Activated Nuclear Metabotropic Glutamate Receptor mGlu5 Couples to Nuclear Gq/11 Proteins to Generate Inositol 1,4,5-Trisphosphate-mediated Nuclear Ca2+ Release*

Recently we have shown that the metabotropic glutamate 5 (mGlu5) receptor can be expressed on nuclear membranes of heterologous cells or endogenously on striatal neurons where it can mediate nuclear Ca2+ changes. Here, pharmacological, optical, and genetic techniques were used to show that upon activation, nuclear mGlu5 receptors generate nuclear inositol 1,4,5-trisphosphate (IP3) in situ. Specifically, expression of an mGlu5 F767S mutant in HEK293 cells that blocks Gq/11 coupling or introduction of a dominant negative Gαq construct in striatal neurons prevented nuclear Ca2+ changes following receptor activation. These data indicate that nuclear mGlu5 receptors couple to Gq/11 to mobilize nuclear Ca2+. Nuclear mGlu5-mediated Ca2+ responses could also be blocked by the phospholipase C (PLC) inhibitor, U73122, the phosphatidylinositol (PI) PLC inhibitor 1-O-octadecyl-2-O-methyl-sn-glycero-3-phosphorylcholine (ET-18-OCH3), or by using small interfering RNA targeted against PLCβ1 demonstrating that PI-PLC is involved. Direct assessment of inositol phosphate production using a PIP2/IP3 “biosensor” revealed for the first time that IP3 can be generated in the nucleus following activation of nuclear mGlu5 receptors. Finally, both IP3 and ryanodine receptor blockers prevented nuclear mGlu5-mediated increases in intranuclear Ca2+. Collectively, this study shows that like plasma membrane receptors, activated nuclear mGlu5 receptors couple to Gq/11 and PLC to generate IP3-mediated release of Ca2+ from Ca2+-release channels in the nucleus. Thus the nucleus can function as an autonomous organelle independent of signals originating in the cytoplasm, and nuclear mGlu5 receptors play a dynamic role in mobilizing Ca2+ in a specific, localized fashion.

Many cells produce Ca 2ϩ signals both in the cytoplasm as well as the nucleus. Nuclear Ca 2ϩ is thought to play a vital role in a variety of nuclear functions such as cell division, proliferation, protein import, apoptosis, and gene transcription (1).
Nuclear Ca 2ϩ may be mobilized from a number of sources including diffusion of cytosolic Ca 2ϩ waves through nuclear pore complexes (2), release into the nucleoplasm from the nuclear lumen (3,4), or release from the so-called nucleoplasmic reticulum, invaginations of the nuclear envelope into the nucleoplasm itself (5). Presumably Ca 2ϩ release from the lumen of the nuclear membrane and/or the nucleoplasmic reticulum would either amplify Ca 2ϩ signals arriving via the nuclear pore complex and/or independently generate nucleoplasmic Ca 2ϩ transients. Such a model of nuclear Ca 2ϩ release is bolstered by data documenting the presence and activation of the inositol 1,4,5-trisphosphate receptors (IP 3 Rs) 2 and ryanodine receptors (RyRs) on inner nuclear membranes and the nucleoplasmic reticulum (3, 4, 6 -9). Thus these Ca 2ϩ -release channels are perfectly poised to independently regulate nucleoplasmic Ca 2ϩ levels.
Despite the importance of IP 3 Rs and RyRs in controlling the release of nuclear Ca 2ϩ , very little information is available regarding how these receptors themselves are activated. One of the most prevalent models is one in which IP 3 produced in the cytoplasm diffuses into the nucleus thereby activating nuclear IP 3 Rs (10). This premise is supported by studies showing that extracellular insulin-like growth factor-1 (IGF-1) activates its own receptor as well as phospholipase C (PLC) to stimulate IP 3 mobilization and the ensuing activation of both cytoplasmic and nuclear IP 3 Rs (11,12). Alternatively, IP 3 might be synthesized within the nucleus itself. This hypothesis is consistent with reports showing that the nucleus contains all of the enzymatic machinery necessary to synthesize IP 3 (13). Moreover, the nucleus also has the components necessary for the production of the RyR agonists, cyclic ADP-ribose (14) or nicotinic acid adenine dinucleotide phosphate (7). Thus, the nucleus could potentially function as an independent signaling unit controlling Ca 2ϩ signals in a specific, localized fashion.
How might nuclear IP 3 be generated? Typically, cell surface G protein-coupled receptors (GPCRs) associated with G q -like proteins (G q and its close homolog, G 11 ) (15) activate phospholipases such as PLC to hydrolyze phosphatidylinositol 4,5bisphosphate (PIP2) leading to IP 3 production. GPCRs coupling with G i/o -like proteins can also increase IP 3 levels via dissociated ␤/␥ subunit activation of PLC (16,17). These same mechanisms might also exist in the nucleus where several GPCRs have been recently localized not only to the plasma membrane but also intracellularly on nuclear membranes (18 -20). For example, receptors known to couple to G q/11 such as the prostaglandin EP receptor EP 1 , as well as the platelet-activating factor and lysophosphatidic acid (LPA-1) receptors have been found on the nuclear envelope (21). Similarly, we have shown that the Group 1 metabotropic glutamate receptors, mGlu1 and mGlu5, both of which couple to G q/11 , are endogenously expressed on the inner nuclear membranes of many different types of neurons (18,19,22). G q itself is present on nuclear membranes (23). Moreover, we have previously shown that a GFP-tagged G q protein co-localizes with mGlu5 and the inner nuclear membrane protein, lamin B 2 , in heterologous cell types (18). Thus, signaling systems are in place that could generate the requisite second messengers within the nucleoplasm itself.
Surprisingly, few studies have directly examined signal transduction pathways associated with nuclear GPCRs. Given the importance of nuclear Ca 2ϩ signaling and the unlikelihood, at least in neurons, that IP 3 could diffuse long distances from its presumed site of synthesis in neuronal processes, it seems reasonable to propose that IP 3 is generated in situ via the actions of GPCRs located on the inner nuclear membrane. Here we show that, like plasma membrane receptors, activated nuclear mGlu5 receptors couple to G q/11 and phosphatidylinositol (PI)-PLC in mGlu5-expressing HEK cells as well as striatal nuclei to generate IP 3 -mediated release of Ca 2ϩ via Ca 2ϩ release channels in the nucleus. Taken together, these data point to a novel mode of nuclear Ca 2ϩ generation, independent of cytosolic Ca 2ϩ , mediated through activated nuclear GPCRs. Cell Culture and Plasmids-HEK293 cells and the mGlu5 stable HEK cell line were maintained as described (18,24). Primary striatal cultures were prepared and maintained exactly as described (19). The mGlu5 mutant (F767S) was generated by recombinant polymerase chain reaction (25) using sense and antisense primers containing the relevant phenylalanine to serine mutation and wild type mGlu5 construct (24) as template. The mutation was confirmed by sequencing. The mGlu5 mutant F767S stable HEK cell line was generated using standard transfection techniques followed by repetitive rounds of limiting dilution (24). The construct containing the pleckstrin homology domain of PLC␦1 fused to enhanced green fluorescent protein (pEGFP-C1-PLC␦1-PH) was a gift from Dr. T. Meyer, Stanford University, Stanford, CA (26). pDsRed2 encoding DsRed2, a red fluorescent protein, was obtained from Clontech Laboratories, Inc., Mountain View, CA, and subcloned into the pcDNA3 vector. The dominant negative G␣ q construct (pcDNA3.1ϩG␣ q containing mutations Q209L/ D277N) was obtained from the University of Missouri cDNA Resource Center, Rolla, MO, and was used as described (27).
Subcellular Fractionation and Preparation of Nuclei-Plasma membrane and nuclear fractions were prepared from HEK cells as described (18). Nuclei were prepared from postnatal day 10 (P10) rat striata as described (19) and further purified using a 25-35% (w/v) iodixanol gradient as per the manufacturer's instructions. Aliquots from each fraction were used for gel electrophoresis and membrane binding. Protein concentrations of each fraction were determined using the Bradford assay (Bio-Rad). Purity of nuclei was assessed by loading HEK cells or striatal neurons with the cytoplasmic mitochondrialselective stain MitoTracker Deep Red 633 as well as the DNAspecific vital dye, Hoechst 33258 as described (29). After a 20-min preincubation, subcellular fractionation was performed and nuclear staining was monitored.
Fluorescent Measurements of Ca 2ϩ in Intact Cells and Isolated Nuclei-For whole cell measurements, mGlu5 wild type or mutant expressing HEK cells were washed with serum-free medium, incubated with Oregon Green 488 BAPTA-1AM (Molecular Probes) and imaged as described (18). Similarly, primary cultured striatal neurons transiently transfected with DsRed2 or co-transfected with DNG␣ q /DsRed2, scrambled siRNA/DsRed2, or PLC␤1 siRNA/DsRed2 (10:1 ratio in all cotransfections) were prepared as described (19). To measure Ca 2ϩ changes in individual nuclei, nuclei from HEK cells or rat P10 striatal tissues were prepared and processed as described (18,19). Drugs at ϫ100 concentration were added to the side of the dish and allowed to diffuse at room temperature. Extra-and intracellular buffers used for Ca 2ϩ measurements on intact cells or isolated nuclei were as described (19). Following image collection, cells/nuclei were fixed, stained with anti-mGlu5, lamin B 2 , and/or PLC␤1, and field relocated.
Single Cell and Nuclear Imaging of IP 3 -Dissociated striatal neurons were transiently transfected with pEGFP-C1-PLC␦1-PH (2 g 35-mm dishes Ϫ1 ). After 24 -48 h cells were washed with Neurobasal medium (Invitrogen) and used for monitoring IP 3 generation in real time. Nuclei prepared from mGlu5/HEK cells transiently transfected with pEGFP-C1-PLC␦1-PH were imaged as described (18,19). To monitor IP 3 generation and Ca 2ϩ changes simultaneously, nuclei were loaded with Calcium Crimson AM (Molecular Probes).
Confocal Microscopy and Data Analysis-All Ca 2ϩ measurements and IP 3 imaging were done using laser scanning confocal microscopes FluoView 500 and/or FluoView 1000 (Olympus, Center Valley, PA) as described previously (18,19). Simultaneous IP 3 and Ca 2ϩ measurements were done using bandpass barrier filter settings at 483-536 nm for EGFP and 590 -680 nm for Calcium Crimson (31). Analysis of changes in GFP fluorescence in striatal neurons or HEK nuclei, in real time, was performed as described (32). Increases in IP 3 were detected by measuring the translocation of EGFP from the plasma membrane to the cytosol or from nuclear membrane to the nucleoplasm. Images were processed with MetaMorph (version 5.0.7) Professional Image Analysis software. Data are presented as ratio of change in fluorescent intensity at a given time to initial fluorescence and expressed as percentage (⌬F/F o , %).
[ 3 H]IP Measurements-IP was measured in cells expressing mGlu5 as described (33,34). For measuring IP levels in nuclei, cells were treated as described (33,34), nuclei were prepared and then resuspended in intracellular buffer (in mM: 125 KCl, 2 KH 2 PO 4 , 2 MgCl 2 , 0.3 CaCl 2 , 10 D-glucose, 1 ATP and 40 HEPES pH 7.0) before being processed further (34). Data calculated as the ratio of labeled IP or IP 3 to total radioactivity (inositol, IP 1 , IP 2 , and IP 3 ) were expressed as percentage of response in unstimulated HEK cells or nuclei expressing mGlu5.

mGlu5 Receptors Are Localized at Both the Plasma and
Nuclear Membranes-We have previously shown that mGlu5 is expressed on plasma membranes as well as on many intracellular membranes including nuclear membranes (18,19). To confirm and extend these studies, we expressed N-terminal hemagglutinin-tagged mGlu5 in HEK cells and used anti-hemagglutinin antibody and differential permeabilization techniques to verify the presence of mGlu5 on plasma and nuclear membranes (supplemental Fig. S1). In other studies immunogold EM with antibodies directed against the C-terminal portion of mGlu5 was used to show that immunogold particles were abundantly associated with endoplasmic reticulum and nuclear membranes (supplemental Fig. S1). Previously we showed that agonists such as glutamate and quisqualate reach nuclear receptors via both sodium-dependent transporters and cystine glutamate exchangers (19). Here, we confirm that quisqualate is taken up by HEK cells, that the mGlu5 agonist 3,5dihydroxyphenylglycine is an impermeable agonist, and that the drugs LY395053 and LY367366 are impermeable antagonists (35) (supplemental Fig. S2). Thus, as an impermeable agonist, 3,5-dihydroxyphenylglycine cannot activate nuclear mGlu5 receptors in isolated nuclei derived from mGlu5 expressing HEK cells (mGlu5/HEK) or endogenously expressed on striatal nuclear membranes, whereas quisqualate can (supplemental Fig. S3, E-H). Finally, using differential labeling with vital dyes such as MitoTracker, which is excluded from the nucleus and Hoechst, which is retained in the nuclear compartment, we provide visible evidence of the purity of these nuclear preparations (supplemental Fig. S3, A-D). Therefore mGlu5 receptors are present and functional on nuclear membranes in heterologous cells as well as endogenous striatal nuclei.
Nuclear mGlu5 Couples to Nuclear G q/11 Proteins-Certain nuclear GPCRs appear to mediate nucleoplasmic Ca 2ϩ changes via pertussis toxin-sensitive pathways suggesting a G i/o -driven mechanism (21). However, we have previously shown that in HEK293 cells stably expressing mGlu5, pertussis toxin does not affect mGlu5-mediated cytoplasmic or nuclear Ca 2ϩ oscillations (18), thus we hypothesized that nuclear mGlu5 couples to nuclear G q/11 proteins to activate downstream signaling components. To test this idea, an mGlu5 mutant was constructed in which phenylalanine at position 767 in the third intracellular loop was replaced with serine (F767S), leading to the loss of G-protein coupling (36,37). Like wild type mGlu5, the F767S mutant was localized on both plasma and nuclear membranes where it co-localized with the nuclear membrane marker, lamin B 2 (Fig. 1, A and B). Similarly, nuclei isolated from mGlu5 or F767S stable HEK cell lines co-expressed either receptor together with lamin B 2 confirming expression on nuclear membranes ( Fig. 1, C and D). Using subcellular fractionation, both wild type and mutant receptors were clearly expressed on the plasma membrane as well as nuclear fractions as indicated by the membrane-specific markers, Na ϩ K ϩ -ATPase and lamin B 2 , respectively ( Fig. 1, E (18), bath application of glutamate-induced Ca 2ϩ oscillations in both the cytoplasm and nucleoplasm of wild type cells that were inhibited by the membrane-permeable mGlu5specific antagonist MPEP ( Fig. 2A). In contrast, no such oscillations were observed in cells expressing F767S (Fig. 2B). More directly, agonist induced Ca 2ϩ oscillations in nuclei derived from wild type cells, whereas no such induction was observed in nuclei isolated from F767S cells (Fig. 2, C versus D). The quantitation of data from three to five independent experiments is shown in Fig. 2, E and F. Thus although the F767S mutant bound ligand and was expressed on the same membranes as the wild type receptor, it was completely inactive in either intact cells or isolated nuclei. Given that the F767S mutation prevents G-protein coupling, and that pertussis toxin was ineffective in blocking mGlu5-mediated nuclear Ca 2ϩ oscillations (18) as well as that HEK293 cells do not express G o (36), these data demonstrate that mGlu5-induced Ca 2ϩ oscillations in the nucleus are a consequence of functional nuclear G q/11 coupling.
Nuclear mGlu5 Stimulates Nuclear PI-PLC-On the plasma membrane, activation of G q/11 -coupled receptors leads to PLC activation and PIP2 hydrolysis. Because both the substrate PIP2 (40) as well as PLC (␤1 isoform) (41) are known to be in the nucleus (13), we tested whether nuclear mGlu5 mediates nuclear Ca 2ϩ changes due to PLC stimulation. Using the same experimental paradigm, mGlu5-expressing nuclei were treated with glutamate to induce real time nuclear Ca 2ϩ oscillations prior to bath application of the PLC inhibitor U73122 (3 M) or its inactive analog U73343 (3 M). This concentration was chosen because others have shown that 3 M U73122 specifically inhibits PLC activity without affecting intracellular Ca 2ϩ levels FIGURE 1. mGlu5 wild type and F767S mutant exhibit nuclear localization. Stable HEK cell lines expressing either wild type mGlu5 or the F767S mutant, were fixed, permeabilized, and processed for immunocytochemistry using receptor-specific antibodies as well as anti-lamin B 2 . Cells were analyzed by confocal microscopy to detect receptor localization (red) or lamin B 2 distribution (green). Photographs represent single optical sections of 0.4 m merged such that yellow indicates co-localization of the specific antigens. Receptors expressed by intact cells (A and B; TL, transmitted light) or their isolated nuclei (C and D) exhibited pronounced co-localization with lamin B 2 . Subcellular fractionation of HEK cell lines expressing either wild type mGlu5 (E) or the F767S mutant (F) shows that wild type or mutant receptors can be detected in fractions containing either the nuclear (N) or plasma membranes (PM). Equal amounts of protein (30 g) from each fraction were separated on reducing SDS gels and transferred to nylon membranes (T, total cell lysates). The same blot was sequentially probed with antibodies against mGlu5, the inner nuclear marker, lamin B 2 , and the plasma membrane marker, Na ϩ K ϩ -ATPase. MAY 16, 2008 • VOLUME 283 • NUMBER 20 JOURNAL OF BIOLOGICAL CHEMISTRY 14075 (27,42). As shown in Fig. 3, A and F, the active analog U73122 inhibited glutamate-induced Ca 2ϩ oscillations, whereas the inactive analog, U73343, did not (Fig. 3, B and F). Moreover, pre-treatment with U73122 (3 M for 120 s) prevented glutamate-induced Ca 2ϩ oscillations in mGlu5-expressing HEK nuclei, whereas pre-treatment with either U73343 or the drug vehicle, 0.1% DMSO, did not prevent Ca 2ϩ oscillations (n Ͼ 10; data not shown).

Nuclear mGlu5 Couples to G q/11 /PI-PLC/IP 3 Pathway
Given that U73122 inhibits all types of PLCs, we further tested whether PI-PLC or phosphatidylcholine-PLC (PC-PLC) was involved. To do so, mGlu5-responding nuclei were treated either with the PI-PLC inhibitor ET-18-OCH 3 or the PC-PLC selective inhibitor D609 (42). Only the PI-PLC inhibitor blocked mGlu5-mediated nuclear Ca 2ϩ oscillations (Fig. 3, C,  D, and F). Finally, because studies have shown that IGF-1 induces nuclear Ca 2ϩ transients through G i/o subunits and PLC as well as via PI3K (12), we tested whether the PI3K inhibitor LY294002 could block agonist-induced Ca 2ϩ responses. Bath application of LY294002 had no effect on mGlu5-mediated nuclear responses (Fig. 3, E and F). Taken together, these data reveal that like plasma membrane receptors, nuclear mGlu5 receptors also couple to G q/11 and PLC. Inasmuch as all of the experiments were done in acutely isolated nuclei, it is clear that all of the necessary enzymatic machinery is available for these responses independent of cellular components.
Nuclear mGlu5 Generates Nuclear IP 3 -Due to the transient and readily metabolizable nature of IP 3 , it can be difficult to accurately measure in the cytosol let alone the nucleoplasm. Indeed, only one study has reported IP 3 changes in isolated nuclei (33). Moreover, mGlu5 is known to exhibit high constitutive activity in heterologous cell types (43). Not surprisingly then, biochemical assays to measure agonist-induced changes in inositol phosphates (IP) or more specifically, IP 3 , revealed small yet significant increases in IP and IP 3 levels as compared with untreated controls. Specifically, 41 and 21% increases in IP levels were observed in glutamate-treated mGlu5/HEK cells and nuclei, respectively (cytoplasmic IP levels were normalized at 100.0 Ϯ 0.1% without glutamate and 141.2* Ϯ 5.3% in the presence of 10 M glutamate. Nuclear IP levels were 100.0 Ϯ 0.3%, in the absence of glutamate and 120.8* Ϯ 5.2% in its presence; n ϭ 3, *, p Ͻ 0.05). Separate experiments examining IP 3 changes revealed an ϳ25% increase in glutamate-treated mGlu5/HEK cells and about a 15% increase in isolated mGlu5/HEK nuclei (cytoplasmic IP 3 levels were normalized at 100.0 Ϯ 2.5% in the absence of glutamate, whereas glutamate-treated cytoplasmic IP 3 levels were 125.0* Ϯ 5.2%. IP 3 levels in isolated nuclei were 100.0 Ϯ 1.3% in untreated controls and 115.3* Ϯ 3.2% for glutamate-treated; n ϭ 3, *, p Ͻ 0.05). Consistent with the notion that mGlu5 is constitutively active, the inverse agonist, MPEP, which locks the receptor into its inactive state (44), reduced basal IP levels by ϳ5-fold in the absence (19.2 Ϯ 0.7%) or presence of glutamate (23.5 Ϯ 7.6%). Moreover, IP levels were about 17% in F767S/HEK cells regardless of treatment.
To circumvent the limitations of the biochemical assay, we used a well established construct "pEGFP-C1-PLC␦1-PH" in which the pleckstrin homology (PH) domain of PLC␦1 with its high affinity for the polar group of PIP2 has been tagged with GFP (26,45). This probe is bound to PIP2 in the plasma membrane and the increase in IP 3 is indicated by the translocation of the fusion protein from the plasma membrane to the cytoplasm. Because this probe not only depends upon IP 3 but also on the PIP2 concentration in the plasma membrane, it is perhaps more aptly referred to as a PIP2/IP 3 biosensor (46). Therefore, mGlu5/HEK cells were transiently transfected with the PIP2/IP 3 biosensor, nuclei were isolated, and GFP-expressing nuclei were imaged in real time (Fig. 4). Under basal conditions the PIP2/IP 3 biosensor is located at the inner nuclear membrane due to its affinity for PIP2 (Fig. 4A, second panel). Upon glutamate treatment and PLC activation, the biosensor moved off the membrane into the nucleoplasm because of its 20-fold higher affinity for IP 3 (Fig. 4A, remaining panels) (45). Glutamate-induced biosensor responses were oscillatory and completely blocked by MPEP (Fig. 4, A and B). Isolated HEK nuclei expressing the PIP2/IP 3 biosensor or F767S mutant HEK nuclei co-expressing the biosensor never showed this response (data not shown). When analyzed by expressing the peak increases in nucleoplasmic GFP as compared with basal fluorescence, significant changes in F/F o were observed (Fig. 4C).
To demonstrate more clearly that nuclear mGlu5-induced IP 3 leads to Ca 2ϩ changes in the nucleus, we loaded mGlu5/ HEK nuclei expressing the PIP2/IP 3 biosensor with the Ca 2ϩ fluorophore, Calcium Crimson-AM. Bath application of glutamate leads to marked biosensor movement, which was concomitant with the rise in Ca 2ϩ in mGlu5/HEK nuclei (Fig. 4, D and E). Collectively, these data demonstrate that activated nuclear mGlu5 receptors generate nuclear IP 3 .
Nuclear mGlu5-mediated Ca 2ϩ Changes Originate from IP 3 R and RyR-Numerous studies have shown that upon activation IP 3 R and RyR located on the inner nuclear membrane can release Ca 2ϩ from the nuclear lumen into the nucleoplasm (7,9). Therefore, we used specific IP 3 R and RyR inhibitors to test the hypothesis that nuclear mGlu5 receptors coupled to G q/11 and PLC activate inner nuclear membrane Ca 2ϩ channels to mediate Ca 2ϩ changes. Glutamateinduced Ca 2ϩ oscillations in isolated mGlu5/HEK nuclei  were blocked by both the IP 3 R inhibitor, 2-APB (100 M), and the RyR antagonist, ryanodine (100 M) (Fig. 5). Pretreating with either drug prevented the induction of mGlu5 responses in nuclei isolated from the mGlu5/HEK cell line (not shown). Another IP 3 R antagonist, the highly specific xestospongin C (2 M) (47,48) also inhibited glutamateinduced Ca 2ϩ oscillations in isolated mGlu5/HEK nuclei by ϳ75% (Fig. 5, B and D). Thus, mGlu5-mediated Ca 2ϩ responses arise via IP 3 R and RyR channels.
Striatal Nuclear mGlu5 Also Uses the G q/11 /PI-PLC/IP 3 Pathway to Mediate Nuclear Ca 2ϩ Changes-To extend these findings to a more physiological system, we examined nuclear mGlu5 signal transduction pathways in dissociated striatal neurons or in nuclei acutely isolated from striatal tissue that we have previously shown to express functional nuclear mGlu5 receptors (19). The role of G q/11 in mediating mGlu5 responses was determined by transfecting striatal neurons with dominant-negative G␣ q (DNG␣ q ) together with DsRed2 or with DsRed2 only. Two days later neurons were loaded with Oregon Green BAPTA-1AM, imaged to acquire base-line Ca 2ϩ changes and then treated with 10 M quisqualate. As quisqualate also activates ␣-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid channels and mGlu1 receptors, it was bath applied in the presence of 5 M GYKI53655, an ␣-amino-3hydroxy-5-methyl-4-isoxazolepropionic acid antagonist, and 20 M CPCCOEt, an mGlu1 antagonist. Consistent with previous results (19), neurons transfected with DsRed2 alone showed Ca 2ϩ increases in both the cytoplasm as well as nucleus consisting of two phases, an initial rapid rise followed by a sustained plateau. Both responses could be terminated by addition of MPEP (Fig. 6, A and C). In contrast, neurons co-transfected with DNG␣ q and DsRed2 failed to show similar changes in cytoplasmic or nuclear Ca 2ϩ levels (Fig. 6, B and D). Specifically, there was approximately an 80% reduction in the nuclear Ca 2ϩ levels in cells co-transfected with DsRed2 and DNG␣ q as compared with DsRed2 only following quisqualate treatment (Fig. 6E). Cell surface mGlu5 receptors served as an internal positive control of DNG␣ q efficacy; hence cytoplasmic Ca 2ϩ levels were also reduced by 80 -85% in cells co-transfected with DsRed2 and DNG␣ q versus DsRed2-only following quisqualate treatment (Fig. 6). For further support of a predominant role of G q/11 in mGlu5mediated nuclear Ca 2ϩ increases, striatal cultures were pretreated with pertussis toxin for 18 h. Like mGlu5/HEK cells, pertussis toxin did not affect striatal mGlu5-mediated cytoplasmic or nuclear Ca 2ϩ  responses ruling out a G i/o -mediated response (n Ͼ 15; data not shown).
To determine whether PLC␤ was involved in nuclear mGlu5mediated Ca 2ϩ responses, siRNA targeted against PLC␤1, the most prevalent nuclear isoform (41), was used to knockdown expression. Western blotting of striatal lysates prepared 48 and 72 h after siRNA transfection showed that there was at least a 60 -70% knockdown of PLC␤1 using the targeted siRNA versus a scrambled control (Fig. 7, A and B). Immunostaining of transfected cultures further revealed a knockdown of PLC␤1 following the introduction of PLC␤1-specific siRNA versus the scrambled control (Fig. 7, C and D, last panels). Scrambled siRNA together with DsRed2 had no effect on agonist-induced mGlu5 Ca 2ϩ responses (Fig. 7, C, E, and G), whereas neurons co-transfected with PLC␤1 siRNA exhibited a 78% reduction in cytoplasmic and an 83% reduction in nuclear Ca 2ϩ following quisqualate treatment (Fig. 7, D, F, and G). Pharmacological support of these data comes from acutely isolated striatal nuclei experiments in which nuclei were loaded with the Ca 2ϩ fluorophore followed by quisqualate to induce a sustained nuclear Ca 2ϩ response. This response could be blocked by MPEP (19) (data not shown), U73122 (Fig. 8, A and F), or ET-18-OCH 3 (Fig. 8, C and F) but not U73343 (Fig. 8, B and F), D609 (Fig.  8, D and F), or LY294002 (Fig. 8, E  and F). Thus, PI-PLC but not PC-PLC or PI3K are required for nuclear Ca 2ϩ signaling following quisqualate activation of mGlu5 nuclear receptors.
To test whether mGlu5-induced striatal nuclear Ca 2ϩ changes were mediated via IP 3 generation, neurons were transfected with the PIP2/IP 3 biosensor. Images taken from a single optical plane revealed that GFP fluorescence was associated with both cell surface (Fig. 9, blue arrows) and nuclear membranes (Fig. 9, red arrows) when compared with the corresponding transmitted light images (Fig. 9A) or with the lamin B 2 staining (Fig. 9B). These results are consistent with the HEK data (Fig. 4) and with reports that PLC␦1 accumulates in nuclear compartments (49,50). Upon quisqualate stimulation, the PIP2/IP 3 biosensor moved away from either membrane exhibiting a sustained fluorescent increase akin to observed Ca 2ϩ changes (Fig. 9, C and D). Peak increases in cytoplasmic or nucleoplasmic fluorescence expressed as a percent of basal fluorescence showed similar responses (Fig. 9E). To ensure that these were mGlu5-mediated responses, cultures were fixed, stained with an anti-mGlu5 antibody, and subsequently fieldrelocated (not shown). Moreover, quisqualate-mediated Ca 2ϩ responses could be blocked by 2-APB, xestospongin C, or ryanodine (Fig. 10). Taken together, these data demonstrate that endogenous striatal mGlu5 receptors activate essentially the same G q/11 /PI-PLC/IP 3 pathway in the nucleus as they do on striatal plasma membranes (Fig. 11).

DISCUSSION
Given the many components of G-protein signaling described in the nucleus (13,(51)(52)(53)(54), the presence of both IP 3 R and RyR on the inner nuclear membrane (3,4), as well as the recent demonstrations of functional GPCRs on nuclear membranes such as mGlu5 (18,19) and mGlu1 (22), we hypothesized that canonical plasma membrane-based signaling components serve similar functions at nuclear membranes. Using FIGURE 7. Knockdown of PLC␤1 leads to reduction in striatal mGlu5-mediated Ca 2؉ changes. A, primary striatal neurons were transiently transfected with either scrambled siRNA (S) or PLC␤1 siRNA (P). At the times indicated, cells were lysed and proteins were separated on reducing SDS gels and transferred to nylon membranes. The same blot was sequentially probed with antibodies against PLC␤1 (Santa Cruz) and ␤-actin. B, the relative abundance of PLC␤1 was measured by Western blotting; the efficiency of knockdown is expressed in arbitrary units compared with the PLC␤1 levels in scrambled siRNA-transfected neurons. The data shown are compiled from three independent experiments. *, p Ͻ 0.005 when compared with PLC␤1 protein levels in scrambled siRNA-transfected cells. On the 12 th day in vitro, striatal neurons were transiently co-transfected with scrambled siRNA and DsRed2 (C) or with PLC␤1 siRNA and DsRed2 (D). Two days later, neurons were loaded with Oregon Green BAPTA-1AM (second panels) and quisqualate (Quis; 10 M) was bath applied (third panels). Representative traces are shown for cytoplasmic (blue line) or nuclear (red line) Ca 2ϩ responses from scrambled (E) or PLC␤1 siRNA (F) following quisqualate and subsequently, 1 M MPEP administration (black line). Imaged neurons were post hoc identified using mGlu5 (fourth panels) or PLC␤1 antibodies (last panels). G, compiled data from the maximum response of initial peak (⌬F/F o , %) from either scrambled (n ϭ 13) or PLC␤1 (n ϭ 17) siRNA-transfected neurons from three independent experiments. **, p Ͻ 0.0001 when compared with Ca 2ϩ responses from control siRNA and DsRed2-transfected cells.
optical, pharmacological, and genetic techniques, the present findings confirm this hypothesis showing that nuclear mGlu5 couples to G q/11 to activate nuclear PI-PLC, hydrolysis of PIP2, and generation of nuclear IP 3 . The latter leads to the release of Ca 2ϩ from the nuclear envelope in heterologous cell types as well as in striatal neurons. Taken together, these data suggest that signals generated at the inner nuclear membrane might amplify second messengers arriving via the nuclear pore complex and/or independently regulate nuclear function.
The basic signaling components ascribed to nuclear mGlu5 receptors are supported by a number of observations. First, mutations that block mGlu5 coupling to G q/11 prevented cytoplasmic and nuclear Ca 2ϩ changes in heterologous cells and their isolated nuclei despite normal agonist binding (Fig. 2). Second, when expressed in striatal neurons, dominant negative G␣ q abolished mGlu5-mediated nuclear Ca 2ϩ increases (Fig.  6). Third, low concentrations of the widely used PLC inhibitor, U73122, blocked mGlu5-mediated increases of nuclear Ca 2ϩ in both mGlu5/HEK and striatal nuclei (Figs. 3 and 8). Moreover, knockdown of PLC␤1 in striatal neurons significantly reduced mGlu5-mediated nuclear Ca 2ϩ responses as well (Fig. 7). Finally, in situ IP 3 production was revealed following mGlu5 activation in both heterologous and striatal nuclei using a sensitive optical PIP2/IP 3 biosensor approach (Figs. 4 and 9). Taken together, these data strongly support a model in which nuclear mGlu5 receptors lead to the activation of G␣ q/11 , PLC, and IP 3 to generate changes in nuclear Ca 2ϩ levels.
The traditional idea that GPCRs signal only from the cell surface is gradually being refined by studies showing that even internalized receptors can serve as scaffolds for signaling molecules (55) or, more directly, intracellular receptors can couple to various intracellular G proteins. Thus if mechanisms exist by which a receptor might be activated, signaling molecules are available to transmit the signal. Receptor activation might be accomplished in a variety of ways. For example, a large number of GPCRs such as the prostaglandin, platelet-activating factor, and LPA receptors, whose ligands are bioactive lipids derived from membrane hydrolysis, are located on nuclear membranes (for review, see Ref. 21). As ligand-generating enzymes are also present on nuclear membranes and because such ligands readily diffuse through lipid bilayers, PGE 2 , platelet-activating factor, and LPA can easily activate their cognate receptors. In contrast, mGlu5 or mGlu1 ligand-binding domains are within the nuclear lumen such that agonists must traverse both the cell surface lipid bilayer as well as the outer nuclear membrane for receptor activation (18,19,22). Mechanistically agonist transport is achieved via  the sodium-dependent glutamate transporter and/or the cystine, glutamate xCT exchanger (19,22). Although other mechanisms might also lead to intracellular mGlu5 or mGlu1 receptor activation, direct transfer of ligand is at least one effective means of delivering agonist to such receptors. There appear to be many different mechanisms involved in mobilizing Ca 2ϩ from storage in the nuclear envelope. Many of the described nuclear GPCRs couple to several different G proteins including G q/11 and G i/o . To date, G i/o -mediated signaling pathways seem to predominate regardless of whether a particular GPCR couples to G q/11 at the plasma membrane. For instance, although PGE 2 EP1 receptors couple to G q/11 at the cell surface, nuclear EP1 receptors generate nucleoplasmic Ca 2ϩ signals in a pertussis toxin-sensitive fashion (56,57). Similarly, although the LPA-1 receptor is known to interact with G i/o , G q/11 , and G 12/13 proteins at the cell surface, it is also G i/o -coupled on nuclear membranes (20). IGF-1 stimulation is also known to mobilize Ca 2ϩ from nuclear stores. Results suggest that extracellular IGF-1 activates IGF-1R, which in turn activates a pertussis toxin-sensitive G protein. The latter stimulates PI3K and subsequently PLC to generate cytoplasmic/ perinuclear IP 3 , which diffuses into the nuclear lumen (11,12). The possibility that cytoplasmic IP 3 diffuses into the nucleus via nuclear pore complexes can be ruled out here because the present experiments utilize pure isolated nuclei. Thus, for nuclear mGlu5 receptors, signal transduction appears to be via the same canonical G q/11 /PLC/IP 3 pathway that is also found at the plasma membrane (58).
Interestingly, both Ca 2ϩ channels, IP 3 R and RyR, contribute to the mGlu5-mediated nuclear Ca 2ϩ rises because adding their respective antagonists blocked the Ca 2ϩ changes. Although activation of IP 3 Rs is consistent with numerous studies showing that mGlu5 generates IP 3 , participation of RyR channels in the same process is intriguing. One possibility is that IP 3 -mediated Ca 2ϩ release leads to the activation of the RyR. In contrast, ryanodine may directly inhibit IP 3 -mediated Ca 2ϩ signals (59). Which of these models, direct or indirect inhibition of IP 3evoked Ca 2ϩ release, holds true for mGlu5-mediated Ca 2ϩ changes remains to be tested. Finally, because IP 3 Rs are present on the inner nuclear membrane, the translocation of IP 3 from its site of synthesis into the nucleoplasm as inferred from the PIP2/IP 3 biosensor experiments, is puzzling (Fig. 4A). Conceivably, IP 3 generated at the nuclear membrane might translocate to the nucleoplasmic reticulum that also expresses IP 3 Rs to further release nuclear Ca 2ϩ in specialized subdomains (6,60). In support of this idea, translocation of the biosensor was never uniform in neuronal nuclei; biosensor movement was skewed toward one side of the nucleus or another (Fig. 9C). Thus, spatial constraints, diffusion characteristics, and/or intranuclear buffering capabilities may regulate specific local Ca 2ϩ signals in nuclear subdomains.
It is widely believed that the amplitude, duration, and/or frequency of Ca 2ϩ fluctuations can differentially regulate downstream effectors such as transcriptional regulators, coactivators, and/or modifying enzymes (61-63). In the FIGURE 10. Striatal mGlu5 receptors release nuclear Ca 2؉ via Ca 2؉ release channels. A-C, representative traces of nuclear (black line) Ca 2ϩ responses in isolated striatal nuclei represented as the fractional change in fluorescence relative to the basal level. Isolated nuclei were treated with 10 M quisqualate (Quis) and the indicated antagonists. D, bar graph shows compiled data from the maximum response of initial peak (⌬F/F o , %) for nuclear responses from n Ͼ 10 for either antagonist, from three independent experiments. After treatment with antagonists the Ca 2ϩ responses were significantly different when compared with Ca 2ϩ responses in the presence of quisqualate (Quis) only (*, p Ͻ 0.001).
FIGURE 11. Proposed model of the signal transduction pathway associated with nuclear mGlu5 receptors. Endogenous striatal nuclear mGlu5 receptors, like plasma membrane receptors, couple to the G q/11 /PI-PLC/IP 3 pathway. EAAT, sodium-dependent transporter; xCT, cystine-glutamate transporter; DAG, diacylglycerol; ER, endoplasmic reticulum; NR, nucleoplasmic reticulum. nucleus differential regulation of Ca 2ϩ is crucial because blocking increased nuclear Ca 2ϩ prevents Ca 2ϩ -induced developmental processes and/or long term plasticity (62, 64 -66). For example, using an in vivo genetic approach, Limback-Stokin et al. (67) demonstrated that nuclear Ca 2ϩ signaling pathways, not cytoplasmic, were responsible for converting short term memory into long term memory. Clearly, one consequence of nuclear mGlu5 activation is prolonged nuclear Ca 2ϩ responses (see Ref. 19 and studies herein). Presumably, sustained mobilization of nuclear Ca 2ϩ activates different signaling pathways including nuclear Ca 2ϩ /calmodulin-activated kinase IV, a kinase known to phosphorylate the cAMP response element-binding protein (CREB) (68). Indeed, we have previously shown that agonist treatment can directly activate CREB in mGlu5-expressing isolated striatal nuclei (19). CREB may, in turn, lead to de novo gene transcription. Although these notions remain to be tested, activation of nuclear receptors creates yet one more way in which a cell can nuance its responses to a given stimuli.