GABA Receptor ρ1 Subunit Interacts with a Novel Splice Variant of the Glycine Transporter, GLYT-1*

Ionotropic γ-aminobutyric acid (GABAA and GABAC) receptors mediate fast synaptic inhibition in the central nervous system. GABACreceptors are expressed predominantly in the retina on bipolar cell axon terminals, and are thought to mediate feedback inhibition from GABAergic amacrine cells. Utilizing the yeast two-hybrid system, we previously identified MAP1B as a binding partner of the GABAC receptor ρ1 subunit. Here we describe the isolation of an additional ρ1 interacting protein: a novel C-terminal variant of the glycine transporter GLYT-1. We show that GLYT-1 exists as four alternatively spliced mRNAs which encode proteins expressing one of two possible intracellullar N- and C-terminal domains. Variants containing the novel C terminus efficiently transport glycine when expressed in COS cells, but with unusual kinetics. We have confirmed the interaction between the novel C terminus and ρ1 subunit and demonstrated binding in heterologous cells. This interaction may be crucial for the integration of GABAergic and glycinergic neurotransmission in the retina.

In the central nervous system, fast inhibitory neurotransmission is mediated predominantly by the activation of ionotropic ␥-aminobutyric acid (GABA) 1 receptors and glycine receptors localized to synaptic sites. The synaptic activity of these transmitters is terminated by sodium-dependent transport into neurons or glial cells in the vicinity of the synapse. Two categories of ionotropic GABA receptor are expressed in the central nervous system: GABA A and GABA C receptors (1)(2)(3)(4). Both are GABA-gated chloride channels, but are classified as different subtypes because of their distinct pharmacological and biophysical properties, as well as expression patterns. GABA A receptors are expressed throughout the central nervous system (1,2), whereas GABA C receptors are expressed predominantly in the retina, with lower levels detected in other central nerv-ous system regions (5)(6)(7). GABA C receptors are enriched on bipolar cell axon terminals where they receive GABAergic input from amacrine cells (8,9). Activation of these receptors inhibits glutamate release onto retinal ganglion cells (9 -11), and is likely to be important in tuning the temporal resolution and spatial contrast of responses to visual stimuli in the retina (12).
GABA C receptors are thought to be formed from subunits which are localized to retinal bipolar cells (13). Subunits expressed in Xenopus oocytes exhibit the same pharmacological and electrophysiological properties as GABA C receptors studied in the retina (8,14,15). Three subunits have been cloned, and each forms functional homo-oligomeric receptors when expressed in Xenopus oocytes (12,13). However, there is pharmacological and biochemical evidence to suggest that 1 and 2 subunits can also form functional hetero-oligomers (16,17). In contrast, GABA A receptors are more complex hetero-oligomers, and are formed from ␣ (1-6), ␤ (1-3), ␥ (1-3), ␦, ⑀, and subunits, with the majority of GABA A receptors comprising at least one ␣, one ␤, and one ␥ subunit (2).
Although ligand-gated ion channels expressed in heterologous, non-synaptic systems display essentially native electrophysiological and pharmacological properties, it is becoming increasingly clear that they do not exist simply as isolated macromolecules at synaptic membranes in vivo. Yeast twohybrid screens using intracellular domains of ionotropic glutamate receptor subunits as baits, have isolated various proteins as binding partners for these receptors. These proteins may function to maintain the clustered distribution of the receptor at synaptic sites, or for bringing other important proteins into close proximity for intracellular signaling up-or downstream from the receptor (18 -20). These studies show that the postsynaptic density at excitatory synapses is a complex array of proteins which interact with ionotropic glutamate receptors, either directly or via scaffolding proteins (20).
More recently, similar studies aimed at understanding the architecture of the inhibitory synapse have used the large intracellular loop between transmembrane regions (TM) three and four of ionotropic GABA receptor subunits as yeast twohybrid baits. These studies have identified a novel tubulinbinding protein, GABARAP, which interacts with the ␥2 subunit of GABA A receptors and may function to localize the receptor at synaptic sites (21). Similarly, GABA C receptors have been shown to interact with MAP1B, a microtubule-associated protein which can influence the subcellular distribution of the receptor (22).
To investigate further the identity of GABA C receptor interacting proteins, we have used the 1 subunit intracellular loop between TM3 and TM4 as a yeast two-hybrid bait for additional screening of a retinal library. Here we report that a novel C-terminal splice variant of the glycine transporter GLYT-1 interacts specifically with the 1 subunit bait and show that this interaction occurs in a cellular environment. We describe the tissue distribution and functional characterization of GLYT-1 variants isolated from the retina. Finally, we propose a role for co-localization of these two proteins in the integration of GABA and glycinergic signaling, and the modulation of glutamatergic neurotransmission at bipolar cell-retinal ganglion cells synapses. This is the first report to demonstrate a direct protein-protein interaction between a neurotransmitter receptor and a neurotransmitter transporter.

EXPERIMENTAL METHODS
Yeast Two-hybrid System-Approximately 3 million clones from a bovine retinal cDNA library constructed in the Gal4 activation domaincarrying vector pACTII (CLONTECH) were screened with a bait encoding the intracellular domain of the human 1 subunit cloned in the Gal4-DNA-binding domain-carrying vector pAS2-1 (CLONTECH), as described previously (22). 1, 2, and GABA A receptor intracellular domain bait plasmids were constructed by PCR amplification of cDNA fragments, followed by in-frame ligation into pAS2-1, as described previously (22).
Isolation of bGLYT-1 cDNA Clones-DNA sequencing was carried out on both strands using Sequence TM Version 2.0 (U. S. Biochemical Corp.). The insert of "clone 6" isolated from the yeast-two hybrid screen was 32 P-labeled and used to screen an adult bovine retinal cDNA library (Stratagene) by hybridization as described by Blondel et al. (23). A single clone, designated C6pA1, was obtained that contained a 1779-bp open reading frame similar to mGLYT-1, followed by 96 bp of 3Ј-untranslated sequence and a poly(A) tail. To determine if the bovine retina expressed GLYT-1 variants containing the previously characterized C terminus (D, Fig. 2) retinal cDNA using a bovine specific upstream primer located in the core region (GenBank U52689) 5Ј-CAGC-CATTGTGGATGAGGTAG-3Ј, and a GLYT-1 downstream primer to a region of 3Ј-untranslated sequence common to human and mouse (Gen-Bank X67056) 5Ј-TCATATCCGGGAGTCCTGGAA-3Ј. Amplification was using standard conditions (24) and PCR products were blunt-end cloned into the HincII site of pGEM-4Z (Promega), and sequenced. All clones contained a 667-bp insert encoding the bovine homologue of the mGLYT-1 C terminus.
Additional 5Ј end sequences were isolated by ligation anchor (LA-) PCR of bovine retinal cDNA as described by Trout et al. (25), using a bGLYT-1 downstream primer (GenBank U52689) 5Ј-GCGATGCAGAT-GACCACGTTGTAGTA-3Ј, and an upstream primer specific to the anchor sequence. Bovine retinal cDNA was amplified, blunt-end ligated into the HincII site of pGEM-3Z, and sequenced. Sequences were confirmed in two independent LA-PCR experiments.
Expression Pattern of GLYT-1 Variants by RT-PCR-Expression of mRNA encoding GLYT-1 variants in retina, kidney, hypothalamus, or adrenal was examined by RT-PCR using specific primer pairs. These primers amplified a sequence common to all forms of bGLYT-1 (universal primers), or were specific for transcripts containing N termini "A" or "B", or C termini "D" or "F," where one primer of a pair used was specific for a terminus, and the other primer used was to sequence in the "core" region (see Fig. 2). The up-and downstream primer sequences (5Ј to 3Ј) and the length of their products (in bp) are: universal (210) CTGTGC-CACCAGTGTCTATGCC and CCCAGCAAGATGAGCATGAA; N terminus A (235) TGGACAGGGAGTCCAGAGCC and CGCATGATGAAG-TAGGGGAAC; N terminus B (257) GGATGGCGGCGGCTCAGGGA and CGCATGATGAAGTAGGGGAAC; C terminus D (258) TCATG-GCTCTGTCCTCTGTCAT and GGCTGGAGCCATTACTGCC; C terminus "F" (262) CTTCCAGGACATCCAGATGATG and GGACACACGGA-CACTCCAGG. RT-PCR of cytoplasmic ␤-actin was used as a amplification control. The primers for ␤-actin were (5Ј to 3Ј) CCAGAG-CAAGAGAGGCATCCT and CCAGGTCCAGACGCAGGATG, generating a 357-bp PCR product. Tubes containing 100 pg of the relevant bGLYT-1 expression construct DNA and tubes which did not have DNA added were used as positive and negative controls, respectively. PCR products were separated on a 5% polyacrylamide gel, dried, and exposed to a PhosphorImager plate for 14 h. Plates were scanned and analyzed using a PhosphorImager and ImageQuant software (Molecular Dynamics).
bGLYT-1 Eucaryotic Expression Constructs-Full-length expression constructs of GLYT-1 variants were constructed as follows: LA-PCR products containing 80 bp of 5Ј-untranslated and the N terminus A or B were ligated to the C6pA1 clone (containing C terminus E) and subcloned into the eucaryotic expression vector pCB6, to obtain expression constructs GLYT-1E and -1F. Expression constructs GLYT-1A and -1B were constructed by removing part of the core region and C terminus E from bGLYT-1E and -1F, respectively, and replacing them with the corresponding region from PCR product containing C terminus D.
GLYT-1 Expression and Radioflux Assay-COS-7 cells at 50% confluence in 60-mm dishes were transfected with 7 g of bGLYT-1A, -1B, -1E, or -1F DNA using 30 l of Lipofectin (Life Technologies, Inc.). Twenty-four h after transfection cells were collected by trypinsization, plated onto two 100-mm dishes, and grown at 37°C, 5% CO 2 in high glucose Dulbecco's modified Eagle's medium with 400 g/ml G418 (Life Technologies, Inc.), 100 units/ml penicillin G, 100 g/ml streptomycin, and 10% fetal bovine serum. Medium was changed every third day for 2 weeks. Individual G418-resistant colonies were selected and used to establish independent stable cell lines for each variant. Glycine transport activity was measured on cells at passages 2-6. GLYT-1 activity was measured by radioflux assay. H]glycine and uptake media with equimolar amounts of either choline or n-methylglucamine substituted for sodium, or equimolar amounts of acetate and gluconate substituted for chloride. Wells were incubated for 3 min in ion-substituted or sodium uptake/inhibitor media before the addition of radiolabel. After incubation wells were washed three times with cold choline uptake media and solubilized in 600 l of 1% SDS. Concentration experiments were run on three cell passages in duplicate or quadruplicate for all other experiments. Protein concentrations were determined using a modified Lowry assay (Bio-Rad, DC protein kit). Nonspecific [ 3 H]glycine uptake was determined in cells transfected with vector alone and subtracted from the total uptake to yield specific uptake.
Overlay Assay-In vitro overlay assays were performed as described (22). GST fusion proteins were produced by PCR cloning of 1 cDNA fragments into pGEX-2TK (Amersham Pharmacia Biotech) for 32 P radiolabeling using primers 5Ј-CGTGGATCCGAGTATGCGGCCGT-CAAC-3Ј and 5Ј-GAATTCTCACCTGGAGTATTTATCAAT-3Ј. GSTclone 6 was produced by excision from pAS2-1 using BamHI and EcoRI, and ligation into pGEX-4T3 (Amersham Pharmacia Biotech). Subsequent purification of GST fusions was carried out as described previously (26). 10 g of GST and GST-clone 6 were separated by SDS-PAGE, transferred to nitrocellulose membrane, and subjected to a guanidine denaturing-renaturing process in overlay buffer (10 mM HEPES, pH 7.5, 70 mM KCl, 5 mM EDTA). After blocking by incubation in overlay buffer containing milk, the filter was incubated with 32 Plabeled GST-1 overnight at 4°C. After washing, bound 32 P-labeled 1 was detected by autoradiography.
Construction of myc-tagged 1 Subunit Construct-Addition of the myc tag was carried out as described (27), using the mutagenic primer: COS Cell Transfection and Immunofluorescence-COS cells were cultured at 37°C, 5% CO 2 , in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) containing 10% fetal calf serum (Life Technologies, Inc.), 2 mM L-glutamine (Life Technologies, Inc.), penicillin, and streptomycin. Cells were grown to 50 -70% confluency, and transfected by electroporation. 10 g of cytomegalovirus promoter-driven expression constructs were used per transfection. After electroporation, cells were plated on coverslips coated with poly-L-lysine for immunofluorescence, or on plastic dishes for biochemical analysis, and incubated overnight. Immunofluorescence staining was carried out as described previously (27) except permeabilization was carried out with 0.1% Nonidet P-40.
Affinity Purification (Pull-down) Assays-After overnight expression, 1Mry-transfected COS cells were lysed in 20 mM HEPES, pH 7.5, 100 mM KCl, 10 mM EDTA, 1% Triton X-100, plus protease inhibitors phenylmethylsulfonyl fluoride, aprotinin, leupeptin, antipain, and pepstatin. Nuclei were removed by centrifugation. The lysate was incubated with 25 g of GST fusion proteins bound to glutathione-agarose beads for 2 h rotating at 4°C. After washing four times in lysis buffer, bound proteins were separated by SDS-PAGE, and detected by Western blotting.
[ 35  After washing four times in lysis buffer, the first two washes supplemented with 0.5 M NaCl, bound proteins were separated by SDS-PAGE, and detected by autoradiography.
Co-immunoprecipitation-After overnight expression, co-transfected COS cells were lysed, and treated as described above using 5 g of anti-GLYT-1E/F, or nonimmune IgG. Bound proteins were detected by Western blotting with anti-Myc 9E10.

Isolation of a Novel GLYT-1 Variant as a 1 Subunit Binding
Partner by Yeast Two-hybrid Screening-Using the yeast twohybrid system with the large intracellular domain between TM3 and TM4 of the 1 subunit as a bait we screened a bovine retinal cDNA library to identify GABA C receptor interacting proteins. In addition to the microtubule-associated protein MAP1B (22), we isolated a novel clone, termed clone 6 ( Fig. 1A).
To determine if the interaction between clone 6 and 1 is specific, or a general feature of ionotropic GABA receptor subunits, we tested clone 6 against intracellular domains from a range of GABA receptor subunits in the yeast two-hybrid system. Fig. 1B shows that the interaction is specific to the GABA C receptor 1 subunit, since no interaction is seen with GABA A receptor subunits nor the GABA C receptor 2 subunit.
Hybridization screening of an adult bovine retinal cDNA library with a 32 P-labeled insert from clone 6 isolated a clone containing a 1779-bp open reading frame, termed C6pA1. The deduced amino acid (a.a.) sequence of this clone showed 94.9% identity to the mouse glycine transporter mGLYT-1A in a region corresponding to the N-terminal and core regions (28). The C-terminal of C6pA1, identical in sequence to clone 6, shares only 10% a.a. identity with the mGLYT-1 C terminus. Taken together these data indicate that C6pA1 is a bovine homolog of GLYT-1 with a novel C terminus (Fig. 2). Bovine retinal cDNA clones containing a C terminus homologous to mGLYT-1, 93.1% identity, were obtained by RT-PCR. Since clone C6pA1 contained only a few bases of 5Ј-untranslated sequence, we used ligation anchor PCR of bovine retinal cDNA to ensure we had isolated the complete open reading frame. We identified three species of mRNA encoding bGLYT-1 5Ј ends, two encoding N terminus A with different 5Ј-untranslated regions, and one encoding N terminus B. The complexity of bGLYT-1 5Ј splice products is similar to that reported for the mouse GLYT-1 (29). The bovine N termini are termed A and B, according to the mGLYT-1 nomenclature, and the C termini are termed D and E. Bovine clones homologous to mGLYT-1C, an alternatively spliced N terminus, were not found. Additional PCR experiments using upstream primers specific for either N terminus and downstream primers specific for either C terminus showed the presence of transcripts that give rise to four potential GLYT-1 variants. These variants are termed GLYT-1A and -1B if they contain C-terminal D in conjunction with N-terminal A or B, respectively, or GLYT-1E and -1F if they contain C-terminal E in conjunction with N-terminal A or B, respectively. Thus the intracellular domain of 1 subunit can interact with the C-terminal intracellular domain of a novel splice variant of the glycine transporter, termed GLYT-1E/F. Expression of Alternatively Spliced Forms of GLYT-1 mRNA-The tissue distribution of GLYT-1 mRNAs containing N termini A or B, or C termini D or E was examined by RT-PCR (Fig. 3). The overall level of GLYT-1 transcript expression, as determined using primers to the core region, was similar in adrenal, hypothalamus, kidney, and retina (Fig. 3, universal). The expression levels of N-terminal A and C-terminal D containing transcripts were also similar in all tissues. By contrast, N-terminal B containing transcripts were present at higher levels in hypothalamus and retina than adrenal and kidney. Transcripts containing the 1-interacting C-terminal E (as determined by the presence of the expected 263-bp band) occurred at highest levels in the hypothalamus, but also at significant levels in the retina and kidney, and was absent in the adrenal. Additional PCR products were obtained with amplification of C-terminal E primers: the sequences of these products were not determined.
Confirmation of 1-GLYT-1E/F Interaction by Overlay Assay-To demonstrate a direct interaction between the 1 intracellular domain and GLYT-1E/F C terminus (clone 6), GST fusion proteins were constructed encoding both of these polypeptides for use in overlay assays in which no other proteins are present. GST-1 was constructed in pGEX-2TK, which has an Arg-Arg-Ala-Ser motif engineered just C-terminal to GST, allowing in vitro labeling of the fusion protein with [␥-32 P]ATP by the catalytic subunit of PKA. Gel overlay assays indicated that 32 P-labeled GST-1 bound to GST-clone 6, but not to GST alone, demonstrating that 1 intracellular domain binds directly to GLYT-1E/F C terminus, and no further factor is required for the interaction (Fig. 4).
Functional Characterization of Different Forms of GLYT-1-Constructs encoding GLYT-1A, -1B, -1E, and -1F were transfected into COS cells and the kinetic and pharmacological characteristics of glycine uptake measured (Fig. 5). Variation in the intracellular N-terminal domain had no significant effect on the kinetic properties of GLYT-1, as demonstrated by the similar properties of GLYT-1A compared with B, and of GLYT-1E compared with F (data not shown). However, variation in the C-terminal domain has a dramatic effect on the kinetic properties of the expressed transporter. GLYT-1B has a K m for glycine of 59 Ϯ 8 M and a V max of 4.7 Ϯ 0.5 nmol/mg/ min. In contrast, GLYT-1F showed complex nonlinear kinetics which cannot be fit using a single Michaelis-Menten equation. A K m could not be reliably estimated for GLYT-1F, yet it had an estimated V max of 1.8 Ϯ 0.2 nmol/mg/min. The time course of 10 M [ 3 H]glycine uptake by GLYT-1B was linear for the first 10 min, whereas uptake by GLYT-1F was linear over at least 1 h. The average initial velocity of glycine uptake was 95 Ϯ 10 pmol/mg/min by cells expressing GLYT-1B and 24 Ϯ 4 pmol/ mg/min for GLYT-1F. Although variation in the C-terminal of GLYT-1 affects kinetic properties, it does not alter the sodium or chloride ion dependence of glycine uptake by GLYT-1E/F. No specific uptake was detected when sodium was substituted with either choline or NMG, or when chloride was substituted with acetate or gluconate. Nor do the pharmacological properties differ between bGLYT-1 variants. Specific uptake of radiolabeled glycine by all four variants was inhibited 85-90% by unlabeled glycine and sarcosine, whereas L-alanine, GABA, and MeAIB had no effect. These data are consistent with the properties of other GLYT-1 transporters. Proline, reportedly an inhibitor of the human GLYT-1 (25), has no effect on bovine GLYT-1 variants.
Construction and Characterization of Anti-GLYT-1E/F-specific Antibodies-Polyclonal antibodies were raised against a GST fusion protein of clone 6 (GLYT-1E/F C-terminal intracellular domain) and tested for specificity in a number of assays. COS cells transfected with a construct encoding full-length GLYT-1E were prepared for immunofluorescence and stained with GLYT-1E/F antiserum. Fig. 6A shows some plasma membrane and predominantly endoplasmic reticulum staining in permeablized GLYT-1E transfected cells, but not in untransfected cells. The staining was abolished by pre-absorbtion of the antiserum with GST-clone 6, the polypeptide used to raise the antibodies. No signal was detected in non-permeabilized transfected cells stained with anti-GLYT-1E/F, suggesting that the C terminus is intracellular and not accessible to the anti- serum. Fig. 5 shows specific uptake of [ 3 H]glycine by intact GLYT-1E-transfected COS cells providing evidence of correct processing and cell surface expression of the transporter. To further analyze the antiserum, we metabolically labeled GLYT-1E-transfected COS cells with [ 35 S]methionine, extracted protein with Triton X-100, and immunoprecipitated using anti-GLYT-1E/F bound to protein A-Sepharose. Proteins were separated by SDS-PAGE and visualized by autoradiography (Fig. 6B). Anti-GLYT-1E/F recognizes a protein of ϳ65 kDa in GLYT-1E transfected cells, but not in untransfected cells. The band was not seen when the antiserum was pre-absorbed with GST-clone 6. Unfortunately, this antiserum was unsuitable for Western blotting.
Interaction of 1 and GLYT-1E/F in Heterologous Cells-GST-clone 6 was also used in affinity purification (pull-down) assays to analyze its binding to full-length 1 from cell lysate. COS cells were transfected with a construct encoding fulllength 1 subunit, modified by addition of the Myc epitope between residues 4 and 5 at the N terminus of the protein.
Previous studies have demonstrated that this addition is functionally silent for a range of GABA A receptor subunits (27). Staining of 1 Myc -transfected COS cells with anti-Myc 9E10 in both non-permeabilized and permeabilized conditions, shows that 1 Myc reaches the cell surface efficiently (Fig. 7A). In permeabilized cells, endoplasmic reticulum staining predominates, due to the very high levels of protein expression in COS cells. No signal was seen in untransfected cells stained with anti-Myc 9E10. Western blotting analysis using anti-Myc 9E10 shows a 60-kDa band present in 1 Myc transfected cells, that is not seen in untransfected cells (Fig. 7B). To analyze the interaction between 1 Myc and the GLYT-1E/F C terminus, 1% Triton X-100 extracts of 1 Myc -transfected COS cells were incubated with GST-clone 6 bound to glutathione-agarose beads. As a positive control, 1 Myc binding to GST-MAP1B (1-binding region) is also shown (22). Fig. 7C shows that 1 Myc binds to GST-clone 6 as well as GST-MAP1B, but not to GST alone.
To demonstrate that both full-length proteins interact in a cellular environment, we co-expressed 1 Myc and GLYT-1E in COS cells. After permeabilization with 0.1% Nonidet P-40 these cells were co-stained with anti-GLYT-1E/F and anti-Myc 9E10 and analyzed by immunofluorescence. Fig. 8A demonstrates that GLYT-1E shows striking co-localization with 1 Myc in the endoplasmic reticulum, but the subcellular localization of either protein is unaffected. Immunoprecipitation with anti-GLYT-1E/F was carried out on a Triton X-100 extract of cotransfected cells, and the presence of 1 Myc determined by Western blotting using anti-Myc 9E10. As shown earlier, anti-GLYT-1E/F efficiently immunoprecipitates GLYT-1E from transfected COS cells (Fig. 6B). Fig. 8B shows that 1 Myc is present in anti-GLYT-1E/F immunocomplexes, but not nonimmune IgG complexes. These data demonstrate that these two full-length proteins interact in COS cells. DISCUSSION We have identified a novel C-terminal variant of the sodiumdependent glycine transporter GLYT-1, which specifically interacts with the 1 subunit of GABA C receptors in recombinant systems. Expression of GLYT-1 variants containing the novel C terminus produces glycine transporters with identical pharmacological, but different kinetic properties to GLYT-1 variants containing the previously characterized C terminus. We show that the novel C-terminal variant is expressed in the retina, the predominant site of GABA C receptor expression. In the yeast two-hybrid system, the novel C terminus specifically interacts with the 1 subunit, showing no interaction with 2 or GABA A receptor subunits. Gel overlay assays show that this interaction requires no other factors, and affinity purification and immunoprecipitation show the interaction occurs in a cellular environment. Thus, we provide the first evidence of a direct interaction between a neurotransmitter receptor and a neurotransmitter transporter.
GLYT1-E/F Represents a Novel Variant of Glycine Transporter-Glycine transporters are members of the Na ϩ /Cl Ϫ -dependent family of neurotransmitter transporters, which have a proposed membrane topology of 12 TMs and intracellular N and C termini (30). Alternative splicing of the GLYT-1 N terminus has been reported for human, rat, and mouse homologues. These variants differ in their pattern of expression, but produce transporters with identical kinetic and pharmacological properties (28,31,32). Our results show that the same is true for bovine GLYT-1 N-terminal variants.
C-terminal splicing effects both the pattern of expression and kinetic properties of the transporter, but has no effect on either its ion dependence or pharmacology. Variants containing the novel C terminus, GLYT-1E and -1F, have complex nonlinear kinetics that cannot be fit with a single Michaelis-Menten equation, whereas the kinetics of variants containing the known C terminus, GLYT-1A and -1B, are linear and fit by a single Michaelis-Menten equation. The basis for the nonlinear kinetics displayed by GLYT-1E/F is unknown, but may be due to the presence of multiple binding sites with different substrate affinities or a single site whose affinity is altered by modulators. Alternatively, this complex response could arise from a combination of inward and outward transport, where outward flux is more significant for the GLYT-1E/F variants than for GLYT-1A/B. These data also suggest that the regions of the GLYT-1 protein necessary for determining its ion dependence and pharmacological properties are located outside of the C terminus.
When compared with the previously characterized GLYT-1A/B C terminus, the novel, 1 interacting C terminus is 45 a.a. residues shorter. This C terminus shares little a.a. identity with the GLYT-1A/B C terminus and does not contain an amphipathic region corresponding to TM12. Previously it was shown that progressive truncation of the GLYT-1A/B C terminus produces a progressive decrease in transport activity (33). Expression of the largest C-terminal truncation in COS cells failed to localize transporters to the plasma membrane, and reconstitution experiments showed it was non-functional. These data suggest the C terminus is important for normal function and protein targeting. Despite the sequence divergence and apparent lack of TM12, expression of GLYT-1E/F in COS cells produces a functional glycine transporter indicating the novel C terminus provides the information necessary for function and cell surface expression. Immunocytochemistry of GLYT-1E/F transfected COS cells clearly indicates that this novel C terminus is located intracellularly, suggesting that either this C terminus crosses the plasma membrane using a non-amphipathic domain, or that the proposed topology for these transporters is incorrect.
GLYT-1E/F Interacts with 1 Subunit of GABA C Receptors-Yeast two-hybrid studies using glutamate or GABA A receptor subunit intracellular domains as baits have identified protein interactions which mediate signaling functions and/or affect receptor localization (18 -20). Previously we reported the specific interaction between GABA C receptor 1 subunit and microtubule-associated protein 1B (MAP1B), which can influence the subcellular localization of the receptor (22). In this study we show that a novel GLYT-1 C-terminal variant and the large intracellular loop of the GABA C receptor 1 subunit interact in a cellular environment but do not have a significant effect on the localization of either protein. Taken together these data show that the intracellular domain between TM3 and TM4 of the GABA C receptor 1 subunit interacts directly with at least two proteins, namely MAP1B and GLYT-1E/F. It is currently unknown if the interaction of 1 subunit with these two proteins occurs concurrently or sequentially, and if the interactions are cooperative or mutually exclusive. Concurrent interaction between 1, MAP1B, and GLYT-1E/F may serve as a means of targeting or anchoring this transporter to the synapse. Further studies are necessary to determine if the interaction between 1 and GLYT-1E/F influences the functional properties of either protein. For example, it would be of great interest to investigate whether the single-channel characteristics of the GABA C receptor are affected by the binding, and possibly even the activity of GLYT-1E/F.

What Possible Role Is There for a Co-localization between GLYT-1E/F and GABA C Receptors in Retinal Physiology?-
Retinal bipolar cells receive glutamatergic inputs from photoreceptors at their dendrites, GABA-/glycinergic inputs from amacrine cells at their axon terminal, and deliver glutamatergic output to retinal ganglion cells (34). GABA C receptors are expressed predominantly in bipolar cell axon terminals, and colocalization with GLYT-1E/F in this region could effect communication between bipolar and ganglion cells by modulating: 1) the amount of glutamate released from bipolar cells, and 2) the responsiveness of ganglion cell NMDA receptors by controlling the synaptic concentration of glycine, an NMDA receptor coagonist (35).
Bipolar cells are non-spiking, i.e. they do not carry action potentials, and the amount of glutamate released from the bipolar cell terminal is graded according to the extent of depolarization. GABA C receptors are ligand-gated chloride channels that hyperpolarize the cell when activated, causing a reduction in Ca 2ϩ influx and a consequent decrease in glutamate release (11,12). GLYT-1 mediated glycine influx is electrogenic, resulting from the net movement of one positive charge into the cell per glycine molecule, causing the cell to depolarize (36). The magnitude of transporter-mediated depolarization depends upon both the rate of influx and transporter number. GABA C receptor and GLYT-1E/F transporter colocalization in bipolar cell axon terminals could thus modulate the membrane potential bi-directionally and regulate glutamate release.
NMDA receptor channel gating requires allosteric glycine binding in addition to glutamate (35). GLYT-1E/F localization to bipolar cell terminals would control the glycine concentration at the synapse. GLYT-1 mediated influx lowers the synaptic concentration of glycine, resulting in a reduction in NMDA receptor channel gating (37). In addition, Na ϩ -dependent neurotransmitter transporters can function in reverse to release transmitter into synapses (38). Depolarization and raised internal [Na ϩ ] of the bipolar terminal could result in transporter-mediated glycine release, as well as conventional vesicular release of glutamate. Concurrent release of glutamate and glycine into the synapse would favor NMDA receptor channel activation. When the bipolar cell terminal hyperpolarizes, glutamate release is diminished and transporter-mediated glycine uptake is favored. This would lower the concentration of glutamate and glycine in the synaptic cleft, reducing NMDA receptor activation. Colocalization of GABA C receptors and GLYT-1E/F in bipolar cell axon terminals may play a role in the integration of GABA and glycinergic signaling and the modulation of glutamatergic neurotransmission between bipolar cells and retinal ganglion cells.
It should be noted that GLYT-1 does not transport GABA (this study and Ref. 39), and also that homomeric 1 GABA C receptors are particularly unresponsive to glycine, showing a 10,000-fold lower affinity to glycine compared with GABA, and a very low response to even high doses of glycine (40). However, these authors report a significant potentiating effect of glycine on GABA-activated 1 receptor currents. It is conceivable that colocalization of GLYT-1E/F with 1 subunits is important for the regulation of glycine concentrations in the vicinity of the GABAC receptor and thus the fine-tuning of GABAC-mediated responses.
In conclusion, we have isolated a novel GLYT-1 C-terminal variant that interacts with GABA C receptors. We show that heterologous expression of variants containing the novel C terminus produce functional glycine transporters pharmacologically identical but kinetically different from known GLYT-1 transporters, and discuss the implications this novel variant has on modeling transporter topology. This is the first report to demonstrate a direct protein-protein interaction between a neurotransmitter receptor and a neurotransmitter transporter.