[1] 20w The g-aminobutyric acid receptor B, but not the metabotropic glutamate receptor type-1, associates with lipid rafts in the rat cerebellum

[1] 171w Recent evidence suggests that specialized microdomains, called lipid rafts, exist within plasma membranes. These domains are enriched in cholesterol and sphingolipids and are resistant to non-ionic detergent-extraction at 48C. They contain speci®c populations of membrane proteins, and can change their size and composition in response to cellular signals, resulting in activation of signalling cascades. Here, we demonstrate that both the metabotropic g-aminobutyric acid receptor B (GABA B receptor) and the metabotropic glutamate receptor-1 from rat cerebellum are insoluble in the non-ionic detergent Triton X-100. However, only the GABA B receptor associates with raft fractions isolated from rat brain by sucrose gradient centrifugation. Moreover, increasing the stringency of isolation by decreasing the protein : detergent ratio caused an enrichment of the GABA B receptor in raft fractions. In contrast, depletion of cholesterol from cerebellar membranes by either saponin or methyl-b-cyclodextrin treatment, which solubilize known raft markers, also increased the solubility of the GABA B receptor. These properties are all consistent with an association of the GABA B receptor with lipid raft microdomains.

[1] 160w Initial studies were performed on the detergent solubility of the GABA B R and mGluR1 by incubating rat cerebellar homogenates with 1% Triton X-100 prior to centrifugation at 14,000 g. Following centrifugation, almost all of the GABA B R1/2 and mGluR1a/b were localised in the pellet (Fig. 1, pellet), indicating that the receptors did not solubilize in Triton X-100 under these conditions. Both Thy-1 and GM1, which are commonly used raft markers (Brown and London 1998;Hooper 1999), were also found almost exclusively in the precipitate. In contrast, the integral membrane protein Na 1 /K 1 ATPase, which is not raft associated (Martens et al. 2000), was found almost exclusively in the supernatant. Since lipid raft components are insoluble in Triton X-100 because of their high cholesterol and sphingolipid content, these results could indicate raft association of the receptors. Alternatively their insolubility could be due to cytoskeletal association. To distinguish between these possibilities raft fractions were isolated by sucrose gradient centrifugation.

[2] 112w A widely used approach for isolating lipid rafts is to treat membranes with detergent followed by ¯oatation of the detergent-insoluble glycolipid-enriched domains (DIGs) on sucrose gradients (Brown and London 1998;Hooper 1999). When Triton X-100 solubilized membranes are loaded at the bottom of a sucrose gradient and centrifuged at high speed, DIGs and their associated proteins migrate to the upper, low-density, region of the gradient, because of their high lipid content. Detergent-soluble material will remain in the lower fractions and the detergent-insoluble cytoskeletal components pellet at the bottom of the tube. Proteins that are DIG-associated after detergent extraction are presumed to have been raft-associated in the intact cell (Brown and London 1998;Hooper 1999).

[3] 187w Rat cerebellum was solubilized in Triton X-100 at a ratio of 3 : 1 protein to detergent and centrifuged as described in `materials and methods'. Fractions were collected from the top of the gradient. Comparing these for protein content, 90% of the protein was found in the lower fractions with approximately 20% present in the insoluble pellet. In contrast, only about 5±8% of protein was present in the lighter DIG fractions (Fig. 2A). However, more than 80% of alkaline phosphatase activity, a raft-associated enzyme, was found in the DIG fractions (Fig. 2B). Similarly, the raft markers Thy-1 and GM1 distributed almost exclusively to Fig. 3 GABA B R, but not mGluR1, is present in detergent-insoluble, glycolipid-enriched domains isolated from rat cerebellum. Following solubilization in 1% Triton X-100 at a protein to detergent ration of 3 : 1 the lysate was centrifuged on a sucrose gradient as described in Materials and methods. Equal volumes of the sucrose gradient fractions were analysed by SDS±PAGE and immunoblotting. The immunoblots were probed for the metabotropic neurotransmitter receptors GABA B R1/2, mGluR1a and mGluR1b. Comparable results were obtained in ®ve separate experiments.

[4] 132w Fig. 4 GABA B R, but not mGluR1, is enriched in detergent-insoluble, glycolipid-enriched domains when the protein:detergent ratio is decreased. Cerebellar homogenates were prepared at the different protein:detergent ratios indicated prior to sonication and sucrose gradient centrifugation. (a) Equal amounts of protein (2 mg) from cerebellar total membrane (TM) and from DIGs isolated at a protein to detergent ratio of 10 : 1, 7 : 1 and 3 : 1 were analysed by SDS± PAGE and immunoblotting for the raft markers GM1 and Thy-1, the non-raft Na 1 /K 1 ATPase, and the metabotropic neurotransmitter receptors GABA B R1/2, mGluR1a and mGluR1b. (b) Cerebellar homogenates prepared at a protein : detergent ratio of 7 : 1 were centrifuged on a sucrose gradient and analysed by SDS±PAGE and immunoblotting for mGluR1a and mGluR1b.

[5] 21w DIG fractions, whereas the non-raft associated enzyme Na 1 /K 1 ATPase, was found in the lower fractions only (Fig. 2C).

[6] 129w The metabotropic GABA B R containing the R2 subunit together with R1a or R1b showed a mixed distribution, with 30% to 40% present in the DIG fractions and the remainder being found mainly in the detergent-soluble fractions (Fig. 3; comparable results obtained in ®ve separate experiments). This distribution was consistently observed for all three subunits analysed, suggesting that similar ratios of GABA B R1a/R2 and GABA B R1b/R2 associate with lipid rafts in cerebellar membranes. In contrast, mGluR1a or mGluR1b were found exclusively in the soluble fractions and the insoluble pellet. Similar results were obtained when using synaptosomes, rather than homogenates, as starting material (data not shown). Together these results indicate that a signi®cant fraction of the GABA B R is found in the raft domains in the cerebellum.

[7] 157w GABA B R is enriched in detergent-insoluble, glycolipidenriched domains when the protein : detergent ratio is decreased Proteins that are almost exclusively raft-associated should be enriched in DIGs when compared to total membrane fractions. This is indeed the case for Thy-1 as shown in Fig. 4(A) where equal amounts of protein from differently extracted cerebellum are compared. The ganglioside GM1 is similarly enriched in these fractions. However, GABA B R, which is only partially raft-associated, was not enriched to the same degree compared to Thy-1 and GM1 (Fig. 4A) at all the different protein to detergent ratios tested. However, as the ratio of protein to detergent was decreased, thereby increasing the stringency of the raft isolation procedure, the amount of GABA B R present in the DIG fractions became a greater proportion of the total DIG fraction (Fig. 4A). This indicates that, while only a fraction of the receptor associates with lipid rafts, this association is speci®c.

[8] 82w Interestingly, at less stringent protein to detergent ratios we found some mGluR1a and b in the upper light fractions of the sucrose ¯oatation gradients (Fig. 4B, where a protein:detergent ratio of 7 : 1 was used, and Fig. 4A). This probably re¯ects contamination of DIG fractions with nonraft proteins at the high protein to detergent ratios. This is consistent with detection of traces of Na 1 /K 1 ATPase in the lighter fractions at these protein:detergent ratios (Fig. 4A and not shown).

[9] 238w Cholesterol-depletion increases the detergent-solubility of GABA B R Saponin is a detergent that complexes with membrane cholesterol and can increase the solubility of the raft associated proteins (Cerneus et al. 1993). Cerebellar membranes were therefore incubated with 0.5% saponin for 30 min on ice prior to treatment with 1% Triton X-100 and centrifugation at 14,000 g for 30 min. In the samples not treated with saponin, most of the raft markers Thy-1 and GM1 were pelleted by centrifugation, con®rming that they were not soluble in Triton X-100 (Fig. 5, `control' lanes). However, in the saponin-treated samples, almost all of the Thy-1 and GM1 was found in the supernatants (Fig. 5, `1 saponin' lanes). Therefore, reducing the cholesterol content in membranes solubilized these raft-associated proteins. In contrast, most of the Na 1 /K 1 ATPase was soluble in 1% Triton X-100. The amount of soluble GABA B R increased signi®cantly after saponin treatment indicating that indeed a proportion of the receptor is in the raft fraction. The fact that not all of the receptor was solubilized meant that the remaining GABA B R behaved like the mGluRs which were not solubilized following saponin treatment (Fig. 5). The relative dif®culty of solubilizing the mGluRs and a proportion of the GABA B R probably re¯ects their association with cytoskeletal elements, in the case of the mGluR1a via an association with Homer proteins and/or tubulin (Brakeman et al. 1997;Ciruela et al. 1999a).

[10] 146w To investigate the effect of cholesterol depletion on raftassociation of GABA B R, cerebellar homogenates were treated with methyl-b-cyclodextrin (MBC) prior to Triton X-100 solubilization and centrifugation. MBC removes cholesterol from membranes without binding to or inserting into the membrane (Klein et al. 1995). In lymphocytes, MBC has been shown to release substantial amounts of Fig. 5 Saponin-treatment increases the detergent-solubility of the GABA B receptor. Cerebellar membranes were incubated for 30 min on ice with or without 0.5% (w/v) saponin prior to extraction with 1% Triton X-100 and centrifugation at 14,000 g for 30 min The pellets (P) were resuspended to the original volume and equal volumes of these and the corresponding supernatants (S) were analysed by SDS±PAGE and immunoblotting for the raft markers GM1 and Thy-1, the non-raft Na 1 /K 1 ATPase and the metabotropic neurotransmitter receptors GABA B R1/2, mGluR1a and mGluR1b.

[11] 94w Thy-1 and GM1 from raft membranes (Ilangumaran and Hoessli 1998). Following MBC treatment, Thy-1 distributed to lower fractions on the sucrose density gradients when compared to controls (Fig. 6, top), showing its raft association was disrupted by the cholesterol depletion. Similarly, GABA B R no longer ¯oated to the upper regions of the gradient following MBC treatment when compared to controls (Fig. 6, centre and bottom) Thus, as with the saponin treatment, we ®nd that cholesterol depletion by MBC solubilizes GABA B R. All these results con®rm that GABA B R is indeed raft-associated.

[12] 90w G protein subunits G ao and G ai are present in detergent-insoluble, glycolipid-enriched domains What could be the functional implication of the raftassociation of the GABA B R? Lipid rafts are thought to function as signalling platforms and have been shown to cluster some receptors with their downstream signalling pathways in the haematopoietic system. To test whether lipid rafts may also function in clustering the GABA B R to its downstream signalling pathway, we tested the gradient fractions for presence of the G proteins G ao and G ai .

[13] 14w GABA B Rs modulate their effector systems by coupling to these inhibitory G proteins.

[14] 41w We found that approximately 30% of proteins G ao and G ai localize to DIG fractions (Fig. 7). These results may indicate that lipid rafts in rat cerebellum function to cluster the GABA B R and its immediate downstream signalling molecules.

[1] 196w Together all of the data presented in this report provide compelling evidence for the association of a signi®cant fraction (30% to 40%) of the GABA B R with lipid raft domains in rat cerebellum. The stringency of the isolation procedure was such that no detectable Na 1 /K 1 ATPase was found in the isolated DIG fractions nor were the related seven transmembrane spanning receptors mGluR1a and b. In addition, treatment of the cerebellar membranes with saponin or MBC reduced the association of the GABA B R and known raft associated proteins with the DIGs. Thus all of the results are consistent with the isolated DIGs corresponding to raft domains in the tissue, and support the presence of a signi®cant fraction of the GABA B R in a DIG fraction in the rat cerebellum. We also ®nd approximately 30% of the G protein subunits G ao and G ai in the DIG fractions. The GABA B R couples through these inhibitory G proteins to downstream effector systems (Couve et al. 2000), and the presence of both the receptor and the G proteins in lipid rafts may indicate a possible functional role for the receptor's raft association.

[2] 255w The reason for an incomplete association of the GABA B receptor with the raft fraction is not clear, but may re¯ect a dynamic and changing distribution of the receptor in the neuronal membrane. Experiments in the haematopoietic system have suggested strongly that lipid rafts can rapidly recruit other signalling complexes into functional assemblies following appropriate stimulation of the cells (Brown and London 1998;Moran and Miceli 1998;Viola et al. 1999;Simons and Toomre 2000). In the central nervous system raft signalling and recruitment of proteins has been shown to occur in the regulation of axonal growth and fasciculation via ephrins and ephrin tyrosine kinases (Bruckner et al. 1999), and the glial derived neurotrophic growth factor GNDF appears to act through recruitment of its co-receptor GFRa1 and the tyrosine kinase RET into lipid rafts (Tansey et al. 2000). Similarly, cross-linking of the cerebellar neuronal adhesion molecule TAG-1 causes recruitment of the tyrosine kinase Lyn into lipid rafts whilst cross-linking of the adhesion molecules NCAM 120 and F3 in maturing oligodendrocytes similarly recruits the kinase Fyn (Kramer et al. 1999). In the course of this study the AMPA glutamate receptors have been shown to be raft associated, and here too only a proportion of the receptors was present in the raft fraction (Suzuki et al. 2001). Following solubilization in 1% Triton X-100 at a protein to detergent ration of 3 : 1 the lysate was centrifuged on a sucrose gradient as described in Materials and methods. Equal volumes of the sucrose gradient fractions were analysed by SDS±PAGE and immunoblotting.

[3] 138w Alternatively, the association of a proportion of the GABA B receptors with lipid rafts may modulate the signalling of a speci®c set of the GABA B receptors. A modulatory function of raft-association has been shown for the voltage-gated K 1 channel K v 2.1, at least in stably transfected cell lines, where disruption of rafts by cyclodextrin signi®cantly shifts the steady-state inactivation of this channel by almost 40 mV (Martens et al. 2000). It is noteworthy that the same study showed that the related voltage-gated potassium channel K v 4.2 was not associated with lipid rafts. In our study the mGluR1 receptors, which show homology with the GABA B receptors, were also not found in the cerebellar raft fractions, a ®nding that suggests that the association of the GABA B receptor with these membrane domains has physiological relevance.

[4] 190w Biochemical studies are consistent with the GABA B receptor having pre-and post synaptic components, with the R1a and R1b subunits present in synaptic plasma membranes, but with only R1a being highly enriched in the postsynaptic density preparations (Benke et al. 1999). In our study whilst both the GABA B R1a and R1b receptor subunits are found in the raft fractions, saponin treatment of the cerebellar membranes preferentially solubilized GABA B R1b. This could suggest that GABA B receptors formed from the R1a and R2 subunits may be in a population of lipid rafts different from those containing the R1b and R2 subunits, which could contribute to the differential targeting of the differently spliced R1 subunits. Interestingly, GABA B receptors formed from R1a and R2 subunits appear to couple differently from those formed from R1b and R2 in transfected HEK 293 cells (Leaney and Tinker 2000). In these cells, the R1a/R2 receptors utilized both G ia2 and G oaa for signalling, whereas R1b/R2 receptors coupled only through the former. This behaviour of the different splice forms of GABA B R1 subunits could be due to their targeting into different raft populations.

[5] 143w It has been also suggested that axonal targeting of some proteins may be dependent on their association with lipid rafts. Thus, the targeting of Thy1 to the axon in cultured rat hippocampal neurones seemed to be affected by cholesterol depletion of the cultures (Ledesma et al. 1998) and the axonal targeting of the voltage gated potassium channel K v 1.4 was found to be dependent on its association with PSD-95. This interaction is dependent on the palmitoylation of PSD-95 which itself undergoes a redistribution to the axon in the presence of K v 1.4, and palmitoylation is a modi®cation of proteins that can promote raft association (Topinka and Bredt 1998). The formation of lipid rafts and synthesis of sphingomyelin and cholesterol have also been reported to be important for the maturation of axons in cultures of rat hippocampal neurones (Ledesma et al. 1999).

[6] 98w Relatively little is known about the lipid raft association of neurotransmitter receptors in the brain. Apart from the AMPA receptor (Suzuki et al. 2001), only the a7-subunit nicotinic acetylcholine receptor in the chick ciliary ganglion (Bruses et al. 2001) has been shown to associate with lipid rafts. No function was ascribed to the AMPA receptor raft association but, for the acetylcholine receptor, clustering in the ciliary cells was suggested to be regulated by raft association. It will be interesting to see if other central nervous system neurotransmitter receptors are found to be present in these specialized membrane domains.

[7] 102w In summary, we ®nd that GABA B R associates with lipid raft fractions isolated from rat cerebellum, whereas the related mGluR1 does not. Furthermore, we ®nd the inhibitory G protein G ao and G ai subunits in lipid raft fractions, indicating perhaps that lipid rafts in rat cerebellum function to cluster the GABA B R and its immediate downstream signalling pathway. These ®ndings increase the potential signalling complexity for this important neurotransmitter receptor. It will now be important to identify whether additional components of the GABA B R signalling pathway are present in rafts and how raft association may modulate receptor function.

[1] 105w Monoclonal anti-Thy-1 antibody was a kind gift from Prof. N. Barclay (Molecular Immunology Group, University of Oxford, UK), and polyclonal anti G ao and G ai antibodies were kindly donated by Prof. G. Milligan (Division of Biochemistry and Molecular Biology, University of Glasgow, U.K.). Anti-mGluR1 polyclonal antibodies F1 (pan-mGluR1) and F2 (anti-mGluR1a) (Ciruela et al. 1999a) and polyclonal antibodies against GABA B R1a/b and GABA B R2 (White et al. 2000) have been described previously. Ganglioside GM1 was detected with cholera toxin subunit b conjugated to horseradish peroxidase (Sigma). Protein was quanti®ed using the Bradford (Bio-Rad) or Bicinchoninic Acid assay (Pierce) with BSA as standard.

[2] 50w Cerebellum from adult Wistar rats was homogenized as outlined below, and incubated with 1% Triton X-100 for 30 min on ice followed by centrifugation at 14,000 g for 30 min. The detergentinsoluble pellets and their corresponding supernatants were analyzed by SDS±PAGE and immunoblotting as previously described (Ciruela et al. 1999a).

[3] 289w Detergent-insoluble glycolipid fractions (DIGs) were prepared following a protocol adapted from Hooper and co-workers (Parkin et al. 1999). All steps were carried out on ice or at 48C. Cerebellum from adult Wistar rats was homogenized in 10 volumes (w/v) of MES-buffered saline (MBS; 25 mm MES, pH 6.5, and 0.15 m NaCl) and a mixture of protease inhibitors. Homogenization was performed with 20 passes of a Dounce homogenizer followed by ®ve passages through a 23-gauge needle. Samples were assayed for protein and adjusted to 3 mg/mL unless otherwise indicated, and Triton X-100 was added to a ®nal concentration of 1% (w/v). The samples were then sonicated at 25% maximum power for 5 Â 20 s with a 60-s cooling period between each round of sonication using a Soniprep Soni®er. After dilution with an equal volume of 80% (w/v) sucrose in MBS, 1.7 mL of sample was injected under a 10.3-mL 5±30% (w/v) sucrose gradient in MBS. Following centrifugation at 140,000 g for 18 h at 48C in a SW 40.1 rotor (Beckman Instruments), 1 mL fractions were harvested from the top to the bottom of the gradient, and the detergent-insoluble pellet at the base of the gradient was resuspended in 1 mL of MBS. Samples were analyzed using SDS±PAGE and immunoblotting. Ganglioside GM1 was detected subsequent to sample electrophoresis, transfer to Immobilon-p membrane (Sigma) and blocking the membrane with skimmed milk (5% in PBS with 0.05% Tween-20) by incubation with cholera toxin subunit b conjugated to horseradish peroxidase (Sigma) 1 : 10,000 overnight at 48C and visualization using enhanced chemiluminescence (Pierce). Cholera toxin subunit b binds speci®cally to the ganglioside GM1 (Schon and Freire 1989) and is used routinely to visualize GM1 (Wu et al. 1997;Kawabuchi et al. 2000).

[4] 35w Samples were assayed for the GPI-anchored enzyme, alkaline phosphatase, using p-nitrophenyl phosphate as substrate, and the product was quanti®ed spectrophotometrically (Parkin et al. 1999). Enzyme activity was calculated as a percentage of the total activity.

[5] 54w Total membrane fractions from cerebellum were incubated with or without 0.5% (w/v) saponin for 30 min on ice prior to extraction with 1% Triton X-100 and centrifugation at 14 000 g for 30 min. Following resuspension of the pellets to the starting volume, equal volumes of the fractions were analyzed by SDS±PAGE and immunoblotting.

[1] 191w Neurotransmitter receptors are highly concentrated and precisely localized in the plasma membrane. The correct targeting and clustering of receptors is critical for neurotransmission. Several proteins have been identi®ed that interact with neurotransmitter receptors and link them to the cytoskeleton and to speci®c signal transduction pathways. These include the PSD-95/SAP90 family of synaptic proteins for NMDA receptors, GRIP for AMPA receptors and the Homer family of proteins for the metabotropic glutamate receptors (mGluRs; for a review see Fanning and Anderson 1998;Garner et al. 2000;Kennedy 2000;Scannevin and Huganir 2000;Sheng and Pak 2000). The inhibitory neurotransmitter receptors for glycine and GABA also interact with speci®c proteins, namely gephyrin and GABARAP, respectively (for a review see Kirsch 1999). One feature of these receptor-associated proteins is that they can cluster their associated receptors in both heterologously transfected cells and neurones, and they have the potential to interact with additional proteins other than their associated receptors, thereby enabling large signalling complexes to be established. Indeed, the existence of such large postsynaptic protein complexes has been elegantly demonstrated using SDS-gel electrophoresis and mass spectrometry following the isolation of NMDA receptors combined with synaptic components (Husi et al. 2000).

[2] 95w In cultured neurones postsynaptic neurotransmitter receptors and their interacting proteins form clusters, often independently, before synapse formation in culture (Richmond et al. 1996;Rao et al. 1998). Interestingly, some neurotransmitter receptors and their associated interacting proteins seem to be insoluble in Triton X-100 even when the cytoskeleton is disrupted by latrunculin-A (Allison et al. 1998;Allison et al. 2000). Since Triton X-100 insolubility and the formation of clusters are properties of lipid rafts (Brown and London 1998;Hooper 1999;Simons and Toomre 2000), this behaviour of certain receptors suggests that perhaps they can associate with these specialized lipid domains.

[3] 96w Lipid rafts, also termed detergent-insoluble glycolipid fractions (DIGs) or detergent resistant membranes (DRMs), are enriched for cholesterol and sphingolipids and form a distinct, liquid-ordered phase in the lipid bilayer of membranes (Simons and Ikonen 1997;London and Brown 2000). In addition, they contain speci®c subpopulations of membrane proteins often being enriched in glycophosphoinositollinked proteins (GPI-proteins) and speci®c tyrosine kinases. Lipid rafts also appear to concentrate lipid-modi®ed proteins such as Src family kinases, heterotrimeric G-proteins, cholesterol-linked and palmitoylated proteins such as Hedgehog, as well as palmitoylated transmembrane proteins (for reviews see Brown and London 1998;Simons et al. 1999).

[4] 188w Much work on rafts has been done in the haematopoietic system, where they were shown to be important for signalling processes in basophils and mast cells (Baird et al. 1999) and for T cell activation (Moran and Miceli 1998;Viola et al. 1999). As a result lipid rafts have been suggested to serve as assembly and sorting platforms for signalling complexes. Less is known about rafts in the central nervous system, although some proteins have been identi®ed that associate with rafts in neurones. These include GAP-43 and NAP-22 (Maekawa et al. 1999;Laux et al. 2000) the cell adhesion molecules Kilon (Funatsu et al. 1999) and F3 (Olive et al. 1995) and some tyrosine kinase receptors (Wu et al. 1997). However, relatively little is known about the neurotransmitter receptor content of lipid rafts in the brain. Of the neurotransmitter receptors, the a7-subunit nicotinic acetylcholine receptor (Bruses et al. 2001) and the ionotropic AMPA-type glutamate receptor (Suzuki et al. 2001) have been shown to associate with lipid rafts. In addition, the voltage-gated K 1 channel K v 2.1, but not K v 4.2, is associated with rafts (Martens et al. 2000).

[5] 183w The GABA B receptors (GABA B Rs) and the metabotropic glutamate receptors (mGluRs) belong to class C of the seventransmembrane-domain receptors. The GABA B receptor is a heterodimer composed of the homologous GABA B R1 and GABA B R2 subunits, neither of which is fully functional when expressed alone. It is thought that subunit dimerisation promotes proper post-translational processing and membrane targeting (Kaupmann et al. 1997;Kaupmann et al. 1998;White et al. 1998). Since the initial cloning of GABA B R1a and GABA B R1b, several other splice variants have been identi®ed both for GABA B R1 and GABA B R2 (Blein et al. 2000;Couve et al. 2000). The GABA B R couples through the heterotrimeric guanine-nucleotide binding (G) proteins Gi and Go to adenyl cyclase and Ca 21 and K 1 channels (Blein et al. 2000;Couve et al. 2000). It also interacts with members of the 14-3-3 family of signalling molecules (Couve et al. 2001), and the transcription factors ATFx and ATF4 (Bruckner et al. 1999;White et al. 2000). However, no interacting molecules have yet been identi®ed that target or cluster the receptor.

[6] 236w GABA B Rs share sequence similarity with the mGluRs which can be categorised into three subgroups on the basis of their pharmacology and sequence (Pin and Duvoisin 1995). mGluR1 and its splice variants belong to the group-1 mGluRs which are predominantly postsynaptic, and interact with the EVH-domain-containing protein Homer (Brakeman et al. 1997). This interaction may be important for receptor traf®cking, clustering and the formation and stabilization of signalling complexes at the plasma membrane (Tu et al. 1998(Tu et al. , 1999;;Xiao et al. 1998;Ciruela et al. 1999bCiruela et al. , 2000)). The group 1 mGluRs activate the phospholipase C signalling pathway and can also couple to calcium and potassium channels (Pin and Duvoisin 1995). Both the mGluRs and the GABA B Rs modulate central nervous system neurotransmission, albeit in different directions, and have been implicated in a number of pathologies. These include neurodegenerative disorders for the mGluRs (Pin and Duvoisin 1995;Conn and Pin 1997;), and drug addiction, epilepsy and depression for the GABA B Rs (Blein et al. 2000;Billinton et al. 2001;). This together with the fact that lipid rafts can concentrate speci®c G proteins and other proteins into signalling complexes prompted us to investigate the association of these neurotransmitter receptors with lipid rafts in the rat brain. Here we show that, while both the GABA B R and mGluR1 are insoluble in Triton X-100, only the GABA B receptor associates with lipid raft fractions.

[7] 50w Total membranes were collected from cerebellar postnuclear supernatants by centrifugation at 100,000 g for 2 h. DIGs were concentrated by dilution of the relevant sucrose gradient fractions ®vefold with MBS followed by ultracentrifugation at 100,000 g for 2 h. Equal amounts of total protein were analyzed by SDS±PAGE and immunoblotting.

[8] 24w Cerebellar homogenates were incubated either with 50 mm methylb-cyclodextrin or without for 30 min at 48C prior to detergent lysis and sucrose gradient ¯oatation.