Subtype-selective Interaction with the Transcription

THE JOURNAL OF BIOLOGICAL CHEMISTRY VOL. 280, NO. 39, pp. 33566 –33572, September 30, 2005 © 2005 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in the U.S.A.

Subtype-selective Interaction with the Transcription Factor CCAAT/Enhancer-binding Protein (C/EBP) Homologous Protein (CHOP) Regulates Cell Surface Expression of GABAB Receptors* Received for publication, March 30, 2005, and in revised form, July 26, 2005 Published, JBC Papers in Press, August 4, 2005, DOI 10.1074/jbc.M503482200

Kathrin Sauter‡, Thomas Grampp‡, Jean-Marc Fritschy‡, Klemens Kaupmann§, Bernhard Bettler¶, Hanns Mohler‡储, and Dietmar Benke‡1 From the ‡Institute of Pharmacology and Toxicology, University of Zurich, 8057 Zurich, the 储Department of Chemistry and Applied Biosciences, Federal Institute of Technology (ETH), 8093 Zurich, §Neuroscience Research, Novartis Institutes for BioMedical Research, Novartis Pharma AG, 4002 Basel, and the ¶Pharmazentrum, University of Basel, Department of Clinical-Biological Sciences, Institute of Physiology, 4056 Basel, Switzerland

GABAB2 receptors are G-protein-coupled receptors that play an important role in the control of inhibitory neurotransmission. Presynaptically, GABAB receptors modulate neurotransmitter release by inhibiting Ca2⫹ channels, whereas postsynaptically, they induce slow inhibitory potentials by activating K⫹ channels. Structurally, functional GABAB receptors require the heterodimerization of the two subunits GABAB1 and GABAB2. Heterodimerization is in part mediated by leucine zipper motifs present in the intracellular C-terminal domains of GABAB1 and GABAB2, forming a coiled-coiled structure. Although

* 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 To whom correspondence should be addressed: Institute of Pharmacology and Toxicology, University Zu¨rich, Winterthurerstrasse.190, 8057 Zu¨rich. Tel.: 41-44-635-5930; Fax: 41-44-635-6874; E-mail: [email protected]. 2 The abbreviations used are: GABAB, ␥-aminobutyric acid, type B; CHOP, C/EBP homologous protein; C/EBP, CCAAT/enhancer-binding protein; HEK 293, human embryonic kidney 293; WB, Western blotting; PBS, phosphate-buffered saline.

33566 JOURNAL OF BIOLOGICAL CHEMISTRY

GABAB1 harbors the ligand binding site, GABAB2 contains all the molecular determinants for G-protein coupling and is necessary for trafficking of GABAB receptors to the cell surface by shielding the endoplasmic reticulum retention signal of GABAB1 (for a review, see Refs. 1– 4). Two main GABAB receptor subtypes, GABAB1a/GABAB2 and GABAB1b/GABAB2, are expressed in the brain, as defined by the presence of alternative transcription sites in the GABAB1 gene and by differential promoter usage (5). GABAB1a and GABAB1b display strikingly distinct spatial and temporal expression patterns, with GABAB1a being the predominant isoform in the developing brain and GABAB1b being prevalent in the adult brain. However, the functional significance of GABAB receptor subtypes is not yet understood since no unambiguous differences in function or pharmacology have been detected so far (reviewed in Refs. 1– 4). The molecular mechanisms leading to differential targeting and anchoring of GABAB receptor subtypes to specific synaptic and extrasynaptic sites as well as the regulation of the cell surface number of GABAB receptors are poorly understood. Accessory proteins interacting with GABAB receptors in a temporally and spatially defined manner may be involved in these processes, and distinct GABAB receptor subtypes may be regulated differentially by specific sets of interacting proteins. Extensive searches for GABAB receptor-associated proteins using the yeast two-hybrid system yielded a number of potential interacting proteins. Most interestingly, the transcription factor CREB2/ATF4 was shown to directly interact with the leucine zipper of GABAB1, suggesting a novel G-protein-independent mechanism of signal transduction (6 – 8). Furthermore, GABAB1 associates with 14-3-3 proteins via a site that overlaps with its coiled-coil domain (9). Since 14-3-3 competes with GABAB2 for binding to GABAB1 in vitro, it may regulate GABAB receptor heterodimerization (9). Recently, Marlin-1, a novel brain-specific RNA-binding protein, was shown to interact with GABAB1 via the coiled-coil domain (10). Marlin-1 was proposed to regulate the cellular levels of GABAB2 and thereby to affect the number of functional GABAB receptors (10). These findings demonstrate that in particular, the coiled-coiled domain of the GABAB receptor subunits provides a site of interaction for a variety of different protein partners. However, the precise physiological relevance of these interactions remains to be determined. In addition, none of these proteins were shown to interact with specific GABAB receptor subtypes. In this regard, the presence of two sushi repeats (complement control protein modules) in the extracellular domain of GABAB1a, but not in GABAB1b, and their capability to bind

VOLUME 280 • NUMBER 39 • SEPTEMBER 30, 2005

Downloaded from http://www.jbc.org/ by guest on September 23, 2015

The metabotropic ␥-aminobutyric acid, type B (GABAB) receptors mediate the slow component of GABAergic transmission in the brain. Functional GABAB receptors are heterodimers of the two subunits GABAB1 and GABAB2, of which GABAB1 exists in two main isoforms, GABAB1a and GABAB1b. The significance of the structural heterogeneity of GABAB receptors, the mechanism leading to their differential targeting in neurons as well as the regulation of cell surface numbers of GABAB receptors, is poorly understood. To gain insights into these processes, we searched for proteins interacting with the C-terminal domain of GABAB2. Here, we showed that the transcription factor CCAAT/enhancer-binding protein (C/EBP) homologous protein (CHOP) directly interacts with GABAB receptors in a subtype-selective manner to regulate cell surface expression of GABAB1a/GABAB2 receptors upon co-expression in HEK 293 cells. The interaction of CHOP with GABAB1a/ GABAB2 receptors resulted in their intracellular accumulation and in a reduced number of cell surface receptors. This regulation required the interaction of CHOP via two distinct domains with the heterodimeric receptor; its C-terminal leucine zipper associates with the leucine zipper present in the C-terminal domain of GABAB2, and its N-terminal domain associates with an as yet unidentified site on GABAB1a. In conclusion, the data indicated a subtype-selective regulation of cell surface receptors by interaction with the transcription factor CHOP.

Subtype-selective Interaction of CHOP with GABAB Receptors

EXPERIMENTAL PROCEDURES Antibodies—The following primary antibodies were used: rabbit GABAB1 (18) (1:250 for Western blotting (WB)), guinea pig GABAB1 (Chemicon; 1:5000 for immunofluorescence), rabbit GABAB2 (19) (1:100 for WB, 1:500 for immunofluorescence), guinea pig GABAB2 (Chemicon; 1:5000 for WB), mouse CHOP (B3, Santa Cruz Biotechnology; 1:1000 for WB, 1:500 for immunofluorescence), mouse hemagglutinin tag (12CA5, gift from G. Radziwill, Zurich, Switzerland; 1:1000 for WB and immunofluorescence). In addition, a polyclonal antiserum directed against recombinant CHOP protein (amino acids 35–170) overexpressed in Escherichia coli was generated by immunizing rabbits. CHOP antibodies were affinity-purified using the recombinant protein covalently attached to Affi-Gel 15 (Bio-Rad). Specificity of the CHOP antibodies was assessed by Western blotting (1:750) and immunocytochemistry (1:50) using HEK 293 cells expressing CHOP as compared with non-transfected HEK 293 cells. Plasmids—Expression plasmids containing GABAB1a, GABAB1b, GABAB2, GABAB1aI860 (named GABAB1a⌬C in this study), GABAB2T749 (named GABAB2⌬C in this study), GABAB2⌬LZ2, and GABAB2LZ1 were described previously (20 –22). Full-length cDNA of human CHOP was kindly provided by R. Nehring (Go¨ttingen, Germany). Deletion mutants of CHOP were generated by PCR and inserted into pcDNA3.1-TOPO CHOP⌬LZ lacking the last 108 nucleotides (forward primer, 5⬘-GAGAATGAAAGGAAAGTGGCACAG-3⬘ and reverse primer, 5⬘-CTGTGCCACTTTCCTTTCATTCTC-3⬘) and CHOP⌬N lacking the first 42 amino acids (forward primer, 5⬘-ACCATGAATGAAGAGGAAGAATCAAAA-3⬘ and reverse primer, 5⬘-TGCTTGGTGCAGATTCACCAT-3⬘). Yeast Two-hybrid Assay—The yeast two-hybrid screen was performed by Invitrogen. The entire C-terminal domain of human GABAB2 (accession number AF056085, amino acid sequence LITLRT . . . VMVSGL) was used to screen a human brain cDNA library. Cell Culture—HEK 293 cells were maintained in minimum Eagle’s medium containing 10% (v/v) fetal calf serum, 2 mM glutamine and transfected with appropriate plasmids by calcium phosphate precipita-

SEPTEMBER 30, 2005 • VOLUME 280 • NUMBER 39

tion. Transfected cells were used 2 days after transfection for subsequent immunofluorescence or immunoprecipitation studies. Primary cultured neurons were prepared from rat embryos taken at embryonic day 19 from pregnant Wistar rats anesthetized with ether. The hippocampus was dissected on ice in PBS containing 5.5 mM glucose. The tissue was dissociated with trypsin and resuspended in Dulbecco’s modified Eagle’s medium containing 10% fetal calf serum and plated to a density of 50,000 cells/cm2 onto poly-L lysine-coated coverslips and transferred to 12-well plates containing glial feeder cells in a defined serum-free medium (Dulbecco’s modified Eagle’s medium without phenol red containing B27 supplement, 250 ␮g/ml AlbuMAX, 11 mg/ml sodium-pyruvate, 5 mM glutamine, and 50 ␮g/ml gentamycin). Immunoprecipitation—Transfected HEK 293 cells were harvested and homogenized in lysis buffer (PBS containing 5 mM EDTA, 5 mM EGTA, 50 mM sodium fluoride, 1 mM sodium orthovanadate, 25 ␮g/ml aprotinin, 1 mM benzamidine, and 1 mM phenylmethylsulfonyl fluoride). After homogenization, cells were solubilized by the addition of 1% Triton X-100 (final concentration) for 30 min on ice followed by centrifugation at 100,000 ⫻ g for 30 min. The supernatant was subjected to immunoprecipitation using the respective GABAB receptor antibodies covalently immobilized to protein A-agarose or protein G-agarose, respectively. Immunoprecipitation was performed for 4 h at 4 °C. Immune complexes were collected by centrifugation and extensively washed with lysis buffer containing 1% Triton X-100. Bound proteins were released with 2⫻ sample buffer for SDS-polyacrylamide gel electrophoresis, heated, and analyzed by Western blotting. Immunoprecipitation of GABAB receptors from rat brain tissue was performed essentially as described previously (18). Immunocytochemistry—Double and triple labeling immunocytochemistry was performed with transiently transfected HEK 293 cells and primary cultured neurons. HEK 293 cells transfected with the appropriate plasmids were fixed with 4% paraformaldehyde in 150 mM phosphate buffer, pH 6.8 –7.5, permeabilized for 5 min with 0.5% Triton X-100 in PBS and incubated with the primary antibodies diluted to the appropriate concentration (see above) in PBS containing 10% normal goat serum. After washing extensively with PBS, the cells were incubated with secondary antibody coupled to Alexa Fluor 488 (1:1000, Molecular Probes), Cy-3 (1:500, Jackson ImmunoResearch), and/or Cy-5 (1:100, Jackson ImmunoResearch). After three washes for 10 min with PBS, the cells were mounted in 90% glycerol containing 0.1% phenylenediamine, 50 mM Tris, pH 7.5, and 10 mM NaCl. Cells were analyzed by confocal laser scanning microscopy (LSM 510 Meta, Zeiss), and images were processed using Imaris (Bitplane, Zurich, Switzerland). Cell Surface Enzyme-linked Immunosorbent Assay—For labeling of cell surface receptors, living HEK 293 cells transiently transfected with GABAB1a/GABAB2 or GABAB1a/GABAB2/CHOP plated on 96-well plates were incubated with an antibody directed against the N terminus of GABAB2 (18) in buffer A (25 mM HEPES 7.4, 2 mM CaCl2, 2 mM MgCl2, 30 mM glucose, 5 mM KCl, 119 mM NaCl) containing 450 mM sucrose and 10% normal goat serum for 40 min at room temperature. After washing the cells extensively with buffer A, the cells were fixed with 4% paraformaldehyde in 150 mM phosphate buffer, pH 6.8 –7.5, and incubated with horseradish peroxidase-conjugated anti-rabbit antibodies for 60 min. After extensive washes, horseradish peroxidase activity was determined by tetramethylbenzidine as substrate (100 ␮l/well 0.24 mg/ml tetramethylbenzidine, 0.2 M sodium citrate, pH 3.95, 0.006% H2O2). The color reaction was terminated by the addition of 100 ␮l of 1 M H2SO4, and the optical density was recorded at 450 nm in a microplate reader (Synergy HT, Biotek). The abundance of total receptors was determined in parallel cultures after fixation and permeabilization of the

JOURNAL OF BIOLOGICAL CHEMISTRY

33567


the extracellular matrix protein fibulin-2 in vitro (11), provide evidence for the existence of subtype-selective interacting proteins. We used the yeast two-hybrid system to search for novel GABAB receptor-interacting proteins and identified the transcription factor CHOP as a potential GABAB2-associated protein. CHOP (also designated as gadd153 or C/EBP␨) belongs to the CCAAT/enhancer-binding protein (C/EBP) family of transcription factors, which all contain a characteristic basic-leucine zipper motif in their extreme C-terminal domain, whereby the basic region serves for DNA binding and the leucine zipper as a protein-protein interaction module (12). CHOP is strongly up-regulated by factors that interfere with normal function of the endoplasmic reticulum (13) and has been implicated in the control of endoplasmic reticulum stress-induced apoptosis associated with diabetes (14 –16). In addition, CHOP also appears to be involved in ischemia-induced apoptosis since CHOP knock-out mice exhibit strongly reduced ischemia-induced neuronal cell death (17). Here, we identified CHOP as a novel GABAB receptor-interacting protein and demonstrate its co-localization in cultured hippocampal neurons as well as its physical association with GABAB receptors in brain extracts by co-immunoprecipitation. Co-expression in HEK 293 cells revealed that CHOP interacts selectively with GABAB1a/GABAB2 receptors via two distinct binding sites, which results in an intracellular accumulation and thereby in a reduced cell surface expression of the receptors.


cells with 0.5% Triton X-100. Nonspecific antibody reaction was determined in parallel cultures of non-transfected cells.

RESULTS Interaction of CHOP with GABAB Receptors in Neurons—Screening with the yeast two-hybrid system of a brain cDNA expression library using the entire intracellular C-terminal domain of GABAB2 yielded the transcription factor CHOP as a putative GABAB receptor-associated protein. An association of CHOP with native GABAB receptors was tested by co-immunoprecipitation of CHOP with GABAB receptors from brain extracts and by determination of the immunocytochemical distribution of CHOP and GABAB receptors in primary cultured hippocampal neurons. GABAB receptors were immunoprecipitated with antibodies recognizing either GABAB1 or GABAB2 and analyzed for the presence of CHOP. In both GABAB receptor immunoprecipitates, CHOP was detected, demonstrating the physical association of CHOP with native GABAB receptors (Fig. 1A). In agreement with their coimmunoprecipitation, double labeling immunofluorescence microscopy of primary cultured hippocampal neurons revealed an extensive overlap in the subcellular distribution of GABAB receptors and CHOP (Fig. 1B). Both GABAB1 and GABAB2 were distributed in the soma and neurites of the neurons. Similar to GABAB receptors, CHOP exhibited a widespread distribution in the soma and neurites (Fig. 2B). Apparent co-localization of CHOP with GABAB receptors was observed in the soma and in large patches on the neurites (Fig. 1B). In conclusion, these experiments demonstrated that CHOP interacts with GABAB receptors in their native environment, i.e. in the brain. Interaction of CHOP with GABAB2 in HEK 293 Cells—To analyze the interaction of CHOP with GABAB receptors in more detail, the effect of co-expression in HEK 293 cells of CHOP with GABAB2 in the absence of GABAB1 was tested. In contrast to GABAB1, GABAB2 was trafficked to the cell surface on its own. Immunocytochemistry revealed that in the absence of GABAB2, CHOP was specifically localized to the nucleus (Fig. 2A, a). However, co-expression of CHOP with GABAB2 resulted in a cytoplasmic distribution of CHOP, where it co-localized extensively with GABAB2 (Fig. 2A, b). The redistribution of CHOP upon co-expression with GABAB2 argues for a direct interaction of CHOP and GABAB2.


FIGURE 2. Interaction of CHOP with GABAB2 in HEK 293 cells. A, HEK 293 cells transiently expressing CHOP alone (a), CHOP and GABAB2 (b), or CHOP and a deletion mutant of GABAB2 lacking its C-terminal domain (c, GABAB2⌬C) were processed for immunofluorescence staining using antibodies against CHOP (red) and GABAB2 (green). In the absence of GABAB2, CHOP was localized to the nucleus (a), whereas in the presence of GABAB2, CHOP was redistributed to cytoplasmic sites and the plasma membrane, where they co-localized (b, yellow). Deletion of the entire C-terminal domain of GABAB2 resulted in the loss of co-localization and redistribution of CHOP (c), indicating that the CHOPGABAB2 interaction is specifically mediated by the C-terminal domain of GABAB2. B, coimmunoprecipitation of CHOP with GABAB2. HEK 293 cells co-expressing CHOP and GABAB2 were solubilized and subjected to immunoprecipitation using GABAB2 antibodies. The presence of CHOP (left panel) and GABAB2 (right panel) in the immunoprecipitate was analyzed by Western blotting. The very strong CHOP signal in the extract as compared with that in the GABAB2 immunoprecipitate suggests that only a fraction of CHOP expressed in the cell interacts with GABAB2. Extract, Triton X-100 solubilized proteins; IP GABAB2, proteins immunoprecipitated by GABAB2 antibodies; Control, precipitation with protein A-agarose in the absence of GABAB2 antibodies.

Since CHOP was isolated with the C-terminal domain of GABAB2, the specificity of the CHOP-GABAB2 interaction was determined using a deletion mutant of GABAB2 lacking its entire C-terminal domain (GABAB2⌬C). Upon co-expression with the truncated GABAB2, the redistribution of CHOP to cytoplasmic sites was prevented, and consequently, no co-localization of CHOP with GABAB2 was observed (Fig. 2A, c). The physical association of CHOP and GABAB2 in HEK 293 cells was verified by co-immunoprecipitation using an antibody directed against the C terminus of GABAB2. By immunoblotting, CHOP was detected in the GABAB2 immunoprecipitate, demonstrating that CHOP associates with GABAB2 (Fig. 2B). Isoform-selective Interaction of CHOP with GABAB1 in HEK 293 Cells— To investigate whether CHOP interacts with GABAB receptors in a subunit-specific manner, a potential association of CHOP with the two main GABAB1 isoforms was tested. CHOP was co-expressed in HEK 293 cells with GABAB1a or GABAB1b, respectively, and analyzed for physical association by immunoprecipitation. Surprisingly, CHOP co-



FIGURE 1. Co-immunoprecipitation and co-localization of CHOP with GABAB receptors in neurons. A, co-immunoprecipitation of CHOP with GABAB receptors from a rat brain extract. Brain tissue was solubilized with 0.5% deoxycholate, and the extract was subjected to immunoprecipitation using either GABAB1 or GABAB2 antibodies followed by Western blotting using CHOP antibodies. Extract, deoxycholate solubilized proteins; IP GABAB1, proteins immunoprecipitated by GABAB1 antibodies; IP GABAB2, proteins immunoprecipitated by GABAB2 antibodies; Control, precipitation with protein A-agarose in the absence of antibodies. B, co-localization of CHOP (red) with GABAB1 or GABAB2 (green) in primary cultured hippocampal cells by immunofluorescence staining. GABAB1 and GABAB2 were expressed throughout the neuron and co-distributed (yellow) extensively with CHOP in the soma and on the neurites.


immunoprecipitated with GABAB1a but not with GABAB1b (Fig. 3A). Thus, CHOP was not only able to interact with GABAB2 but also with GABAB1a. At the subcellular level, double labeling immunofluorescence microscopy revealed a widespread distribution of GABAB1a throughout the cell with the exception of the nucleus (Fig. 3B, a). This distribution was in line with the endoplasmic reticulum retention of GABAB1 when expressed in the absence of GABAB2 (22–24). Co-expression of GABAB1a with CHOP resulted in a predominant perinuclear localization of CHOP, where CHOP was partially co-localized with GABAB1a (Fig. 3B, a). In contrast, upon co-expression of CHOP with GABAB1b, no co-localization was observed. GABAB1b displayed its expected intracellular distribution in the absence of GABAB2, whereas CHOP was localized to the nucleus (Fig. 3B, b). These observations demonstrated that the interaction of CHOP with GABAB1 is restricted to the GABAB1a isoform. Identification of the CHOP-GABAB Receptor Interaction Domains— The presence of a leucine zipper motif in the C-terminal domains of both CHOP and GABAB2 argued for an association via these structures. Co-expression of CHOP with a GABAB2 deletion mutant lacking its leucine zipper (GABAB2⌬LZ) resulted in the loss of co-localization and redistribution of CHOP (Fig. 4A). Accordingly, co-expression of fulllength GABAB2 with a CHOP deletion mutant lacking its leucine zipper (CHOP⌬LZ) resulted likewise in the loss of co-localization and redistribution of CHOP (Fig. 4B). Thus, the CHOP-GABAB2 interaction was


FIGURE 4. Identification of CHOP-GABAB receptor interaction domains. HEK 293 cells expressing GABAB receptor (green) or CHOP (red) deletion mutants were analyzed for co-localization by immunofluorescence microscopy. A and B, co-expression of a GABAB2 deletion mutant lacking its leucine zipper (GABAB2⌬LZ) with wild-type CHOP (A) or wildtype GABAB2 with a CHOP deletion mutant lacking its leucine zipper (CHOP⌬LZ) (B) resulted in the loss of co-localization and redistribution of CHOP, indicating that the CHOP-GABAB2 interaction is mediated by the respective leucine zippers. C, co-expression of CHOP with a GABAB1 insertion mutant, in which the leucine zipper of GABAB2 was replaced with the leucine zipper of GABAB1 (GABAB2LZ1), resulted likewise in the loss of co-localization and redistribution of CHOP, showing that CHOP specifically associates to the leucine zipper of GABAB2 but not to the leucine zipper of GABAB1. D and E, coexpression of a GABAB1a deletion mutant lacking its entire C-terminal domain (GABAB2⌬C) with wild-type CHOP (D) or wild-type GABAB1a with a CHOP deletion mutant lacking its leucine zipper (E, CHOP⌬LZ) resulted in the co-localization and redistribution of CHOP, demonstrating that neither the C-terminal domain of GABAB1a nor the leucine zipper of CHOP is involved in the CHOP-GABAB1a interaction site. F, co-expression of a CHOP deletion mutant lacking the first 42 N-terminal amino acids (CHOP⌬N) with GABAB1a resulted in the loss of co-localization of CHOP with GABAB1a, indicating that the N-terminal domain of CHOP constitutes the binding site with GABAB1a.

mediated by the leucine zipper motifs present in their C-terminal domains. Since GABAB1 also contains a leucine zipper motif in its C-terminal domain, it was tested whether the leucine zipper of GABAB1 also interacts with CHOP. Co-expression of CHOP with a GABAB2 insertion mutant in which the leucine zipper of GABAB2 was replaced by the leucine zipper of GABAB1 (GABAB2⌬LZ1) resulted in a loss of co-localization and redistribution of CHOP (Fig. 4C). These findings demonstrated that CHOP associates specifically with the leucine zipper of GABAB2 but not with the leucine zipper of GABAB1. To determine the interaction site of CHOP and GABAB1a, HEK 293 cells were transiently transfected with full-length CHOP and with a deletion mutant of GABAB1a lacking its entire cytoplasmic C-terminal domain (GABAB1a⌬C, Fig. 3D) or with CHOP lacking its leucine zipper and full-length GABAB1a (Fig. 3E). In both cases, a co-localization of CHOP and GABAB1a was found. Thus, the association of CHOP with GABAB1a was neither mediated via the leucine zipper of CHOP nor mediated via the C-terminal domain of GABAB1a. However, truncation of the N-terminal amino acids 1– 42 of CHOP (CHOP⌬N) abolished the redistribution of CHOP and its co-localization with GABAB1a, demonstrating that CHOP interacts with its N-ter-


33569


FIGURE 3. Selective association of CHOP with GABAB1a but not GABAB1b. A, co-immunoprecipitation of CHOP with GABAB1a. Detergent extracts of HEK 293 cells expressing either CHOP and GABAB1a (left panels) or CHOP and GABAB1b (right panels) were subjected to immunoprecipitation using GABAB1 antibodies coupled to protein A-agarose and subsequent to Western blotting using CHOP and GABAB1 antibodies. CHOP co-immunoprecipitated with GABAB1a, but not with GABAB1b. Extract, Triton X-100 solubilized proteins; Extract depleted, extract after immunoprecipitation of GABAB1 or GABAB2, respectively, indicating quantitative precipitation of GABAB1; IP GABAB1a, proteins immunoprecipitated by GABAB1 antibodies; IP GABAB1b, proteins immunoprecipitated by GABAB1 antibodies; Control, precipitation with protein A-agarose in the absence of antibodies. B, HEK 293 cells expressing either GABAB1a or GABAB1b (green) and CHOP (red) were analyzed for co-localization by immunofluorescence staining. a, co-expression of CHOP with GABAB1a resulted in a perinuclear localization of CHOP, where it colocalized with GABAB1a. b, upon co-expression of CHOP with GABAB1b, CHOP displayed an entirely nuclear localization and consequently no co-localization with GABAB1b.



FIGURE 5. Interaction of CHOP with heterodimeric GABAB1a/GABAB2 receptors leads to a perturbed localization and a reduction of cell surface receptors. A, immunocytochemical localization of either GABAB1a or GABAB1b (green) with GABAB2 (red) and CHOP (blue) in HEK 293 cells. a, in the absence of CHOP, GABAB1a and GABAB2 were targeted to the cell surface. b, co-expression of GABAB1a/GABAB2 receptors with CHOP resulted in a prominent intracellular accumulation of GABAB1a as well as GABAB2 and frequently resulted in the formation of large GABAB2-positive inclusions in or in the vicinity of the nucleus. A restricted co-localization of CHOP with GABAB receptors at cytoplasmic and perinuclear sites was found. c and d, the intracellular accumulation of GABAB1a/GABAB2 receptors was mediated by CHOP since co-expression of GABAB1a/ GABAB2 with CHOP deletion mutants lacking either the C-terminal leucine zipper (CHOP⌬LZ) that is not able to interact with GABAB2 (c) or the N-terminal amino acids 1– 42 (CHOP⌬N) that cannot interact with GABAB1a (d) resulted in a normal cell surface expression of GABAB1a/GABAB2 receptors. In c, CHOP⌬LZ remained redistributed and co-localized with GABAB receptors at the cell surface. This is most likely due to the interaction of the N-terminal domain of CHOP with GABAB1a. Similarly, in d, CHOP⌬N remained partially redistributed and was co-localized with GABAB2 at the cell surface. e, CHOP did not interact with GABAB1b/GABAB2 receptors, as evidenced by the normal cell surface localization of GABAB1 and GABAB2 and the entirely nuclear localization of CHOP. B, reduced or lack of permanent association of CHOP with heterodimeric GABAB1a/GABAB2 receptors. HEK 293 cells expressing GABAB1a/GABAB2 receptors and CHOP were solubilized and subjected to immunoprecipitation with GABAB1 (right panel) or GABAB2 antibodies (left panel) followed by Western blotting for detection of GABAB1, GABAB2, and CHOP. Only in the GABAB1 immunoprecipitate was a faint CHOP signal detected. Extract, Triton X-100 solubilized proteins; Extract depleted, extract after immunoprecipitation of GABAB1 or GABAB2; IP GABAB1, proteins immunoprecipitated by GABAB1 antibodies; IP GABAB2, proteins immunoprecipitated by GABAB2 antibodies; Control, precipitation with protein A-agarose in the absence of antibodies. C, reduction of cell surface GABAB1a/GABAB2 receptors upon co-expression with CHOP. Cell surface receptors of living HEK 293 cells expressing GABAB1a/ GABAB2 receptors in the presence and absence of CHOP were labeled with antibodies directed against the N-terminal domain of GABAB2 and quantified by enzyme-linked immunosorbent assay. The total amounts of receptors expressed were determined upon fixation and permeabilization of the cells prior to the incubation with the antibody (100%). The values represent the mean ⫾ S.D. from three independent experiments.



minal domain with an as yet unidentified site on GABAB1a (Fig. 4F). Thus, CHOP interacted via different domains with GABAB1a and GABAB2. Cell Surface GABAB1a/GABAB2 Receptors Are Regulated by CHOP— The previous results demonstrated that CHOP interacts with GABAB2 and GABAB1a but not with GABAB1b. Since the expression of functional GABAB receptors requires the heterodimerization of both GABAB1 and GABAB2, the effect of co-expression of CHOP with heterodimeric GABAB receptors was analyzed. When co-expressed in HEK 293 cells, GABAB1a and GABAB2 were targeted to the cell surface, where they are co-localized (Fig. 5A, a). In contrast, co-expression of CHOP with GABAB1a and GABAB2 resulted in a dramatic alteration of receptor distribution. In the presence of CHOP, GABAB1a and GABAB2 displayed, besides their normal cell surface expression, an unusual intracellular localization (Fig. 5A, b). Co-localization of all three proteins was occasionally detected at cytoplasmic and perinuclear sites. In addition, GABAB2, but not GABAB1a, was often found in inclusions in or in the vicinity of the nucleus. CHOP staining was detected in the nucleus and cytoplasmic sites and was partially co-localized with GABAB1a and GABAB2 at cytoplasmic and perinuclear sites (Fig. 5A, c). This is in contrast to the abundant co-localization and association of CHOP with either GABAB2 alone (Fig. 2) or GABAB1a alone (Fig. 3) shown above. To verify a reduced association of CHOP with heterodimeric GABAB1a/ GABAB2 receptors, GABAB1a, GABAB2, and CHOP were co-expressed in HEK 293 cells and subjected to immunoprecipitation using GABAB1or GABAB2-selective antibodies. Immunoblotting revealed the lack of co-immunoprecipitation of CHOP in the GABAB2 immunoprecipitate, whereas a weak signal of CHOP was detected in the GABAB1a immunoprecipitate (Fig. 5B). The weak CHOP signal in the GABAB1a immunoprecipitate is in line with the co-localization of CHOP and GABAB1a observed in the immunocytochemical experiment (Fig. 5A, b). The lack or strongly reduced co-immunoprecipitation of CHOP with GABAB2 and GABAB1a, respectively, and the intracellular accumulation of GABAB1a and GABAB2 suggested that a transient interaction of CHOP with heterodimeric GABAB1a/GABAB2 receptors may induce the internalization of cell surface receptors, or alternatively, disturb the targeting of receptors to the cell surface. To analyze whether the perturbed localization of GABAB1a/GABAB2 receptors is specifically mediated by CHOP, GABAB1a/GABAB2 receptors were co-expressed with CHOP lacking its leucine zipper (CHOP⌬LZ) that is not able to interact with GABAB2. Indeed, co-expression of CHOP⌬LZ with GABAB1a/GABAB2 receptors prevented their intracellular accumulation and resulted in a normal cell surface distribution of both GABAB1a and GABAB2 without the formation of GABAB2-positive inclusions (Fig. 5A, c). Interestingly, CHOP⌬LZ was still co-localized with GABAB1a/GABAB2 receptors at the cell surface (Fig. 5A, c). This co-localization of CHOP⌬LZ with GABAB receptors was most likely due to an interaction with GABAB1a, which binds to the N-terminal domain of CHOP, not involving the leucine zippers (Fig. 4). Similarly, co-expression of the deletion mutant of CHOP lacking the N-terminal amino acids 1– 42 (CHOP⌬N) with GABAB1a and GABAB2 resulted in a normal, cell surface localization of GABAB2 and the absence of intracellular GABAB2-positive inclusions (Fig. 5A, d). These results indicated that the perturbed localization of GABAB1a/GABAB2 receptors and the formation of GABAB2-positive cytoplasmic inclusions was specifically mediated by CHOP and required the interaction of CHOP via two distinct domains with both GABAB1a and GABAB2. Since the immunoprecipitation data indicated a strongly reduced association of CHOP upon co-expression, most likely, a transient interaction of CHOP with heterodimeric GABAB1a/GABAB2 receptors induced the intracellular accumulation of GABAB1a and GABAB2.


DISCUSSION The results of the present study have provided the first example of a subtype-selective regulation of GABAB receptors by an interacting protein. Yeast two-hybrid screens identified the transcription factor CHOP as a potential interacting protein. CHOP interacted via its C-terminal leucine zipper with the leucine zipper present in the C-terminal domain of GABAB2. This interaction was highly specific since replacing the leucine zipper of GABAB2 with the leucine zipper of GABAB1 abolished the association. Unexpectedly, CHOP also interacted through its N-terminal domain with an as yet undefined site of GABAB1a but not with GABAB1b. Since GABAB1a and GABAB1b share identical amino acid sequences in their intracellular domains, GABAB1a and GABAB1b may differ in their conformation, which results in the formation of a binding site for CHOP selectively on GABAB1a. Co-expression of CHOP with heterodimeric GABAB1a/GABAB2 receptors resulted in a perturbed localization of the receptors to intracellular sites along with frequent GABAB2-positive inclusions and a reduction of cell surface receptors. To exert this effect, CHOP must bind to both GABAB1a (via its N-terminal domain) and GABAB2 (via its C-terminal leucine zipper), which explains the GABAB1a/GABAB2 subtype selectivity of the interaction. The intracellular accumulation of GABAB1a/GABAB2 receptors was specifically mediated by CHOP since deletion of either the first 42 N-terminal amino acids interacting with GABAB1a or the C-terminal leucine zipper of CHOP interacting with GABAB2 abolished the perturbed localization of GABAB1a/GABAB2 receptors and the formation of GABAB2-positive inclusions. Currently, it is unclear whether CHOP induces the internalization of cell surface receptors or disturbs the targeting of receptors to the cell surface. The observation that the distribution of GABAB1b/GABAB2 receptors was not affected by co-expression of CHOP indicated that CHOP cannot interact alone via its leucine zipper with the coiled-coil domain of GABAB2 in the presence of GABAB1. Similar findings were reported for ATF4 and 14-3-3, which interact selectively with the leucine zipper of GABAB1 solely in the absence of GABAB2 (7–9). Thus, the coiled-coiled domain of GABAB1 and GABAB2 appeared to provide a site of interaction for a number of different protein partners for which binding should be mutually exclu-


sive. Accordingly, the interaction of CHOP via two distinct sites with GABAB1a/GABAB2 receptors may at least transiently disrupt or weaken the heterodimers. The formation of the large GABAB2-positive inclusions that are largely devoid of staining for GABAB1 further supported this hypothesis. In this regard, the prominent co-localization of CHOP with GABAB1 and GABAB2 in cultured hippocampal neurons as well as their co-immunoprecipitation from brain extracts not only indicated that the interaction of CHOP with GABAB receptors is of physiological relevance but also suggested that GABAB1 and GABAB2 may exist in appreciable pools of unassembled subunits. The mechanism by which CHOP induces the perturbed localization of GABAB1a/GABAB2 remains to be determined. The results of the present study fit well to a hypothetical model in which CHOP transiently interacts selectively with GABAB1a/GABAB2 receptors to induce their intracellular accumulation. This mechanism would allow selective down-regulation of the cell surface expression of a specific GABAB receptor subtype upon up-regulation of CHOP expression. In this regard, it is interesting to note that the expression of CHOP is strongly up-regulated upon endoplasmic reticulum stress (12). In neurons, CHOP expression has been shown to be induced preceding to N-methyl-D-aspartate- or dopamine-induced cytotoxicity (25) as well as transient cerebral ischemia (17, 26). In addition, CHOP was shown to be up-regulated in response to mutations in presenilin-1, which cause early onset familial Alzheimer disease (27). Therefore, induction of CHOP might be a powerful mechanism to regulate the cell surface expression of GABAB1a/GABAB2 receptors under pathophysiological conditions. REFERENCES 1. Bettler, B., Kaupmann, K., Mosbacher, J., and Gassmann, M. (2004) Physiol. Rev. 84, 835– 867 2. Bowery, N. G., Bettler, B., Froestl, W., Gallagher, J. P., Marshall, F., Raiteri, M., Bonner, T. I., and Enna, S. J. (2002) Pharmacol. Rev. 54, 247–264 3. Calver, A. R., Davies, C. H., and Pangalos, M. N. (2002) NeuroSignals 11, 299 –314 4. Couve, A., Calver, A. R., Fairfax, B., Moss, S. J., and Pangalos, M. N. (2004) Biochem. Pharmacol. 68, 1527–1536 5. Steiger, J. L., Bandyopadhyay, S., Farb, D. H., and Russek, S. J. (2004) J. Neurosci. 24, 6115– 6126 6. Nehring, R. B., Horikawa, H. P. M., El Far, O., Kneussel, M., Brandstatter, J. H., Stamm, S., Wischmeyer, E., Betz, H., and Karschin, A. (2000) J. Biol. Chem. 275, 35185–35191 7. Vernon, E., Meyer, G., Pickard, L., Dev, K., Molnar, E., Collingridge, G. L., and Henley, J. M. (2001) Mol. Cell. Neurosci. 17, 637– 645 8. White, J. H., McIlhinney, R. A. J., Wise, A., Ciruela, F., Chan, W.-Y., Emson, P. C., Billintion, A., and Marschall, F. H. (2000) Proc. Natl. Acad. Sci. U. S. A. 97, 13967–13972 9. Couve, A., Kittler, J. T., Uren, J. M., Calver, A. R., Pangalos, M. N., Walsh, F. S., and Moss, S. J. (2001) Mol. Cell. Neurosci. 17, 317–328 10. Couve, A., Restituito, S., Brandon, J. M., Charles, K. J., Bawagan, H., Freeman, K. B., Pangalos, M. N., Calver, A. R., and Moss, S. J. (2004) J. Biol. Chem. 279, 13934 –13943 11. Blein, S., Ginham, R., Uhrin, D., Smith, B. O., Soares, D. C., Veltel, S., McIlhinney, R. A., White, J. H., and Barlow, P. N. (2004) J. Biol. Chem. 279, 48292– 48306 12. Ramji, D. P., and Foka, P. (2002) Biochem. J. 365, 561–575 13. Oyadomari, S., Araki, E., and Mori, M. (2002) Apoptosis 7, 335–345 14. Oyadomari, S., Koizumi, A., Takeda, T., Gotoh, T., Akira, S., Araki, E., and Mori, M. (2002) J. Clin. Investig. 109, 525–532 15. Oyadomari, S., Takeda, K., Takiguchi, M., Gotoh, T., Matsumoto, M., Wada, I., Akira, S., Araki, E., and Mori, M. (2001) Proc. Natl. Acad. Sci. U. S. A. 98, 10845–10850 16. Zinszner, H., Kuroda, M., Wang, X. Z., Batchvarova, N., Lightfoot, R. T., Remotti, H., Stevens, J. L., and Ron, D. (1998) Genes Dev. 12, 982–995 17. Tajiri, S., Oyadomari, S., Yano, S., Morioka, M., Gotoh, T., Hamada, J.-I., Ushio, Y., and Mori, M. (2004) Cell Death Differ. 11, 403– 415 18. Benke, D., Honer, M., Michel, C., Bettler, B., and Mohler, H. (1999) J. Biol. Chem. 274, 27323–27330 19. Benke, D., Michel, C., and Mohler, H. (2002) J. Recept. Signal Transduc. 22, 253–266 20. Kaupmann, K., Huggel, K., Heid, J., Flor, P. J., Bischoff, S., Mickel, S. J., McMaster, G., Angst, C., Bittiger, H., Froestl, W., and Bettler, B. (1997) Nature 386, 239 –246 21. Kaupmann, K., Malitschek, B., Schuler, V., Heid, J., Froestl, W., Beck, P., Mosbacher, J., Bischoff, S., Kulik, A., Shigemoto, R., Karshin, A., and Bettler, B. (1998) Nature 396, 683– 687


33571


To test whether CHOP induces a reduction of GABAB1a/GABAB2 receptors on the cell surface, the abundance of cell surface receptors was quantified by enzyme-linked immunosorbent assay. HEK 293 cells expressing GABAB1a/GABAB2 in the absence or presence of CHOP were seeded onto 96-well plates, and cell surface receptors were determined by incubating living cells with an antibody directed against the N terminus of GABAB2. The abundance of total receptors expressed was measured by staining fixed and permeabilized cells of parallel cultures. Quantification of the colorimetric enzymatic reactions revealed a 27 ⫾ 9% (n ⫽ 3) reduction of cell surface receptors (Fig. 5C). This value might be underestimated since not all HEK 293 cells expressing GABAB1a/ GABAB2 receptors co-express CHOP. In contrast to the co-expression of CHOP with GABAB1a/GABAB2 receptors, co-expression of CHOP with GABAB1b/GABAB2 receptors resulted in a normal cell surface expression of GABAB1b and GABAB2 with CHOP being entirely localized to the nucleus (Fig. 5A, e). Thus, CHOP was not able interact with GABAB1b/GABAB2 receptors. Although CHOP associates with GABAB2 when expressed in the absence of GABAB1 (Fig. 2A), the presence of GABAB1b apparently prevents its interaction with GABAB2. This may be due to a competition of binding between the leucine zippers of GABAB1 and CHOP for GABAB2. This finding supported the observation that the effect of CHOP on GABAB1a/GABAB2 requires the interaction with both GABAB1a and GABAB2.

Subtype-selective Interaction of CHOP with GABAB Receptors 22. Pagano, A., Rovelli, G., Mosbacher, J., Lohmann, T., Duthey, B., Stauffer, D., Ristig, D., Schuler, V., Meigel, I., Lampert, C., Stein, T., Prezeau, L., Blahos, J., Pin, J. P., Froestl, W., Kuhn, R., Heid, J., Kaupmann, K., and Bettler, B. (2001) J. Neurosci. 21, 1189 –1202 23. Couve, A., Filippov, A. K., Connelly, C. N., Bettler, B., Brown, D. A., and Moss, S. J. (1998) J. Biol. Chem. 273, 26361–26367 24. Margeta-Mitrovic, M., Jan, Y. N., and Jan, L. Y. (2000) Neuron 27, 97–106

25. Stokes, A. H., Freeman, W. M., Mitchell, S. G., Burnette, T. A., Hellmann, G. M., and Vrana, K. E. (2002) Neurotoxicology 23, 675– 684 26. Paschen, W., Gissel, C., Linden, T., Althausen, S., and Doutheil, J. (1998) Mol. Brain Res. 60, 115–122 27. Milhavet, O., Martindale, J. L., Camandola, S., Chan, S. L., Gary, D. S., Cheng, A., Holbrook, N. J., and Mattson, M. P. (2002) J. Neurochem. 83, 673– 681




Mechanisms of Signal Transduction: Subtype-selective Interaction with the Transcription Factor CCAAT/Enhancer-binding Protein (C/EBP) Homologous Protein (CHOP) Regulates Cell Surface Expression of GABA B Receptors

J. Biol. Chem. 2005, 280:33566-33572. doi: 10.1074/jbc.M503482200 originally published online August 4, 2005

Access the most updated version of this article at doi: 10.1074/jbc.M503482200 Find articles, minireviews, Reflections and Classics on similar topics on the JBC Affinity Sites. Alerts: • When this article is cited • When a correction for this article is posted Click here to choose from all of JBC's e-mail alerts This article cites 27 references, 12 of which can be accessed free at http://www.jbc.org/content/280/39/33566.full.html#ref-list-1


Kathrin Sauter, Thomas Grampp, Jean-Marc Fritschy, Klemens Kaupmann, Bernhard Bettler, Hanns Mohler and Dietmar Benke