Anchoring proteins cluster receptors and ion channels at postsynaptic membranes in the brain. They also act as scaffolds for intracellular signaling molecules including synGAP and NO synthase. Here we report a new function for intracellular anchoring proteins: the regulation of synaptic ion channel function. A neuronal G protein‐gated inwardly rectifying K+ channel, Kir3.2c, can not be activated either by M2‐muscarinic receptor stimulation or by Gβγ overexpression. When coexpressed with SAP97, a member of the PSD/SAP anchoring protein family, the channel became sensitive to G protein stimulation. Although the C‐terminus of Kir3.2c bound to the second PDZ domain of SAP97, functional analyses revealed that the guanylate kinase (GK) domain of SAP97 is crucial for sensitization of the Kir3.2c channel to G protein stimulation. Furthermore, SAPAP1/GKAP, which binds specifically to the GK domain of membrane‐associated guanylate kinases, prevented the SAP97‐induced sensitization. The function of a synaptic ion channel can therefore be controlled by a network of various intracellular proteins.
The specific localization of ion channels, receptors and signaling molecules in the postsynaptic region is essential for effective synaptic transmission. The targeting of these proteins to the postsynaptic membrane is achieved by different anchoring and scaffold proteins, such as PSD‐95, SAP97 and syntrophins (for review, see Froehner, 1993; Hall and Sanes, 1993; Saras and Heldin, 1996; Sheng, 1996; Ziff, 1997; Craven and Bredt, 1998; Gee et al., 1998). However, the functional roles of these anchoring proteins on synaptic signal‐transduction have not been fully elucidated.
G protein‐gated K+ (KG) channels are directly activated by G proteins and coupled to various inhibitory receptors, including M2‐muscarinic, A1‐purinergic, D2‐dopaminergic, 5‐HT1A, μ‐ and δ‐opioid, GABAB and somatostatin receptors (Yamada et al., 1998). Thus, they play a key role in generation of slow inhibitory postsynaptic potentials in the brain (Lüscher et al., 1997). We recently found that KG channels in dopaminergic neurons of substantia nigra are localized specifically at the postsynaptic membrane of the dendrites (Inanobe et al., 1999). These KG channels are either heteromultimers of Kir3.2a and Kir3.2c or homomultimers of Kir3.2c alone. However, when the KG channels composed of Kir3.2c alone were expressed in Xenopus oocytes, they did not respond to G protein stimulation at all, although the channel proteins seemed to be generated and transported to the cell membrane as judged from the localization of green fluorescence protein (GFP) tagged to Kir3.2c (Inanobe et al., 1999). Therefore, the homomeric Kir3.2c channel would be produced in vivo but not be functional. It is, however, also possible that unidentified mechanisms might exist that confer functionality to the deaf KG channel.
In this study, because Kir3.2c could bind to PSD‐95 and SAP97 proteins (Inanobe et al., 1999), we hypothesized that the anchoring proteins might affect the function of Kir3.2c channels, and found that they confer the G protein sensitivity to homomeric Kir3.2c channels. Furthermore, the guanylate kinase (GK) domain of the anchoring proteins was indispensable for the SAP97‐induced sensitization of Kir3.2c. Therefore, the anchoring proteins may regulate not only the distribution but also the function of synaptic ion channels in the brain.
The effects of SAP97 on the KG channels composed of Kir3.2c
To investigate the function of KG channels composed of Kir3.2c alone, we injected the cRNA of Kir3.2c into Xenopus oocytes with either that of M2‐muscarinic receptor (M2R) or those of G protein β1 and γ2 subunits (Gβγ). These oocytes exhibited neither background nor acetylcholine (ACh)‐induced Kir currents, although the GFP fluorescence suggested that Kir3.2c proteins were localized at the plasma membrane (Figure 1A). The postembedding immunogold electron microscopy further showed that the gold particles of Kir3.2c immunoreactivity were associated with the invaginated plasma membrane of oocytes (Figure 1A, right panel). When we further coinjected the cRNA of either SAP97 or PSD‐95 into the M2R oocytes, the oocytes expressed a significant Ba2+‐sensitive background Kir current, which was markedly enhanced by ACh (10 μM). The amplitude of the Ba2+‐sensitive Kir current in the presence of 10 μM ACh was 5.81 ± 0.65 μA (n = 4) at −120 mV for the SAP97‐injected oocytes (Figure 1) and 5.70 ± 0.55 μA (n = 4) for those injected with PSD‐95. Gβγ could also fully activate the Kir3.2c channel in the presence of SAP97 (6.45 ± 1.02 μA, n = 4; Figure 1). The intensity and distribution of Kir3.2c–GFP fluorescence as well as the density of Kir3.2c immunoreactivity detected with gold particles were not significantly affected by coexpression of either SAP97 (Figure 1A) or PSD‐95 (data not shown). Therefore the coexpression of anchoring proteins was required for the full functional expression of Kir3.2c ion currents, which were activated by muscarinic receptor stimulation or more directly by overexpression of Gβγ. During this series of experiments for >6 months, we obtained this effect of SAP97 and PSD‐95 on the Kir3.2c channel in oocytes from most frogs, but in some frogs we could not see the effect. Therefore, in the following experiments, we always performed the control experiments using wild‐type SAP97 or PSD‐95 to confirm the oocytes being used to exhibit the SAP97 or PSD‐95‐induced sensitization of Kir3.2c to G protein stimulation.
Kir3.2c possesses the PDZ domain binding motif, ESKV, at its C‐terminus. The interaction between the C‐terminus of Kir3.2c and PDZ‐containing anchoring proteins might therefore be involved in the sensitization of Kir3.2c to G protein stimulation. Kir3.2a and Kir3.2c are splicing variants derived from a single gene where Kir3.2c is 11 amino acids longer than Kir3.2a at its C‐terminus (Figure 1). This region is the only divergent part of the two isoforms and contains the ESKV sequence. Kir3.2a did not bind to PSD‐95 or SAP97 (Inanobe et al., 1999; see also Figure 2B). The oocytes injected with Kir3.2a cRNA expressed background and ACh (10 μM)‐induced Kir currents even without coexpression of either SAP97 or PSD‐95. The Ba2+‐sensitive component of the Kir3.2a channel in the presence of ACh was 6.23 ± 0.75 μA (n = 4) (Figure 1). Gβγ also fully activated the Kir3.2a channel (6.45 ± 0.6 μA, n = 4; Figure 1). The coexpression of SAP97 or PSD‐95 did not affect and certainly did not further increase the ACh‐ or Gβγ‐induced Kir3.2a currents (n = 4 for each; data not shown). We constructed a deletion mutant of Kir3.2c that lacked the final four ESKV amino acids of the C‐terminus. When Kir3.2c‐ΔC4 was expressed in oocytes, the properties of the ion currents generated were identical to those of Kir3.2a. The amplitude of the Ba2+‐sensitive Kir current of Kir3.2c‐ΔC4 in the presence of ACh (10 μM) was 6.47 ± 0.29 μA (n = 4). Thus, the four amino acids, ESKV, at the C‐terminus are responsible for the insensitivity of the Kir3.2c channel to G protein stimulation in the absence of the anchoring proteins.
Identification of critical domains in SAP97‐induced sensitization of Kir3.2c to G protein.
To examine which region(s) of the anchoring proteins is crucial for the sensitization of Kir3.2c to G protein stimulation, we concentrated upon SAP97. We constructed different truncated mutants of SAP97 and coexpressed them with Kir3.2c in the presence of either M2R or Gβγ in Xenopus oocytes (Figure 2A). Although with the wild‐type SAP97, Kir3.2c was consistently activated by M2R stimulation or Gβγ overexpression, none of the truncated mutants of SAP97 that lacked potential Kir3.2c‐interacting PDZ domain(s), i.e. ΔPDZ‐1, ΔPDZ‐12, ΔPDZ‐123, SH3 and GK, could confer G protein sensitivity to the Kir3.2c channel (Figure 2A). Unexpectedly, the sensitization of Kir3.2c was also not achieved by coexpression of either PDZ‐123 or ΔGK, which contained all the PDZ domains but not the GK domain. Therefore, to examine the actual interactions between Kir3.2c and SAP97, we performed a ‘pull‐down’ assay in HEK293T cells (Figure 2B).
SAP97 expressed in HEK293T cells bound to the C‐terminal region of the Kir3.2c glutathione S‐transferase (GST) fusion protein (GST–Kir3.2c‐C) (Figure 2Ba). This interaction was abolished when the four amino acids (ESKV) at the C‐terminus of Kir3.2c were truncated (GST–Kir3.2c‐ΔC4‐C). Neither the GST fusion protein of the C‐terminus of Kir3.2a (GST–Kir3.2a‐C) nor that of its N‐terminus (GST–Kir3.2a/c‐N) bound to SAP97. Next we examined the interaction between Kir3.2c and different parts of SAP97 (Figure 2Bb). The GST fusion protein of full‐length SAP97 (GST–full) clearly bound to Kir3.2c. Consistent with the electrophysiological results (see Figure 2A), this interaction was abolished when the PDZ domains were deleted from SAP97 (GST–ΔPDZ‐12 and GST–ΔPDZ‐123). Of the three PDZ domains of SAP97, Kir3.2c bound to the second PDZ domain (GST–PDZ‐2) but to neither the first nor the third domain (GST–PDZ‐1 and GST–PDZ‐3). In addition, Kir3.2c did not bind to either the SH3 domain (GST–SH3) or the GK domain (GST–GK) of SAP97. We also found no binding between SAP97 and either M2R, Gα or Gβγ under our experimental conditions (data not shown). These results suggest that Kir3.2c specifically binds to the second PDZ domain of SAP97 through the four amino acids at its C‐terminus.
Because the truncated forms of SAP97, PDZ‐123 and ΔGK could bind to Kir3.2c (Figure 2B; data not shown), it was rather a surprise that neither of them could sensitize the Kir3.2c channel expressed in Xenopus oocytes (Figure 2A). This suggests the possibility that the GK domain might be involved in the SAP97‐induced sensitization of Kir3.2c to G protein stimulation. Thus, we examined the effects of different overexpressed parts of SAP97 on the wild‐type SAP97‐induced sensitization of Kir3.2c (Figure 2C). Overexpression of ΔPDZ‐123, which was composed of SH3 and GK but did not contain PDZ domains, almost completely prevented ACh induction of Kir3.2c current even in the presence of wild‐type SAP97 (0.17 ± 0.07 μA, n = 3) (Figure 2C). Overexpression of only the GK domain also completely suppressed ACh induction, while that of the SH3 domain had no inhibitory effect. These results suggest that the GK domain plays an indispensable role in the SAP97‐induced sensitization of Kir3.2c to G protein stimulation, even though we could not detect any physical binding between the GK domain and Kir3.2c under our experimental conditions.
The effects of SAPAP1 on SAP97‐induced sensitization of Kir3.2c
SAPAPs/GKAP/DAP are the proteins expressed at the postsynaptic site that bind specifically to the GK domain of the PSD/SAP protein family (Kim et al., 1997; Naisbitt et al., 1997; Satoh et al., 1997; Takeuchi et al., 1997). To examine further the possible functional roles of the GK domain of SAP97, we examined the effect of SAPAP1 upon ACh induction of Kir3.2c channel current in the presence of wild‐type SAP97 (Figure 3). In the oocytes coexpressing SAPAP1, we could see neither background nor ACh‐induced Kir currents (n = 5). Thus, SAPAP1 clearly prevented the SAP97‐induced sensitization of Kir3.2c to G protein, probably by binding to the GK domain, although it was reported that SAPAPs increase the recruitment of PSD‐95 to the plasma membrane (Takeuchi et al., 1997). This result further supports the notion that the GK domain of SAP97 plays a critical role in the sensitization of the Kir3.2c channel.
The primary function of the PSD‐95/SAP90 family of anchoring proteins is to cluster ion channels and receptors at postsynaptic sites. The PDZ domains of the anchoring proteins are mainly responsible for this function. The anchoring proteins also act as scaffolds by arranging signal molecules in a certain order that is determined by the specific binding of signal molecules to different PDZ domains. Recent studies have revealed that the GK domain of PSD/SAP family proteins, although it has no guanylate kinase activity (Kistner et al., 1995), acts as a binding domain for SAPAPs/GKAP/PAP (Kim et al., 1997; Satoh et al., 1997; Takeuchi et al., 1997), MAP1A (Brenman et al., 1998), BEGAIN (Deguchi et al., 1998) or the kainate receptor (Garcia et al., 1998). However, the functional role of the GK domain has remained largely unclarified. We show here that a deaf KG channel composed of homomeric Kir3.2c only reacts to G protein stimulation when bound to SAP97, where the GK domain of SAP97 plays a critical role.
SAP97 might affect Kir3.2c channel activity either by recruiting the channels from the intracellular pool to the plasma membrane or by directly controlling the channel function. The first possibility is unlikely, because (i) in both GFP fluorescence and immunoelectron microscopic measurements, even when expressed alone, Kir3.2c seemed to be localized on the plasma membrane of Xenopus oocytes and its localization was unaffected by the coexpression of SAP97 (Inanobe et al., 1999; see Figure 1); (ii) overexpression of the GK domain prevented the effect of wild‐type SAP97 even though the GK domain itself does not bind to Kir3.2c; and (iii) SAPAP1, which binds specifically to the GK domain, prevented the effect of SAP97. Therefore, it may be reasonable to postulate that SAP97 by directly controlling channel function confers the G protein sensitivity to the Kir3.2c channel. Theoretically, the whole‐cell channel current can be enhanced by increase of either functional channel number, individual channel open probability or unitary channel conductance. Although preliminary studies using the single channel recording technique have suggested that the third possibility was unlikely (not shown), the first two possibilities were difficult to discriminate. Further detailed studies on the properties of the expressed channel currents are needed to clarify this point.
None of the ΔPDZ‐1, ΔPDZ‐12 and ΔPDZ‐123 mutants sensitized Kir3.2c channels (Figure 2A). Because Kir3.2c may bind to the second PDZ domain, the channel may not bind to ΔPDZ‐12 and ΔPDZ‐123 mutants. The ΔPDZ‐1 mutant, however, lacks only the first PDZ domain and the N‐terminus of SAP97, and would interact with Kir3.2c. Because it was shown that the N‐terminal end of SAP97 was mandatory for the sorting of the protein to the cell membrane in epithelial cells (Wu et al., 1998), it is also possible that all three mutants including ΔPDZ‐1 were not transported to the oocyte plasma membrane and could not interact with the channel. Neither PDZ‐123 nor ΔGK mutants could sensitize Kir3.2c channels (Figure 2A), although these mutants possessed all three PDZ domains and the N‐terminal end of SAP97. These results may indicate that the binding of the second PDZ domain of SAP97 and the C‐terminal end of Kir3.2c is necessary but insufficient to sensitize the channel. The essential role of GK domain in this task is, therefore, obvious from the present results. It remains unclear, however, how the GK domain opens the door of Kir3.2c for the access of Gβγ. There may be several possibilities for this mechanism. For example, the GK domain plays an essential role in forming the structural arrangement of SAP97, which is needed for Kir3.2c sensitization to G proteins by SAP97. It was reported that the GK domain bound to SH3 but not any PDZ domains within PSD‐95 as well as DLG molecules (Brenman et al., 1998; McGee and Bredt, 1999). This intramolecular interaction between SH3 and GK domains may be a candidate for the mechanism of the structural arrangement of SAP97. However, it was shown that the overexpressed GK domain alone can not associate with either SH3‐GK protein or full‐length PSD‐95. Because the SAP97‐induced sensitization of Kir3.2c channels was prevented by coexpression of the GK domain (Figure 2C), the intramolecular interaction between the SH3 and GK domains can not explain the present results. However, it is still possible that weak association between the GK domain and the Kir3.2c channel or other parts of SAP97 may exist and cause Kir3.2c sensitization through the conformational change of Kir3.2c and/or SAP97. The second possibility is that the GK domain may interact with unidentified molecules intrinsically expressed in Xenopus oocytes, which is essential for SAP97‐induced sensitization of Kir3.2c. During this series of experiments for >6 months, we noticed that in the oocytes from certain frogs the SAP97‐induced sensitization of the Kir3.2c channel to G protein stimulation did not occur, while in those from most frogs we could clearly see the effect as described previously. This may suggest that the second possibility is likely. Further studies are required to clarify this point.
This study shows for the first time that PSD/SAP family anchoring proteins directly control the function of synaptic ion channels. This indicates a possibility that the ion channels so far reported as nonfunctional in the heterologous expression assay may be functional in vivo by associating with other molecules including MAGUKs. For example, δ glutamate receptor subunit expressed alone could not form functional channels (Lomeli et al., 1993). However, the glutamate receptor was shown to bind to PSD‐93 (Roche et al., 1999), and the effect of coexpression of the PDZ‐anchoring protein on the subunit function should be examined. Because the function of SAP97 could be modulated by SAPAP1 (Figure 3), synaptic ion channels seem to be controlled by a network of intracellular proteins. SAPAPs are expressed differentially among various cultured hippocampal neurons (Rao et al., 1998); therefore, they might be one physiological factor controlling the properties of KG channels expressed in different neurons. This system could be a novel target for various cell signals to modulate synaptic transmission in the brain.
Materials and methods
Functional expression in Xenopus oocytes and electrophysiological measurements
Rat SAP97 and PSD‐95 cDNAs were obtained by RT–PCR cloning from a rat brain cDNA library. The cDNA that we obtained was a splicing variant of rat SAP97 reported previously (Müller et al., 1995), i.e. amino acids residues 161–194, 407–421 and 695–701 were deleted, and AFRKNH (312–317), S (406) and H (694) were substituted for PASEKIM, T and R, respectively. The truncated clone of Kir3.2c, Kir3.2c‐ΔC4, was constructed using the GeneEditor™ in vitro site‐directed mutagenesis system (Promega Corp., Madison, WI). Various types of deletion mutants of SAP97 were constructed with PCR methods (all mutants except for PDZ‐123 and ΔGK) or the GeneEditor™ system (PDZ‐123 and ΔGK) (Promega) and inserted into the pGEMHE vector as reported previously (Inanobe et al., 1999). The pGEMHE vectors expressing the mutants of SAP97 contain the SAP97‐cDNA sequence encoding the following amino acids: ΔPDZ‐1, 282–857; ΔPDZ‐12, 407–857; PDZ‐123, 1–507; ΔGK, 1–621; ΔPDZ‐123, 533–857; SH3, 533–621; GK, 653–857. The methods of preparation of oocytes, cRNA injection and electrophysiological measurements have been described previously (Sugimoto et al., 1992; Inanobe et al., 1999). Each cDNA transcript was obtained with the mRNA capping kit (Stratagene, La Jolla, CA). Fifty nanograms of each cRNA dissolved in 50 nl of sterile water were manually injected into defoliculated Xenopus oocytes. After injection, oocytes were incubated in a modified Bath's solution at 18°C, and electrophysiological studies were undertaken 72–96 h later.
Two‐electrode voltage–clamp experiments were performed with a commercially available amplifier (model TEC 01C; Turbo Clamp, Tamm, Germany) with microelectrodes that had resistances of 0.5–1.5 MΩ when filled with 3 M KCl. Oocytes were bathed in a solution that contained (in mM) 90 KCl, 3 MgCl2, 5 HEPES–KOH pH 7.4 and 150 μM niflumic acid to block endogenous chloride current. Voltage steps (1.2 s in duration) from the holding potential of 0 mV to potentials between +60 and −120 mV with −20 mV increments were delivered every 3 s. Experiments were performed at room temperature (20–22°C). Electrophysiological data were stored on video tapes using a PCM data recording system (NF Electronic Design, Tokyo, Japan) and subsequently replayed for computer analysis (Patch Analyst Pro; MT Corporation, Hyogo, Japan). The data were expressed in means ± SE.
Affinity‐purified anti‐G2A‐5 antibody, which was raised against the amino acid sequences in the C‐terminus of Kir3.2a and Kir3.2c (WSVSSKLNQHAELE), was prepared as reported previously (Inanobe et al., 1999). A mouse monoclonal anti‐dlg antibody, which specifically recognizes the mouse, rat and human dlg/SAP97, and an anti‐GIRK2 antibody were obtained commercially (Transduction Laboratories, Lexington, KY and Alamone Laboratory, Jerusalem, Israel, respectively).
Histological analyses of Kir3.2 isoforms in Xenopus oocytes
GFP‐tagged Kir3.2a and Kir3.2c were constructed as described previously (Inanobe et al., 1999). Electrophysiological properties of these clones were identical to those of the wild type (not shown). The oocytes were fixed with 4% (w/v) paraformaldehyde (PFA) for 1 h at 4°C. The oocytes were cut into 20 μm sections, mounted on glass slides with 10% (v/v) glycerol in phosphate‐buffered saline (PBS) and examined with a confocal microscope (model MRC‐1024; Bio‐Rad, Hercules, CA).
Immunogold electron microscopy was performed as described previously (Gotow et al., 1995). The oocytes were fixed with 4% PFA and 0.1% (w/v) glutaraldehyde overnight at 4°C and dehydrated in 2.3 M sucrose containing 0.1 M sodium phosphate pH 7.4, and frozen in liquid nitrogen. Cryothin sections were cut on a microtome equipped with cryo‐attachment (OmU4, Reichert, Vienna, Austria) and collected on Formvar carbon‐coated grids. The cryothin sections on grids were treated with 1% BSA in PBS and incubated with anti‐GIRK2, and then goat anti‐rabbit IgG coupled to 5 nm colloidal gold particles (Amersham Pharmacia Biotech., Uppsala, Sweden). The sections were again fixed with 2% glutaraldehyde and post‐fixed with 1% OsO4, stained with 0.5% uranyl acetate, dehydrated in ethanol and embedded in London Resin white.
Fusion proteins with GST of specific regions of Kir3.2c, Kir3.2a and SAP97 were constructed by subcloning PCR‐amplified DNA fragments directionally into the EcoRI–EcoRI or BamHI–EcoRI sites of pGEX‐2T (Amersham Pharmacia Biotech.). Vectors expressed GST‐fusion proteins containing the following amino acids: GST–Kir3.2c‐C, 192–425; GST–Kir3.2c‐ΔC4‐C, 192–421; GST–Kir3.2a‐C, 192–414; GST–Kir3.2a/c‐N, 1–96 in Kir3.2c; GST–ΔPDZ‐12, 407–857; GST–ΔPDZ‐123, 533–857; GST–PDZ‐1, 188–285; GST–PDZ‐2, 282–421; GST–PDZ‐3, 407–507; GST–PDZ‐123, 188–507; GST–SH3, 533–621; GST–GK, 653–857 in SAP97. Each fusion protein was expressed in Escherichia coli and purified on glutathione–Sepharose. HEK293T cells transfected with Kir3.2c or SAP97 were homogenized and solubilized in a lysis buffer [40 mM Tris–HCl pH 7.4, 0.15 M NaCl, 10 mM EDTA, 0.2 mg/ml benzamidine, 30 kallikrein inhibitory units of aprotinin/ml, 1 mM phenylmethylsulfonyl fluoride, and 1% (w/v) Triton X‐100] and then centrifuged at 100 000 g for 30 min at 4°C. Solubilized homogenates were incubated for 1 h with 2.5 μg of purified fusion protein bound to 20 μl of glutathione–Sepharose. Samples were washed four times with PBS containing 0.4 M NaCl, resolved by SDS–8.5% polyacrylamide gel electrophoresis (SDS–PAGE), and then analyzed by Western blotting using anti‐G2A‐5 or anti‐dlg antibodies.
We thank Dr Ian Findlay (Tours, France) for his critical reading of this manuscript, Ms Mari Imanishi and Ms Kazue Takahashi for their technical assistance, and Ms Keiko Tsuji for secretarial work. This work was supported by the grants to Y.K. from the Ministry of Education, Culture, Sports and Science of Japan, from the Research for the Future Program of the Japan Society for the Promotion of Science (96L00302), and from the Human Frontier Science Program (RG0158/1997‐B).
- Copyright © 2000 European Molecular Biology Organization