The GLYT1 (glycine transporter-1) regulates both glycinergic and glutamatergic neurotransmission by controlling the reuptake of glycine at synapses. Trafficking to the cell surface of GLYT1 is critical for its function. In the present paper, by using mutational analysis of the GLYT1 C-terminal domain, we identified the evolutionarily conserved motif R575L576(X8)D585 as being necessary for ER (endoplasmic reticulum) export. This is probably due to its capacity to bind Sec24D, a component of the COPII (coatomer coat protein II) complex. This ER export motif was active when introduced into the related GLYT2 transporter but not in the unrelated VSVG (vesicular-stomatitis virus glycoprotein)–GLYT1 protein in which this motif was mutated but was not transported to the plasma membrane, although this effect was rescued by co-expressing these mutants with wild-type GLYT1. This behaviour suggests that GLYT1 might form oligomers along the trafficking pathway. Cross-linking assays performed in rat brain synaptosomes and FRET (fluorescence resonance energy transfer) microscopy in living cells confirmed the existence of GLYT1 oligomers. In summary, we have identified a motif involved in the ER exit of GLYT1 and, in analysing the influence of this motif, we have found evidence that oligomerization is important for the trafficking of GLYT1 to the cell surface. Because this motif is conserved in the NSS (sodium- and chloride-dependent neurotransmitter transporter) family, it is possible that this finding could be extrapolated to other related transporters.
- coatomer coat protein II (COPII)
- endoplasmic reticulum (ER)
- glycine transporter-1 (GLYT1)
- neurotransmitter transporter
- protein traffic
Beside its well-characterized role as an inhibitory neurotransmitter, glycine is a co-agonist of NMDAR [NMDA (N-methyl-D-aspartate) receptor], which is necessary for ion channel opening and, probably, for the internalization of the receptor from the cell surface [1,2]. While it was initially believed that the concentration of glycine in the synaptic cleft would be high enough to saturate the glycine sites on NMDAR, recent pharmacological and electrophysiological evidence indicates that due to the activity of the GLYT1 (glycine transporter-1), this might be not the case. GLYT1 seems to fulfil a dual role in neurotransmission. First, it is highly expressed in glycinergic areas of the nervous system where it is predominantly found in glial cells and is associated with glycinergic inhibitory neurotransmission [3,4]. Secondly, GLYT1 is present in neuronal elements closely associated with the glutamatergic pathways throughout the brain . Functional and anatomical evidence strongly support a role of GLYT1 in regulating NMDAR-mediated neurotransmission [5–12].
The mechanisms responsible for the insertion of GLYT1 into glutamatergic or glycinergic synapses are unknown. However, recent studies ( and references cited therein) indicate that the movement of transporters within the cell is highly organized and that a number of ancillary proteins control their intracellular trafficking, interacting with targeting motifs in the sequence of the transporter. Indeed, GLYT1, like other neurotransmitter transporters, is asymmetrically distributed in polarized cells [13,14]. The asymmetrical distribution of NSSs (sodium- and chloride-dependent neurotransmitter transporters) requires a number of steps that commence with their efficient exit from the ER (endoplasmic reticulum). This is followed by sorting processes in the Golgi complex, insertion into the plasma membrane and retention of the transporter at functional synaptic sites. Moreover, the amount of transporter in the plasma membrane is also regulated by endocytosis and recycling mechanisms. Among these steps, export of polytopic proteins from the ER, such as NSSs, is poorly characterized. Such an ER export does require specific signals in the amino acid sequence and post-translational modifications that probably include glycosylation and oligomerization. These mechanisms might help to pass the stringent quality control in the ER, which rejects misfolded proteins (reviewed in ). We have shown that half of the C-terminal domain of GLYT1 that lies closest to the plasma membrane is necessary for ER export of this protein . In addition, mutations in the PDZ-interacting motif Ser-Arg-Ile present in the C-terminus of GLYT1 delays its delivery to the plasma membrane . Similarly, ER export of the DAT (dopamine transporter) and the GAT1 [GABA (γ-aminobutyric acid) transporter 1] is also dependent on the co-operation of two discontinuous segments located in the C-terminal domain: one corresponding to the last three residues (also a PDZ-binding motif) and another to a sequence segment adjacent to the twelfth transmembrane helix [17–19]. Extensive mutagenesis of this second region in DAT revealed that mutating Gly585, Lys590 or Asp600 led to the protein retention in the ER . A further step in the understanding of ER export of this family of transporters has been recently reported Sitte and co-workers  who found that mutations in the sequence R566L567, located in the C-terminus of the GAT1, disturb the interaction of this transporter with Sec24D, a component of the COPII (coatomer coat protein II) complex.
Oligomer formation of newly synthesized transporters has been suggested as an essential step in ER exit. Biochemical experiments including radiation inactivation, chromatography filtration of detergent solubilized proteins, co-immunoprecipitation and cross-linking studies have provided evidence that SERT (serotonin transporter), DAT and the NET (noradrenaline transporter) form oligomeric complexes [22–28]. Moreover, assays using dominant-negative forms of these three transporters have produced data compatible with the existence of transporter oligomers [18,29,30]. Additional evidence for the formation of oligomers for the GAT1, SERT and DAT has also been obtained in living cells by FRET (fluorescence resonance energy transfer) microscopy [25,31]. These transporters not only exist as oligomers in the plasma membrane, but also through the biosynthetic and protein trafficking pathway. Hence, only properly assembled transporters might be able to recruit the COPII needed for ER export. However, biochemical analysis of GLYT1 and GLYT2 expressed in Xenopus oocytes did not find evidence for oligomerization of these two GLYTs in the cell surface, although the existence of intracellular oligomers was reported . Additionally, biochemical studies using purified GLYT2 revealed that this protein exists as a monomer in the pig brain . Since the members of the NSS family seem to be subjected to similar trafficking mechanisms, the existence of two exceptions, like GLYT1 and GLYT2, brings into question the influence of oligomerization on specific NSS trafficking. Alternatively, it is possible that oligomers of GLYT are very labile and readily disrupted, making their detection difficult through standard biochemical analysis.
In the present study, we have studied the export of GLYT1 from the ER, showing the conservation of the export mechanisms along the NSS family. Through a mutational analysis, we have identified an ER export signal in the C-terminus of GLYT1. This signal might be involved in the recruitment of the COPII component Sec24D. In addition, by using FRET microscopy in living cells, we show that GLYT1 can form oligomers along the trafficking pathway that it follows to the plasma membrane, similarly to that described for other members of this transporter family.
[3H]Glycine, glutathione–Sepharose 4B, the pGEX-5X plasmid, protein standards for SDS/PAGE Rainbow markers and ECL® Western blotting detection reagents were all obtained from Amersham Biosciences (Little Chalfont, Bucks., U.K.). The Lipofectamine™ PLUS, Lipofectamine™ 2000 and the pCDNA3 plasmid were purchased from Invitrogen (Carlsbad, CA, U.S.A.), whereas PMSF, the Expand High Fidelity PCR system (Taq polymerase) and all restriction enzymes were from Roche (Mannheim, Germany). The QuikChange® site-directed mutagenesis kit was from Stratagene Cloning Systems (La Jolla, CA, U.S.A.). Nitrocellulose sheets were obtained from Bio-Rad (Richmond, CA, U.S.A.) and fetal calf serum was supplied by Gibco (Paisley, Renfrewshire, Scotland, U.K.). The rabbit anti-calnexin serum was from Stressgen Bioreagents (Victoria, BC, Canada), whereas the goat anti-rabbit and goat anti-mouse coupled with Alexa Fluor® 488 or Alexa Fluor® 555 were from Molecular Probes (Eugene, OR, U.S.A.). Vectashield was obtained from Vector Laboratories (Burlingame, CA, U.S.A.) and the reagents DSS (disuccinyl suberate), BS3, [bis(sulfosuccinimidyl) suberate] and EZ-Link™ Sulfo-NHS-SS-Biotin were from Pierce (Rockford, IL, U.S.A.). The pGEM-T Easy cloning vector was purchased from Promega (Madison, WI, U.S.A.) and the oligonucleotides used were synthesized by Isogen (Utrecht, The Netherlands). All other chemicals were obtained from Sigma Chemical (St. Louis, MO, U.S.A.). The CyPetm and YPetm cDNAs, optimized FRET versions of CFP (cyan fluorescent protein) and YFP (yellow fluorescent protein), respectively, were gifts from Dr Patrick Daugherty (Department of Chemical Engineering, Biomolecular Science and Engineering, University of California at Santa Barbara, Santa Barbara, CA, U.S.A.).
Cell growth and transfections
COS-7 and MDCK cells (Madin–Darby canine kidney cells) (A.T.C.C., Manassas, VA, U.S.A.) were grown at 37 °C and 5% CO2 in high-glucose Dulbecco's modified Eagle's medium supplemented with 10% (v/v) fetal bovine serum. Transient expression in COS-7 or MDCK cells was carried out using Lipofectamine™ Plus or Lipofectamine™ 2000 respectively following the procedures indicated by the supplier. Cells were incubated for 48 h at 37 °C and then used for immunofluorescence, biochemical and/or transport assays.
The various GLYT1 mutants used in the present study were prepared using the rat GLYT1b in pCDNA3 as a template (derived from the rb20 clone and obtained from Dr Richard L. Weinshank, Synaptic Pharmaceutical Corporation, Paramus, NJ, U.S.A.) and the QuikChange® site-directed mutagenesis kit, according to the manufacturer's instructions. The construct GLYT2Δct was generated by PCR, inserting a stop codon at residue 748 of the rat GLYT2 sequence in the reverse primer. Restriction sites for HindIII and BglII were introduced in the forward and reverse primers respectively and the PCR fragment was cloned into the HindIII/BamHI sites of pCDNA3. The GLYT2–GLYT1ct chimaera was made by PCR in two steps. First, the sequence corresponding to residues 1–747 of GLYT2 was amplified and cloned as indicated for GLYT2Δct (avoiding the stop codon in the reverse primer), creating the GLYT2-1/147 pCDNA3 construct. Subsequently, the sequence corresponding to residues 565–638 of GLYT1b was amplified with primers containing restrictions sites for BamHI in the forward primer and for XbaI in the reverse primer. The fragment was cloned into the BglII/XbaI-digested GLYT2-1/147 pCDNA3. The VSVG (vesicular-stomatitis virus glycoprotein)–GLYT1ct chimaera was also made in two steps. First, the sequence encoding residues 1–498 was amplified by PCR with primers containing restriction sites for HindIII and BamHI, and it was then cloned into the equivalent sites of pCDNA3. The GLYT1ct tail was then cloned into the BamHI/XbaI sites of this construct, maintaining the correct reading frame. GST (glutathione transferase)–GLYT1Ct was also cloned using PCR into pGEX-5X as described previously . To insert the YFP or CFP tags into these constructs, cDNAs for the fluorescent proteins were amplified by PCR using pYPet-His or pCyPet-His as the template . The forward primer was GCGCGGATCCACCATGGTGAGCAAAGGCGAAGAG and, while for CyPetm the reverse primer was GCGCGAATTCTTATTTGTACAGTTCGTCCATGCC, the reverse sequence for YPetm was GCGCGAATTCTTATTTGTACAATTCATTCATCCCTC. The PCR products were then cloned into the BamHI/EcoRI sites of pCDNA3. Full-length GLYT1 was amplified to eliminate the stop codon and cloned in front of the fluorescent proteins (HindIII/BamHI) to create the chimaeras GLYT1–YFP and GLYT1–CFP in pCDNA3.
Solubilization and reconstitution in proteoliposomes
Transfected cells from one 10 cm diameter dish were used for each reconstitution experiment. Cells were scraped and collected by centrifugation, and the protein concentration was adjusted to 5–10 mg/ml with PBS (137 mM NaCl, 2.7 mM KCl, 4.3 mM Na2HPO4·7H2O and 1.4 mM KH2PO4, pH 7.3). Cells were solubilized in sodium cholate at a 1:1 detergent/protein ratio. After 10 min on ice, solubilized proteins were reconstituted with asolectin/brain lipids , and proteoliposomes were maintained on ice until use.
Protein concentration was determined using the Bio-Rad protein determination kit using BSA as the standard.
Cross-linking assays and sucrose-gradient velocity centrifugation
Chemical cross-linking was performed using a fraction of the plasma membrane isolated from COS cells or in rat brain synaptosomes isolated as described previously . After incubation at room temperature (25 °C) for 30 min with the cross-linking agents DSS or BS3 (5 mM), the reaction was terminated by adding Tris/HCl (pH 7.4) to 50 mM for 15 min. Synaptosomal proteins were solubilized by incubation in RIPA buffer [150 mM NaCl, 2 mM EDTA, 50 mM Tris/HCl, pH 8.0, 0.3% SDS, 1% NP40 (Nonidet P40) and 0.5% deoxycholate] for 30 min at 4 °C. The samples were centrifuged at 100000 g for 30 min to remove insoluble material and then layered on to a 10 ml 1–15% sucrose gradient on the top of a 1.0 ml 25% sucrose cushion. The gradient was centrifuged using an SW40.1 rotor in an ultracentrifuge (Beckman Instruments, Palo Alto, CA, U.S.A.) for 16 h at 28000 rev./min. After centrifugation, samples were collected from the bottom of the tube in 0.5 ml fractions. Proteins were then precipitated with trichloroacetic acid and resuspended in 100 μl of loading buffer (70 mM Tris/HCl, 2% SDS, 2.5% 2-mercaptoethanol and 10% glycerol, pH 6.8). A sample of 30 μl was analysed by SDS/PAGE and immunoblotting.
COS cells were transfected with an expression vector for Myc–Sec24D and, 2 days later, cells were solubilized in ice-cold lysis buffer (150 mM NaCl, 2 mM EDTA, 50 mM Tris/HCl, pH 8.0, 0.1% SDS, 1% NP40 and 0.5% deoxycholate) for 30 min at 4 °C. The solubilized material was centrifuged at 10000 g for 20 min, and the supernatant was precleared with 100 μl of a 50% (v/v) glutathione–Sepharose bead suspension for 1 h at 4 °C with constant rotation. After preclearing, the supernatants were transferred to a clean tube containing GST, GST–GLYT1ct or the mutated fusion protein GST–GLYT1ctR575A coupled with glutathione–Sepharose and incubated for 1 h at room temperature with constant rotation. Subsequently, the beads were washed twice with ice-cold lysis buffer and three times with PBS. Bound proteins were eluted with 10 mM glutathione in 50 mM Tris/HCl (pH 8.0). Finally, 50 μl of SDS/PAGE sample buffer was added to each sample and proteins were resolved by SDS/PAGE on 10% gels, blotted and probed as indicated above using a primary antibody that recognized the 6×His tag.
Cell surface biotinylation
Cells were plated at 50% confluence on to 60 mm cell culture plates and transfected as indicated above. After 2 days, cell surface proteins were labelled having first washed the cells with ice-cold PBS. The cells were then incubated for 20 min at 4 °C in a 1 ml solution containing the non-permeable Sulfo-NHS-SS-Biotin reagent (1 mg/ml in PBS). The cells were washed with 2 ml of PBS plus 100 mM lysine for 20 min to quench the reagent. After three additional washes with PBS, the cells were lysed with 1 ml of lysis buffer (150 mM NaCl, 5 mM EDTA, 50 mM Hepes/Tris, 0.25% deoxycholate, 1% Triton X-100 and 0.1% SDS, pH 7.4) for 30 min, and the lysate was cleared by centrifugation at 14000 g for 10 min. The biotinylated proteins were finally recovered by incubating the cleared lysate for 2 h at room temperature with streptavidin–agarose beads. After three washes of the beads with 1 ml of lysis buffer, the protein bound to the beads was eluted with 2×Laemmli sample buffer, separated by SDS/PAGE and immunoblotted. Biotinylated GLYT1 was visualized with anti-GLYT1 serum.
Electrophoresis and blotting
SDS/PAGE was performed in 6% (w/v) polyacrylamide gels in the presence of 2-mercaptoethanol. Samples from cross-linking experiments were resolved in gradient polyacrylamide gels (4–7.5% gels). After electrophoresis, the protein samples were transferred to a nitrocellulose membrane in a semidry electroblotting system at 1.2 mA/cm2 for 2 h (LKB) and using transfer buffer containing 192 mM glycine and 25 mM Tris/HCl (pH 8.3). Non-specific binding to the membrane was blocked by incubating the filter with 5% non-fat milk protein in 10 mM Tris/HCl (pH 7.5) and 150 mM NaCl for 4 h at 25 °C. The blot was then probed (overnight at 4 °C) with the diluted primary antibody, which after washing, was visualized using an anti-rabbit or anti-mouse IgG peroxidase-linked secondary antibody. The labelled bands were revealed by ECL (enhanced chemiluminescence) and quantified by densitometry (Molecular Dynamics Image Quant v. 3.0).
Immunofluorescence in cultured cells
MDCK cells grown on poly(L-lysine)-treated glass coverslips were transfected with the corresponding expression vectors using Lipofectamine™ 2000 according to the manufacturer's instructions. The cells were rinsed, 2 days later, with PBS and fixed for 20 min with 4% (w/v) paraformaldehyde in PBS. After washing with PBS, the cells were permeabilized and blocked at room temperature for 1 h in PBS containing 1% BSA and 0.02% digitonin. The cells were incubated overnight at 4 °C with anti-GLYT1  in PBS containing 0.1% BSA and 0.02% digitonin. After three washes with PBS, the cells were then incubated with the goat anti-rabbit secondary antibodies conjugated to Alexa Fluor® 488 (1:250) for 1 h at room temperature. Finally, the cells were washed exhaustively with PBS and the coverslips were mounted in Vectashield. The staining was visualized and the images were captured on a confocal Microradiance (Bio-Rad) coupled with an Axioscop2 microscope (Carl Zeiss, Jena, Germany) or in the FRET microscope described below.
FRET efficiency between CyPetm and YPetm fused to GLYT1 was measured in live cells plated on glass-bottommed culture dishes by sensitized emission, 24–48 h after transfection. An Axiovert 200M (Zeiss) inverted microscope equipped with Atto-Arc 2 (HBO 100 W) epifluorescence illumination, excitation and emission filter wheels (Lambda 10-2; Sutter) and a Ropper Scientific CoolSnap FX monochrome camera was used. Microscope control, image acquisition and processing were achieved using Metamorph 6.2.6 (Universal Imaging). Donor, acceptor and FRET images were acquired sequentially (200 ms exposure, bin 2) with a PlanApo ×63 oil-immersion objective (1.4 NA) using a fixed polychroic mirror (86002v2bs; Chroma Technology Corp.), the excitation filters (CFP: S430/×25, and YFP: S500/×20) and the emission filters (CFP: S470/30 and YFP: S535/30). After background subtraction and shade correction, images were registered using Metamorph. FRET efficiency analysis was performed using ImageJ software [W.S. Rasband, ImageJ, NIH (National Institutes of Health), Bethesda, MD, U.S.A., http://rsb.info.nih.gov/ij/, 1997–2006] and the PixFRET plug-in . In all FRET experiments, positive and negative FRET controls were analysed after transfection of a CFP–YFP tandem (positive control) or co-transfection of CyPetm and YPetm or co-transfection of CyPetm-EAAT2 and YPetm-GLYT1 (negative controls). Bleed-through coefficients were calculated using fret/donor or fret/acceptor image stacks captured from cells expressing only the donor or acceptor respectively. Coefficients were averaged from 20 separate stacks for each experiment. Typically, donor bleed-through in the FRET channel averaged 38% and acceptor cross-excitation in the FRET channel averaged 4–5%. The PixFRET plug-in renders pixel-by-pixel FRET efficiency values (expressed as percentage) and typical images are shown. For calculations and statistical evaluation, mean values of intensity histograms constructed from FRET efficiency images were averaged. A minimum of 40 different cell stacks were taken in each experiment.
The C-terminal domain of GLYT1 contains trafficking sequences transplantable to another neurotransmitter transporter
Previous studies of GLYT1 deletion mutants revealed the importance of residues located in the cytoplasmic C-terminus for the correct trafficking of the transporter from the ER to the plasma membrane [13,15]. To investigate whether the domain corresponding to residues 565–638 of GLYT1b (GLYT1ct) contains sufficient information to promote ER export, acting as an autonomous export signal, we analysed the capacity of GLYT1ct to promote ER exit when placed in the context of other proteins (see the scheme in Figure 1J). Like GLYT1, deletion of the C-terminus (residues 748–799) of the related GLYT2 produced a protein (GLYT2Δct) incapable of reaching the plasma membrane of transfected MDCK cells (Figures 1A–1C). This contrasted with the capacity of YFP–EAAT2 to reach the plasma membrane, a glutamate transporter that was co-transfected with GLYT2Δct to control for the integrity of the ER export pathway and to clearly identify the membrane compartment. A chimaera in which the C-terminus of GLYT2 was replaced with GLYT1ct (GLYT2–GLYT1ct) was properly inserted in the plasma membrane and co-localized with YFP–EAAT2 in most of the transfected cells (Figures 1D–1F). A quantitative estimation of the amount of these constructs that reached the plasma membrane was obtained by measuring their glycine uptake capability (Figure 1K). While glycine uptake measured in cells transfected with the construct GLYT2Δct was only 2% of that observed with GLYT2wt, the uptake after transfection of the chimaera GLYT2–GLYT1ct was 53% of the wild-type value (Figure 1K).
Differently from the GLYT2–GLYT1ct construct, a chimaera containing residues 1–498 of the vesicular stomatitis virus glycoprotein VSVG plus GLYT1ct, was unable to exit the ER (Figures 1G–1I). This is entirely consistent with previous reports on the traffiking of VSVG which showed that a C-terminal diacidic motif (residues 500–511) of wild-type VSVG controls its trafficking to the plasma membrane . This signal was absent from the VSVG–GLYT1ct chimaera and could not be replaced, at a functional level, by the addition of the GLYT1ct tail. Thus the export signal of GLYT1ct seems to only operate in the context of related proteins, suggesting that it needs additional export determinants provided by upstream sequences not contained in the transplanted peptide.
Identification of the intracellular residues important for GLYT1 exit from the ER
The capacity of the GLYT1ct export signal to substitute for that of GLYT2ct indicates that it is evolutionarily conserved. To more precisely identify residues involved in the exit of GLYT1 from the ER, we compared the amino acid sequence of the GLYT1ct with that of related NSS family transporters. In this gene family, only five amino acids are highly conserved in this particular region of the transporter: Gly569, Leu572, Arg575, Leu576 and Pro582 (Figure 2A). Additionally, residue Asp585 is conserved in most NSS transporters although in betaine transporter, TAUT (taurine transporter) and SERT, non-conservative substitutions can be found (proline or serine) (Figure 2A). We replaced residues Gly569, Arg575, Leu576, Pro582 and Asp585 (arrows in Figure 2A) using site-directed mutagenesis, introducing either alanine or other conserved or non-conserved residues. Subsequently, the function of these mutants was analysed in transport assays in transiently transfected COS cells (Figure 2B). Two of these mutated residues (Arg575 and Leu576) did not admit the alanine substitution and produced inactive transporters. Moreover, while conservative substitutions at these sites did yield active transporters (R575K and L576I respectively), non-conservative substitutions (R575D and L576A respectively) abolished the capacity to transport glycine. In addition, mutations in position Asp585 had a drastic effect on the transporter activity since mutant D585A lost approx. 80% of the transport capability, and its mutation to arginine reduced the activity by 75%. Previously, it was shown that the equivalent residues to Arg575 and Leu576 in the related GAT1 play an essential role in the traffic of this transporter to the plasma membrane . However, the equivalent residue to Asp585 was not investigated in that study. Thus we analysed how mutations in this position might affect the transport of substrates through GAT1 (residue Asp576) and the GLYT2 (residue Asp759). Mutations in GAT1 had a small effect on the GABA uptake (reduced by 7% in GAT1-D576A and 15% in GAT1-D576R). However, this residue was more important for GLYT2, where glycine uptake was reduced by 39 and 67% in mutants GLYT2-D759A and GLYT2-D759R respectively (Figure 2C).
The G569A and P582A mutants of GLYT1 transported glycine with an efficiency similar to that of wild-type GLYT1, suggesting that these positions are not crucial for GLYT1 function (Figure 2B). The conserved Leu572 residue has already been studied in a previous publication and was therefore not analysed further. Indeed, the L572A mutant was functional although it was missorted in polarized cells .
The inactive mutants Arg575 and Arg576 and the partially active D585A are retained in the ER
To elucidate why some of these mutants displayed impaired activity, we analysed their subcellular distribution by immunofluorescence and through biotinylation assays. Accordingly, we found that the inactive mutants R575A (Figure 3B), R575D (Figure 3C) and L576A (Figure 3E) were retained in the intracellular compartment, whereas the active mutants R575K (Figure 3D), L576I (Figure 3F), G569A (Figure 3G) and P582A (Figure 3H) displayed a similar subcellular distribution to that of the wild-type transporter (Figure 3A), with clear labelling of the plasma membrane. The partially active mutant D585A was largely retained in the intracellular compartment, but a weak staining of the plasma membrane was seen in some cells (Figure 3I). These results were concordant with those observed in biochemical assays in which membrane proteins were biotinylated with the membrane-impermeable reagent Sulfo-NHS-SS-Biotin. When the biotinylated proteins were recovered using streptavidin–agarose beads and analysed by immunoblotting, the inactive mutants (Arg575 and Leu576) were clearly not accessible to biotinylation due to their intracellular retention. In contrast, the active mutants were biotinylated, whereas mutant D585A that has a low transport capability was weakly accessible to biotinylation (Figure 4A). As reported previously, two GLYT1wt bands are detected in immunoblots corresponding to the total cell lysate . The faster migrating form corresponds to a partially glycosylated transporter that resides in the ER, whereas the slower one corresponds to the fully glycosylated GLYT1, and is accessible to biotinylation. As expected, mutants retained in the intracellular compartment were only detected as a single, faster migrating band, whereas both bands could be identified in the mutants that progressed to the plasma membrane. It has to be noted that the relatively low protein level detected in the intracellularly retained mutants, in part, might be due to the tendency of these forms of the protein to aggregate, yielding adducts that were not properly resolved in SDS/PAGE. Additionally, forms of the protein retained in the intracellular compartment might be rapidly degraded. To rule out the possibility that defective membrane targeting was due to mutant GLYT1 misfolding, glycine transport was assayed in reconstituted proteoliposomes obtained from COS cells transfected with mutants or wild-type GLYT1. After their reconstitution into proteoliposomes, the mutants R575A, R575D and L576A all displayed a greater percentage of activity than that measured in native COS cells (Figure 4B), indicating that the retained mutants were intrinsically capable of mediating glycine uptake. Moreover, a kinetic analysis of the glycine uptake by these reconstituted mutants revealed no significant changes in the Km of the mutants in comparison with the wild-type transporter (Km values were 31±5 μM for R575A, 24±4 μM for R575D, 28±5 μM for L576A and 18±3 μM for GLYT1wt). Hence, we could consider that these mutants were properly folded.
Since the R575A and L576A and D585A largely co-localized with the ER marker calnexin, they were clearly retained in the ER (Figures 5A–5C; see Supplementary Figure 1 at http://www.BiochemJ.org/bj/409/bj4090669add.htm). The staining of the wild-type form was clearly dissociated from calnexin in most of the cells, since it was mainly detected in the plasma membrane (Figures 5D–5F). However, occasionally, some cells overexpressing GLYT1wt also showed intracellular localization of GLYT1wt, which co-localized with calnexin (results not shown). The importance of the R575L576(X8)D585 motif for ER export was reinforced by the intracellular retention of the R575A mutation in the context of the GLYT2–GLYT1ct chimaera (Figures 5J–5L), in contrast with the chimaera that contained the wild-type form of GLYT1ct (Figures 5G–5I and Figures 1D–1F). A GLYT2–GLYT1ct chimaera containing mutations L576A or D585A was also retained in the ER (results not shown).
Recently, the dibasic [RK](X)[RK] motif has been proposed to be involved in ER export of the Golgi-resident glycosyltransferases . Because residue 577 of GLYT1 is a lysine, the sequence R575L576K577 could fit into this class of export signal. However, the K577A mutant was exported to the plasma membrane and transported glycine as efficiently as GLYT1wt (results not shown), ruling out this possibility. Similarly, the D568G569D570 sequence of GLYT1ct could also be considered as such a diacidic ER export motif (see Figure 2A) . However, again the D570A mutant was delivered normally to the plasma membrane (results not shown), equally ruling out the participation of this motif in ER export of GLYT1.
Sec24 interacts with the RL(X8)D motif of GLYT1ct
ER export involves recruitment of cargo proteins to ER-derived vesicles by the COPII. The first step in the capture of the cargo seems to be accomplished by the co-ordinate binding of the COPII components Sar1 GTPase, the Sec23–Sec24 complex, followed by the Sec13–Sec31 complex. Previously, the homologous RL motif of the GATs was shown to interact with Sec24D . Thus, to investigate the possible involvement of the RL(X8)D motif in binding to COPII, we analysed the capacity of GST fusion proteins containing the wild-type GLYT1ct (GST–G1ct in Figure 6) or with mutations in the Arg575, Leu576 or Asp585 positions (GST–G1ctR575A, GST–G1ctL576A and GST–G1ctD585A) to pull down a Myc-tagged form of Sec24D (Myc–Sec24D) expressed in transfected COS cells. While some non-specific Myc–Sec24D binding to control GST beads was observed (Figure 6A), there was a clear increase in the amount of Myc–Sec24D that was precipitated by GST–G1ct but not by the fusion proteins mutated at position 575 or 576 of GLYT1ct (GST–G1ctR575A and GST–G1ctL576A). A quantitative estimation of the binding capability was obtained by densitometric analysis of immunoblots (Figure 6B). Although mutant GST–G1ctD585A conserved some capability to pull down Myc–Sec24D (3.8 times over the GST control), this was clearly lower than GST–G1ct (6.4 times over the GST control), while the binding to GST–G1ctR575A and GST–G1ctL576A was similar to that of GST (Figures 6A and 6B). Hence, the RL(X8)D motif appears to participate in the interaction with the COPII complex component Sec24D.
Trafficking of Arg575 and Leu576 mutants can be partially rescued by co-expression with wild-type GLYT1
Immunofluorescence assays indicated that when the inactive R575A or Leu576 mutants were co-expressed in MDCK cells with wild-type GLYT1, a proportion of the mutant protein was able to reach the plasma membrane (Figures 7A–7C). In these experiments, the GLYT1 constructs were either tagged at the N-terminus with GFP (green fluorescent protein) or with the HA (haemagglutinin) epitope, and they were then visualized by double immunofluorescence. Control experiments revealed that the inclusion of these tags into GLYT1wt or mutants had no functional consequences either on transport or on the targeting processes (results not shown). However, when GFP-tagged forms of the mutant R575A (G-R575A) were co-expressed with HA-tagged GLYT1wt (HA–GLYT1wt), the mutant protein was now detected in the plasma membrane of cells with a low expression level of the mutant (Figures 7D–7F). The recruitment to the plasma membrane of the R575A mutant did not occur in a control experiment performed by co-expressing G-R575A with the glutamate transporter (HA–EAAT2) (Figures 7G–7I). Similar observations were made for L576A mutant (Figures 7J–7O)
GLYT1 oligomer formation
These immunocytochemical experiments suggested the existence of an interaction between the mutated forms of the transporter (R575A and L576A) and the native form of GLYT1 along the biosynthetic pathway, possibly involving oligomerization. Indeed, other members of the NSS transporter family are known to form oligomers, and an oligomerization step has been suggested to be part of the ER quality control mechanism for these closely related transporters. Nevertheless, previous experiments performed in Xenopus oocytes with cross-linker reagents failed to demonstrate oligomerization of GLYT1 . In our hands, chemical cross-linkers favoured the formation of high-molecular-mass aggregates containing the transporter in transfected COS cells (results not shown), but the interpretation of these experiments was complicated by the tendency of GLYT1 to aggregate non-specifically during the solubilization procedures in transfected cells. Non-specific aggregation was not observed in rat brain membranes and therefore, we studied the possible existence of GLYT1 oligomers in a preparation of synaptosomes. Putative oligomers were stabilized by treating the synaptosomal fraction with cross-linker reagents. Two reagents were used, the membrane-permeable bifunctional lysine cross-linker DSS or its non-permeable analogue BS3 (sulfo-DSS) . The cross-linked adducts were solubilized and resolved by sucrose-gradient velocity sedimentation, followed by SDS/PAGE and immunoblot analysis of the fractions obtained from the sucrose gradient (Figures 8A and 8B). Immunoreactivity for GLYT1 was detected in several fractions, with most of the protein migrating as a monomer with both reagents. The peak of this form of the protein was found in fraction 5, which corresponded to 3.3% sucrose (Figure 8). Densitometric analysis of the immunoblots indicated that 88.0±2.3% (n=3) of the total protein treated with DSS that had been loaded in the gradient migrated as a monomer in SDS/PAGE, while in samples reacted with BS3, the monomer represented 92.8±2.5% (n=3) of the total protein (Figure 8C). A form of GLYT1 with the electrophoretic mobility expected for a dimer had a peak around fraction 7 (4.7% sucrose) and represented 4.9±0.4% and 4.5±0.4% of the total protein in the presence of DSS or BS3 respectively (Figure 8C). Since it was equally accessible to both reagents, it seems to be located mainly at the cell surface. Another form of GLYT1, with the mobility expected for a tetramer, had a peak around the fraction 8 (5.3% sucrose). In this case, the amount of adduct measured in the presence of the permeable reagent DSS was significantly higher than that observed in the presence of BS3 (7.1±0.6 and 2.7±0.3%, P<0.01), suggesting that tetrameric arrangements of GLYT1 are more abundant in the intracellular compartment (Figure 8C). Additional adducts of higher density were also resolved in the sucrose gradients, but their electrophoretic mobility could not be measured with precision. Although the flotation characteristics and the electrophoretic mobility of the different GLYT1 adducts are compatible with the oligomerization of this protein, these results obtained in synaptosomes cannot rule out the possibility that these complexes would be formed by GLYT1 and other interacting proteins instead of GLYT1 protomers. Thus, to avoid the complications of the use of cross-linkers and to investigate the possibility that labile GLYT1 oligomers might also form along their biosynthetic pathway, we decided to use also a less invasive technique by measuring the efficiency of FRET between GLYT1wt with CyPetm or YPetm fused to its C-terminus. CyPetm and YPetm are optimized FRET versions of CFP and YFP . We used a wide-field microscope to measure FRET as an increase in acceptor fluorescence resulting from donor excitation (i.e. sensitized emission FRET). Images collected in the FRET channel were corrected for CFP and YFP spectral bleed-through and normalized for expression levels (NFRET), as indicated in the Experimental section. Figures 9(A)–9(C) provide representative images showing subcellular distribution of NFRET in COS cells co-transfected with GLYT1-CyPetm plus GLYT1-YPetm. Normalized FRET values are represented in a pseudocolour scale and clearly reveal the existence of FRET in the plasma membrane of transfected cells (Figures 9A and 9B). In those cells where ER forms of GLYT1 were observable, FRET was also appreciated in the intracellular compartment (Figure 9C). To perform the statistical analysis presented in Figure 9(D), the mean values of the intensity, constructed from the FRET efficiency images recorded over the whole surface delimited by the cell contour, were averaged and presented as a histogram. These values were compared with a negative control obtained by the co-expression of CyPetm and YPetm, and a positive control with a CFP–YFP tandem construct. Cells transfected with GLYT1-CyPetm plus GLYT1-YPetm showed a FRET efficiency significantly higher than that of the negative control [15.52±1.30% for GlyT1 compared with 2.73±0.63% for the negative CyPetm+YPetm control and 48.21±1.85% for the positive control (CFP–YFP), P<0.001, paired t test with negative control]. The FRET efficiency between the pair GLYT1-CyPetm and GLYT1-YPetm was also higher than that measured using as a donor the CFP-tagged form of the glutamate transporter EAAT2 (EAAT2-CyPetm) and GLYT1-YPetm as the acceptor (0.39±0.15%) (Figure 9D), despite that both proteins co-localize in the plasma membrane, indicating that the measured FRET was not due to non-specific aggregation of overexpressed plasma membrane proteins.
The precise localization of neurotransmitter transporters in specific compartments of the neuronal and perhaps the glial plasma membrane seems to be critical for the correct function of these proteins. One of the first steps in the correct targeting of the transporter is the exit from the ER. In those NSS transporters so far analysed, this involves both the oligomerization of the transporter and the interaction of specific residues of the C-terminus with the ER export machinery . Previous evidence suggested that the GLYT1 might use an ER export mechanism that could differ significantly from that used by other members of the NSS family, since it was reported that oligomerization was absent. Consequently, in the present study, we investigated the mechanisms of exit of the GLYT1 from the ER. However, our findings support a mechanism of ER export of GLYT1 similar to that of other NSS transporters. Here, we have identified the R575L576(X8)D585 motif, which is evolutionarily conserved across the NSS transporter family and is necessary for targeting GLYT1 to the plasma membrane. The RL moiety of the motif also mediates the ER export of the GAT1 . Moreover, RL operates in GAT1 by a similar mechanism since it also mediated the interaction of the transporter with the COPII complex component Sec24D (see below). Consistently with the evolutionary conservation of this sequence, mutations of Lys590 of the DAT, equivalent to Arg575 in GLYT1, also provoked the retention of DAT in the ER . Evolutionary conservation of Asp585 is less strict. Nevertheless, it is required for efficient expression not only of GLYT1, but also for GLYT2 (Asp759) (the present study) and for DAT (Lys600) . It seems to play a minor role for GAT1 expression. The RL(X8)D motif of GLYT1 differs from the three types of ER export motifs previously reported (the diacidic, the dihydrophobic and the dibasic motifs) . The diacidic motif is required for efficient export of the VSVG protein in mammalian cells  and it has also been found in Kir2.1 potassium channels, among other proteins . Interestingly, a diacidic motif that is not conserved in the NSS gene family can be found in GLYT1ct (residues D568G569D570; Figure 2A). However, this motif does not seem to regulate ER export of GLYT1 since the D570A mutant reached the plasma membrane (results not shown). Moreover, this sequence was unable to replace the endogenous diacidic motif of the viral protein VSVG, despite its inclusion in the transplanted fragment of GLYT1, suggesting that it is not in the appropriate context to promote ER export. Another ER export motif is the dihydrophobic motif that is used by proteins like ERGIC-53 (ER–Golgi intermediate compartment-53) or p24 [44,45]. The C-terminus of GLYT1 also contains two dileucine motifs that could be considered to fit into the dihydrophobic class of ER export motif. However, previous studies from our laboratory ruled out this possibility since transporters with mutations in this motif reached the plasma membrane, although their distribution was altered in polarized MDCK cells. Accordingly, this motif might act in sorting GLYT1 to the basolateral surface in the trans-Golgi network. ER export of glycosyltransferases is dependent on a dibasic motif in their cytoplasmic tails that fits to the consensus sequence [RK](X)[RK] . Since a lysine residue (Lys577) follows the RL half of the R575L576(X8)D585 motif, this GLYT1 sequence could be a variant of the dibasic motif. However, it appears to behave differently since the K577A mutant was efficiently exported to the plasma membrane. Moreover, although this second lysine residue is present in GLYT2, in the GAT2 and in TAUT, it is not conserved in other transporters of the NSS family (Figure 2A).
Surprisingly, mutations at two other evolutionarily conserved positions of GLYT1, Gly569 and Pro582, produced a transporter that underwent apparently normal trafficking and which displayed normal transport activity. Indeed, the G585A mutation in DAT, the equivalent of Gly569 in GLYT1, produced a dramatic effect on DAT export from the ER . It is possible that mutation of these residues to alanine is not sufficiently deleterious to alter the properties of GLYT1. Thus less conservative mutations would be necessary to evaluate the real importance of these two strictly conserved residues.
The three known ER export motifs interact with components of the COPII complex in order to be incorporated into COPII vesicles. Our results support a similar interaction between the RL(X8)D motif of GLYT1 and Sec24D of the COPII complex. However, this interaction seems not to be sufficient to promote ER export of GLYT1 since it was not transplantable to an unrelated protein like VSVG. These not yet identified additional ER export signals seem to be present in the related transporter GLYT2. Indeed, other export motifs characterized to date are not readily transferred to other proteins, suggesting that multiple determinants are commonly involved in the recruitment of many proteins into COPII vesicles.
The effects of mutations in the RL(X8)D motif were partially rescued by co-expression of mutated proteins with the native form of the transporter, GLYT1wt. This observation suggested that oligomers of GLYT1 form along the secretory pathway, with the GLYT1wt providing the mutant forms with the necessary signal for ER export. Other NSS transporters like GAT1, DAT, NET or SERT have been shown to be capable of forming oligomers and this might be related to their trafficking to the plasma membrane (for a review, see ). However, a previous report failed to identify GLYT1 and GLYT2 oligomers at the cell surface of Xenopus oocytes which had been injected with the mRNA encoding these transporters , and monomers were also detected in another study on the hydrodynamic properties of a GLYT purified from pig brain stem . The fact that two members of the NSS family might be incapable of undergoing oligomerization raised questions about the real importance of oligomerization in the physiology of this family of transporters. Nevertheless, the study in Xenopus oocytes had detected intracellular forms of GLYT1 and GLYT2 with an electrophoretic mobility that was consistent with the existence of oligomers . Thus we re-evaluated the capacity of GLYT1 to form oligomers in the rat brain by using chemical cross-linkers and in transfected living cells by using FRET microscopy. Results obtained by using these two approaches are compatible with the formation of GLYT1 oligomers. While FRET microscopy indicates that GLYT1 is able to form oligomers both at the cell surface and in the intracellular compartment, the cross-linking experiments in native rat brain membranes suggest that such complexes exist also in vivo, although in limited amounts (approx. 12% of the total GLYT1). The relatively low level of GLYT1 oligomers in these native membranes or their absence at the cell surface of Xenopus oocytes indicates that oligomerization may not be an essential determinant for substrate transport but rather that oligomerization influences trafficking. The preferential localization of tetrameric arrangements in the intracellular compartment that we detected with the use of differentially permeable cross-linker reagents, and the referred oligomeric intracellular forms detected in Xenopus oocytes  would be compatible with this idea. The identification of GLYT1 oligomers indicates that oligomerization is, indeed, a general feature of NSS transporters, although the biological significance of their formation remains unclear. Our results are compatible with the hypothesis that oligomerization might bring together several ER COPII-binding motifs, thereby increasing the efficiency of ER export of this family of transporters [15,19].
GLYT1 is a transporter that is enriched in glial cells in the glycinergic areas of the nervous system as well as in glutamatergic neurons, where it is strategically placed close to NMDARs, thereby regulating both glycinergic and glutamatergic activity in the nervous system. GLYT1 is exported from the ER through a multistep process that involves glycosylation , oligomerization and an interaction with the COPII complex through a specific and evolutionarily conserved motif present in the carboxyl tail of the protein. After escaping this stringent quality control of the ER, GLYT1 is sorted in the Golgi apparatus and delivered to the plasma membrane, where it is inserted by the exocyst and anchored to scaffold proteins in the cell surface to regulate extracellular levels of the neuroactive glycine [6,13,46].
We thank E. Núñez for expert technical help and Carlos Sánchez (Confocal Microscopy Department, Centro de Biología Molecular ‘Severo Ochoa’). We also thank Dr Patrick Daugherty for CyPetm and YPetm plasmids and Dr Wan Jin Hong (Institute of Molecular and Cell Biology of Singapore, Singapore) for the Myc–Sec24D plasmid. This work was supported by grants from the Spanish Dirección General de Investigación Científica y Técnica (SAF2005-03185), the Comunidad Autónoma de Madrid and by an institutional grant from the Fundación Ramón Areces.
Abbreviations: BS3, bis(sulfosuccinimidyl) suberate; CFP, cyan fluorescent protein; COP, coatomer coat protein; DAT, dopamine transporter; DSS, disuccinyl suberate; ECL, enhanced chemiluminescence; ER, endoplasmic reticulum; FRET, fluorescence resonance energy transfer; GABA, γ-aminobutyric acid; GAT1, GABA transporter 1; GFP, green fluorescent protein; GLYT1, glycine transporter-1; GST, glutathione transferase; HA, haemagglutinin; MDCK cell, Madin–Darby canine kidney cell; NMDA, N-methyl-D-aspartate; NMDAR, NMDA receptor; NP40, Nonidet P40; NSS, sodium- and chloride-dependent neurotransmitter transporter; SERT, serotonin transporter; TAUT, taurine transporter; VSVG, vesicular-stomatitis virus glycoprotein; YFP, yellow fluorescent protein
- © The Authors Journal compilation © 2008 Biochemical Society