The Stargazin C Terminus Encodes an Intrinsic and Transferable Membrane Sorting Signal*

Activity-dependent plasticity of α-amino-3-hydroxy-5-methylisoxazole-4-propionic acid receptors is regulated by their auxiliary subunit, stargazin. Association with stargazin enhances α-amino-3-hydroxy-5-methylisoxazole-4-propionic acid receptor surface expression and modifies the receptor's biophysical properties. Fusing the cytoplasmic C terminus of stargazin to the C-terminal domains of either GluR1 or the gonadotropin-releasing hormone receptor permits efficient trafficking from the endoplasmic reticulum and sorting to the basolateral membrane without altering other properties of either receptor.


MATERIALS AND METHODS
Generation of Constructs-The cytomegalovirus expression plasmids (pRK) for GluR1 were provided by Dr. Peter Seeburg (Max Planck Institute for Medical Research, Heidelberg, Germany). R1i 81 -YFP was generated using overlapping PCR to make an in-frame fusion protein with YFP, using pEYFP (Clontech) as a template. Stargazin cloned from Rattus norvegicus in the pcDNA3 vector was a generous gift from Dr. David Bredt (University of California, San Francisco). STG-YFP was created by inserting YFP in-frame at the BglII site in stargazin but truncating after amino acid 269. R1i 46 -YFP was made by inserting the restriction site MluI (ACGCGT) between amino acids GGG and SGE of the C-terminal domain using a QuikChange II XL site-directed mutagenesis kit (Stratagene, La Jolla, CA). The final sequence at the fusion was GGGTR(YFP). R1i 46 cSTG-YFP was created by fusing the stargazin cytoplasmic tail, amino acids 203-323, beginning DRHK and ending TTPV, followed by YFP. The final fusion had MluI sites between the C terminus of R1i 46 and the 5Ј end of the stargazin cytoplasmic tail and the 3Ј end of the stargazin cytoplasmic tail and YFP. The cSTG-CFP fusion protein was created using the pECFP-N1 vector and inserting amino acids 203-323 of the stargazin cytoplasmic tail preceded by ATG at the XhoI and HindIII restriction sites. GnRHR-GFP (13) was created by inserting amino acids 203-323 of stargazin between GnRHR and pEGFP at the BamHI site.
Transfections for Confocal Microscopy-Collagen-or poly-Dlysine-coated 14-mm glass bottom culture dishes (MatTek Corp., Ashland, MA) were incubated with ECL attachment matrix (Upstate Cell Signaling Solutions, Lake Placid, NY) for 1 h at 37°C and then washed with complete MEM before plating cells. Cells were transfected using Lipofectamine 2000 when 60 -90% confluent and incubated under identical conditions as cells used for standard electrophysiology. For each transfection 70 l of MEM was incubated with 3 l of Lipofectamine, and in another tube 30 l of MEM was incubated with 0.1-3 g of total cDNA and thoroughly mixed. Contents of the tubes were combined, and after 20 -30 min the solution was added to the cells. 6 -24 h later, the solution was exchanged with fresh complete MEM with 10 -20 M 2,3-dioxo-6-nitro-1,2,3,4-tetrahy-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1  drobenzo(f)quinoxaline-7-sulfonamide. All cells were imaged at room temperature 2-3 days after transfection. Immediately before imaging, the solution was exchanged with complete MEM containing no phenol red.
Confocal Imaging-Transfected cells were imaged using an LSM 510 META laser scanning confocal microscope (Zeiss, Thornwood, NY) with a Plan Achromat ϫ63/1.4 oil differential interference contrast objective. YFP was excited with sweeps of the 514-nm line of an argon laser operated at 6.1-6.3 A and directed to the cell via a 458/514-nm dual dichroic mirror. The emitted yellow fluorescence was directed to a photomultiplier equipped with a 530-nm long-pass filter. CFP was excited similarly but at 458 nm, and the emission was measured with a 460 -500-nm bandpass filter. Green fluorescence was excited at 488 nm and measured with a 530-nm long-pass filter.
Sorting Assay-Madin-Darby canine kidney cells were grown in high glucose Dulbecco's modified Eagle's medium (Mediatech Inc., Herndon, VA) supplemented with 2 mM L-glutamine, 100 units/ml penicillin, 100 g/ml streptomycin, 25 mM HEPES, and 5% fetal bovine serum. Cells were plated on Transwell filters (0.4-m pore size, Corning 3401) and transfected 4 days later using Effectene (Qiagen) according to the method described (15). 72 h after transfection, cells were incubated in 10 g/ml Alexa 594-conjugated concanavalin A in phosphatebuffered saline for 15 min at 4°C, washed twice with cold phosphate-buffered saline, and fixed in 4% paraformaldehyde. After fixation, filters were removed by cutting with a surgical blade. Cells and filters were mounted on slides with Prolong Antifade (Molecular Probes) and imaged by confocal laser scanning microscopy.

RESULTS AND DISCUSSION
We fused the C terminus of stargazin (amino acids 203-323) in-frame with the C terminus of GluR1 and continued in-frame with YFP. This construct (R1i 46 cSTG-YFP) (Fig. 1a) was expressed heterologously, and confocal microscopy was used to assess the trafficking pattern. R1i 46 cSTG-YFP was effectively trafficked from the ER to the plasma membrane as determined by identifying cells in which the surface expression was much greater than the cytosolic expression. In two separate experiments, 69% (43/62 cells) and 55% (55/100 cells) of cells expressing R1i 46 cSTG-YFP had predominant surface expression. This was unlike GluR1 in the absence of stargazin (Յ1% cells with surface expression for R1i 46 -YFP) and more comparable with GluR1 co-expressed with stargazin (5-35% surface expression for R1i 46 -YFPϩSTG (12)) (Fig. 1b). In contrast, a soluble cSTG-CFP fusion protein that contained amino acids 203-323 of the stargazin cytoplasmic tail did not traffic AMPA receptors to the plasma membrane (data not shown). We confirmed that R1i 46 cSTG-YFP ion channel density was greater than wild-type channel density by measuring current amplitude in outside-out patches of cells with visually enhanced surface expression (708 Ϯ 130 pA) versus R1i 81 -YFP (206 Ϯ 77 pA) (Fig. 1c). This value was similar to what we found previously for R1i 81 -YFP and R1i 36 -YFP (12). Without selecting for R1i 46 cSTG-YFP enhanced surface expression, we did not see a significant current amplitude increase (337 Ϯ 220 pA). These results are similar to what is found upon co-transfecting stargazin and AMPA receptors (12).
To demonstrate that the C terminus of stargazin modulated only the trafficking and not other aspects of AMPA receptor function, we assessed the activity of the fusion proteins. The desensitization kinetics (3.0 Ϯ 0.4 ms) (Fig. 1c) were typical of wild-type GluR1(flip) (3.7 Ϯ 0.4 ms) and different from stargazin-modulated wild-type GluR1 (8.5 Ϯ 0.8 ms) (Ref. 12; see also Refs. 6,8,and 9). This provides strong evidence that trafficking by stargazin of AMPA receptors occurs independently of its modulation of AMPA receptor kinetics. We also determined that R1i 46 -CFP could have a marked increase in surface expression when co-expressed with R1i 46 cSTG-YFP, presumably by hetero-oligomerization (Fig. 1d).
To test whether the reduced ER retention was related to a GluR1-specific interaction, we fused the same C-terminal domain of stargazin to the GnRHR (GnRHRcSTG-GFP). Although a member of the G-protein-coupled receptor superfamily, the GnRHR is inefficiently trafficked to the plasma membrane and, like AMPA receptors, is largely retained in the ER in both homologous and heterologous cells (16,17). The C-terminal fusion of stargazin dramatically enhanced GnRHR transport to the cell surface (Fig. 1e), with nearly every cell demonstrating robust surface expression. This is significant as it reveals trafficking information that is intrinsic to stargazin and transferable to an unrelated plasma membrane receptor.
The enhanced trafficking of the GnRHR by GnRHRcSTG-GFP did not impair protein function. We tested ligand-induced phosphorylation of ERK, a well established intracellular target of GnRH signaling (Fig. 1f). Also, specific binding of GnRH agonist was not affected by the C-terminal stargazin fusion, and consistent with the imaging analysis, total cell-surface binding was significantly greater for cells expressing GnRHRcSTG-GFP compared with GFP alone or wild-type GnRHR-GFP (Fig. 1g).
The results from Fig. 1 indicate that the C terminus of stargazin encodes a membrane sorting signal that can operate independently of the rest of the stargazin protein. It is likely that stargazin-mediated trafficking of AMPA receptors therefore occurs by virtue of this signal, such that the AMPA receptors are pulled to the surface through their specific interaction with stargazin in a region that is not the stargazin cytoplasmic tail. Furthermore, this ability is not receptor-dependent, as the C-terminal domain acts on the GnRHR, a protein with which it does not associate.
Because stargazin has been shown to direct AMPA receptors not only to the plasma membrane, but more specifically to the synapse (7,18,19), we tested whether the C-terminal domain was involved with protein sorting. Stargazin was heterologously expressed in Madin-Darby canine kidney cells, a model system for neuronal trans-Golgi network sorting, where basolateral sorting in Madin-Darby canine kidney cells may reflect somatodendritic sorting in neurons (20). STG-YFP expression was restricted to the basolateral membrane (Fig. 2). GnRHRs did not demonstrate membrane sorting. However, when the C terminus of stargazin was fused to the GnRHR, it was trafficked specifically to the basolateral membrane, suggesting that stargazin encodes a sorting signal that either directs sorting vesicles to the basolateral membrane or permits enhanced endocytosis from the apical membrane. A similar effect was seen for R1i 46 -YFP, although again, the ability of stargazin to direct AMPA receptor sorting was not as robust as its effect on the GnRHR.
Several protein motifs have been identified as positive ER export signals: LL, DXE, RXR, VXXSL, and hydrophobic motifs, including FNX 2 L 2 , FXXXFXXXF, and FCYENE (21). Amino acids 270 -323 of the stargazin C terminus may be truncated without affecting its ability to traffic: within the remaining C-terminal FIGURE 1. Fusion with the C terminus of stargazin traffics GluR1 or GnRHRs from the ER. a, schematic representations of stargazin with a C-terminal truncation at amino acid 269 fused in-frame with YFP (left) and fluorophore-tagged GluR1 (middle) and GnRHR (right) proteins fused in-frame with the complete cytosolic C terminus (amino acids 203-323) of stargazin (cSTG). b, confocal images of GluR1-YFP, STG-YFP, GluR1i 46 -YFP co-expressed with stargazin, and R1i 46 cSTG-YFP in HEK cells (arrows show regions of significant surface expression compared with cytosolic expression; expression of YFP is shown with green pseudo-color). Confocal experiments were performed as described previously (12). c, current trace of a response of GluR1i 46 cSTG-YFP evoked by 10 mM glutamate, measured in an outside-out patch from a transfected HEK293 cell. Introduction of the cSTG sequence did not slow the kinetics of desensitization. Shown are mean current amplitudes from outside-out patches of cells expressing wild-type R1i-YFP (n ϭ 4) or R1i 46 cSTG-YFP (n ϭ 5 each for Ϫ and ϩ ring conditions). For R1i 46 cSTG-YFP, amplitudes were either measured on a population of cells that were selected to have visible surface expression (ϩrings) or not (Ϫrings). Cells that had obvious surface expression had a significantly larger mean current amplitude (**, p Ͻ 0.02). d, confocal images of cells co-transfected with R1i 46 -CFP and R1i 46 cSTG-YFP, demonstrating membrane localization of both proteins. In this case the cell on the right in panel d shows surface expression of R1i 46 -CFP, presumably through hetero-oligomerization with R1i 46 cSTG-YFP, which rescues poor GluR1 trafficking. e, confocal images of the enhanced GFP-tagged GnRHR (localized to the ER), which is trafficked to the plasma membrane when fused with the C terminus of stargazin (GnRHRcSTG-GFP). f, ERK phosphorylation (p) by GnRHRcSTG-GFP and wild-type GnRHR in transiently transfected HEK293 cells treated with 100 nM GnRH is equivalent. g, binding of 125 I-labeled D-Ala 6 -GnRH1 in HEK293 cells expressing GFP, GnRHR-GFP, and GnRHRcSTG-GFP, expressed as fraction of total bindable counts. GnRHR binding and ERK activation protocols were done as described previously (16). ***, p Ͻ 0.001. domain (amino acids 203-269) there are several possible dibasic motifs, the first of which is proximal to stargazin transmembrane domain 4 (amino acids 204 -206), an obvious target as an ER export site. We mutated this motif in the GnRHR fusion and found that neutralization of the charges did not reverse the effects of the C-terminal domain of stargazin on GnRHR trafficking (data not shown).
We assumed previously that stargazin-mediated surface expression of AMPA receptors in HEK cells was variable because of differences in the amount of stargazin per cell. We found that with increasing stargazin per constant AMPA receptor cDNA concentration, we could get a surface expression enhancement in ϳ50% of cells, but this value approached an asymptotic line (refer to Fig. 1c of Ref. 12). Surprisingly, despite having homomeric AMPA receptor chimeras in this study, we still observe ϳ40% of the cell population that does not demonstrate an enhancement of surface expression or current amplitude. These results suggest that there is likely intrinsic cell variability in protein expression. The differences in the way these proteins may interact with AMPA receptors and keep them from the surface membrane are not present in GnRHRs. Sorting the two populations of cell types for microarray analysis to determine gene candidates that hinder or allow AMPA receptor trafficking to surface membranes may be instructive.
In summary, we have shown that the C terminus of stargazin directs effective trafficking of itself from the ER, as well as of AMPA receptors complexed with it, or of other receptors to which its C terminus is directly fused. Unexpectedly, we also found that the C terminus of stargazin encodes a sorting signal, directing trafficking through the trans-Golgi network to specific membrane compartments. It remains to be shown whether the sorting is regulated in an activity-dependent manner.