Multifunctional Basic Motif in the Glycine Receptor Intracellular Domain Induces Subunit-specific Sorting*

The strychnine-sensitive glycine receptor (GlyR) is a ligand-gated ion channel that mediates fast synaptic inhibition in the vertebrate central nervous system. As a member of the family of Cys-loop receptors, it assembles from five homologous subunits (GlyRα1–4 and -β). Each subunit contains an extracellular ligand binding domain, four transmembrane domains (TM), and an intracellular domain, formed by the loop connecting TM3 and TM4 (TM3–4 loop). The TM3–4 loops of the subunits GlyRα1 and -α3 harbor a conserved basic motif, which is part of a potential nuclear localization signal. When tested for functionality by live cell imaging of green fluorescent protein and β-galactosidase-tagged domain constructs, the TM3–4 loops of GlyRα1 and -α3, but not of GlyRα2 and -β, exhibited nuclear sorting activity. Subunit specificity may be attributed to slight amino acid alterations in the basic motif. In yeast two-hybrid screening and GST pulldown assays, karyopherin α3 and α4 were found to interact with the TM3–4 loop, providing a molecular mechanism for the observed intracellular trafficking. These results indicate that the multifunctional basic motif of the TM3–4 loop is capable of mediating a karyopherin-dependent intracellular sorting of full-length GlyRs.

In the vertebrate nervous system, signal transmission at chemical synapses is mediated by ionotropic and metabotropic neurotransmitter receptors. Ligand-gated ion channels harbor an intrinsic channel pore that is opened almost instantly upon ligand binding. Among the ligand-gated ion channels, the superfamily of Cys-loop receptors comprises the nicotinic acetylcholine receptor, the 5-hydroxytryptamine type 3 receptor, the ␥-aminobutyric acid type A/C receptor, and the GlyR. 2 Besides sequence homology, Cys-loop receptors share a com-mon pentameric rosette-like composition of homologous subunits and a characteristic subunit topology (Fig. 1, A and B). The N-terminal extracellular domain forms a twisted ␤-sandwich structure contributing to the intersubunit ligand binding pockets (1). Each subunit contains four ␣-helical TMs, where TM2 lines the channel pore. The intracellular domain of the receptor is mainly formed by the TM3-4 loop; its protein conformation has not been resolved. The C-terminal part of the TM3-4 loop, however, is thought to form an intracellular cavity that might contribute to ion selectivity (2).
Spinal GlyRs play an important role in the neuronal control of muscle tone. Upon glycine binding, the intrinsic chloride channel opens, resulting in hyperpolarization of the postsynaptic membrane of ␣ motoneurons, thereby reducing their activity. GlyR defects underlay excessive reflex responses, as in symptomatic hyperekplexia (OMIM 149400), stiff-person syndrome (OMIM 184850), or strychnine intoxication (3). Up to five GlyR genes have been described in vertebrates (4), four of which code for subunits that are active as homomers in heterologous systems (␣1-4). Among these, the ␣2 subunit predominates during embryonic development, forming homomeric, extrasynaptic receptors that are thought to be important for calcium-dependent synaptogenesis (5,6). The adult receptors, however, are mostly heteromers of 2 GlyR␣ and 3 GlyR␤ subunits, both of which contribute to ligand binding (7).
The TM3-4 loop is the domain with the highest sequence variability between different subunits, possibly regulating trafficking, anchoring, and protein interactions. The best characterized interaction takes place between gephyrin and a 13-residue motif in the TM3-4 loop of GlyR␤ (8). Gephyrin forms hexagonal lattices at the postsynaptic density, allowing for synaptic anchoring of heteromeric receptors (9,10). As a postsynaptic scaffold for glycinergic synapses, gephyrin also recruits a number of interaction partners to the synapse (11). By interaction with the dynein complex, gephyrin finally mediates vesicular retrograde transport of the receptor along microtubules (12). For the TM3-4 loops of GlyR␣ subunits, a number of protein-protein interactions that are guided by consensus motifs have been described (13,14). However, the roles of these interactions and consensus motifs in trafficking of the receptor remain elusive.
Here, we report the presence of a multifunctional basic motif that is part of a potential nuclear localization signal (NLS) within the TM3-4 loop of adult, synaptic GlyR␣ subunits, i.e. GlyR␣1 and -␣3. When functionality of this NLS was tested, full-length GlyR subunits, but not fragments, were shown to be imported into the nucleus. Karyopherins ␣3 and ␣4 were identified as interaction partners of the GlyR in vitro. In addition to nuclear sorting, this interaction may also underlie other functions of the basic motif, including preservation of correct protein topology or membrane integration.

EXPERIMENTAL PROCEDURES
Sequence Analysis and Cloning-For multiple sequence alignment, the sequences and boundaries of the TM3-4 loops were taken from the annotation in the Uniprot data base (15). Alternative variants of the TM3-4 loops of human (Homo sapiens, hs), rat (Rattus norvegicus, rn), and mouse (Mus musculus, mm) GlyR subunits were subjected to a multiple sequence alignment using the T-COFFEE algorithm (16). Prediction of subcellular localization and nuclear import was performed with the TM3-4 loop sequence of hsGlyR␣1. Cloning was carried out with standard protocols using plasmid DNA as template and overlap extension PCR for mutagenesis. Fusion proteins with green fluorescent protein (GFP) were generated using the pEGFP system (N1, C1-3; Clontech, Mountain View, CA). Because of high sequence conservation, orthologous GlyR subunits were used equivalently. The cloned TM3-4 loops of human and murine GlyR subunits were identical to the sequences obtained from the Uniprot data base, whereas for historical reasons the TM3-4 loops of rat GlyR␣ subunits include the additional N-terminal sequence of AAVNFVS and the TM3-4 loop of rat GlyR␤ carries the additional N-terminal sequence of VVQVML, which are annotated as TM3 in the Uniprot data base. For enlargement of the final fusion constructs, ␤-galactosidase full-length cDNA was inserted in-frame into pEGFP-N1 making use of NheI and BglII restriction sites to preserve the multiple cloning site for insertion of the TM3-4 loop. Karyopherin full-length cDNAs were obtained from imaGenes, Berlin. For GST pulldown assays, the cloning vectors pET21a and pET41a were used (EMD, La Jolla, CA).
Yeast Two-hybrid Screening and GST Pulldown Assay-8,000,000 cDNAs from an adult human brain Matchmaker cDNA library in pACT2 (Clontech; HL4004AH) were screened with the TM3-4 loop of rnGlyR␣3L as bait. In specificity tests, interaction of single clones with other GlyR subunit TM3-4 loops was tested. A GST pulldown assay was performed essentially as described (17).
Neuronal Cell Culture-Hippocampi were isolated from Wistar rats at embryonal day 18, meninges were removed, and tissue was pooled. Trypsin digestion was carried out with 1 mg/ml trypsin, 0.1 mg/ml DNase I in neurobasal medium (Invitrogen) at 37°C for 30 min and stopped by the addition of 10% (v/v) fetal calf serum. Cells were dissociated by trituration, washed, and plated in neurobasal medium with 1% B-27 (Invitrogen) in 3-cm cell culture dishes or in 24-well plates on polylysine-coated coverslips. Transfection was carried out using Lipofectamine (Invitrogen) according to the manufacturer's instructions.
Immunocytochemistry, GFP Imaging, and Electrophysiology-For immunocytochemistry, HEK293 cells or neurons were grown on polylysine-coated coverslips, transfected if necessary, fixed with 3% paraformaldehyde in phosphate-buffered saline (PBS), and permeabilized as well as blocked with 0.1% Triton X-100, 5% donkey or sheep serum (Dianova) in PBS for 30 min at room temperature. Primary and secondary antibodies were diluted 1:200 -400 in 5% serum. Cells were incubated with primary antibodies for 1 h at room temperature and washed 3 times with PBS. This procedure was repeated with secondary antibodies. Finally, cells were embedded in Mowiol and subjected to confocal microscopy on a DMIRE2 confocal microscope (Leica, Wetzlar. Germany). If required, serial z-scans were performed. Medial z-slices were used to define central nucleoplasmic regions of interest analyzed for fluorescence intensity using Leica Confocal Software. For GFP-imaging, HEK293 cells were grown on chamber slides (ibidi, Martinsried, Germany) and transfected by calcium phosphate precipitation. After 12-36 h, medium was exchanged for PBS followed by live cell imaging. Alternatively, transfected neurons grown on polylysine-coated coverslips were fixed using 5% acetate in methanol for 10 min at Ϫ20°C, embedded in Mowiol, and subjected to either confocal or standard fluorescence microscopy, the latter with an Axioskop microscope (Carl Zeiss, Jena, Germany). Electrophysiology was carried out essentially as described (18).
Immunoprecipitation and Western Blot-For immunoprecipitation, either cellular compartments were used or wholecell lysates were prepared. The latter was accomplished by homogenization of tissue or cultured cells in 50 mM Tris, 150 mM NaCl, 2% (v/v) Triton X-100, 5 mM EDTA, and complete protease inhibitors (Hoffmann-La Roche), pH 7.4, and clearance by centrifugation for 1 h at 100,000 ϫ g and 4°C. To the respective solution, antibodies with 4 -10 g/ml final concentration were added, mixed, and agitated overnight at 4°C. 20 l of Pansorbin-Agarose (EMD) was added, and agitation on 4°C was continued for 2 h. Then, the beads were washed four times with PBS before they were subjected to Western blot analysis. Primary antibodies were diluted 1:2000, and secondary horseradish peroxidase-coupled antibodies were diluted 1:20000. Chemiluminescence substrates were ECLϩ (GE Healthcare) and SuperSignal West Femto (Thermo Fischer Scientific, Waltham, MA) for normal and weak signals, respectively. Cellular Fractionation Techniques-Multiple cell fractions were obtained with Compartmental Protein Extraction kit according to the manufacturer's instructions (Millipore). Alternatively, cells were grown on a 10-cm cell culture dish, washed twice in PBS, harvested in 400 l of Buffer A (10 mM HEPES, 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, 2 mM dithiothreitol, and complete protease inhibitors (Hoffmann-La Roche), pH 7.9) and incubated on ice for 5 min. By the addition of 1% (final concentration) Nonidet P-40 and mixing, cells were lysed. The supernatant after centrifugation for 30 s at 20,000 ϫ g and 4°C was taken as the extranuclear compartment. Subsequently, the pellet was resuspended in 200 l of Buffer B (Buffer A including 400 mM NaCl), mixed, and agitated at 4°C for 15 min and centrifuged for 5 min at 20,000 ϫ g and 4°C. The supernatant was taken as the nuclear compartment. Membrane preparations were performed essentially as described (18). For alkaline extraction, 100 l of membrane preparation was centrifuged at 100,000 ϫ g and 4°C for 1 h. The pellet was resuspended in 100 l of 100 mM Na 2 CO 3 , pH 11.5, by sonification with a microtip (15 ϫ 3 s on ice), incubated on ice for 30 min, and centrifuged at 100,000 ϫ g and 4°C for 1 h. The supernatant (corresponding to the membrane associated proteins) was saved, and the pellet (corresponding to integral membrane proteins) was resuspended in 100 l of PBS.

Sorting Prediction and Sequence
Analysis-To predict potential sorting motifs within the TM3-4 loop of human GlyR␣1 (hsGlyR␣1), we used different algorithms (Table 1). Curiously, both PSORTII (19) and HSLPred (20) predicted the nucleus to be the most probable sorting destination of the TM3-4 loop. Next, we applied two algorithms specialized for nuclear proteins. PredictNLS (21), which is restricted to published NLSs in a given polypeptide sequence, did not provide a significant hit. NucPRED, however, confirmed the nuclear sorting with considerable specificity and identified the basic motif 346 RRKRR 350 as a potential NLS (22). By multiple sequence alignment, we determined that this motif was highly conserved in GlyR␣ but not ␤ subunits (Fig. 1C).
Nuclear Accumulation of the TM3-4 Loop-To determine the actual subcellular localization of the GlyR␣1 TM3-4 loop, we generated fusion constructs ( Fig. 2) with GFP and performed live cell imaging of HEK293 cells transfected with these plasmids (Fig. 2, A and C). As expected, GFP distributed evenly between cytoplasm and nucleus. By contrast, fusion constructs of GFP with the TM3-4 loops from human or mouse GlyR␣1 accumulated in the nucleus. This is consistent with the high sequence similarity between species orthologs seen in multiple sequence alignments (Fig. 1C). Nuclear accumulation was confirmed, applying cell fractionation by differential centrifugation and Western blot using an antibody against GFP (Fig. 2B).
When the basic motif (RRKRR) was mutated to a polyalanine sequence in the mouse GlyR␣1 TM3-4 loop construct, the resulting fusion protein distributed evenly between nucleus and cytoplasm, as observed for GFP (Fig. 2B). This indicated that the TM3-4 loop holds a five-amino acid basic sorting motif that is necessary for the promotion of nuclear accumulation.
To test whether the TM3-4 loop containing the motif RRKRR was also active as a sorting signal in nerve cells, we transfected primary rat hippocampal neurons with these constructs (Fig. 3). As evident from nuclear accumulation, the cellular activity of the basic sorting motif in neurons was similar to that of HEK293 cells. Apparently, HEK293 cells provide an appropriate model to study this sorting mechanism.
Only Selected GlyR Subtypes Harbor a Functional NLS-To test the impact of sequence diversity between different GlyR subunits on nuclear accumulation, we transfected HEK293 cells with GFP and GFP C-terminal-fused to TM3-4 loops of rnGlyR␣1, rnGlyR␣2, rnGlyR␣3K, rnGlyR␣3L, or rnGlyR␤. In live confocal microscopy, large intercellular variability of absolute and relative nuclear fluorescence intensity was observed for all TM3-4 loop fusion constructs. Nevertheless, calculation of nuclear to cytoplasmic fluorescence intensity ratios indicated pronounced nuclear accumulation for the TM3-4 loops of rnGlyR␣1 and rnGlyR␣3L and little nuclear accumulation for rnGlyR␣3K (Fig. 4A). As expected from the multiple sequence alignment, which revealed little conservation of the required basic motif RRKRR (Fig. 1C), the TM3-4 loop of rnGlyR␤ showed no tendency to nuclear accumulation. Surprisingly, the TM3-4 loop of GlyR␣2 also failed to induce nuclear accumulation. This may be related to the interruption of the basic motif by a glutamine residue in the sequence of GlyR␣2 (RRRQK).
Subunit-specific Active Nuclear Import-Given their small size of Ͻ40 kDa, it cannot be excluded that nuclear accumulation of the recombinant fusion proteins resulted from diffusion through the nuclear pore and subsequent intranuclear retention. Diffusion through the nuclear pore, however, can be restricted by enlargement of the molecular weight of the fusion protein (23). To this end, we fused ␤-galactosidase to the reporter constructs, obtaining molecular masses Ͼ150 kDa (Fig. 5A). Although the ␤-galactosidase-GFP fusion was efficiently excluded from the nucleus, we observed nuclear import for fusion constructs of TM3-4 loop of GlyR␣1 with ␤-galactosidase and GFP (Fig. 5B). However, a construct composed of merely the basic motif 346 RRKRR 350 fused between ␤-galactosidase and GFP with terminal spacer sequences was excluded from the nucleus. This suggests that the basic motif is required but not completely sufficient for nuclear import. Subsequently, we tested the TM3-4 loop sequences of different GlyR subunits including the long (L) and short (K) splice variants of GlyR␣3    (24) for the presence of a functional NLS. As observed for GlyR␣1, the TM3-4 loops of the GlyR␣3L and ␣3K (potential NLS: RRKRK) also mediated nuclear import of ␤-galactosidase-GFP fusion proteins. By contrast, the TM3-4 loops of GlyR␣2 and -␤ failed to show nuclear import (Fig. 5C) essentially confirming the results obtained with GFP-TM3-4-loop fusion constructs (Fig. 4). Thus, the TM3-4 loops of GlyR␣1, -␣3L, and -␣3K subunits each contain a functional NLS, which accounts for the nuclear accumulation observed previously (Fig. 4).
Detection of Full-length Nuclear GlyR-The detection of low amounts of nuclear GlyR antigen using monoclonal antibodies mAb4a or mAb2b directed against the N-terminal domain of all GlyR␣ subunits (mAb4a) or GlyR␣1 (mAb2b), respectively, was hindered by considerable nonspecific nuclear reactivity of the antibodies (data not shown). To improve the signal-to-background ratio and to facilitate direct detection of the intracellu-lar domain, we introduced small epitope tags into the TM3-4 loop. To conserve structure and function of the receptor, we inserted short and uncharged HA epitope and T7 major capsid protein epitope tags into the variable regions (Fig. 1C) of the TM3-4 loop of the full-length GlyR␣1 subunit, obtaining the construct GlyR␣1-HA-T7. Subsequently, GlyR␣1-HA-T7 was tested for expression, sorting, and function, revealing no gross differences from the wild-type counterpart (supplemental Fig. 1).
Construct GlyR␣1-HA-T7 was used to detect nuclear reactivity with confocal immunofluorescence-tomography (Fig. 6). To clearly visualize the boundaries of the nucleus, co-staining with an antibody directed against lamin A, a protein enriched in the nuclear envelope, was performed. In section-planes parallel to the coverslip through medial cellular levels, nuclear regions of interest medial to the anti-lamin A peak-immunoreactivity  were marked, and the relative fluorescence intensity of anti-HA immunoreactivity was quantified. Nuclear immunoreactivity for the GlyR-HA-T7 antigen was significantly higher in transfected (18.3 Ϯ 7.0%) as compared with untransfected HEK293 cells (2.2 Ϯ 0.9%), indicating that nuclear import occurs upon transfection of the full-length receptor.
Even though full-length receptor constructs were transfected, it is possible that the nuclear immunoreactivity observed in confocal immunofluorescence-tomography stems from fragment peptides that are generated by posttranslational cleavage of the subunit. To show that the observed nuclear immunoreactivity corresponds to the full-length receptor, it has to be ruled out that fragments are generated that hold the NLS and, thus, might also be imported into the nucleus.
In Western blot analysis of lysate from GlyR␣1-HA-T7transfected HEK293 cells, we detected a fragment of about 9 kDa (Fig. 7A). This fragment was stained only with the antibody against the T7 tag but not with the antibody against the HA tag, indicating probable cleavage between the tags in the central part of the TM3-4 loop. Such a cleavage would generate a fragment of 7.5-12 kDa that holds TM4 but not the NLS as the NLS is N-terminal of the HA tag (supplemental Figs. 1A). To further corroborate the identity of this fragment, we prepared a membrane fraction from transfected HEK293 cells using alkaline extraction, pH 11.5. This method is used to distinguish between membrane-associated and membrane-integral proteins. A protein that cannot be solubilized from the membrane fraction by alkaline extraction is generally regarded to be a membrane integral protein, e.g. with a transmembrane domain. Consistent with the size of the fragment and the site of the cleavage, the fragment was not soluble after extraction, suggest-   FEBRUARY 5, 2010 • VOLUME 285 • NUMBER 6

JOURNAL OF BIOLOGICAL CHEMISTRY 3735
ing that, indeed, it harbors TM4 (Fig. 7B). Altogether, these results show that the nuclear immunoreactivity observed in confocal microscopy probably did not correspond to this fragment.
To conclude that the observed nuclear immunoreactivity resulted from full-length subunits, it is essential to rule out that such a soluble fragment was generated as a potential soluble fragment might be easily imported into the nucleus. Although no such fragment was detected in high sensitivity Western blot analysis, we attempted to even further enhance analytical sensitivity. An assay was designed to detect even very low amounts of soluble TM3-4 loop fragments that would be imported into the nucleus. This was achieved by nuclear compartmental extraction, subsequent immunoprecipitation with an antibody directed against the T7-tag, and Western blot analysis using an antibody directed against the HA tag (Fig. 7, C and D). Importantly, the full-length subunit, but no immunolabeled fragment, was detectable even with extended exposure times, indicating that the nuclear epitopes in immunocytochemistry solely represent fulllength GlyRs. It should be noted that most of the precipitated fulllength GlyRs in the nuclear compartment are probably ER contaminants. Because the outer nuclear membrane is continuous with the rough endoplasmic reticulum, it cannot be separated fully from nuclear compartmental extractions. This leads to an overestimation of the amount of nuclear GlyR if judged by band intensity in the Western blot. Nevertheless, together with immunocytochemical data, which clearly distinguish between nuclear envelope and nucleoplasmic staining (Fig. 6), it can be concluded that the minor intranuclear staining observed with the HA antibody represents fulllength subunits.
Karyopherins Interact with the TM3-4 Loop-In an independent yeast two-hybrid assay, we screened for potential interaction partners of the TM3-4 loop of rnGlyR␣3L. Of 71 positive clones, 51 corresponded to either karyopherin ␣3 or ␣4 (data not shown). Karyopherin ␣-subunits bind to karyopherin ␤1 via their autoinhibitory domains, thereby opening their NLS binding pockets. This pocket holds multiple, partially overlapping binding sites for   (lanes 1 and 3) and GlyR␣1-HA-T7 (lanes 2 und 4, see the scheme in Fig. 6A). The expected molecular mass for the full-length receptor is ϳ49 kDa. A low molecular mass fragment generated from GlyR␣1-HA-T7 is marked (gray arrow). As this fragment is only recognized by the anti-T7 antibody and not the anti HA antibody, it corresponds to a cleavage between these epitopes (compare Fig. 6A). B, alkaline extraction of membrane preparations of hsGlyR␣1-HA-T7-transfected HEK293 cells is shown. Soluble (S) and insoluble (I) proteins after extraction are shown. The low molecular mass fragment (gray arrow) is insoluble after alkaline extraction and, thus, as expected, holds TM4. C, immunoprecipitation with anti-T7-antibody from nuclear (N) or extranuclear (EX) compartments of HEK293 cells transfected with hsGlyR␣1 (lanes 2 and 4) or hsGlyR␣1-HA-T7 (lanes 1 and 3). In this Western blot with anti-HA-antibody the heavy (upper arrowhead) and the light chain (lower arrowhead) of the antibody and the unprocessed hsGlyR␣1-HA-T7 (black arrow) are marked. D, controls for enrichment of different compartments with anti-glyceraldehyde-3-phosphate dehydrogenase (GAPDH) antibody (extranuclear) and anti-histone deacetylase 1 (HDAC) antibody (nuclear). Taken together, these results exclude that significant amounts of soluble fragments of the GlyR are generated in transfected HEK293 cells.
NLSs. Trimeric complexes of ␣-karyopherins, karyopherin ␤1, and NLS-harboring proteins are transferred through the nuclear pore complex in an energy-dependent process. In vitro, in the absence of free karyopherin ␤1, the affinity of ␣-karyopherins to their cargo is reduced by 3 orders of magnitude (25). For GST pulldown assays, we therefore generated full-length and truncated karyopherin ␣3 and ␣4 lacking the N-terminal autoinhibitory domain (Fig. 8A). All four constructs were pulled down with GST fused to the TM3-4 loop of GlyR␣1 but not with GST alone (Fig. 8, B-D). This indicates a high in vitro affinity of GlyR␣1 to karyopherin ␣3 and ␣4. Taken together, the yeast two-hybrid assay and GST pulldown detected molecular interactions for the TM3-4 loops of rnGlyR␣1, -␣3L, and -␣3K but not for ␣2 and ␤. For GlyR␣3L, these interactions were found to be dependent on the basic NLS motif.
To test the possibility of GlyR-karyopherin interactions in neurons, we performed confocal immunocytochemistry of spinal cord neuronal cultures using antibodies against GlyR␣1 (mAb2b), synaptophysin (as synaptic marker) and karyopherin ␣3 or ␣4 (data not shown). Only minimal association of GlyR␣1 and karyopherins ␣3 or ␣4 was observed, indicating that there is no strong static interaction in vivo. Nevertheless, this observation does not rule out the presence of a potentially short-lived complex between karyopherins and the GlyR.

DISCUSSION
Structural comparisons of GlyR subunit polypeptides showed that a conserved basic motif is located within the TM3-4 loop of GlyR␣1 and -␣3, whereas the homologous sequence varies slightly in the ␣2 and ␤ subunits. Sequence analysis of the basic motif present in the GlyR␣1 and -␣3 subunits predicted this sequence to correspond to a nuclear localization signal. However, basic motifs may be functionally ambiguous; in addition to the nuclear sorting of proteins, the basic motif present in the TM3-4 loops of GlyR␣1 and -␣3 may be multifunctional and also mediate the interaction with a molecular chaperone (26). When tested for functionality in HEK293 cells and primary hippocampal neurons, the intracellular domain of the GlyR was, indeed, subject to nuclear sorting.
To achieve nuclear sorting, the basic motif 346 RRKRR 350 in the conserved intracellular region C1 was required but not sufficient. As shown by the use of large molecular weight constructs that restrict diffusion through the nuclear pore, nuclear import was mediated by functional NLS motifs in GlyR␣1 and -␣3 subunits. A similar trafficking process has been observed for the N-methyl-D-aspartate receptor variant NR1. The intracellular domain of NR1 is imported into the nucleus via a functional NLS, as analyzed by recombinant expression of GFP fusion constructs (27). Likewise, the isolated intracellular domain constructs of the receptor-tyrosine kinase notch (28) and the L-type voltage-gated calcium channel Ca V 1.2 (29) are subject to nuclear sorting, indicating that various types of transmembrane proteins harbor nuclear trafficking motifs. For these proteins, however, the proportion of nuclear antigen generated from recombinant or endogenous full-length proteins was low, demonstrating that the detection of nuclear antigen originating from transmembrane proteins requires sensitive detection techniques (27,29,30).
Indeed, the detection of nuclear GlyR antigen by monoclonal antibodies (mAb2b, mAb4a) was limited by low signal to background ratios for nuclear staining. To enhance sensitivity and to allow for detection of intracellular domain fragments, we introduced affinity tags into the TM3-4 loop of the full-length GlyR␣1 subunit. Using high laser intensities in confocal imaging, we succeeded in the detection of nucleoplasmic immunoreactivity when HEK293 cells were transfected with full-length tagged GlyR constructs, indicating nuclear import.
The transmembrane proteins NR1, notch, and Ca V 1.2 undergo proteolytic cleavage before their intracellular domains are translocated into the nucleus (27,29,31). Making use of affinity tags, we analyzed the proteolytic integrity of transfected GlyR constructs in HEK293 cells. A fragment was identified carrying the T7 tag in the C-terminal part of the TM3-4 loop, indicating proteolytic cleavage. As deduced from the apparent molecular weight and resistance to alkaline extraction, this fragment most likely also held the TM4 region as well as the C terminus of the subunit. The proteolytic cleavage probably occurred between the conserved regions C1 and C3 of the TM3-4 loop. This localization is consistent with a study reporting ubiquitination and subsequent cleavage of a C-terminal-tagged GlyR construct expressed in Xenopus oocytes (32). The T7-tagged fragment we observed, however, neither harbored the basic motif 346 RRKRR 350 essential for nuclear import nor did it hold the further N-terminal HA tag used for detection of the GlyR in the nucleus. Thus, it could not have contributed to the observed nucleoplasmic GlyR immunoreactivity. As detection of nuclear soluble fragments might have been limited by sensitivity, we enriched the antigen by nuclear compartmental extraction and immunoprecipitation. Still, we found no evidence for a fragment containing both affinity tags, whereas fulllength GlyR subunits were readily detectable in the nuclear compartment of transfected HEK293 cells. These results indicate that the epitopes present in the nucleus corresponded to full-length GlyR subunits.
Nuclear import relies on an intricate cellular machinery, where translocation of NLS harboring proteins is frequently mediated by karyopherins. Indeed, interaction of the GlyR intracellular domains with karyopherins ␣3 and ␣4 was strong in in vitro studies involving yeast two-hybrid screening and GST pulldown assays. Immunocytochemistry of primary neuronal cultures, however, did not indicate a strong interaction in vivo. Nevertheless, the interaction of karyopherins with the TM3-4 loops of GlyR␣1 and -␣3, but not of GlyR␣2 and -␤, closely resembled the subunit specificity of the nuclear import in our reporter systems. This suggests that the interaction with karyopherins underlies the observed nuclear import of the TM3-4-loop constructs. Besides, the specificity of karyopherin interaction may represent one of the mechanisms resulting in differential trafficking of the GlyR subunits. The basic motif 346 RRKRR 350 was required but not sufficient to mediate nuclear sorting. Resembling the situation for NR1 (27), the interaction motif located within the TM3-4 loop might, therefore, be a bipartite NLS, potentially also involving basic residues in C3.
In addition to nuclear import, extranuclear functions of the interaction between GlyRs and karyopherins are conceivable. Indeed, ␣-karyopherins have been described to mediate transport functions in pre-and postsynaptic compartments of neurons (33)(34)(35)(36). It has been proposed that karyopherin-mediated signaling from the synapse to the nucleus represents a general mechanism for regulation of synaptic plasticity (37). The GlyR might, therefore, serve as a synaptic anchor liberating karyopherins upon channel opening and chloride influx. Eventually, karyopherin ␣3 and ␣4 might mediate the interaction of GlyRs to the dynein transport complex via karyopherin ␤1, representing a second link in addition to gephyrin (38,39).
The basic motif RRKRR has been shown to be part of an important determinant of GlyR␣1 topology, namely RXR-RKRR, in oocyte systems (40). The authors concluded that the basic residues are required to counteract the propensity for intracellular retention of the TM2-3 loop resulting from three basic amino acids therein. Recently, we showed that RRKRR, as a part of RRKRRH, greatly enhances the potential of truncated oscillator GlyR␣1 subunits to be rescued with tail domains, indicating an important role in protein-protein interaction (41). Here, we present the interaction with karyopherins and the resulting regulation of intracellular sorting as a novel role for this multifunctional basic motif. Consistent with its multifunctional role, the basic motif may also mediate interaction to other proteins such as the molecular chaperone VCP/p97, which has been reported to bind short, basic motifs (26). These interactions might represent the molecular mechanism underlying both correct membrane topology and efficient functional rescue. Furthermore, sequence diversity within this motif, as seen for GlyR␣2, may underlie differential regulation of single activities of GlyR subunit variants.