cells. A similar analysis and GluR2/3 antibodies; some astrocytes showed GluR4 im- demonstrated that the basal phosphopeptide was also present munostaining (data not shown). Becauseof the individual sub- in theserecombinant receptors; in addition, several minor basal unit specificity of the GluRl antibodies and the relative abun- phosphopeptideswere noted in the recombinant receptor pro- dance of this subunit in these neurons, we chose to evaluate tein (Fig. 2). phosphorylation of the GluRl receptor subunit. By using met- The modulation of the phosphorylation state of GluRl was 7588 Blackstone et al. l Glutamate Receptor Phosphorylation

Figure 2. Activity-independent basal phosphorylation of GluR 1. L&i, GluR 1 was isolated by immunoprecipitation from j2 P- orthophosphate-labeled transfected 293 cells or cultured cortical neurons in which the spontaneous synaptic activity of the cultures had been Neurons 293 Cells suppressed with TTX (1 PM) and MK- 205 - 801 (3 PM). The immunoprecipitates were resolved by SDS-PAGE, and 3zP- labeled GluRl was visualized using au- A toradiography. The sizes of molecular 116- 2 mass standards (in kDa) are indicated. P Right, Gel slices containing GluR 1 were B excised and digested with trypsin, and z the resulting phosphopeptides were E separated by thin-layer chromatogra- 66- I? phy on cellulose layers in two dimen- 4 sions as shown. The origins are identified with circles. The major basal phosphopeptide common to both re- 45- combinant and neuronal GluRl is in- 4 w dicated (arrowheads). - Electrophoresis -I- then studied in cultured neurons prelabelled with 32P-ortho- indicating that this additional phosphorylation of recombinant phosphate. Stimulation of synaptically inactive cultures with GluRl may be due to basal PKC activity in the 293 cells. In the adenylyl cyclase activator forskolin and the phosphodies- some cases, such as phosphopeptide 1 in Figure 4C-E, it is terase inhibitor IBMX induced a 60% f 6% (n = 3) increase possiblethat several phosphopeptideswith similar migrations above basal levels of 32Piincorporated into GluRl, as deter- may appear as one phosphopeptide. However, when the elec- mined by Cerenkov counting of the excised GluR 1-containing trophoresis was conducted until the markershad migrated 8 cm gel slice. Stimulation of the inactive cultures with the phorbol in the first dimension (insteadof 6 cm) no additional distinctions ester PDA resulted in an increasein phosphorylation of 44% f were evident (data not shown). 19% (n = 3). Finally, spontaneoussynaptic activity resulted in Given the large increasein phosphorylation at a site distinct a 38% f 9% (n = 7) increasein phosphorylation compared to from the basalsite in responseto forskolin treatment in neurons, synaptically inactive neurons. yet the relatively small increasenoted in GluRl -transfected 293 As thesereceptors are basally phosphorylated, we sought to cells in previous studies(Moss et al., 1993), we examined the determine whether the observed modulation of thesereceptors ability of heterologously expressedPKA to direct phosphory- involved the sameor different residues.The amino acid residues lation ofrecombinant GluRl (Fig. 5). Cellscoexpressing GluRl phosphorylated were evaluated by phosphoamino acid analysis and PKA Ca showed a robust (> IO-fold) increase in GluRl (Fig. 3). The basal phosphorylation was exclusively on serine phosphorylation above basal levels. In Figure 5, the 99 kDa residues,consistent with previous findings for the recombinant phosphoprotein representsan unglycosylated precursor form of GluRl receptor (Moss et al., 1993). The increasein phosphor- GluRl (Blackstone et al., 1992b), while the more intensely la- ylation in responseto forskolin and IBMX was overwhelmingly beled 106 kDa phosphoprotein representsmature glycosylated on serine residues, although slight increasesin threonine and GluRl. Both signalswere completely abolishedby preadsorp- tyrosine phosphorylation were noted as well. Spontaneoussyn- tion of the antibodies with an excess(50 &ml) of synthetic aptic activity or treatment of inactive cultures with the phorbol peptide prior to immunoprecipitation. Two-dimensional phos- ester PDA showed increasesin serine and threonine phosphor- phopeptide map analysis revealed that this increased GluRl ylation (Fig. 3). Individual sites of phosphorylation were re- phosphorylation in responseto PKA Ca cotransfection wascon- vealed by two-dimensional phosphopeptidemapping of tryptic fined to the samephosphopeptide as in the forskolin-stimulated GluRl peptides (Fig. 4A-D). Increasesin responseto forskolin neurons.Tryptic digestion generatedeither tight clustersof sev- treatment were confined to a tight cluster of slowly migrating eral distinct phosphopeptides(as in native receptor PKA-de- peptides (compare Fig. 4AJ). Smaller but consistent increases pendent phosphorylation shown in Fig. 5) or a single major were observed with phorbol ester treatment, which resulted in phosphopeptide (as in recombinant receptor PKA-dependent several less intense, scattered phosphopeptides(compare Fig. phosphorylation shown in Fig. 5). However, both patterns were 4A,C). Interestingly, spontaneoussynaptic activity of the neu- seenwith recombinant as well as native receptors over several rons (Fig. 40) resultedin a phosphopeptidemap pattern almost different experiments, suggestingthat the additional phospho- identical to that seenwith phorbol ester stimulation. These re- peptides reflect incomplete digestion or additional phosphory- sults are summarized in Figure 4F and in Table 1. Analysis of lation events within the sametryptic fragment. This PKA phos- recombinant GluRl from 293 cells activated with forskolin (50 phopeptide cluster was much more highly phosphorylated with PM), IBMX (75 PM), and the phorbol esterTPA (100 nM) resulted respectto the basalphosphopeptide in recombinant GluRl from in a phosphorylation pattern (Fig. 4E) similar to that seenin 293 cells relative to that observed in GluRl isolated from cul- neurons,accounting for all phosphopeptidesobserved in GluRl tured cortical neurons (Fig. 5), most likely due to the lower isolated from the neurons. The minor basal phosphopeptides stoichiometry of basalphosphorylation in the recombinant pro- from unstimulated 293 cells (Fig. 2) resemblethose from neu- tein. rons that appearupon stimulation with phorbol ester (Fig. 4C), We examined the time course of GluRl phosphorylation in The Journal of Neuroscience, December 1994, 14(12) 7589

Table 1. Summary of the effects of various treatments on GluRl phosphorylation in cultured cortical neurons

Treatment Phospho- Phorbol Synaptic 205- peptide Basal Forskolin ester activity 1 ++++ ++++ ++++ ++++ 2 - + ++ ++ 3 - + + ++ 116- 4 - - ++ ++

\;1.. 5 -/+ ++++ ++ ++ 97) 6 - - ++ +++ Phosphopeptide numbering is derived from Figure 4~. Signal intensity: -, un- 66- detectable; +, low; + +, moderate; + + + , high; + + + +, intense.

response to strong synaptic burst stimulation. Cortical neurons were prelabeled with 32P-orthophosphate,and the amplitude 45- and duration of synaptic bursts were suppressedby lowering the extracellular CaCI, to 0.5 mM and increasing MgSO, to 2 mM as described in Materials and Methods. Extracellular calcium was then restored to 2.25 mM for times ranging from 10 set to 20 min (at room temperature) before terminating the reaction with ice-cold lysis buffer. The GluRl protein was then immunoprecipitated and resolved by SDS-PAGE. Gel slicescontain- ing GluR 1 were excised, and the radioactivity incorporated was determined by Cerenkov counting. Synaptic bursting resulted in a rapid increase in phosphorylation of GluRl, with >50% of the increasein phosphorylation occurring within the first 30 sec. By 10 min, the increasein phosphorylation had plateaued. Phosphopeptide map analysis of tryptic digestsof GluRl con- Ser- sistently revealed that these activity-dependent increasesin 0 phosphorylation were on the same tryptic peptides as shown previously in Figure 40 (data not shown). Restoration of calcium levels for up to 10 min in the presenceof TTX did not Thr- lead to increasedGluRl phosphorylation (data not shown), sug- gesting that activation is associatedwith augmented synaptic Tyr - activity and not the solution changesthemselves.

Discussion In this study, we have evaluated the phosphorylation state of Figure 3. Second messenger and synaptic activity-dependent phos- the GluR 1 AMPA receptor subunit in cultured cortical neurons phorylation of GluRl in cortical neurons. Top, Neuronal cultures in which the spontaneous synaptic activity was suppressed with TTX and in responseto various stimuli. These included the spontaneous MK-801 were prelabeled with )*P- orthophosphate and then treated synaptic activity of the culturesas well as membrane-permeable with either forskolin and IBMX or phorbol ester (PDA) for 20 min. activators of intracellular secondmessenger systems and protein Where indicated, the endogenous synaptic activity of the cultures was kinases.In neuronsin which the spontaneoussynaptic activity not blocked, and the incubation was continued for 20 min. GluR 1 was isolated by immunoprecipitation, resolved by SDS-PAGE, and visu- was suppressed(inactive neurons), the GluRl protein (most alized by autoradiography. Where indicated, immunoprecipitation from likely including both flip and flop alternative splice forms) was synaptically active cultures was conducted using antibodies which had phosphorylated on serine residueswithin a single tryptic phos- been preadsorbed with an excess of synthetic peptide. Sizes of molecular phopeptide. Previous electrophysiological studies in cultured mass standards are shown in kilodaltons. The GluRl phosphoprotein hippocampal neuronsreported a “rundown” of kainate-evoked is identified with an arrowhead. Bottom, The GluRl -containing gel slices were excised, and GluRl was digested with trypsin and hydrolyzed with currents when electrode solutions without ATP were usedthat 6 M HCl. The resulting phosphoamino acids were then separated by was likely due, at least in part, to the decreasedability of the electrophoresis as described in Materials and Methods. The migrations neuronsto maintain intracellular phosphorylation (Wang et al., of ninhydrin-stained phosphoserine (Ser), phosphothreonine (Thr), and 199 1). The strong basal phosphorylation we observe would be phosphotyrosine (Tyr) phosphoamino acid standards are indicated (cir- consistent with theseobservations. Furthermore, the ability to cles). Forskolin-stimulated Thr and Tyr as well as phorbol ester-stimulated Thr phosphorylation were revealed more clearly on longer ex- observephosphorylation of this samesite in recombinant GluR 1 posures. receptors should make it possibleto identify the site, remove it by site-specific mutagenesis,and note any direct functional changesthat result. Identification of the residue may also pro- vide clues as to the identity of the protein kinase responsible for this basalphosphorylation. 7590 Blackstone et al. l Glutamate Receptor Phosphotylation

4 + - Electrophoresis f

Figure 4. Phosphopeptide map analysis of GluRl . A-D, Cortical neurons were prelabeled with 32P-orthophosphate and treated as described. The GluRl protein was immunoprecipitated and resolved by SDS-PAGE. Gel pieces were excised and digested with trypsin. The resulting tryptic GluRl fragments were separated in two dimensions by thin-layer chromatography, and ‘*P-labeled GluRl phosphopeptides were revealed by autoradiography. The origins are indicated (circles). A, Synaptically inactive cultures (basal). /3, Synaptically inactive cultures + forskolin, IBMX. C’, Synaptically inactive cultures + phorbol ester (PDA). D, Synaptically active cultures. E, For confirmation that all of these phosphopeptides in A-D were derived from the GluRl receptor,GluRl-transfected 293 cellswere prelabeled with 32P-orthophosphateasabove and then treatedwith phorbolester (TPA), forskolin,and IBMX. Immunoprecipitation,SDS-PAGE, tryptic digestion,and phosphopeptidemapping were performed as in A-D. F, Schematicdiagram showing the major phosphopeptidespresent in A-E.

In addition to the strong phosphorylation at this location, a is at a site not conforming to the ideal consensussequence and robust increasein phosphorylation on another phosphopeptide that native conformation, subunit composition, or accessory was observed in responseto forskolin stimulation, also on serine proteins are important for phosphorylation, or alternatively that residues. The same phosphopeptide was detected in receptor the phosphorylation is not direct but occurs through a kinase isolated from 293 cells cotransfected with the catalytic subunit cascade.In the latter case,the samecascade would also have to of PKA, strongly supporting a role for PKA-mediated phos- be operative in kidney epithelial cells (i.e., 293 cells).The ability phorylation of GluRl in neurons. In fact, in hippocampal neu- to phosphorylate the recombinant receptor to very high levels rons kainate-gated currents mediated by AMPA receptors are by cotransfection with PKA Carshould make theseissues ame- enhancedby CAMP analogsand PKA, and they are diminished nable to study by identification of the site and its replacement by inhibitors of PKA (Greengardet al., 1991; Wang et al., 1991). by site-directed mutagenesis. In white perch retinal horizontal cells, a similar increasein the Another significant finding of this study is the increase in kainate-evoked current was observed (Liman et al., 1989). phosphorylation observed in responseto brief bursts of synaptic Moreover, CAMP analogsalso potentiated the current and cal- activity. As determined by phosphopeptidemap analysis,these cium influx through recombinant AMPA receptors consisting sites are essentially the same as those phosphorylated in re- of GluRl and GluR3 subunits expressedin Xenopus oocytes sponseto phorbol ester stimulation, suggestingthe involvement (Keller et al., 1992). Interestingly, there is no ideal consensus of PKC either directly or as part of a kinase cascade.In fact, site for PKA phosphorylation (i.e.,-Arg-Arg-XSer-) anywhere synaptic activity in thesecultures has previously been shown to in the GluRl protein (Hollmann et al., 1989), and in vitro studies activate the phosphoinositide second messengersystem (Mur- have failed to showPKA phosphorylation of GluR 1 (McGlade- phy et al., 1992b), which results in the stimulation of PKC, McCulloh et al., 1993). It may be that phosphorylation of GluRl however, this synaptic activity also results in a rapid increase The Journal of Neuroscience, December 1994, 14(12) 7591

Neurons 293 Cells 205 -

116- P 0 97- B ;;; E 66- e 6

45- 1 b - Electrophoresls +

293 Cells

Figure 5. PKA-mediated phosphorylation of GluRl in transfected 293 cells. Left, Human embryonal kidney 293 cells were transfected with GluR 1 or cotransfected with GluRl and PKA Car and prelabeled with 32P-orthophosphate. The GluR 1 protein was isolated by immunoprecipitation using anti-GluR 1 antipeptide antibodies and resolved by SDS-PAGE. Basal phosphorylation in the GluRl lane was revealed at longer exposures. Where indicated, the antibodies were first preadsorbed with an excess of synthetic peptide. Sizes of molecular mass standards (in kilodaltons) are indicated. Right, Gel slices containing GluRl that had been isolated from 293 cells cotransfected with PKA Cal were excised and digested with trypsin. The resulting phosphopeptides were separated in two dimensions on thin-layer cellulose plates and visualized by autoradiography. The origin is indicated circle). A phosphopeptide map of trypsin-digested GluRl isolated from cultured cortical neurons which had been stimulated with the adenylyl cyclase activator forskolin (as in Fig. 4B) is included for comparison. The PKA-dependent phosphorylation site common to both neuronal and recombinant GluRl is indicated (arrowheads). The basal phosphorylation site (as in Fig. 2) is clearly seen in neuronal GluRl at this exposure, but is evident in recombinant GluRl only on longer exposures. in the activity of another kinase,CaMKII. The increasein GluRl (Blackstone et al., 1992b; Petralia and Wenthold, 1992; Craig phosphorylation in responseto synaptic activity was very rapid, et al., 1993; Martin et al., 1993). It may be that most receptors with > 50% of the stimulation occurring within the first 30 sec. are accessibleto the protein kinase responsiblefor basalphos- In thesecultures, 90% of maximal CaMKII activity occurswith- phorylation as well as to PKA, but that synaptic activity-de- in the first 10 set of synaptic activity (Murphy et al., 1994). pendent kinase activation results in phosphorylation of only Thus, it remainspossible that CaMKII is involved in this phos- synaptic receptors. Thus the relative stoichiometry of the activ- phorylation. Supporting this notion are previous studiesshow- ity-dependent phosphorylation may appear artificially low on ing that CaMKII phosphorylates GluRl in vitro, and that ac- phosphopeptidemaps. Even so, the relatively low stoichiometry tivated CaMKII enhances kainate-gated currents in cultured of phorbol ester-stimulated phosphorylation is not likely to be hippocampal neurons (McGlade-McCulloh et al., 1993). Pre- due to spatial restriction of PKC activation. It is possible,how- liminary studiessuggest that the activity-dependent phosphor- ever, that PKC phosphorylation is relatively slow and incom- ylation in cortical neurons lasts for many minutes after the plete or that PKC is part of a kinase cascadeinvolving a more activity is terminated, changesare diminished but still evident spatially restricted kinase. Identification of the kinase(s) in- after 10 min of inactivity following 20 min of activity (C. Black- volved should help to clarify this issue. stone, T. H. Murphy, J. M. Baraban, and R. L. Huganir, un- In future studies,it will be interesting to examine other AMPA published observations). Thus, this phosphorylation may be receptor subunits (GluR2-4 and splice variants) to determine involved in somemechanisms of short-term changesin synaptic whether there is any subunit variation in phosphorylation. Dif- efficacy (Kullmann et al., 1992) that may involve modulation ferential regulation by protein phosphorylation has been dem- of AMPA receptor sensitivity. Interestingly, 30 set of synaptic onstrated for GABA, receptors, nicotinic acetylcholine recep- bursting, which leadsto an increasein GluRl phosphorylation, tors, and glycine receptorsthrough protein kinasesacting at sites also results in an increase in network synaptic activity which in the proposedmajor intracellular loop between putative trans- declinesover a 20 min period (T. H. Murphy and J. M. Baraban, membrane domains III and IV (Swope et al., 1992). Unlike unpublished observations). subunitsfrom theseother receptor families the proposedmajor One possibleconcern is that the stoichiometry of the activity- intracellular loops are nearly identical among GluRla (Kei- dependent and phorbol ester-stimulated phosphorylation ap- nanen et al., 1990), and it might be expectedthat there may not pears low, particularly relative to basal and PKA-mediated be differential phosphorylation among the AMPA receptor sub- phosphorylation. Though the changesmay in fact be smaller, it units. However, recent studieswith the NMDA receptor subunit is also possible that the result is deceiving. Many immunocy- NMDARl suggestthat, contrary to proposedtopology models, tochemical studiesevaluating rat brain sectionsand neurons in the C-terminal domain is in fact intracellular (Tingley et al., culture have demonstratedthat although theseAMPA receptors 1993). Immunocytochemical findings usingantibodies directed are most enriched in spines and postsynaptic densities, there against specific regions of AMPA receptor subunits are also are extensive nonsynaptic intracellular pools of receptor as well consistent with an intracellular location of this domain for 7592 Blackstone et al. * Glutamate Receptor Phosphorylation

GluR l-4 (Petralia and Wenthold, 1992; Craig et al., 1993; Mar- subunit: cellular and subcellular distribution in the rat forebrain. Neu- tin et al., 1993; Molnar et al., 1993). As this region is quite roscience 53:307-326. Monaghan DT, Bridges RJ, Cotman CW (1989) The excitatory amino divergent among GluR1-4, the prospects for differential regu- acid receptors: their classes, pharmacology, and distinct properties in lation by protein phosphorylation may be greater than previ- the function of the central nervous system. Annu Rev Pharmacol ously suspected. Studies aimed at identifying specific residues Toxic01 29:365-402. at which receptor phosphorylation occurs and examining the Moss SJ, Blackstone CD, Huganir RL (1993) Phosphorylation of re- functional significance of these modifications by site-directed combinant non-NMDA glutamate receptors on serine and tyrosine residues. Neurochem Res 18: 105-l 10. mutagenesis techniques in heterologous expression systems may Murphy TH, Baraban JM (1990) Glutamate toxicity in immature help to clarify molecular mechanisms underlying synaptic plas- cortical neurons precedes development ofglutamate receptor currents. ticity. Dev Brain Res 57:146-150. Murphy TH, Blatter LA, Wier WG, Baraban JM (1992a) Spontaneous synchronous synaptic calcium transients in cultured cortical neurons. J Neurosci 12:483ti845. References Murphy TH, Wright DD, Baraban JM (1992b) Phosphoinositide tum- Blackstone CD, Levey AI, Martin LJ, Price DL, Huganir RL (1992a) over associated with synaptic transmission. J Neurochem 59:2336- Immunological detection of glutamate receptor subtypes in human 2339. central nervous system. Ann Neurol 31:680-683. Murphy TH, Blatter LA, Bhat RV, Fiore RS, Wier G, Baraban JM Blackstone CD, Moss SJ, Martin LJ, Levey AI, Price DL, Huganir RL (1994) Differential regulation of calcium calmodulin dependent pro- (1992b) Biochemical characterization and localization of a non-N- tein kinase II and p42 MAP kinase activity by synaptic transmission. methyl-D-aspartate glutamate recentor in rat brain. J Neurochem 58: J Neurosci 14:1320-1331. 111811126.- - Nakanishi S (1992) Molecular diversity of glutamate receptors and Blackstone CD, Raymond L, Moss SJ, Huganir RL (1992~) Regulation implications for brain function. Science 258:597-603. of non-NMDA glutamate receptors by protein phosphorylation. In: Petralia RS, Wenthold RJ (1992) Light and electron immunocyto- Fidia Research Foundation symposium series, Vo19, Excitatory ami- chemical localization of AMPA-selective glutamate receptors in the no acids (Simon RP, ed), pp 15-20. New York: Thieme. rat brain. J Comp Neurol 3 18:329-354. Bliss TVP, Collingridge GL (1993) Synaptic model of memory: long- Raymond LA, Blackstone CD, Huganir RL (1993a) Phosphorylation term potentiation in the hippocampus. Nature 361:31-39. and modulation of recombinant GluR6 glutamate receptors by CAMP- Craig AM, Blackstone CD, Huganir RL, Banker G (1993) The dis- dependent protein kinase. Nature 361:637-641. tribution of glutamate receptors in cultured rat hippocampal neurons: Raymond LA, Blackstone CD, Huganir RL (1993b) Phosphorylation postsynaptic clustering of AMPA-selective subunits. Neuron 10: 1055- of amino acid neurotransmitter receptors in synaptic plasticity. Trends 1068. Neurosci 16:147-153. Gallo V, Upson LM, Hayes WP, Vyklicky L Jr, Winters CA, Buonanno Roche KW, Raymond LA, Blackstone C, Huganir RL (1994) Trans- A (1992) Molecular cloning and developmental analysis of a new membrane topology of the glutamate receptor subunit GluR6. J Biol glutamate receptor isoform in cerebellum. J Neurosci 12: 10 1O-l 023. Chem 269:11679-11682. Gasic GP, Hollmann M (1992) Molecular neurobiology of glutamate Rogers SW, Hughes TE, Hollmann M, Gasic GP, Deneris ES, Heine- receptors. Annu Rev Phvsiol 54:507-536. mann S (1991) The characterization and localization of the aluta- Greengard P, Jen J, Nairn AC, Stevens CF (1991) Enhancement of mate receptor subunit GluRl in the rat brain. J Neurosci 11:27 13- the glutamate response by CAMP-dependent protein kinase in hip- 2724. pocampal neurons. Science 253: 1135-l 138. Seeburg PH ( 1993) The TINS/Tips lecture: the molecular biology of Hollmann M, Heinemann S (1994) Cloned glutamate receptors. Annu mammalian glutamate receptor channels. Trends Neurosci 16:359- Rev Neurosci 17:3 l-108. 365. Hollmann M, O’Shea-Greenfield A, Rogers SW, Heinemann S (1989) Shaw PJ (1993) Excitatory amino acid receptors, excitotoxicity, and Cloning by functional expression of a member of the glutamate re- the human nervous system. Curr Opin Neurol Neurosurg 6:414-422. ceptor family. Nature 342:643-648. Sheng M, Cummings J, Roldan LA, Jan YN, Jan LY (1994) Changing Ito M (1989) Lona-term denression. Annu Rev Neurosci 12:85-102. subunit composition of heteromeric NMDA receptors during devel- Keller BU, Hollmamr M, Hememann S, Konnerth A (1992) Calcium opment of rat cortex. Nature 368: 144-147. influx through subunits GluRl/GluR3 of kainate/AMPA receptor Siegelbaum SA, Kandel ER (199 1) Learning-related synaptic plastic- channels is regulated by CAMP dependent protein kinase. EMBO J ity: LTP and LTD. Curr Opin Neurobiol 1: 113-120. 11:891-896. Sommer B, Seeburg PH (1992) Glutamate receptor channels: novel Kullmann DM, Perkel DJ, Manabe T, Nicoll RA (1992) Ca*+ entry properties and new clones. Trends Pharmacol Sci 13:29 l-296. via postsynaptic voltage-sensitive Ca 2+ channels can transiently po- Sommer B, Keinlnen K, Verdoom TA, Wisden W, Bumashev N, Herb tentiate excitatory transmission in the hippocampus. Neuron 9: 1175- A, Kijhler M, Takagi T, Sakmann B, Seeburg PH (1990) Flip and 1183. flop: a cell-specific functional switch in glutamate-operated channels Liman ER, Knapp AG, Dowling JE (1989) Enhancement of kainate- of the CNS. Science 249: 1580-I 585. gated currents in retinal horizontal cells by cyclic AMP-dependent Sommer B, Kiihler M, Sprengel R, Seeburg PH (199 1) RNA editing protein kinase. Brain Res 481:399-402. in brain controls a determinant of ion flow in glutamate-gated chan- Linden DJ (1994) Long-term synaptic depression in the mammalian nels. Cell 67:11-19. brain. Neuron 12:457-472. Sugihara H, Moryoshi K, Ishii T, Masu M, Nakanishi S (1992) Struc- Lipton SA, Rosenberg PA (1994) Mechanisms of disease: excitatory tures and properties of seven isoforms of the NMDA receptor gen- amino acids as a final common pathway for neurologic disorders. N erated by alternative splicing. Biochem Biophy Res Commun 185: Engl J Med 330:613-622. 826-832. Lomeli H, Sprengel R, Laurie DJ, Kijhr G, Herb A, Seebura PH. Wisden Swope SL, Moss SJ, Blackstone CD, Huganir RL (1992) Phosphor- W (1993) The rat delta- 1 and delta-2 subunits.extend the excitatory ylation of neurotransmitter receptors: a possible mode of synaptic amino acid recentor familv. FEBS Lett 3 15:3 18-322. plasticity. FASEB J 6:25 14-2523. Martin LJ, Blacksione CD, -Levey AI, Huganir RL, Price DL (1993) Tingley WG, Roche KW, Thompson AK, Huganir RL (1993) Reg- AMPA glutamate receptor subunits are differentially distributed in ulation of NMDA receptor phosphorylation by alternative splicing rat brain. Neuroscience 53~327-358. of the C-terminal domain. Nature 364:7&73. McGlade-McCulloh E, Yamamoto H, Tan S-E, Brickey DA, Soderling Uhler MD, Carmichael DF, Lee DC, Chrivia JC, Krebs EG, M&night TR ( 1993) Phosphorylation and regulation of glutamate receptors GS (1986) Isolation of cDNA clones coding for the catalytic subunit by calcium/calmodulin-dependent protein kinase II. Nature 362:640- of CAMP-dependent protein kinase. Proc Nat1 Acad Sci USA 83: 642. 1300-1304. Molnk E, Baude A, Richmond SA, Pate1 PB, Somogyi P, McIlhinney Wang L-Y, Salter MW, MacDonald JF (199 1) Regulation of kainate, RAJ (1993) Biochemical and immunocytochemical characteriza- receptors by CAMP-dependent protein kinase and phosphatases. Sci- tion of antipeptide antibodies to a cloned GluR 1 glutamate receptor ence 253:1132-1135. The Journal of Neuroscience, December 1994, 14(12) 7593

Wang L-Y, Tavema FA, Huang X-P, MacDonald JF, Hampson DR and assembly properties of GluR6 and KA2, two members of the ( 1993) Phosphorylation and modulation of a kainate receptor (GluR6) kainate receptor family, determined with subunit-specific antibodies. by CAMP-dependent protein kinase. Science 259: 1173-l 175. J Biol Chem 269: 1332-l 339. Wenthold RJ, Yokotani N, Doi K, Wada K (1992) Immunochemical Yamazaki M, Araki K, Shibata A, Mishina M (1992) Molecular clon- characterization of the non-NMDA glutamate receptor using subunit- ing of a cDNA encoding a novel member of the mouse glutamate specific antibodies. J Biol Chem 267:501-507. receptor channel family. Biochem Biophys Res Commun 183:886- Wenthold RJ, Trumpy VA, Zhu W-S, Petralia RS (1994) Biochemical 892.