Life Sciences, Vol. 66, No. 12, pp. 1063-1070, 2OOQ Copyright 0 2ooO Elsevier Science Inc. Printed in the USA. All rights reserved 0024-3205/00/S-see front matter ELSEVIER
MINIREVIEW INTRACELLULAR TRAFFICKING OF GABA* RECEPTORS Eugene M. Barnes, Jr.
Marrs McLean Department of Biochemistry and Division of Neuroscience, Baylor College of Medicine, Houston TX 77030 (Received in final form September 28, 1999)
Summarv Some of the mechanisms that control the intracellular trafficking of GABA, receptors have recently been described. Following the synthesis of a, g, and y subunits in the endoplasmic reticulum, ternary receptor complexes assemble slowly and are inefficiently inserted into surface membranes of heterologous cells. While p3, p4, and y2S subunits appear to contain polypeptide sequences that alone are sufficient for surface targeting, these sequences are neither conserved nor essential for surface expression of heteromeric GABA, receptors formed from alp or alPy subunits. At the neuronal surface, native GABA, receptor clustering and synaptic targeting require a y2 subunit and the participation of gephyrin, a clustering protein for glycine receptors. A linker protein, such as the GABA, receptor associated protein (GABARAP), may be necessary for the formation of GABAA receptor aggmgates containing gephyrin. A substantial fraction of surface receptors are sequestered by endocytosis, another process which apparently requires a GABA, receptor y2 subunit. In heterologous cells, constitutive endocytosis seems to predominate while, in cortical neurons, internalization is evoked when receptors are occupied by GABA, agonists. After constitutive endocytosis, receptors are relatively stable and can be rapidly recycled to the cell surface, a process that may be regulated by protein kinase C. On the other hand, a portion of the intracellular GABA, receptors derived from ligand-dependent endocytosis is apparently degraded. The clustering of GABA, receptors at synapses and at coated pits are two mechanisms that may compete for a pool of diffusable receptors, providing a model for plasticity at inhibitory synapses. Key Words: GABA, receptor subtype, subcellular targeting, receptor clustering, receptor endocytosis, receptor recycling In the vertebrate central nervous system, GABAA receptors are prominent conduits for fast inhibitory postsynaptic currents and, via binding sites for drugs such as benzodiazepines, barbiturates, and neurosteroids, are important for therapeutic intervention. The central chloride channel of vertebrate
Correpondingauthor: Phone: (713) 798-4523, FAX: (713) 798-7854, Email: [email protected]
GABA, Receptor Trafficking
Vol. 66, No. 12,2OCO
GABA, receptors is formed by a heteropentameric arrangement of polypeptide subunits from three families, cll-~26, pl-p4, and y l-y4 (1,2). Within these families, additional variants arise through alternative splicing. Single-member classes of GABA* receptor subunits, 8, E, and A have also been identified. This great diversity of receptor subunits leads to profound differences in tissue and subcellular distribution, ontogeny, pharmacology, and regulation of GABA, receptors. A recent review (3) has considered the role of GABA, receptor subunit gene disruption in advancing our understanding of some of these relationships. The current article is focused on mechanisms for the intracellular traffkking of GABA, receptors. These mechanisms (Fig. 1) control the assembly of appropriate subunits into receptors and their targeting to synaptic and extrasynaptic membranes. Furthermore, after reaching the neuronal surface, GABA,, receptors are subject to endocytosis, a process which may contribute to the synaptic remodeling which underlies physical dependence on benzodiazepines and GABAmimetic agents (4,5). Lately, substantial advances have been made in understanding how these subcellular processes regulate the intracellular traffic of GABA, receptors.
Model for intracellular trafiicking of GABA, receptors. Individual receptor subunits are synthesized de novo on the endoplasmic reticulum membrane and assembled into receptor complexes (step 1). Unassembled receptor subunits ark degraded (step 2), while assembled receptors are targeted via the Golgi apparatus to the plasma membrane (step 3). The surface receptors may either aggregate to form synaptic clusters (step 4) or cluster in clathrin-coated pits (step 5) which produce coated vesicles by endocytosis (step 6). Via endosomal pathways, the internalized receptors can recycle to the surface (step 7) or be degraded (step 8).
Vol. 66, No. 12.2000
Investigation of the assembly of GABA, receptors in neural tissues is hampered by the diversity both of receptor subunits and cellular specialization, explaining the need for model systems. One simplification has been to focus on co-assembly of al, l32, and y2 subunits into GABA, receptors because such complexes represent a major form in the mammalian brain (6). Coexpression of these subunits in heterologous cells provides a means to limit the substrates available for assembly and to introduce epitope tags on their amino-terminal regions, facilitating topological analysis with antibodies. Such experiments have revealed that al, l32, and y2 subunits co-oligomerize in the endoplasmic reticulum (ER), in association with the chaperone proteins BiP and calnexin (78). These co-assemblies of GAB A, receptor a 1p2y2 subunits are transported to the Golgi apparatus and then are exported to the surface membrane as functional GABA-gated channels. In order to determine some of the rules that govern assembly of GABA, receptors in the ER, individual subunits were expressed alone and in various combinations in human embryonic kidney (HEK) cells. When expressed alone, al or l32subunits are retained within the ER, while al l32- and al l32y2-subunit combinations assemble into oligomers and access the surface as gated channels (7). Insight into some of the structural motifs that may influence surface targeting has been provided by studies of y2 subunits (9). When expressed alone, y2S subunits can reach the plasma membrane, although neither high-molecular weight oligomers nor channels are formed. On the other hand, unitary expression of the y2L subunit, which differs from the y2S subunit by the presence of the insert LeuLeu-Arg-Met-Phe-Ser-Phe-Lys in the large intracellular region, leads to ER retention. Since the level of surface expression of the y2S subunit is reduced by co-transfection with al or /32 subunits, the occurrence of y2S-subunit surface monomers in the brain seems improbable. Additional information about GABA, receptor assembly motifs has been provided by studies of the p3 subunit which, in contrast to the l32 subunit, is capable of producing homomeric, pentobarbitalactivated channels on the surface of HEK cells and Xenopus oocytes (10,ll). Using chimeras of l32 and l33 subunits, Taylor et al. (12) have identified four amino acids in the N-terminal region of the mouse p3 subunit (Gly- 171, Lys- 173, Glu- 179, Arg- 180) that by mutation will enable the 02 subunit to form homomeric channels. Although this could constitute an “assembly signal” for P3-subunit homomers, it is not essential for the association of a 1p3 subunits (12) and is not conserved in chicken p4 subunits that form homomeric GAB A-gated channels (13). How efficient is the process of GABA, receptor assembly? In order to examine the rates of assembly, the formation of receptor oligomers in heterologous cells has been analyzed by sedimentation velocity (12,14). This revealed that l32subunits migrate as a 5s species and thus fail to oligomerize, while l33 subunits (as well as alp2 and alp3 combinations) form 9s complexes. The 5s species from single subunits proved to be relatively unstable in baby hamster kidney (BHK) cells, with half-lives of approximately 2 hr for both the al and the p2 subunit (14). These investigations also show that receptor assembly was relatively inefficient in BHK cells, with slightly more than one-third of the available a 1 and p2 subunits being converted to binary 9s complexes. In addition to the inefficiency of assembly, the appearance of GABA,, receptor complexes on BHK cell surfaces was also relatively slow. The nascent all3Zsubunit combination arrived on the plasmalemma more that 6 hr after formation of the 9s receptor complex. Much less information is currently available about the assembly and stability of GABA, receptors in neurons. Using a combination of quantitative immunoblotting and ligand binding, it was estimated that only 20% of al subunits were assembled into receptor complexes in cultured cortical neurons, while the unassembled subunits degraded with a half-life of 7.7 hr (15). Since al subunits appear to be synthesized in a large excess compared to the other GABA, receptor subunits in the neuronal cultures, the accumulation of al subunits may reflect a relative deficiency of 8 and y subunits, rather than an
GABA, Receptor Tmfficking
Vol. 66. No. 12.2000
inefficiency of receptor assembly. Furthermore, in these neurons, 690% of the assembled receptors have a surface localization (16). Thus, any delays that may occur between neuronal assembly and surface targeting of GABA, receptors are not reflected at steady state. After GABA, receptor assembly in the ER, the processes that regulate receptor maturation and surface targeting are poorly understood. N-linked glycosylation was shown to increase the efficiency of receptor expression on the surface HEK cells, but was not required for assembly of receptor subunits (7). Another important regulatory process is activation of the insulin receptor tyrosine kinase that increases the number of functional GABA, receptor alP2/3y2-subunit complexes on the surface of HEK cells (17). This stimulatory effect of insulin occurred through recruitment of GABA, receptors from an intracellular pool rather than through subunit synthesis. Similar insulin-evoked mechanisms appear to operate in hippocampal neurons in culture and in slices (17). A third kind of regulatory process occurs during kindling epileptogenesis which results in an increased number of surface GABA, receptors at hippocampal inhibitory synapses (18). In this form of synaptic plasticity, a greater density of synaptic GABA, receptors and expansion of the synaptic junctional area produced an enhancement of elementary GABA-gated synaptic currents. The signaling pathway that couples kindling to GABA, receptor regulation is unknown.
In the vertebrate brain, GABA, receptors are usually found on postsynaptic densities, dendrites, and somata. The determinants for the sorting process reside on specific GABA, receptor subunits and on receptor clustering proteins. As an example of the former, GABA, receptor receptors containing a5 subunits are found mainly on somatic and dendridic membranes of hippocampal pyramidal neurons, while a2 subunits are localized to the axon initial segment (19,20). Thus, within the same cell, a subunits appear to participate in subcellular targeting. That the p subunits may also play a role in sorting to specific membranes is indicated by studies of polarized Madin-Darby canine kidney (MDCK) cells. A non-polarized (apical and basolateral) distribution was found for all31 subunit complexes, but a 1B2 and a 1p3 combinations were localized to the basolateral membrane ( 10). In this expression system, dendritic proteins usually appear in basolateral membranes, while apical sorting in MDCK cells may lack a clear neuronal correlate (21). The linker proteins that presumably participate in the membrane sorting of GABA, receptor a or p subunits have not been identified. Studies of neonatal mice with targeted disruption of the GABA, receptor y2-subunit gene have demonstrated a requirement for this subunit in cortical receptor clustering and synaptic localization (22). This clustering mechanism appears also to involve gephyrin, a tubulin-binding protein that is necessary for synaptic targeting of glycine receptors (23) and for synthesis of a cofactor for molybdoenzymes in nonneural tissues (24). Interestingly, the yZsubunit deficient mice also showed a lack of gephyrin clusters at cortical synapses. Furthermore, depletion of gephyrin in hippocampal neurons from normal animals, produced by treatment with antisense oligonucleotides, also reduced the synaptic accumulation of GABA, receptor a2 and y2 subunits (22). With the possible exception of the p3 subunit (23), a direct interaction of gephyrin with GABA, receptor subunits has not been demonstrated, so this targeting mechanism may also require a linker protein. One candidate for such a protein is GABARAP (GABA,-receptor associated protein) identified in a yeast two-hybrid screen using the y2 subunit as bait (25). GABARAP binds in vitro with yZsubunit fusion proteins, coimmunoprecipitates with GABA, receptor subunits from brain extracts, and colocalizes with p2f3 subunits in surface clusters on cultured cortical neurons. The presence of a putative tubulin-binding motif in GABARAP suggests a possible role in surface targeting and/or clustering. Like gephyrin, GABARAP is also found in many peripheral tissues and presumably performs additional functions unrelated to GABA, receptor binding.
Vol. 66, No. 12, 2000
Surface GABA, receptors can also cluster on clathrin-coated pits which invaginate during the process of endocytosis. The endocytosis of GABA* receptors can be either ligand-dependent or ligandindependent. The ligand-dependent process can be evoked by exposure of cortical neurons either to GABA, or benzodiazepine agonists for 2 hr at 37°C (4). The resulting receptor endocytosis could be prevented by co-incubation with the respective antagonists. In comparison, much less sequestration was observed in untreated neurons or antagonist-treated neurons. Some intracellular GABA, receptors are found on clathrin-coated vesicles from normal bovine, rat, and mouse brain. Such vesicles are believed to be the main vehicle for endocytosis of a variety of neural and non-neural receptors. That ligand-dependent internalization of GABA, receptors occurs in vivo was demonstrated by chronic administration of lorazepam to mice. This treatment produced increased numbers of flunitrazepam binding sites and enhanced immunoreactivity of GABA, receptor al subunits in coated vesicles (26). Compared to their counterparts on synaptic membranes from bovine brain, GABA, receptors on clathrin-coated vesicles show impairments to the allosteric coupling mechanisms that link the site for GABA binding to the sites for benzodiazepine binding and picrotoxin binding (27). Western blot analysis of these membrane fractions showed that the ratio of receptor al/P2 subunits on coated vesicles was also substantially less than that on synaptic membranes. For the GABA, receptors on coated vesicles, it was suggested that either a selective retrieval of receptor subunits or subunit proteolysis could account for these modifications. Ligand-independent endocytosis of GABA, receptors appears to predominate in heterologous cells. In Xenopus oocytes, acute treatment with phorbol ester, an activator of protein kinase C (PKC), caused a decline in surface receptors formed from alp2y2L subunits (28). This result, obtained by electrophysiological and fluorescence techniques, is consistent with ligand-independent GABA, receptor internalization. In HEK cells, y2S subunits expressed alone or alfi2y2S- or alp2y2L-subunit combinations have, in addition to surface localization, a perinuclear distribution that colocalizes with transferrin receptors, a marker for endocytosis (9). Similarly, for BHK cells expressing surface y2S subunits alone, immunogold electron microscopy revealed many of these polypeptides in clathrincoated pits (9). Thus, the experiments with HEK and BHK cells are also consistent with ligandindependent endocytosis of GABA, receptors containing y2 subunits. On the other hand, surface a 1P2-subunit combinations did not colocalize with transferrin receptors in HEK cells, nor did this receptor combination collect in the coated pits of BHK cells. This suggests a role for the y2 subunit in the internalization process. No adapter or linker proteins have been identified which could provide associations of GABA, receptors with coated-pit proteins. The ligand-independent endocytosis and subsequent recycling of GABA, receptors has recently been investigated by antibody labeling of intact HEK cells (C.N. Connolly, J.M. Uren, P. Thomas, J.T. Kittler, T. Smart, and S.J. Moss, submitted for publication). After incubating the labeled cells for up to 2 hr at 37 “C, receptors composed of a 1p2 subunits had low amounts of internalization and showed little loss from the surface, while a 1p2y2S or a 1p2y2L subunit complexes showed greater amounts of sequestration and surface depletion. Wortmannin, an inhibitor of transferrin receptor recycling (29), enhanced the depletion of surface GABA, receptors, revealing a constitutive, dynamic cycle of receptor endocytosis and recycling (Fig. 1, steps S-7). Exposure of the HEK cells to phorbol ester also increased the depletion of surface alp2y2L receptors. The effect of phorbol ester was blocked by staurosporine (a PKC inhibitor), which failed however to influence the ability of wortmannin to deplete surface receptors. Since significant amounts of internalization occurred in the absence of phorbol ester or wortmannin, Connolly et al. (submitted) have suggested that PKC activation attenuates recycling rather than enhancing the endocytosis of GABA, receptors. In addition, mutation of the PKC phosphorylation sites on GABA, receptors had no effect on the receptor down-regulation observed in I-IEK cells (Connolly et al., submitted) or oocytes (28). Thus, the critical phosphorylation sites appear to be on proteins that regulate trafficking rather than on receptor polypeptides.
GABA, Receptor Trafficking
Vol. 66, No. 12.2000
In addition to GABA, receptor recycling, another possible fate of sequestered receptors is degradation. In cultured cortical neurons, surface [‘2SI]al-subunits that were sequestered during GABA treatment undergo degradation with an approximate half-life of 2 hr (4,16). This is concordant with a GABAinduced decline in al-subunit immunoreactivity and receptor ligand-binding sites in these cultures (4,15). A similar degradation process is implicated in cerebellar granule cells which show a rapid, benzodiazepine-evoked decline in al- and P2/3-subunit immunoreactivity (30). It is presumed that, after internalization on coated vesicles, receptors targeted for degradation are transferred to late endosomes and then to lysosomes. Proteolysis can occur in either of these compartments (31). In contrast to neurons, GABA, receptors in HEK cells were much more stable, showing no GABAevoked reduction in ligand-binding sites or in al- or pl-subunit immunoreactivity (32). In addition, the al p2y2 surface complexes which were sequestered by ligand-independent endocytosis in HEK cells also appeared to be metabolically stable over a 4 hr period (Connolly et al., submitted). Thus, it appears that the fate of internalized GABA, receptors also differs in neurons and heterologous cells. At this point it may be helpful to summarize this discussion with that in the previous section. To date, the studies of neurons show that surface GABA, receptors are internalized by ligand-dependent endocytosis and then degraded. In contrast, GABA, receptors on heterologous cells are seqestered by ligand-independent endocytosis and then recycled to the surface. The underlying reasons for this discrepancy are not understood. Aside from substantial differences in experimental methodology, another clear possibility is the lack of neuron-specific proteins in the heterologous cells. This may prove to be an advantage in the search for the accessory or linker proteins that guide the intracellular trafficking of GABA, receptors.
One particularly large gap in our knowledge concerns the biological purpose of GABA, receptor endocytosis in the brain. It can be envisioned (Fig. 1) that the clustering of GABA, receptors at synapses (step 4) and in coated pits (step 5) may represent two mechanisms that compete for a pool of diffisable receptors, representing a model for synaptic plasticity. In the absence of receptor ligands, constitutive endocytosis (step 6) could deliver receptors for degradation (step 8), insuring a constant turnover, while receptor recycling (step 7) could provide substrates for intracellular regulation, for example, receptor phosphorylation or dephosphorylation. Ligand-evoked endocytosis of GABA, receptors, by contrast, could furnish a pathway for controlling the number of surface receptors in response to tonic levels of GABA. Such sustained extracellular pools of GABA may be particularly significant in the developing nervous system and in temporal lobe epilepsy. Futhermore, during chronic administration of benzodiazepines, the development of tolerance and dependence is also likely to involve GABA, receptor down-regulation (4). After endocytosis, some of the internalized pool of receptors could be diverted to degradation (step 8), while the remainder could be subject to regulated recycling in neurons, as already described for heterologous cells. Interest in receptor recycling is also piqued by the observation, in cerebellar granule cells, that the rapid, benzodiazepine-evoked downregulation of GABA, receptor subunits is blocked by inhibitors of PKC (30). Clearly, further attention needs to be given to the mechanisms that control the endocytosis and recycling of GABA, receptors in neurons and to the proteins that target receptor subunits to coated pits and endocytic compartments. Finally, consideration should be given to the hypothesis that internalized GABA, receptor subunits may provide signals that contribute to regulation of receptor gene expression. In cerebellar granule cells, which have a characteristic trophic response to GABA, GABA, receptor occupancy slowly induces al- and PZsubunit mRNA biogenesis (33,34). In cortical neurons, prolonged occupancy of surface GABA, receptors by GABA represses the synthesis of al- and P2-subunit mRNA and al-
Vol. 66. No. 12.2000
GABA, Receptor Trafficking
subunit polypeptides (4,15). The signaling pathway from surface to nucleus is undefined, but long periods of GABA, receptor occupancy seem to be required. It is conspicuous that the repression of receptor gene expression in cortical neurons becomes detectable only after the occurrence of GABAevoked internalization and partial degradation of surface GABA, receptors (15). Establishing an important precedent, the studies of Lefkowitz (35) show that agonist-dependent endocytosis of &adrenergic receptors is a necessary step in the activation of MAP kinases and the mitogenic response. This lends indirect support to the hypothesis (15) that GABA, receptors may regulate their own biosynthesis through putative intracellular signals provided in part by GABA-evoked endocytosis. Although this is a speculative notion, it may be offered as an additional stimulus to future investigations of GABA, receptor trafficking.
The author would like thank the investigators that provided manuscripts prior to publication, Dr. Hulusi Cinar for helpful comments on the manuscript, and the National Institutes of Health for support of research in the author’s laboratory.
2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22.
R.L. MACDONALD and R.W. OLSEN, Ann. Rev. Neurosci. 17 569-602 (1995). L.E. RABOW, S.J. RUSSEK, and D.H. FARB, Synapse 21 189-274 (1995). H. MGHLER, B. LUSCHER, J.-M. FRITSCHY, D. BENKE, J. BENSON and U. RUDOLPH, Life Sci. 62 1611-1615 (1998). E.M. BARNES, JR., Int. Rev. Neurobiol. 39 53-76 (1996). P. POISBEAU, S.R. WILLIAMS and I. MODY, J. Neurosci. 17 3467-3475 (1997). A.L. DE BLAS, Mol. Neurobiol. 12 55-71 (1996). C.N. CONNOLLY, B.J. KRISHEK, B.J. MCDONALD, T.G. SMARTand S.J. MOSS, J. Biol. Chem. 271 89-96 (1996). W. WISDEN and S.J. MOSS, Biochem. Sot. Trans. 25 820-824 (1997). C.N. CONNOLLY, J.M. UREN, P. THOMAS, G.H. GORRIE, A. GIBSON, T.G. SMART and S.J. MOSS, Mol. Cell. Neurosci. 13 259-271 (1999). C.N. CONNOLLY, J.R.A. WOOLTORTON, T.G. SMART and S.J. MOSS, Proc. Nat. Acad. Sci. U.S.A. 93 9899-9904 (1996). J.R.A. WOOLTORTON, B.J. MCDONALD, S.J. MOSS and T.G. SMART, J. Physiol. 505.3 633-640 (1997). P.M. TAYLOR, P. THOMAS, G.H. GORRIE, C.N. CONNOLLY, T.G. SMART and S.J. MOSS, J. Neurosci., in press. S.-C. LIU, L. PARENT, R.J. HARVEY, M.G. DARLISON, and E.M. BARNES, JR., Eur. J. Pharmacol. 354 253-259 (1998). G.H. GORRIE, Y. VALLIS, A. STEPHENSON, J. WHITFIELD, B. BROWNING, T.G. SMART and S.J. MOSS, J. Neurosci. 17 6587-6596 (1997). J.D. MIRANDA and E.M. BARNES, JR., J. Biol. Chem. 272 16288-16294 (1997). P.A. CALKIN and E.M. BARNES, JR., J. Biol. Chem. 269 1548-1553 (1994). Q. WAN, Z.G. XIONG, H.Y. MAN, C.A. ACKERLEY, J. BRAUNTON, W.Y. LU, L.E. BECKER, J.F. MacDONALD and Y.T. WANG, Nature 388 686-690 (1997). Z. NUSSER, N. HAJOS, P. SOMOGYI and I. MODY, Nature 395 172-177 (1998). J.-M. FRITSCHY, D.K. JOHNSON, H. MGHLER and U. RUDOLPH, Neurosci. Lett. 249 99-102 (1998). Z. NUSSER, W. SIEGHART, D. BENKE, J.-M. FRJTSCHY and P. SOMOGYI, Proc. Nat. Acad. Sci. USA 93 11939-l 1944 (1996). M. JAREB and G. BANKER, Neuron 20 855-867 (1998). C. ESSRICH, M. LGREZ, J.A. BENSON, J.-M. FRITSCHY and B. LUSCHER, Nature Neurosci. 1 563-571 (1998).
23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35.
GABA, Receptor Traftkking
Vol. 66, No. 12,2OMl
J. KIRSCH, J. KUHSE, and H. BETZ, Mol. Cell Neurosci. 6 450-461 (1995). G. FENG, H. TINTRUP, J. KIRSCH, M.C. NICHOL, J. KUHSE, H. BE’IZ and J. R. SANE& Science 282 1321-1324 (1998). H. WANG, F.K. BEDFORD, N.J. BRANDON, S.J. MOSS and R.W. OLSEN, Nature 397 69-72 (1999). M.H.J. TEHRANI and E.M. BARNES, JR., J. Pharmacol. Exp. Ther. 283 384-390 (1997). M.H.J. TEHRANI, B.J. BAUMGARTNER and E.M. BARNES, JR, Brain Res. 776 195-203 (1997). R. CHAPPELL, O.F. BUENO, X. ALVAREZ-HERNANDEZ, L.C. ROBINSON and N.J. LEIDENHEIMER, J. Biol. Chem. 273 3259532601 (1998). T.J. JESS, C.M. BELHAM, F.J. THOMSON, P.H. SCOTT, R.J. PLEVIN and G.W. GOLD, Cell Signal. 8 297-304 (1996). J.D. JOHNSTON, S.A. PRICE and D.R. BRISTOW, Brit. J. Pharmacol. 124 1388-1340 (1998). C.R. HOPKINS, A. GIBSON, M. SHIPMAN, D.K. STRICKLAND, and IS. TROWBRIDGE, J. Cell. Biol. 125 1265-1274 (1994). R.L. KLEIN, M.P. MASCIA, P.C. HARKNESS, K.L. HADINGHAM, P.J. WHITING and R.A. HARRIS, J. Pharmacol. Exp. Ther. 274 1484-1492 (1995). H.Y. KIM, D.W. SAPP, R.W. OLSEN and A.J. TOBIN, J. Neurochem. 612334-2337 (1993). L. ELSTER, G.H. HANSEN, B. BELHAGE, J.M. FRITSCHY, H. MdHLER and A. SCHOESBOE. Int. J. Dev. Neurosci. 13,417-428 (1995). R.J. LEFKOWITZ, J. Biol. Chem. 273 18677-18680 (1998).