Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 7-Day Trial for You or Your Team.

Learn More →

Number and Stoichiometry of Subunits in the Native Atrial G-protein-gated K+ Channel, IKACh

Number and Stoichiometry of Subunits in the Native Atrial G-protein-gated K+ Channel, IKACh THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 273, No. 9, Issue of February 27, pp. 5271–5278, 1998 © 1998 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in U.S.A. Number and Stoichiometry of Subunits in the Native Atrial Channel, I * G-protein-gated K KACh (Received for publication, November 17, 1997, and in revised form, December 6, 1997) Shawn Corey‡, Grigory Krapivinsky§, Luba Krapivinsky§, and David E. Clapham§¶ From the ‡Neuroscience Program, Mayo Foundation, Rochester, Minnesota 55905 and the §Howard Hughes Medical Institute, Children’s Hospital and Harvard Medical School, Boston, Massachusetts 02115 The G-protein-regulated, inwardly rectifying K lization, and biochemical characterization difficult. To date, (GIRK) channels are critical for functions as diverse as there is no crystal structure or high resolution electron micro- heart rate modulation and neuronal post-synaptic inhi- graphic image of a K -selective channel. bition. GIRK channels are distributed predominantly Of the K -selective channels, the most structural informa- throughout the heart, brain, and pancreas. In recent tion is known about the voltage-gated K -selective channels years, GIRK channels have received a great deal of at- (Kv). Kv channels have been proposed to be tetramers of four bg (G ) regulation. tention for their direct G-protein bg identical or highly homologous subunits, based on a wide vari- is composed of GIRK1 and GIRK4 Native cardiac I KACh ety of methods: toxin binding studies (10), covalently linked subunits (Krapivinsky, G., Gordon, E. A., Wickman, K. A., constructs (11), low resolution electron microscopic imaging Velimirovic, B., Krapivinsky, L., and Clapham, D. E. (12), and cross-linking studies (13). Each subunit is thought to (1995) Nature 374, 135–141). Here, we examine the qua- consist of 6 transmembrane domains with a loop contributing using a variety of complemen- ternary structure of I KACh to the pore (P-loop) region located between the fifth and sixth tary approaches. Complete cross-linking of purified transmembrane domains. Less is known about the inwardly protein formed a single adduct with a total atrial I KACh rectifying K -selective (Kir) channels. Kir channels are dis- molecular weight that was most consistent with a tet- tantly related to the Kv channels and contain regions equiva- ramer. In addition, partial cross-linking of purified lent only to the fifth transmembrane domain, the P-loop, and produced subsets of molecular weights consistent KACh the sixth transmembrane domains of the Kv channels. Despite with monomers, dimers, trimers, and tetramers. Within these similarities, the regions critical for assembly are likely to the presumed protein dimers, GIRK1-GIRK1 and GIRK4- differ between Kir and Kv channels . Although the N-terminal GIRK4 adducts were formed, indicating that the tet- domain has been implicated in Kv assembly (14), the second ramer was composed of two GIRK1 and two GIRK4 sub- transmembrane domain and the proximal C-terminal domains units. This 1:1 GIRK1 to GIRK4 stoichiometry was have been implicated in Kir channel assembly (15). Several confirmed by two independent means, including densi- tometry of both silver-stained and Western-blotted na- groups have suggested that, like Kv channels, Kir channels are . Similar experimental results could po- tive atrial I tetramers (16 –18). KACh tentially be obtained if GIRK1 and GIRK4 subunits We are interested in the unique subfamily of Kir channels assembled randomly as 2:2 and equally sized popula- that are ligand-gated by direct G binding (19, 20), the G- bg tions of 3:1 and 1:3 tetramers. We also show that GIRK protein-regulated, inwardly rectifying K channels (GIRKs). subunits may form homotetramers in expression sys- Using a combination of size exclusion chromatography and tems, although the evidence to date suggests that GIRK1 sucrose density gradients, it was originally proposed that homotetramers are not functional. We conclude that the GIRK1, in combination with an unknown subunit, contained channel, I , a prototyp- inwardly rectifying atrial K KACh three to five subunits of unknown stoichiometry (21). Later, by ical GIRK channel, is a heterotetramer and is most likely studying the biophysical properties of concatenated subunits, composed of two GIRK1 subunits and two GIRK4 Silverman et al. (22) suggested that GIRK1 and GIRK4 formed subunits. a tetramer in a 1:1 stoichiometry. However, Silverman et al. were unable to determine a preferred subunit arrangement around the pore and suggested that more than one arrange- Since the initial cloning of the Shaker K channel in 1987 ment may be viable. Tucker et al. (23) found that the GIRK1- (1–3), a wealth of K -selective channels have been cloned and GIRK4-GIRK1-GIRK4, rather than the GIRK1-GIRK1-GIRK4- their electrophysiologic properties characterized. Detailed GIRK4 arrangement, produced a higher ratio of agonist- structural information on K -selective channels, however, has induced to basal current. lagged considerably behind. Typically, K -selective channels To date, all of the studies on the stoichiometry of Kir chan- exist at low densities of 1–10 channels/mm (4 – 6), compared nels have relied upon the formation of multimeric concatemers. with ; 10,000 channels/mm for the nicotinic receptor (7–9). Briefly, a single protein is formed by the translation of a single Because they are membrane proteins, they require detergents mRNA encoding several artificially concatenated subunits. to keep them solubilized. This has made purification, crystal- 1 1 The abbreviations used are: Kv, voltage-gated K -selective channel; * The costs of publication of this article were defrayed in part by the GIRK, G-protein-regulated, inwardly rectifying K channel; DTSSP, payment of page charges. This article must therefore be hereby marked dithiobis(sulfosuccinimidylpropionate); DSS, disuccinimidyl suberate; “advertisement” in accordance with 18 U.S.C. Section 1734 solely to SSADP, sulfosuccinimidyl (4-azidophenyldithio)propionate; DMA, di- indicate this fact. methyl adipimidatez2HCl; DMS, dimethyl suberimidatez2HCl; WGA, To whom correspondence should be addressed: Cardiovascular Di- wheat germ agglutinin; NHS, N-hydroxysuccinimide; PAGE, polyacryl- vision, Children’s Hospital, 1309 Enders, 320 Longwood Ave., Boston, amide gel electrophoresis; Kir, inwardly rectifying K -selective chan- MA 02115. Tel.: 617-355-6163; Fax: 617-730-0692; E-mail: clapham nel; IP, immunoprecipitation; Ab, antibody; CHAPS, 3-[(3-cholamido- @rascal.med.harvard.edu. propyl)dimethylammonio]-1-propanesulfonic acid. This paper is available on line at http://www.jbc.org 5271 This is an Open Access article under the CC BY license. 5272 I Channels Are GIRK Tetramers KACh eluted with 1 mg/ml CIRN2 antigenic peptide for 150 min at 22 °C with Several combinations of concatenated subunits are examined, three eluate exchanges. For the anti-GIRK4 IP, the eluate was added to and conclusions are drawn from their varying properties. These Protein A-Sepharose and a 1:50 dilution of ascites fluid containing AU5 studies have an advantage in that they define the smallest monoclonal antibodies (BabCO, Berkeley Antibody Co., Richmond, CA) functional unit, whereas biochemical studies can define only was incubated at 4 °C for 150 min. The Protein A-Sepharose-Ab-GIRK the smallest physically associating unit. However, concatemer- complexes were washed five times (1-ml amounts) with IP buffer. GIRK based studies assume that: 1) combinations of tandemly linked heteromultimers were eluted with 0.25 mg/ml AU5 antigenic peptide at 22 °C with three 100-ml eluate exchanges for 1 h each. subunits that yield the most current are the most representa- Chemical Cross-linking—Protein to be cross-linked was treated for tive of the native channel configuration, 2) tandemly linked 1 h on ice with 100 mM dithiothreitol and dialyzed against 1% CHAPS, subunits do not coassemble with other tandemly linked sub- 10 mM HEPES, 400 mM NaCl, pH 8.5 (cross-linking buffer). To block units, and 3) tandemly linked subunits are completely trans- free sulfhydryl groups, dialyzed aliquots were treated with 25 mM lated and are not proteolytically cleaved between linked sub- iodoacetamide (Sigma-Aldrich Inc.) for1hon ice. Aliquots to be cross- units. Unfortunately, there are instances in which each of these linked via disulfides were not treated with iodoacetamide. Typical re- action volumes were 15 ml for the pure, atrial I and 75 ml for the assumptions has been proven to be incorrect (18, 22, 24, 25). KACh recombinant protein. Increasing the reaction volumes .10-fold had no Silverman et al. (22) were forced to assume that two trimeric effect on the adducts formed. Approximately 0.1 ng of pure I ,10 mg KACh constructs combine to form a functional channel to explain the of crude solubilized atrial protein, or 10 mg of solubilized COS-7 mem- high currents produced when only trimeric constructs were brane protein was used in each reaction. expressed. In addition, these studies have been complicated by For complete cross-linking, solutions containing 3 mM dithiobis(sul- the presence of endogenous oocyte subunits. Despite these fosuccinimidylpropionate) (DTSSP, Pierce), 1 mM disuccinimidyl suber- ate (DSS, Pierce), or 3 mM sulfosuccinimidyl (4-azidophenyldithio)pro- drawbacks, in recent years, the use of tandemly linked sub- pionate (SSADP, Pierce) were used. Immediately prior to use, cross- units has dominated the field of channel stoichiometry and has linking reagents were prepared as 10 3 stock solutions in cross-linking remained virtually unchecked by other independent methods. buffer. The water-insoluble DSS was prepared as a 9 3 stock solution in Clearly, alternative means must be used to test the validity of dimethyl sulfoxide (Me SO). Unless specified otherwise, the reactions this approach. were allowed to proceed for1hon ice. Reactions were terminated for 30 In this report, we present the first purification of a native min with 50 mM Tris, pH 7.5, or 15 mM iodoacetamide, when either H O or iodine was used. The hetero-bifunctional SSADP was first used mammalian K channel to homogeneity, the first cross-linking 2 2 like the other reagents, and then the azide was photoactivated by a study examining inwardly rectifying K channel quaternary 1-min exposure in a CL-1000 ultraviolet cross-linker (UVP, Upland, structure and stoichiometry, and the first densitometry study CA). examining GIRK stoichiometry. Using multiple independent For partial cross-linking, dimethyl adipimidatez2HCl (DMA, Pierce) methods, our data indicate that GIRK channels are tetramers. and dimethyl suberimidatez2HCl (DMS, Pierce) were also used. These We hypothesize that native I is composed of two GIRK1 reagents were used at 10 mM final concentrations in 100 mM HEPES KACh containing cross-linking buffer. For DTSSP, SSADP, and iodine, a 10- subunits and two GIRK4 subunits. fold dilution over what was used to completely cross-link the channel EXPERIMENTAL PROCEDURES generally created a laddering pattern. When necessary, trichloroacetic Purification of Native I and Recombinant GIRK1-GIRK4 Hetero- acid (Sigma-Aldrich Inc.) precipitation was used to concentrate the KACh multimers—Bovine atrial plasma membranes were isolated as de- samples prior to SDS-PAGE analysis. scribed (26). Membranes were solubilized in 1.0% CHAPS-HEDN Electrophoresis and Immunoblotting—Native I or recombinant KACh buffer, pH 7.5 (in mM: 10 HEPES, 1 EDTA, 1 dithiothreitol, and 100 GIRK protein was resuspended in Laemmli sample buffer containing NaCl). The protease inhibitors leupeptin (50 mg/ml, Sigma-Aldrich either 50 mM dithiothreitol (reducing conditions) or 25 mM iodoacet- Inc.), phenylmethylsulfonyl fluoride (100 mg/ml, Sigma-Aldrich Inc.), amide (non-reducing conditions) for 30 min at 50 °C. 10% separating aprotinin (1 mg/ml, Sigma-Aldrich Inc.), and pepstatin (2 mg/ml, Sig- and 3% stacking, 3–10% separating and 3% stacking, and precast ma-Aldrich Inc.) were used in all steps of the purification. Approxi- 2–15% (ISS) gels were all utilized. Samples were analyzed by fluorog- mately 150 mg of solubilized atrial proteins were loaded onto a Toyo- raphy with Amplify (Amersham Corp.), Gel Code™ silver staining pearl Red™ affinity column. Flow rates were 0.1 ml/min and 1 ml/min (Pierce), or by immunoblotting with anti-GIRK1 antibodies and/or anti- during the binding and elution steps, respectively. Bound protein was GIRK4 antibodies. Transfer times for Western blot analysis were ex- eluted with the same buffer containing 1 M NaCl. Fractions were as- tended to .2 h at 15 V to ensure transfer of the larger cross-linked sayed for I subunit content by Western blot. Fractions containing complexes. When sequential probing with antibodies was necessary, KACh both I subunits were pooled and dialyzed against 400 mM NaCl polyvinylidene fluoride membranes (Millipore, Bedford, MA) were KACh containing 1.0% CHAPS-HEDN buffer (pH 7.5). The equilibrated frac- stripped with 62.5 mM Tris-HCl, 2% SDS, 100 mM 2-mercaptoethanol tions were concentrated in Centriprep-50™ (Amicon, Inc., Beverly, MA) for 45 min at 50 °C. When quantitation was required, X-Omat AR film concentrators to ,2.0 ml and loaded onto a HighLoad 16/60 Superdex™ (Eastman Kodak Co.) was preflashed (Sensitize™, Amersham Corp.) to (Pharmacia Biotech Inc., Uppsala, Sweden) size exclusion chromatog- 0.15 OD above background and exposed at 270 °C. A GS-700 imaging raphy column at a flow rate of 0.4 ml/min. Pooled I fractions were densitometer (Bio-Rad) was used to analyze the protein gels and im- KACh dialyzed against immunoprecipitation (IP) buffer (1% CHAPS, 10 mM munoblots. Molecular weights were calculated using densitometry pro- HEPES, 100 mM NaCl, 5 mM EDTA at pH 7.5), loaded onto a wheat files from a combination of prestained high molecular weight markers germ agglutinin (WGA, Sigma-Aldrich Inc.) affinity column, and eluted (Bio-Rad) and low and high molecular weight markers (Pharmacia with 0.25 M N-acetylglucosamine. Finally, the eluate was immunopre- Biotech Inc.). In a portion of the gels, thyrogloblin (Pharmacia Biotech cipitated for 150 min at 4 °C with anti-GIRK4 peptide antibody (anti- Inc.) was added to ensure linearity through at least 330 kDa. When CIRN2 was generated against amino acids 19 –32; Refs. 19 and 27) and lanes from several gels were presented together (Figs. 4 – 6), the molec- washed three times (1-ml amounts) for 1 min each, three times (1 ml) ular weight markers corresponded to lanes 1 and 2. The remaining for 5 min each, and eluted with 1 mg/ml antigenic peptide three times lanes were approximately aligned with the molecular weight markers. (100 ml) for 30 min each at 22 °C. In addition, the molecular weights of all the adducts are presented in Plasma membrane proteins containing epitope-tagged GIRK1-AU5 Tables I and II. and GIRK4-AU1 were isolated from COS7 cells as described previously Antibody Standardization—Purified [ S]methionine-labeled recom- (28). Samples were precleared for1hat4 °C with 20 ml of Protein binant GIRK1 and/or GIRK4 subunits were purified to homogeneity, as A-Sepharose (Pharmacia Biotech Inc.). The two-step purification con- described previously. The pure GIRK1-GIRK4 heteromultimers were sisted of sequential immunoprecipitations in which samples were first divided into two aliquots, which were analyzed separately by SDS- immunoprecipitated with anti-GIRK4 antibodies and were then immu- PAGE. The first lane was fixed and exposed to film (Fig. 3B), and the noprecipitated with anti-GIRK1 antibodies. For the anti-GIRK4 IP, second lane was transferred to a polyvinylidene fluoride membrane and Protein A-Sepharose was preincubated with 3 mg of anti-CIRN2 (27, Western-blotted along with several lanes of decreasing quantities of 28). After a 30-min preincubation, solubilized proteins from a single solubilized atrial sarcolemma membranes (Fig. 3A). Autoradiography 100-mm dish were added and incubated at 4 °C for 150 min. The (Fig. 3B) revealed a GIRK1:GIRK4 band intensity ratio of 1.8:1.0 and Protein A-Sepharose-antibody (Ab)-GIRK complexes were washed five 1.5:1.0 after a correction for the methionine content of each protein (see times (1-ml amounts) with IP buffer. The GIRK heteromultimers were Equation 1). The intensity ratio represents the stoichiometry of the I Channels Are GIRK Tetramers 5273 KACh recombinant protein, but not necessarily the stoichiometry of the sub- units in native atrial I . KACh GIRK1 radiometric counts Recombinant stoichiometry 5 GIRK4 radiometric counts number of methionines in GIRK4 3 (Eq. 1) number of methionines in GIRK1 The aliquot of recombinant GIRK1-GIRK4 that had been Western- blotted was then analyzed by densitometry. The GIRK1:GIRK4 band intensity ratio was 8:1 and will be referred to as the Ab stoichiometry (apparent stoichiometry as detected by antibody). The Ab stoichiometry is measured from the ratio of the intensity of Western blot bands and is a product of the number of moles of each individual subunit and the number of Abs bound to each epitope. Dividing the recombinant stoichiometry by the Ab stoichiometry yields a useful term, which we will refer to as the Ab standardization factor (Equation 2). recombinant stoichiometry Ab standardization factor 5 (Eq. 2) Ab stoichiometry In our case, the Ab standardization factor was 0.19 (0.19 5 1.5/8). The Ab standardization factor multiplied by the GIRK1:GIRK4 band inten- sity ratio of the Western-blotted native channel yields the true stoichi- ometry of the native channel (Equation 3). True stoichiometry of native I 5 measured Ab stoichiometry KACh FIG.1. Purification schemes designed to select for heteromul- timeric channels. A, purification of native I . A combination of KACh 3 Ab standardization factor (Eq. 3) WGA (wheat germ agglutinin) chromatography and immunoprecipita- tion with anti-GIRK4 antibodies was used to purify native heteromul- In our case, the true stoichiometry of native I was ;1:1 (0.19 3 6 5 KACh timeric channels. B, purification of recombinant GIRK1-GIRK4 hetero- 1.1). multimers. Sequential immunoprecipitation with anti-GIRK4 and anti- The above procedure assumes that there is a direct linear relation- GIRK1-AU5 tag antibodies was used to purify recombinant ship between the Western blot intensity and the total amount of protein heteromultimeric channels. A tetrameric complex is assumed for the on the blot and that this relationship is maintained for both antibodies purpose of the diagram. Question mark (?), either GIRK1 or GIRK4; over the range of the protein concentrations to be tested. We found that branched tails on circles, glycosylated. our antibodies satisfied this criteria; the intensity of the lanes from Western-blotted atrial sarcolemma membrane varied in a direct and linear manner with the amount of protein. GIRK1 when compared with unpurified channels, but did not completely eliminate unglycosylated GIRK1 (Fig. 2A, 54-kDa RESULTS band). Presumably, some proportion of the native I het- KACh Purification of Native I and Recombinant GIRK1-GIRK4 eromultimers contain both glycosylated and unglycosylated KACh Heteromultimers—We have purified native bovine I and GIRK1 subunits. KACh recombinant GIRK1-GIRK4 heteromultimeric channels to near Recombinant GIRK1-GIRK4 heteromultimeric channels homogeneity. Both purification procedures were specifically were purified from transiently transfected COS7 cells using designed to purify only heteromultimeric channels composed of sequential anti-GIRK1 and anti-GIRK4 immunoprecipitations GIRK1 and GIRK4 subunits (Fig. 1). Potential homomultimeric (see Fig. 1B and “Experimental Procedures”). The sequential channels, or dissociated monomers, were not purified with immunoprecipitations assured that only heteromultimeric these procedures. channels were purified. The purified protein was analyzed by Native I was purified from isolated bovine atrial plasma SDS-PAGE, followed by either Western blotting or autoradiog- KACh membranes (Fig. 1A). The membranes were solubilized in 1% raphy (Fig. 2B). All of the protein bands that appeared on the CHAPS and subjected to the following purification steps: 1) autoradiogram were also recognized by anti-GIRK1 or anti- Toyopearl Red™ affinity chromatography, 2) size exclusion GIRK4 antibodies. chromatography, 3) WGA affinity chromatography, and 4) IP Densitometry of Silver-stained Native I and [ S]Methi- KACh with anti-GIRK4 antibodies followed by elution with antigenic onine-labeled Recombinant Proteins Suggests a 1:1 GIRK1 to peptide. WGA affinity chromatography was specific for GIRK1, GIRK4 Subunit Stoichiometry for Native I —Two inde- KACh because GIRK4 is not glycosylated (39). Thus, the combination pendent methods based on densitometry were used to examine of WGA affinity chromatography and immunoprecipitation GIRK1:GIRK4 stoichiometry. In the first method, silver- with anti-GIRK4 antibodies ensured that no homomultimeric stained SDS-PAGE gels of purified native I were analyzed KACh channels or dissociated monomers were purified. The final by densitometry (Fig. 2A, lane 1). The ratio of band intensities product, purified to greater than 95% homogeneity, was native for GIRK1:GIRK4 was 1.2:1. When this ratio was corrected by bovine atrial I . Aliquots of purified native I were multiplying it by the predicted unglycosylated molecular KACh KACh analyzed by SDS-PAGE and silver-stained (Fig. 2, lane 1)or weight of GIRK4(47 kDa)-GIRK1(56 kDa), a molar ratio of 1:1 immunoblotted with anti-GIRK4 antibodies (Fig. 2, lane 2) and (n 5 3) for GIRK1:GIRK4 was obtained. The advantage of this then stripped and reimmunoblotted with anti-GIRK1 antibod- procedure is that it examines GIRK1:GIRK4 stoichiometry as it ies (Fig. 2, lane 3). The predominant bands in the silver-stained exists in native atrial tissue. GIRK1 and GIRK4 share 57% lane were also recognized by anti-GIRK1 and anti-GIRK4 an- overall amino acid identity, and the silver-stained bands rep- tibodies when Western-blotted. The bands correspond to resenting GIRK1 and GIRK4 proteins varied by less than 3-fold GIRK4 (48 kDa), GIRK1 (54 kDa), and glycosylated GIRK1 in intensity. These GIRK1 and GIRK4 similarities minimize (56 –76 kDa). Interestingly, the WGA affinity chromatography the potential inaccuracies of protein quantification by silver enhanced the proportion of glycosylated to unglycosylated staining. 5274 I Channels Are GIRK Tetramers KACh FIG.2. Purified native I and re- KACh combinant GIRK1-GIRK4 heteromul- timers. Purified native I and recom- KACh binant GIRK1-GIRK4 heteromultimers were analyzed by 10% SDS-PAGE. A, pu- rified native I was silver-stained or KACh immunoblotted with anti-GIRK4 antibod- ies (aGIRK4) or anti-GIRK1 antibodies (aGIRK1). Presumably, some channel complexes contained at least one glycosy- lated (56 –76 kDa) and one unglycosylated GIRK1 (54 kDa) subunit, which allowed the unglycosylated GIRK1 subunit to be co-purified. Densitometry analysis of lane 1 was consistent with a 1:1 GIRK1:GIRK4 stoichiometry. B, purified recombinant GIRK1-GIRK4 heteromultimers were im- munoblotted simultaneously with anti- GIRK1 and anti-GIRK4 antibodies (aGRIK1/4, lane 1) or autoradiographed (lane 2). The majority of bands labeled by silver staining or [ S]methionine ([ S]Met) were also recognized by anti- body. gly, glycosylated. The second independent method used to examine GIRK1: GIRK4 stoichiometry was based on the [ S]methionine label- ing of purified recombinant COS7 GIRK1 and GIRK4 hetero- multimers. Because the number of counts emitted by a radiolabeled subunit is directly proportional to its methionine content, this method more accurately quantifies GIRK1 and GIRK4. Ideally, we could determine the native atrial GIRK1: GIRK4 stoichiometry by comparing S-labeled GIRK1 and GIRK4 (Fig. 2B). However, the stoichiometry may vary with the amount of RNA injected into oocytes (29, 30). If the stoi- chiometry varied in COS-7 cells, as well, this approach would be inadequate for determining native stoichiometry. Indeed, we found that the GIRK1:GIRK4 stoichiometry of heteromultim- FIG.3. Immunoblotting native I with standardized anti- KACh eric channels did vary directly with the ratio of GIRK1:GIRK4 bodies yields 1:1 GIRK1:GIRK4 stoichiometry. [ S]Methionine- DNA used to transfect the COS-7 cells (data not shown). We labeled recombinant GIRK1 and GIRK4 were used to standardize anti- were unable to force the GIRK1:GIRK4 stoichiometry beyond GIRK1 and anti-GIRK4 antibodies. The standardized antibodies were then used to immunoblot native I . A, lane 1,[ S]methionine- KACh 1:3 or 3:1 by varying the ratio of GIRK1:GIRK4 DNA trans- labeled, recombinant GIRK1 and GIRK4; lanes 2–5, 2.0, 1.0, 0.5, 0.25 fected by 30-fold, supporting the conclusion that recombinant mg, respectively, of solubilized atrial membrane protein. All lanes were GIRK1 and GIRK4 subunits form tetramers. analyzed by 10% SDS-PAGE and immunoblotted with anti-GIRK1 and Because the recombinant system does not necessarily reflect anti-GIRK4 antibodies. B,[ S]methionine-labeled recombinant GIRK1 and GIRK4 (rGIRK1 and rGIRK4) were analyzed by 10% SDS-PAGE native stoichiometry, we developed a hybrid approach that 35 followed by autoradiography. combined the accurate qauntitation achievable with [ S]me- thionine labeling of recombinant proteins with the ability of our antibodies to detect native protein. This method involved stan- To address the potential concern that the native complex is dardizing the relative blotting sensitivities of our anti-GIRK1 composed of an integral multiple of the ;235-kDa complex, or and anti-GIRK4 antibodies with a known ratio of labeled re- that associating proteins were lost, native I protein was KACh combinant protein. The standardized antibodies were then treated with an even more highly reactive cross-linking agent. used to probe native atrial protein and the true native stoichi- The heterobifunctional NHS ester/aryl azide SSADP was used ometry was computed as ;1:1 (see “Experimental Procedures” to cross-link native I immediately after solubilization and KACh and Fig. 3). Previously, Huang et al. (31) showed that the ratio prior to any further purification to prevent potential subunit of two antigens could be estimated with confidence using a degradation or dissociation. SSADP cross-linked protein (like similar antibody-blotting method. DTSSP, DSS, and iodine) was detected as a single 224 6 3-kDa Complete Cross-linking of Native Purified I —We exten- (n 5 4) band (Fig. 4, lane 2). Monomers were detected only after KACh sively cross-linked native bovine I (Fig. 4), recombinant extended exposure to film, if at all (data not shown). This KACh GIRK1 homomultimers (Fig. 5), and recombinant GIRK4 ho- indicates that the entire GIRK1-GIRK4 complex was intact momultimers (Fig. 5) using a wide variety of cross-linking after purification because dissociated monomers would have reagents. First, purified native I was treated with the remained at the bottom of the gel. Cross-linking of recombinant KACh highly reactive, amine-specific, N-hydroxysuccinimide (NHS) GIRK1 (Fig. 5A, lane 3) and GIRK4 (Fig. 5A, lane 4) homomul- ester, DTSSP. This reaction produced a multimeric protein timers also produced a single unique complex of 216 6 22 kDa detected as a single 234 6 7-kDa band (n 5 8). This cross- (n 5 4) and 212 6 13 kDa (n 5 3), respectively. If significant linked product was recognized by both anti-GIRK1 (Fig. 4, lane amounts of interchannel cross-linking rather than intrachan- 4) and anti-GIRK4 (Fig. 4, lane 3) antibodies, indicating that nel cross-linking had occurred, a smear would have appeared at both GIRK1 and GIRK4 were cross-linked. The lipid-soluble the top of the gel. A tetramer made up of equal numbers of NHS ester, disuccinimidyl suberate (DSS), yielded a nearly GIRK1 subunits (approximately 65 kDa with glycosylation) identical 235-kDa band (Fig. 4, lane 5). Finally, the entire and GIRK4 subunits (approximately 48 kDa) would have had a complex was cross-linked through simple oxidation with iodine molecular weight of ;226 kDa. Thus, the total molecular (Fig. 4, lane 6), and similar results were obtained. weight of the native cross-linked I complex, ;234 kDa, is KACh I Channels Are GIRK Tetramers 5275 KACh increased with increasing cross-linking times (Fig. 6, lane 1 versus lane 2) or cross-linking agent concentrations (data not shown). By densitometric scanning of Western blots (Fig. 6A, lanes 1 and 4), GIRK1 and GIRK4 antigenicity profiles were developed (Fig. 6B). Cross-linking ladders were also created when native I was treated with DMS, iodine, or DSS (Fig. KACh 6, C, D, and inset to D). In addition, partial cross-linking of recombinant homomultimeric GIRK4 and GIRK1 channels yielded a laddered pattern which was most consistent with homotetrameric proteins. The mean molecular weights of the various adducts produced by cross-linking of native atrial I KACh and recombinant homomultimeric channels are summarized in Table II. An Examination of Specific Dimer, Trimer and Tetramer Adducts Confirms That Native Atrial I Is a Heterotetramer KACh Composed of Two GIRK1 and Two GIRK4 Subunits—Exami- nation of dimers formed by partial cross-linking of purified native atrial I protein reveals GIRK4-GIRK4, GIRK1- Ch GIRK4, and GIRK1-GIRK1 adducts (Fig. 6). As expected, the relative proportions of the specific dimeric adducts that formed depended on the side chain specificity, cross-linking span, and lipid solubility of the cross-linking agent. The formation of the FIG.4. Complete cross-linking of native I yields products KACh GIRK1-GIRK1 and GIRK4-GIRK4 adducts demonstrated that that are most consistent with tetrameric channel formation. native atrial I heterotetramers are composed of two KACh Native I was treated with a wide variety of cross-linking reagents. KACh GIRK1 and two GIRK4 subunits and corroborates the previous Cross-linked products were then analyzed by SDS-PAGE and immuno- blotted. Lane 1, no cross-link control immunoblotted with anti-GIRK1 densitometry experiments. The trimeric peak in Fig. 6C is and anti-GIRK4 antibodies (aGIRK1 and aGIRK4). Lane 2, solubilized composed of GIRK1-GIRK1-GIRK4 and GIRK4-GIRK4-GIRK1 atrial membrane proteins treated with 3 mM SSADP followed by pho- adducts. As expected, the trimeric peak is broad and the GIRK4 tolysis for 1 min and immunoblotted with anti-GIRK1 antibodies. Lane peak antigenicity is shifted toward the lower molecular weights 3, pure I treated for 1 h with 3 mM DTSSP and immunoblotted with KACh anti-GIRK1 antibodies. Lane 4, lane 3 stripped and reimmunoblotted (;179 kDa), whereas the GIRK1 peak antigenicity is shifted with anti-GIRK4 antibodies. Lane 5, pure I treated for 1 h with 1 KACh toward the higher molecular weights (;188 kDa). On no occa- mM DSS and immunoblotted with anti-GIRK4 antibodies. Lane 6, pure sion did such an antigenicity shift occur in the tetrameric I treated for 1 h with 50% saturated iodine and immunoblotted KACh adduct. The simplest interpretation of this pattern is that the with anti-GIRK4 antibodies. Molecular weight markers were run with native I tetramer is composed of a single population of lanes 1 and 2; lanes 3– 6 were derived from separate gels and aligned KACh based on corresponding molecular weight standards. See Table I for a channels with two GIRK1 and two GIRK4 subunits. summary of averaged molecular weights. DISCUSSION consistent with a tetramer. Similarly, the total molecular We report the purification of a native mammalian K chan- weights of completely cross-linked homomultimeric GIRK1 nel to near homogeneity and provide direct biochemical evi- ;216-kDa channel (222 kDa, predicted) and GIRK4 ;211-kDa dence for I channel stoichiometry and quaternary struc- KACh channel (188 kDa, predicted) are consistent with a tetramer. ture. Using numerous independent methods, we have shown Finally, we found that recombinant GIRK1 homomultimers that GIRK proteins form tetramers and that native I is KACh eluted to a position similar to that for native I during size most likely a tetramer composed of two GIRK1 subunits and KACh exclusion chromatography (Fig. 5B). The size exclusion chro- two GIRK4 subunits. matography and complete cross-linking experiments support a After purification of native I to greater than 95% homo- KACh similar oligomeric structure for GIRK1 homomultimers and geneity, we found that the channel was comprised of GIRK4 (48 the native I heteromultimer. A summary of the various kDa), GIRK1 (54 kDa), and glycosylated GIRK1 (56 –76 kDa) KACh cross-linking reactions is given in Table I. subunits. The complex tightly bound WGA during purification, Partial Chemical Cross-linking of Native I Reveals Mo- indicating that it contained terminal sialic acid residues. The KACh nomeric, Dimeric, Trimeric, and Tetrameric Complexes—An- purified product cross-linked into a single high molecular other approach to testing the tetrameric channel hypothesis weight complex, indicating that the purified channel was an involves analysis of partially cross-linked native I . Previ- intact tetramer. The high degree of native channel protein KACh ously, it was shown that the molecular weight of partially purity was the key to our experiments because it eliminated cross-linked proteins increases in a linear fashion with the potential nonspecific cross-linking between native I and KACh number of cross-linked subunits (32). In this experiment, na- other unrelated membrane proteins. We cannot rule out the tive I was cross-linked with DMA, an imidoester, or with possibility that other populations of GIRK1-GIRK4 heteromul- KACh DTSSP, the more reactive NHS ester. The electrophoretic pat- timers with alternate stoichiometries did not copurify. How- tern produced following SDS-PAGE and Western blotting is ever, immunodepletion experiments illustrate that greater shown in Fig. 6. The blot was first probed with anti-GIRK4 than 90% of GIRK4 is associated with GIRK1 (27) and that antibodies (Fig. 6A, lanes 1–3) and then completely stripped greater than 90% of GIRK1 is associated with GIRK4 (data not and reprobed with anti-GIRK1 antibodies (Fig. 6A, lanes 4 – 6). shown). These immunodepletion experiments verify the lack of Partial cross-linking of native I produced a laddered pat- significant quantities of native homomultimeric complexes, if KACh tern consisting of four main adducts (Fig. 6A). The four adducts they exist at all. represent (from bottom to top) monomeric (41– 61 kDa), dimeric Chemical cross-linking has been widely used in the nearest (94 –138 kDa), trimeric (;185 kDa), and tetrameric (;231 kDa) neighbor analysis of membrane proteins and to study subunit forms of the channel (Fig. 6A). The adducts formed in a manner organization (see Refs. 32–34 for review). The total number of in which the proportion of higher molecular weight adducts subunits in both the glycine receptor (25) and Shaker channels 5276 I Channels Are GIRK Tetramers KACh FIG.5. Native I heteromultimers and recombinant GIRK homomultimers form similar oligomeric structures. Despite the KACh inability of GIRK1 subunits to produce functional channels alone, GIRK1 subunits form oligomeric structures similar to native I heteromul- KACh timers and GIRK4 homomultimers. A, native I and recombinant GIRK1 or GIRK4 homomultimers were chemically cross-linked. Cross-linked KACh products were then analyzed by SDS-PAGE and immunoblotted. The molecular weights of the cross-linked native I , recombinant GIRK1, and KACh recombinant GIRK4 homomultimers were all consistent with tetrameric complex formation. Lane 1, no cross-link control immunoblotted with anti-GIRK1(aGIRK1) and anti-GIRK4(aGIRK4) antibodies. Lane 2, solubilized atrial membrane proteins treated with 3 mM SSADP followed by photolysis for 1 min and immunoblotted with anti-GIRK1 antibodies. Lane 3, recombinant GIRK1 homomultimers treated for1hwith3mM DTSSP and immunoblotted with anti-GIRK1 antibodies. Lane 4, recombinant GIRK4 homomultimers treated for1hwith3mM DTSSP and immuno- blotted with anti-GIRK4 antibodies. Molecular weight markers were run with lanes 1 and 2; lanes 3 and 4 were derived from separate gels and aligned based on corresponding molecular weight standards. See Table I for a summary of averaged molecular weights. B, native I and KACh recombinant GIRK1 elute at similar volumes during size-exclusion chromatography. The portion of the channel that eluted with the void volume (V ) varied between trials. No attempt was made to estimate the molecular weight of the channel due to the complications caused by detergent binding. TABLE I Complete cross-linking of GIRK channels DTSSP DSS Oxidation SSADP Mean kDa Native I 234 6 7 235 244 6 2 224 6 3 234 6 8 KACh (n 5 8) (n 5 2) (n 5 4) (n 5 4) GIRK1 homomultimers 216 6 22 NA NA NA 216 (n 5 4) GIRK4 homomultimers 212 6 13 222 6 11 NA 200 211 6 11 (n 5 3) (n 5 3) (n 5 1) Molecular weights are given in units of kDa as a mean 6 standard deviation (n, number of trials). NA represents combinations not attempted. (13) have been estimated by cross-linking approaches. Here, we SDS-PAGE closely approximated the true molecular weight of have demonstrated that GIRK channels are tetrameric com- heteromultimeric proteins (25, 32, 35, 36). On balance, the plexes by using two chemical cross-linking approaches. In the internal consistency of our results suggests that native I KACh first approach, we cross-linked purified native I hetero- components were purified to homogeneity and cross-linked into KACh multimers, recombinant GIRK1 homomultimers, and recombi- one complete complex. To verify that the products resulting nant GIRK4 homomultimers. Four reagents with different side from complete cross-linking were tetramers, we partially cross- chain specificities, lipid solubilities, and cross-linking spans all linked native atrial I , homomultimeric GIRK4, and homo- KACh produced a single unique adduct, strongly indicating that the multimeric GIRK1 channels. Partial cross-linking formed four channels were purified in an intact state and then completely adducts representing monomers, dimers, trimers, and tetram- cross-linked. Native I heteromultimers, recombinant ers. The molecular weight of the adducts increased with the KACh GIRK1 homomultimers, and recombinant GIRK4 homomul- total number of subunits cross-linked in a linear fashion, again timers formed complexes of ;234, 216, and 211 kDa, respec- supporting the conclusion that native atrial I heteromul- KACh tively, consistent with predicted molecular weights of ;226, timers, recombinant GIRK1 homomultimers, and recombinant 222, and 188 kDa, respectively. The use of cross-linking agents GIRK4 homomultimers all formed tetrameric complexes. and iodoacetamide potentially complicates the interpretation of Three species of dimers, GIRK1-GIRK1, GIRK1-GIRK4, and our results, because both of these reagents can covalently bind GIRK4-GIRK4, were detected when native I was partially KACh the protein and therefore might increase its apparent molecu- cross-linked, based upon interpretation of molecular weights lar weight. These agents may also alter the protein’s mobility and Western blotting. The formation of GIRK1-GIRK1 and characteristics by changing its charge, hydrodynamic proper- GIRK4-GIRK4 cross-linked subunits within the native I KACh ties, or SDS binding. Nonetheless, others have shown that the tetramer indicated that the tetramer is composed of two GIRK1 molecular weight of cross-linked complexes determined by subunits and two GIRK4 subunits in ;1:1 stoichiometry. The I Channels Are GIRK Tetramers 5277 KACh FIG.6. Partial cross-linking of pure, native I is most consistent with a tetramer composed of two GIRK1 subunits and two KACh GIRK4 subunits. Partial cross-linking produces adducts that represent monomers, dimers, trimers, and tetramers. A, pure, native I was KACh treated with 10 mM DMA for 15 min (lanes 1 and 4),2h(lanes 2 and 5), or DTSSP for1h(lanes 3 and 6). The products were analyzed by 3–10% SDS-PAGE and immunoblotted. Lanes 1–3 were immunoblotted with anti-GIRK4 antibodies; lanes 4 – 6 correspond to lanes 1–3 when stripped and immunoblotted with anti-GIRK1 antibodies. B, anti-GIRK1 (aGIRK1) and anti-GIRK4 (aGIRK4) antigenicity profiles were created by densitom- etry scanning of lanes 2 and 4 of A. Likewise, profiles created when native I was treated with either iodine (C), DMS (D), or DSS (D, inset) are KACh shown. gly, glycosylated. 1:1 stoichiometry was supported by two additional experimen- ometry is 1:1, and rule out a fixed 3:1 or 1:3 GIRK1:GIRK4 tal approaches. First, silver-stained gels of purified I stoichiometry. KACh yielded a 1:1 GIRK1:GIRK4 staining intensity ratio after cor- Experimental results similar to those shown here might rection for their respective molecular weights. Both the purifi- have resulted from a random assembly of GIRK1 and GIRK4 cation scheme and cross-linking experiments assured that only subunits. This is an intriguing possibility, considering that in complete tetrameric heteromultimers were examined by densi- our heterologous COS7 expression system, the stoichiometry of tometry. Second, immunoblotting of native I was consist- GIRK1:GIRK4 varied directly with the ratio of DNA trans- KACh ent with 1:1 stoichiometry provided that the blotting antibodies fected, and would provide yet another way to contribute to K were first standardized against a known ratio of recombinant channel diversity. However, this interpretation would require GIRK1 and GIRK4 proteins. Thus, three different experimen- that the 3:1 and 1:3 GIRK1:GIRK4 pools were of equal size to tal methods support the conclusion that GIRK1:GIRK4 stoichi- be compatible with the 1:1 stoichiometry determined by densi- 5278 I Channels Are GIRK Tetramers KACh TABLE II Partial cross-linking of GIRK channels 1, GIRK1; 4, GIRK4. Molecular weights are given as a mean in kDa 6 standard deviation (n, number of trials). Adduct formed Tetramer Trimer 1–1 1–4 4–4 1 4 kDa Native I 237 6 4 180 6 4 131 6 4 113 6295 6362 6342 6 2 KACh (n 5 5) (n 5 5) (n 5 5) (n 5 5) (n 5 5) (n 5 5) (n 5 5) GIRK1 homomultimers 231 179 109 6 10 52 6 3 (n 5 2) (n 5 2) (n 5 3) (n 5 3) GIRK4 homomultimers 223 6 8 159 6 12 92 6343 6 5 (n 5 4) (n 5 4) (n 5 4) (n 5 4) Natl. Acad. Sci. U. S. A. 85, 5723–5727 tometry. Furthermore, large pools of complexes with 1:3 and 4. Soejima, M. & Noma, A. (1984) Pflugers Arch. 400, 424 – 431 3:1 stoichiometries would not be consistent with the observed 5. Koumi, S. J. & Wasserstrom, J. A. (1994) Am. J. Physiol. 266, H1812–H1821 nearly identical anti-GIRK1 and anti-GIRK4 tetrameric adduct 6. Ito, H., Hosoya, Y., Inanobe, A., Tomoike, H. & Endoh, M. (1995) Naunyn- Schmiedebergs Arch. Pharmocol. 351, 610 – 617 profiles. Thus, we currently favor the simplest interpretation of 7. Land, B. R., Salpeter, E. E. & Salpeter, M. M. (1980) Proc. Natl. Acad. Sci. the data, which is a fixed 2:2 GIRK1:GIRK4 stoichiometry U. S. A. 77, 3736 –3740 8. Matthews-Bellinger, J. & Salpeter, M. (1978) J. Physiol. 279, 197–213 rather than a random association model. 9. Fertuck, H. C. & Salpeter, M. M. (1976) J. Cell Biol. 69, 144 –158 Previous work has demonstrated that GIRK4 in expression 10. MacKinnon, R. (1991) Nature 350, 232–235 systems may form homomultimeric ion channels (27, 37, 39), 11. Liman, E., Tytgat, J. & Hess, P. (1992) Neuron 9, 861– 871 12. Li, M., Unwin, N., Stauffer, K., Jan, Y. & Jan, L. (1994) Curr. Biol. 4, 110 –115 whereas putative GIRK1 homomultimers are not functional 13. Schulteis, C., Naomi, N. & Papazian, D. (1996) Biochemistry 35, 12133–12140 (37–39 ). Moreover, GIRK1, by itself, does not localize to the 14. Li, M., Jan, Y. N. & Jan, L. Y. (1992) Science 257, 1225–1230 membrane (28). One possible explanation for these findings is 15. Tinker, A., Jan, Y. & Jan, L. (1996) Cell 87, 857– 868 16. Yang, J., Jan, Y. & Jan, L. (1995) Neuron 15, 1441–1447 that GIRK1 is unable to assemble with itself to form a tet- 17. Clement, J., Kunjilwar, K., Gonzalez, G., Schwanstecher, M., Panten, U., ramer, and is instead shuttled into a degradative pathway as Aguilar-Bryan, L. & Bryan, J. (1997) Neuron 18, 827– 838 18. Pessia, M., Tucker, J., Lee, K., Bond, C. & Adelman, J. (1996) EMBO J. 15, monomers or aggregates upon translation. It thus appears that 2980 –2987 GIRK1 must form heteromultimers with another GIRK family 19. Krapivinsky, G., Krapivinsky, L., Wickman, K. & Clapham, D. E. (1995) member to function. Wischmeyer et al. (30) suggest that GIRK1 J. Biol. Chem. 270, 29059 –29062 20. Wickman, K., Iniguez-Lluhi, J., Davenport, P., Taussig, R. A., Krapivinsky, homomultimers may not exist in vivo due to the spatial conflict G. B., Linder, M. E., Gilman, A. & Clapham, D. E. (1994) Nature 368, of bulky phenylalanines in the pore structure. In this study, we 255–257 show that recombinant GIRK1 subunits can form homotet- 21. Inanobe, A., Ito, H., Ito, M., Hosoya, Y. & Kurachi, Y. (1995) Biochem. Biophys. Res. Commun. 217, 1238 –1244 rameric complexes, although all evidence to date suggests they 22. Silverman, S. K., Lester, H. A. & Dougherty, D. A. (1996) J. Biol. Chem. 271, are not functional. 30524 –30528 In summary, the experiments presented here demonstrate 23. Tucker, S. J., Pessia, M. & Adelman, J. P. (1996) Am. J. Physiol. 271, H379 –H385 that GIRK channels form tetramers and that I is most KACh 24. McCormack, K., Lin, L., Iverson, L., Tanouye, M. & Sigworth, F. (1992) likely made up of two GIRK1 subunits and two GIRK4 sub- Biophys. J. 63, 1406 –1411 units. Although we cannot rule out the possibility that there 25. Langosch, D., Thomas, L. & Betz, H. (1988) Proc. Natl. Acad. Sci. U. S. A. 85, 7394 –73988 exist multiple pools of channels with varying stoichiometries, 26. Slaughter, R. S., Sutko, J. L. & Reeves, J. P. (1983) J. Biol. Chem. 258, the consistency of channel conductances and kinetics from sin- 3183–3190 27. Krapivinsky, G., Krapivinsky, L., Velimirovic, K., Wickman, B., Navarro, B. & gle-channel recordings in numerous species make this unlikely. Clapham, D. E. (1995) J. Biol. Chem. 270, 28777–28779 This study lays the foundation for future biochemical studies 28. Kennedy, M., Nemec, J. & Clapham, D. E. (1996) Neuropharmacology 35, on G binding stoichiometry, and determination of K channel 831– 839 bg 29. Slesinger, P., Reuveny, E., Jan, Y. & Jan, L. (1995) Neuron 15, 1145–1156 subunit arrangement around the pore. 30. Wischmeyer, E., Doring, F., Wischmeyer, E., Spauschus, A., Thomzig, A., Veh, R. & Karschin, A. (1997) Mol. Cell. Neurosci. 9, 194 –206 Acknowledgments—We thank Matt Kennedy and Kevin Wickman for 31. Huang, D. & Maero, S. (1997) BioTechniques 22, 454 – 458 critically reading the manuscript, Matt Kennedy for expertise and for 32. Davies, G. & Stark, G. (1970) Proc. Natl. Acad. Sci. U. S. A. 66, 651– 656 providing epitope-tagged GIRK1 and GIRK4, and Yiping Chen for pro- 33. Gaffney, B. (1985) Biochim. Biophys. Acta 822, 289 –317 viding technical assistance. 34. Richards, F. & Peters, K. (1977) Annu. Rev. Biochem. 46, 523–551 35. Aris, J. P. & Simon, R. D. (1983) J. Biol. Chem. 258, 14599 –14609 REFERENCES 36. Mourrain, P., Lasa, I., Guatreau, A., Gouin, E., Pugsley, A. & Cossart, P. 1. Papazian, D., Schwarz, T., Tempel, B., Jan, Y. & Jan, L. (1987) Science 237, (1997) Proc. Natl. Acad. Sci. U. S. A. 94, 10034 –10039 37. Velimirovic, B., Gordon, E., Lim, N., Navarro, B. & Clapham, D. E. (1996) 749 –753 2. Pongs, O., Kecskemethy, N., Muller, R., Krah-Jentgens, I., Baumann, A., FEBS Lett. 379, 31–37 Koltz, H., Canal, I., Llamazares, S. & Ferrus, A. (1988) EMBO J. 7, 38. Hedin, K., Nancy, L. & Clapham, D. (1996) Neuron 16, 423– 429 1087–1096 39. Krapivinsky, G., Gordon, E. A., Wickman, K. A., Velimirovic, B., Krapivinsky, 3. Iverson, L., Tanouye, M., Lester, H., Davidson, N. & Rudy, B. (1988) Proc. L. & Clapham, D. E. (1995) Nature 374, 135–141 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Biological Chemistry Unpaywall

Number and Stoichiometry of Subunits in the Native Atrial G-protein-gated K+ Channel, IKACh

Corey, Shawn; Krapivinsky, Grigory; Krapivinsky, Luba; Clapham, David E.
Journal of Biological Chemistry – Feb 1, 1998
Download PDF
8 pages

Loading next page...
 
/lp/unpaywall/number-and-stoichiometry-of-subunits-in-the-native-atrial-g-protein-BjtweGaFe6

References

References for this paper are not available at this time. We will be adding them shortly, thank you for your patience.

Publisher
Unpaywall
ISSN
0021-9258
DOI
10.1074/jbc.273.9.5271
Publisher site
See Article on Publisher Site

Abstract

THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 273, No. 9, Issue of February 27, pp. 5271–5278, 1998 © 1998 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in U.S.A. Number and Stoichiometry of Subunits in the Native Atrial Channel, I * G-protein-gated K KACh (Received for publication, November 17, 1997, and in revised form, December 6, 1997) Shawn Corey‡, Grigory Krapivinsky§, Luba Krapivinsky§, and David E. Clapham§¶ From the ‡Neuroscience Program, Mayo Foundation, Rochester, Minnesota 55905 and the §Howard Hughes Medical Institute, Children’s Hospital and Harvard Medical School, Boston, Massachusetts 02115 The G-protein-regulated, inwardly rectifying K lization, and biochemical characterization difficult. To date, (GIRK) channels are critical for functions as diverse as there is no crystal structure or high resolution electron micro- heart rate modulation and neuronal post-synaptic inhi- graphic image of a K -selective channel. bition. GIRK channels are distributed predominantly Of the K -selective channels, the most structural informa- throughout the heart, brain, and pancreas. In recent tion is known about the voltage-gated K -selective channels years, GIRK channels have received a great deal of at- (Kv). Kv channels have been proposed to be tetramers of four bg (G ) regulation. tention for their direct G-protein bg identical or highly homologous subunits, based on a wide vari- is composed of GIRK1 and GIRK4 Native cardiac I KACh ety of methods: toxin binding studies (10), covalently linked subunits (Krapivinsky, G., Gordon, E. A., Wickman, K. A., constructs (11), low resolution electron microscopic imaging Velimirovic, B., Krapivinsky, L., and Clapham, D. E. (12), and cross-linking studies (13). Each subunit is thought to (1995) Nature 374, 135–141). Here, we examine the qua- consist of 6 transmembrane domains with a loop contributing using a variety of complemen- ternary structure of I KACh to the pore (P-loop) region located between the fifth and sixth tary approaches. Complete cross-linking of purified transmembrane domains. Less is known about the inwardly protein formed a single adduct with a total atrial I KACh rectifying K -selective (Kir) channels. Kir channels are dis- molecular weight that was most consistent with a tet- tantly related to the Kv channels and contain regions equiva- ramer. In addition, partial cross-linking of purified lent only to the fifth transmembrane domain, the P-loop, and produced subsets of molecular weights consistent KACh the sixth transmembrane domains of the Kv channels. Despite with monomers, dimers, trimers, and tetramers. Within these similarities, the regions critical for assembly are likely to the presumed protein dimers, GIRK1-GIRK1 and GIRK4- differ between Kir and Kv channels . Although the N-terminal GIRK4 adducts were formed, indicating that the tet- domain has been implicated in Kv assembly (14), the second ramer was composed of two GIRK1 and two GIRK4 sub- transmembrane domain and the proximal C-terminal domains units. This 1:1 GIRK1 to GIRK4 stoichiometry was have been implicated in Kir channel assembly (15). Several confirmed by two independent means, including densi- tometry of both silver-stained and Western-blotted na- groups have suggested that, like Kv channels, Kir channels are . Similar experimental results could po- tive atrial I tetramers (16 –18). KACh tentially be obtained if GIRK1 and GIRK4 subunits We are interested in the unique subfamily of Kir channels assembled randomly as 2:2 and equally sized popula- that are ligand-gated by direct G binding (19, 20), the G- bg tions of 3:1 and 1:3 tetramers. We also show that GIRK protein-regulated, inwardly rectifying K channels (GIRKs). subunits may form homotetramers in expression sys- Using a combination of size exclusion chromatography and tems, although the evidence to date suggests that GIRK1 sucrose density gradients, it was originally proposed that homotetramers are not functional. We conclude that the GIRK1, in combination with an unknown subunit, contained channel, I , a prototyp- inwardly rectifying atrial K KACh three to five subunits of unknown stoichiometry (21). Later, by ical GIRK channel, is a heterotetramer and is most likely studying the biophysical properties of concatenated subunits, composed of two GIRK1 subunits and two GIRK4 Silverman et al. (22) suggested that GIRK1 and GIRK4 formed subunits. a tetramer in a 1:1 stoichiometry. However, Silverman et al. were unable to determine a preferred subunit arrangement around the pore and suggested that more than one arrange- Since the initial cloning of the Shaker K channel in 1987 ment may be viable. Tucker et al. (23) found that the GIRK1- (1–3), a wealth of K -selective channels have been cloned and GIRK4-GIRK1-GIRK4, rather than the GIRK1-GIRK1-GIRK4- their electrophysiologic properties characterized. Detailed GIRK4 arrangement, produced a higher ratio of agonist- structural information on K -selective channels, however, has induced to basal current. lagged considerably behind. Typically, K -selective channels To date, all of the studies on the stoichiometry of Kir chan- exist at low densities of 1–10 channels/mm (4 – 6), compared nels have relied upon the formation of multimeric concatemers. with ; 10,000 channels/mm for the nicotinic receptor (7–9). Briefly, a single protein is formed by the translation of a single Because they are membrane proteins, they require detergents mRNA encoding several artificially concatenated subunits. to keep them solubilized. This has made purification, crystal- 1 1 The abbreviations used are: Kv, voltage-gated K -selective channel; * The costs of publication of this article were defrayed in part by the GIRK, G-protein-regulated, inwardly rectifying K channel; DTSSP, payment of page charges. This article must therefore be hereby marked dithiobis(sulfosuccinimidylpropionate); DSS, disuccinimidyl suberate; “advertisement” in accordance with 18 U.S.C. Section 1734 solely to SSADP, sulfosuccinimidyl (4-azidophenyldithio)propionate; DMA, di- indicate this fact. methyl adipimidatez2HCl; DMS, dimethyl suberimidatez2HCl; WGA, To whom correspondence should be addressed: Cardiovascular Di- wheat germ agglutinin; NHS, N-hydroxysuccinimide; PAGE, polyacryl- vision, Children’s Hospital, 1309 Enders, 320 Longwood Ave., Boston, amide gel electrophoresis; Kir, inwardly rectifying K -selective chan- MA 02115. Tel.: 617-355-6163; Fax: 617-730-0692; E-mail: clapham nel; IP, immunoprecipitation; Ab, antibody; CHAPS, 3-[(3-cholamido- @rascal.med.harvard.edu. propyl)dimethylammonio]-1-propanesulfonic acid. This paper is available on line at http://www.jbc.org 5271 This is an Open Access article under the CC BY license. 5272 I Channels Are GIRK Tetramers KACh eluted with 1 mg/ml CIRN2 antigenic peptide for 150 min at 22 °C with Several combinations of concatenated subunits are examined, three eluate exchanges. For the anti-GIRK4 IP, the eluate was added to and conclusions are drawn from their varying properties. These Protein A-Sepharose and a 1:50 dilution of ascites fluid containing AU5 studies have an advantage in that they define the smallest monoclonal antibodies (BabCO, Berkeley Antibody Co., Richmond, CA) functional unit, whereas biochemical studies can define only was incubated at 4 °C for 150 min. The Protein A-Sepharose-Ab-GIRK the smallest physically associating unit. However, concatemer- complexes were washed five times (1-ml amounts) with IP buffer. GIRK based studies assume that: 1) combinations of tandemly linked heteromultimers were eluted with 0.25 mg/ml AU5 antigenic peptide at 22 °C with three 100-ml eluate exchanges for 1 h each. subunits that yield the most current are the most representa- Chemical Cross-linking—Protein to be cross-linked was treated for tive of the native channel configuration, 2) tandemly linked 1 h on ice with 100 mM dithiothreitol and dialyzed against 1% CHAPS, subunits do not coassemble with other tandemly linked sub- 10 mM HEPES, 400 mM NaCl, pH 8.5 (cross-linking buffer). To block units, and 3) tandemly linked subunits are completely trans- free sulfhydryl groups, dialyzed aliquots were treated with 25 mM lated and are not proteolytically cleaved between linked sub- iodoacetamide (Sigma-Aldrich Inc.) for1hon ice. Aliquots to be cross- units. Unfortunately, there are instances in which each of these linked via disulfides were not treated with iodoacetamide. Typical re- action volumes were 15 ml for the pure, atrial I and 75 ml for the assumptions has been proven to be incorrect (18, 22, 24, 25). KACh recombinant protein. Increasing the reaction volumes .10-fold had no Silverman et al. (22) were forced to assume that two trimeric effect on the adducts formed. Approximately 0.1 ng of pure I ,10 mg KACh constructs combine to form a functional channel to explain the of crude solubilized atrial protein, or 10 mg of solubilized COS-7 mem- high currents produced when only trimeric constructs were brane protein was used in each reaction. expressed. In addition, these studies have been complicated by For complete cross-linking, solutions containing 3 mM dithiobis(sul- the presence of endogenous oocyte subunits. Despite these fosuccinimidylpropionate) (DTSSP, Pierce), 1 mM disuccinimidyl suber- ate (DSS, Pierce), or 3 mM sulfosuccinimidyl (4-azidophenyldithio)pro- drawbacks, in recent years, the use of tandemly linked sub- pionate (SSADP, Pierce) were used. Immediately prior to use, cross- units has dominated the field of channel stoichiometry and has linking reagents were prepared as 10 3 stock solutions in cross-linking remained virtually unchecked by other independent methods. buffer. The water-insoluble DSS was prepared as a 9 3 stock solution in Clearly, alternative means must be used to test the validity of dimethyl sulfoxide (Me SO). Unless specified otherwise, the reactions this approach. were allowed to proceed for1hon ice. Reactions were terminated for 30 In this report, we present the first purification of a native min with 50 mM Tris, pH 7.5, or 15 mM iodoacetamide, when either H O or iodine was used. The hetero-bifunctional SSADP was first used mammalian K channel to homogeneity, the first cross-linking 2 2 like the other reagents, and then the azide was photoactivated by a study examining inwardly rectifying K channel quaternary 1-min exposure in a CL-1000 ultraviolet cross-linker (UVP, Upland, structure and stoichiometry, and the first densitometry study CA). examining GIRK stoichiometry. Using multiple independent For partial cross-linking, dimethyl adipimidatez2HCl (DMA, Pierce) methods, our data indicate that GIRK channels are tetramers. and dimethyl suberimidatez2HCl (DMS, Pierce) were also used. These We hypothesize that native I is composed of two GIRK1 reagents were used at 10 mM final concentrations in 100 mM HEPES KACh containing cross-linking buffer. For DTSSP, SSADP, and iodine, a 10- subunits and two GIRK4 subunits. fold dilution over what was used to completely cross-link the channel EXPERIMENTAL PROCEDURES generally created a laddering pattern. When necessary, trichloroacetic Purification of Native I and Recombinant GIRK1-GIRK4 Hetero- acid (Sigma-Aldrich Inc.) precipitation was used to concentrate the KACh multimers—Bovine atrial plasma membranes were isolated as de- samples prior to SDS-PAGE analysis. scribed (26). Membranes were solubilized in 1.0% CHAPS-HEDN Electrophoresis and Immunoblotting—Native I or recombinant KACh buffer, pH 7.5 (in mM: 10 HEPES, 1 EDTA, 1 dithiothreitol, and 100 GIRK protein was resuspended in Laemmli sample buffer containing NaCl). The protease inhibitors leupeptin (50 mg/ml, Sigma-Aldrich either 50 mM dithiothreitol (reducing conditions) or 25 mM iodoacet- Inc.), phenylmethylsulfonyl fluoride (100 mg/ml, Sigma-Aldrich Inc.), amide (non-reducing conditions) for 30 min at 50 °C. 10% separating aprotinin (1 mg/ml, Sigma-Aldrich Inc.), and pepstatin (2 mg/ml, Sig- and 3% stacking, 3–10% separating and 3% stacking, and precast ma-Aldrich Inc.) were used in all steps of the purification. Approxi- 2–15% (ISS) gels were all utilized. Samples were analyzed by fluorog- mately 150 mg of solubilized atrial proteins were loaded onto a Toyo- raphy with Amplify (Amersham Corp.), Gel Code™ silver staining pearl Red™ affinity column. Flow rates were 0.1 ml/min and 1 ml/min (Pierce), or by immunoblotting with anti-GIRK1 antibodies and/or anti- during the binding and elution steps, respectively. Bound protein was GIRK4 antibodies. Transfer times for Western blot analysis were ex- eluted with the same buffer containing 1 M NaCl. Fractions were as- tended to .2 h at 15 V to ensure transfer of the larger cross-linked sayed for I subunit content by Western blot. Fractions containing complexes. When sequential probing with antibodies was necessary, KACh both I subunits were pooled and dialyzed against 400 mM NaCl polyvinylidene fluoride membranes (Millipore, Bedford, MA) were KACh containing 1.0% CHAPS-HEDN buffer (pH 7.5). The equilibrated frac- stripped with 62.5 mM Tris-HCl, 2% SDS, 100 mM 2-mercaptoethanol tions were concentrated in Centriprep-50™ (Amicon, Inc., Beverly, MA) for 45 min at 50 °C. When quantitation was required, X-Omat AR film concentrators to ,2.0 ml and loaded onto a HighLoad 16/60 Superdex™ (Eastman Kodak Co.) was preflashed (Sensitize™, Amersham Corp.) to (Pharmacia Biotech Inc., Uppsala, Sweden) size exclusion chromatog- 0.15 OD above background and exposed at 270 °C. A GS-700 imaging raphy column at a flow rate of 0.4 ml/min. Pooled I fractions were densitometer (Bio-Rad) was used to analyze the protein gels and im- KACh dialyzed against immunoprecipitation (IP) buffer (1% CHAPS, 10 mM munoblots. Molecular weights were calculated using densitometry pro- HEPES, 100 mM NaCl, 5 mM EDTA at pH 7.5), loaded onto a wheat files from a combination of prestained high molecular weight markers germ agglutinin (WGA, Sigma-Aldrich Inc.) affinity column, and eluted (Bio-Rad) and low and high molecular weight markers (Pharmacia with 0.25 M N-acetylglucosamine. Finally, the eluate was immunopre- Biotech Inc.). In a portion of the gels, thyrogloblin (Pharmacia Biotech cipitated for 150 min at 4 °C with anti-GIRK4 peptide antibody (anti- Inc.) was added to ensure linearity through at least 330 kDa. When CIRN2 was generated against amino acids 19 –32; Refs. 19 and 27) and lanes from several gels were presented together (Figs. 4 – 6), the molec- washed three times (1-ml amounts) for 1 min each, three times (1 ml) ular weight markers corresponded to lanes 1 and 2. The remaining for 5 min each, and eluted with 1 mg/ml antigenic peptide three times lanes were approximately aligned with the molecular weight markers. (100 ml) for 30 min each at 22 °C. In addition, the molecular weights of all the adducts are presented in Plasma membrane proteins containing epitope-tagged GIRK1-AU5 Tables I and II. and GIRK4-AU1 were isolated from COS7 cells as described previously Antibody Standardization—Purified [ S]methionine-labeled recom- (28). Samples were precleared for1hat4 °C with 20 ml of Protein binant GIRK1 and/or GIRK4 subunits were purified to homogeneity, as A-Sepharose (Pharmacia Biotech Inc.). The two-step purification con- described previously. The pure GIRK1-GIRK4 heteromultimers were sisted of sequential immunoprecipitations in which samples were first divided into two aliquots, which were analyzed separately by SDS- immunoprecipitated with anti-GIRK4 antibodies and were then immu- PAGE. The first lane was fixed and exposed to film (Fig. 3B), and the noprecipitated with anti-GIRK1 antibodies. For the anti-GIRK4 IP, second lane was transferred to a polyvinylidene fluoride membrane and Protein A-Sepharose was preincubated with 3 mg of anti-CIRN2 (27, Western-blotted along with several lanes of decreasing quantities of 28). After a 30-min preincubation, solubilized proteins from a single solubilized atrial sarcolemma membranes (Fig. 3A). Autoradiography 100-mm dish were added and incubated at 4 °C for 150 min. The (Fig. 3B) revealed a GIRK1:GIRK4 band intensity ratio of 1.8:1.0 and Protein A-Sepharose-antibody (Ab)-GIRK complexes were washed five 1.5:1.0 after a correction for the methionine content of each protein (see times (1-ml amounts) with IP buffer. The GIRK heteromultimers were Equation 1). The intensity ratio represents the stoichiometry of the I Channels Are GIRK Tetramers 5273 KACh recombinant protein, but not necessarily the stoichiometry of the sub- units in native atrial I . KACh GIRK1 radiometric counts Recombinant stoichiometry 5 GIRK4 radiometric counts number of methionines in GIRK4 3 (Eq. 1) number of methionines in GIRK1 The aliquot of recombinant GIRK1-GIRK4 that had been Western- blotted was then analyzed by densitometry. The GIRK1:GIRK4 band intensity ratio was 8:1 and will be referred to as the Ab stoichiometry (apparent stoichiometry as detected by antibody). The Ab stoichiometry is measured from the ratio of the intensity of Western blot bands and is a product of the number of moles of each individual subunit and the number of Abs bound to each epitope. Dividing the recombinant stoichiometry by the Ab stoichiometry yields a useful term, which we will refer to as the Ab standardization factor (Equation 2). recombinant stoichiometry Ab standardization factor 5 (Eq. 2) Ab stoichiometry In our case, the Ab standardization factor was 0.19 (0.19 5 1.5/8). The Ab standardization factor multiplied by the GIRK1:GIRK4 band inten- sity ratio of the Western-blotted native channel yields the true stoichi- ometry of the native channel (Equation 3). True stoichiometry of native I 5 measured Ab stoichiometry KACh FIG.1. Purification schemes designed to select for heteromul- timeric channels. A, purification of native I . A combination of KACh 3 Ab standardization factor (Eq. 3) WGA (wheat germ agglutinin) chromatography and immunoprecipita- tion with anti-GIRK4 antibodies was used to purify native heteromul- In our case, the true stoichiometry of native I was ;1:1 (0.19 3 6 5 KACh timeric channels. B, purification of recombinant GIRK1-GIRK4 hetero- 1.1). multimers. Sequential immunoprecipitation with anti-GIRK4 and anti- The above procedure assumes that there is a direct linear relation- GIRK1-AU5 tag antibodies was used to purify recombinant ship between the Western blot intensity and the total amount of protein heteromultimeric channels. A tetrameric complex is assumed for the on the blot and that this relationship is maintained for both antibodies purpose of the diagram. Question mark (?), either GIRK1 or GIRK4; over the range of the protein concentrations to be tested. We found that branched tails on circles, glycosylated. our antibodies satisfied this criteria; the intensity of the lanes from Western-blotted atrial sarcolemma membrane varied in a direct and linear manner with the amount of protein. GIRK1 when compared with unpurified channels, but did not completely eliminate unglycosylated GIRK1 (Fig. 2A, 54-kDa RESULTS band). Presumably, some proportion of the native I het- KACh Purification of Native I and Recombinant GIRK1-GIRK4 eromultimers contain both glycosylated and unglycosylated KACh Heteromultimers—We have purified native bovine I and GIRK1 subunits. KACh recombinant GIRK1-GIRK4 heteromultimeric channels to near Recombinant GIRK1-GIRK4 heteromultimeric channels homogeneity. Both purification procedures were specifically were purified from transiently transfected COS7 cells using designed to purify only heteromultimeric channels composed of sequential anti-GIRK1 and anti-GIRK4 immunoprecipitations GIRK1 and GIRK4 subunits (Fig. 1). Potential homomultimeric (see Fig. 1B and “Experimental Procedures”). The sequential channels, or dissociated monomers, were not purified with immunoprecipitations assured that only heteromultimeric these procedures. channels were purified. The purified protein was analyzed by Native I was purified from isolated bovine atrial plasma SDS-PAGE, followed by either Western blotting or autoradiog- KACh membranes (Fig. 1A). The membranes were solubilized in 1% raphy (Fig. 2B). All of the protein bands that appeared on the CHAPS and subjected to the following purification steps: 1) autoradiogram were also recognized by anti-GIRK1 or anti- Toyopearl Red™ affinity chromatography, 2) size exclusion GIRK4 antibodies. chromatography, 3) WGA affinity chromatography, and 4) IP Densitometry of Silver-stained Native I and [ S]Methi- KACh with anti-GIRK4 antibodies followed by elution with antigenic onine-labeled Recombinant Proteins Suggests a 1:1 GIRK1 to peptide. WGA affinity chromatography was specific for GIRK1, GIRK4 Subunit Stoichiometry for Native I —Two inde- KACh because GIRK4 is not glycosylated (39). Thus, the combination pendent methods based on densitometry were used to examine of WGA affinity chromatography and immunoprecipitation GIRK1:GIRK4 stoichiometry. In the first method, silver- with anti-GIRK4 antibodies ensured that no homomultimeric stained SDS-PAGE gels of purified native I were analyzed KACh channels or dissociated monomers were purified. The final by densitometry (Fig. 2A, lane 1). The ratio of band intensities product, purified to greater than 95% homogeneity, was native for GIRK1:GIRK4 was 1.2:1. When this ratio was corrected by bovine atrial I . Aliquots of purified native I were multiplying it by the predicted unglycosylated molecular KACh KACh analyzed by SDS-PAGE and silver-stained (Fig. 2, lane 1)or weight of GIRK4(47 kDa)-GIRK1(56 kDa), a molar ratio of 1:1 immunoblotted with anti-GIRK4 antibodies (Fig. 2, lane 2) and (n 5 3) for GIRK1:GIRK4 was obtained. The advantage of this then stripped and reimmunoblotted with anti-GIRK1 antibod- procedure is that it examines GIRK1:GIRK4 stoichiometry as it ies (Fig. 2, lane 3). The predominant bands in the silver-stained exists in native atrial tissue. GIRK1 and GIRK4 share 57% lane were also recognized by anti-GIRK1 and anti-GIRK4 an- overall amino acid identity, and the silver-stained bands rep- tibodies when Western-blotted. The bands correspond to resenting GIRK1 and GIRK4 proteins varied by less than 3-fold GIRK4 (48 kDa), GIRK1 (54 kDa), and glycosylated GIRK1 in intensity. These GIRK1 and GIRK4 similarities minimize (56 –76 kDa). Interestingly, the WGA affinity chromatography the potential inaccuracies of protein quantification by silver enhanced the proportion of glycosylated to unglycosylated staining. 5274 I Channels Are GIRK Tetramers KACh FIG.2. Purified native I and re- KACh combinant GIRK1-GIRK4 heteromul- timers. Purified native I and recom- KACh binant GIRK1-GIRK4 heteromultimers were analyzed by 10% SDS-PAGE. A, pu- rified native I was silver-stained or KACh immunoblotted with anti-GIRK4 antibod- ies (aGIRK4) or anti-GIRK1 antibodies (aGIRK1). Presumably, some channel complexes contained at least one glycosy- lated (56 –76 kDa) and one unglycosylated GIRK1 (54 kDa) subunit, which allowed the unglycosylated GIRK1 subunit to be co-purified. Densitometry analysis of lane 1 was consistent with a 1:1 GIRK1:GIRK4 stoichiometry. B, purified recombinant GIRK1-GIRK4 heteromultimers were im- munoblotted simultaneously with anti- GIRK1 and anti-GIRK4 antibodies (aGRIK1/4, lane 1) or autoradiographed (lane 2). The majority of bands labeled by silver staining or [ S]methionine ([ S]Met) were also recognized by anti- body. gly, glycosylated. The second independent method used to examine GIRK1: GIRK4 stoichiometry was based on the [ S]methionine label- ing of purified recombinant COS7 GIRK1 and GIRK4 hetero- multimers. Because the number of counts emitted by a radiolabeled subunit is directly proportional to its methionine content, this method more accurately quantifies GIRK1 and GIRK4. Ideally, we could determine the native atrial GIRK1: GIRK4 stoichiometry by comparing S-labeled GIRK1 and GIRK4 (Fig. 2B). However, the stoichiometry may vary with the amount of RNA injected into oocytes (29, 30). If the stoi- chiometry varied in COS-7 cells, as well, this approach would be inadequate for determining native stoichiometry. Indeed, we found that the GIRK1:GIRK4 stoichiometry of heteromultim- FIG.3. Immunoblotting native I with standardized anti- KACh eric channels did vary directly with the ratio of GIRK1:GIRK4 bodies yields 1:1 GIRK1:GIRK4 stoichiometry. [ S]Methionine- DNA used to transfect the COS-7 cells (data not shown). We labeled recombinant GIRK1 and GIRK4 were used to standardize anti- were unable to force the GIRK1:GIRK4 stoichiometry beyond GIRK1 and anti-GIRK4 antibodies. The standardized antibodies were then used to immunoblot native I . A, lane 1,[ S]methionine- KACh 1:3 or 3:1 by varying the ratio of GIRK1:GIRK4 DNA trans- labeled, recombinant GIRK1 and GIRK4; lanes 2–5, 2.0, 1.0, 0.5, 0.25 fected by 30-fold, supporting the conclusion that recombinant mg, respectively, of solubilized atrial membrane protein. All lanes were GIRK1 and GIRK4 subunits form tetramers. analyzed by 10% SDS-PAGE and immunoblotted with anti-GIRK1 and Because the recombinant system does not necessarily reflect anti-GIRK4 antibodies. B,[ S]methionine-labeled recombinant GIRK1 and GIRK4 (rGIRK1 and rGIRK4) were analyzed by 10% SDS-PAGE native stoichiometry, we developed a hybrid approach that 35 followed by autoradiography. combined the accurate qauntitation achievable with [ S]me- thionine labeling of recombinant proteins with the ability of our antibodies to detect native protein. This method involved stan- To address the potential concern that the native complex is dardizing the relative blotting sensitivities of our anti-GIRK1 composed of an integral multiple of the ;235-kDa complex, or and anti-GIRK4 antibodies with a known ratio of labeled re- that associating proteins were lost, native I protein was KACh combinant protein. The standardized antibodies were then treated with an even more highly reactive cross-linking agent. used to probe native atrial protein and the true native stoichi- The heterobifunctional NHS ester/aryl azide SSADP was used ometry was computed as ;1:1 (see “Experimental Procedures” to cross-link native I immediately after solubilization and KACh and Fig. 3). Previously, Huang et al. (31) showed that the ratio prior to any further purification to prevent potential subunit of two antigens could be estimated with confidence using a degradation or dissociation. SSADP cross-linked protein (like similar antibody-blotting method. DTSSP, DSS, and iodine) was detected as a single 224 6 3-kDa Complete Cross-linking of Native Purified I —We exten- (n 5 4) band (Fig. 4, lane 2). Monomers were detected only after KACh sively cross-linked native bovine I (Fig. 4), recombinant extended exposure to film, if at all (data not shown). This KACh GIRK1 homomultimers (Fig. 5), and recombinant GIRK4 ho- indicates that the entire GIRK1-GIRK4 complex was intact momultimers (Fig. 5) using a wide variety of cross-linking after purification because dissociated monomers would have reagents. First, purified native I was treated with the remained at the bottom of the gel. Cross-linking of recombinant KACh highly reactive, amine-specific, N-hydroxysuccinimide (NHS) GIRK1 (Fig. 5A, lane 3) and GIRK4 (Fig. 5A, lane 4) homomul- ester, DTSSP. This reaction produced a multimeric protein timers also produced a single unique complex of 216 6 22 kDa detected as a single 234 6 7-kDa band (n 5 8). This cross- (n 5 4) and 212 6 13 kDa (n 5 3), respectively. If significant linked product was recognized by both anti-GIRK1 (Fig. 4, lane amounts of interchannel cross-linking rather than intrachan- 4) and anti-GIRK4 (Fig. 4, lane 3) antibodies, indicating that nel cross-linking had occurred, a smear would have appeared at both GIRK1 and GIRK4 were cross-linked. The lipid-soluble the top of the gel. A tetramer made up of equal numbers of NHS ester, disuccinimidyl suberate (DSS), yielded a nearly GIRK1 subunits (approximately 65 kDa with glycosylation) identical 235-kDa band (Fig. 4, lane 5). Finally, the entire and GIRK4 subunits (approximately 48 kDa) would have had a complex was cross-linked through simple oxidation with iodine molecular weight of ;226 kDa. Thus, the total molecular (Fig. 4, lane 6), and similar results were obtained. weight of the native cross-linked I complex, ;234 kDa, is KACh I Channels Are GIRK Tetramers 5275 KACh increased with increasing cross-linking times (Fig. 6, lane 1 versus lane 2) or cross-linking agent concentrations (data not shown). By densitometric scanning of Western blots (Fig. 6A, lanes 1 and 4), GIRK1 and GIRK4 antigenicity profiles were developed (Fig. 6B). Cross-linking ladders were also created when native I was treated with DMS, iodine, or DSS (Fig. KACh 6, C, D, and inset to D). In addition, partial cross-linking of recombinant homomultimeric GIRK4 and GIRK1 channels yielded a laddered pattern which was most consistent with homotetrameric proteins. The mean molecular weights of the various adducts produced by cross-linking of native atrial I KACh and recombinant homomultimeric channels are summarized in Table II. An Examination of Specific Dimer, Trimer and Tetramer Adducts Confirms That Native Atrial I Is a Heterotetramer KACh Composed of Two GIRK1 and Two GIRK4 Subunits—Exami- nation of dimers formed by partial cross-linking of purified native atrial I protein reveals GIRK4-GIRK4, GIRK1- Ch GIRK4, and GIRK1-GIRK1 adducts (Fig. 6). As expected, the relative proportions of the specific dimeric adducts that formed depended on the side chain specificity, cross-linking span, and lipid solubility of the cross-linking agent. The formation of the FIG.4. Complete cross-linking of native I yields products KACh GIRK1-GIRK1 and GIRK4-GIRK4 adducts demonstrated that that are most consistent with tetrameric channel formation. native atrial I heterotetramers are composed of two KACh Native I was treated with a wide variety of cross-linking reagents. KACh GIRK1 and two GIRK4 subunits and corroborates the previous Cross-linked products were then analyzed by SDS-PAGE and immuno- blotted. Lane 1, no cross-link control immunoblotted with anti-GIRK1 densitometry experiments. The trimeric peak in Fig. 6C is and anti-GIRK4 antibodies (aGIRK1 and aGIRK4). Lane 2, solubilized composed of GIRK1-GIRK1-GIRK4 and GIRK4-GIRK4-GIRK1 atrial membrane proteins treated with 3 mM SSADP followed by pho- adducts. As expected, the trimeric peak is broad and the GIRK4 tolysis for 1 min and immunoblotted with anti-GIRK1 antibodies. Lane peak antigenicity is shifted toward the lower molecular weights 3, pure I treated for 1 h with 3 mM DTSSP and immunoblotted with KACh anti-GIRK1 antibodies. Lane 4, lane 3 stripped and reimmunoblotted (;179 kDa), whereas the GIRK1 peak antigenicity is shifted with anti-GIRK4 antibodies. Lane 5, pure I treated for 1 h with 1 KACh toward the higher molecular weights (;188 kDa). On no occa- mM DSS and immunoblotted with anti-GIRK4 antibodies. Lane 6, pure sion did such an antigenicity shift occur in the tetrameric I treated for 1 h with 50% saturated iodine and immunoblotted KACh adduct. The simplest interpretation of this pattern is that the with anti-GIRK4 antibodies. Molecular weight markers were run with native I tetramer is composed of a single population of lanes 1 and 2; lanes 3– 6 were derived from separate gels and aligned KACh based on corresponding molecular weight standards. See Table I for a channels with two GIRK1 and two GIRK4 subunits. summary of averaged molecular weights. DISCUSSION consistent with a tetramer. Similarly, the total molecular We report the purification of a native mammalian K chan- weights of completely cross-linked homomultimeric GIRK1 nel to near homogeneity and provide direct biochemical evi- ;216-kDa channel (222 kDa, predicted) and GIRK4 ;211-kDa dence for I channel stoichiometry and quaternary struc- KACh channel (188 kDa, predicted) are consistent with a tetramer. ture. Using numerous independent methods, we have shown Finally, we found that recombinant GIRK1 homomultimers that GIRK proteins form tetramers and that native I is KACh eluted to a position similar to that for native I during size most likely a tetramer composed of two GIRK1 subunits and KACh exclusion chromatography (Fig. 5B). The size exclusion chro- two GIRK4 subunits. matography and complete cross-linking experiments support a After purification of native I to greater than 95% homo- KACh similar oligomeric structure for GIRK1 homomultimers and geneity, we found that the channel was comprised of GIRK4 (48 the native I heteromultimer. A summary of the various kDa), GIRK1 (54 kDa), and glycosylated GIRK1 (56 –76 kDa) KACh cross-linking reactions is given in Table I. subunits. The complex tightly bound WGA during purification, Partial Chemical Cross-linking of Native I Reveals Mo- indicating that it contained terminal sialic acid residues. The KACh nomeric, Dimeric, Trimeric, and Tetrameric Complexes—An- purified product cross-linked into a single high molecular other approach to testing the tetrameric channel hypothesis weight complex, indicating that the purified channel was an involves analysis of partially cross-linked native I . Previ- intact tetramer. The high degree of native channel protein KACh ously, it was shown that the molecular weight of partially purity was the key to our experiments because it eliminated cross-linked proteins increases in a linear fashion with the potential nonspecific cross-linking between native I and KACh number of cross-linked subunits (32). In this experiment, na- other unrelated membrane proteins. We cannot rule out the tive I was cross-linked with DMA, an imidoester, or with possibility that other populations of GIRK1-GIRK4 heteromul- KACh DTSSP, the more reactive NHS ester. The electrophoretic pat- timers with alternate stoichiometries did not copurify. How- tern produced following SDS-PAGE and Western blotting is ever, immunodepletion experiments illustrate that greater shown in Fig. 6. The blot was first probed with anti-GIRK4 than 90% of GIRK4 is associated with GIRK1 (27) and that antibodies (Fig. 6A, lanes 1–3) and then completely stripped greater than 90% of GIRK1 is associated with GIRK4 (data not and reprobed with anti-GIRK1 antibodies (Fig. 6A, lanes 4 – 6). shown). These immunodepletion experiments verify the lack of Partial cross-linking of native I produced a laddered pat- significant quantities of native homomultimeric complexes, if KACh tern consisting of four main adducts (Fig. 6A). The four adducts they exist at all. represent (from bottom to top) monomeric (41– 61 kDa), dimeric Chemical cross-linking has been widely used in the nearest (94 –138 kDa), trimeric (;185 kDa), and tetrameric (;231 kDa) neighbor analysis of membrane proteins and to study subunit forms of the channel (Fig. 6A). The adducts formed in a manner organization (see Refs. 32–34 for review). The total number of in which the proportion of higher molecular weight adducts subunits in both the glycine receptor (25) and Shaker channels 5276 I Channels Are GIRK Tetramers KACh FIG.5. Native I heteromultimers and recombinant GIRK homomultimers form similar oligomeric structures. Despite the KACh inability of GIRK1 subunits to produce functional channels alone, GIRK1 subunits form oligomeric structures similar to native I heteromul- KACh timers and GIRK4 homomultimers. A, native I and recombinant GIRK1 or GIRK4 homomultimers were chemically cross-linked. Cross-linked KACh products were then analyzed by SDS-PAGE and immunoblotted. The molecular weights of the cross-linked native I , recombinant GIRK1, and KACh recombinant GIRK4 homomultimers were all consistent with tetrameric complex formation. Lane 1, no cross-link control immunoblotted with anti-GIRK1(aGIRK1) and anti-GIRK4(aGIRK4) antibodies. Lane 2, solubilized atrial membrane proteins treated with 3 mM SSADP followed by photolysis for 1 min and immunoblotted with anti-GIRK1 antibodies. Lane 3, recombinant GIRK1 homomultimers treated for1hwith3mM DTSSP and immunoblotted with anti-GIRK1 antibodies. Lane 4, recombinant GIRK4 homomultimers treated for1hwith3mM DTSSP and immuno- blotted with anti-GIRK4 antibodies. Molecular weight markers were run with lanes 1 and 2; lanes 3 and 4 were derived from separate gels and aligned based on corresponding molecular weight standards. See Table I for a summary of averaged molecular weights. B, native I and KACh recombinant GIRK1 elute at similar volumes during size-exclusion chromatography. The portion of the channel that eluted with the void volume (V ) varied between trials. No attempt was made to estimate the molecular weight of the channel due to the complications caused by detergent binding. TABLE I Complete cross-linking of GIRK channels DTSSP DSS Oxidation SSADP Mean kDa Native I 234 6 7 235 244 6 2 224 6 3 234 6 8 KACh (n 5 8) (n 5 2) (n 5 4) (n 5 4) GIRK1 homomultimers 216 6 22 NA NA NA 216 (n 5 4) GIRK4 homomultimers 212 6 13 222 6 11 NA 200 211 6 11 (n 5 3) (n 5 3) (n 5 1) Molecular weights are given in units of kDa as a mean 6 standard deviation (n, number of trials). NA represents combinations not attempted. (13) have been estimated by cross-linking approaches. Here, we SDS-PAGE closely approximated the true molecular weight of have demonstrated that GIRK channels are tetrameric com- heteromultimeric proteins (25, 32, 35, 36). On balance, the plexes by using two chemical cross-linking approaches. In the internal consistency of our results suggests that native I KACh first approach, we cross-linked purified native I hetero- components were purified to homogeneity and cross-linked into KACh multimers, recombinant GIRK1 homomultimers, and recombi- one complete complex. To verify that the products resulting nant GIRK4 homomultimers. Four reagents with different side from complete cross-linking were tetramers, we partially cross- chain specificities, lipid solubilities, and cross-linking spans all linked native atrial I , homomultimeric GIRK4, and homo- KACh produced a single unique adduct, strongly indicating that the multimeric GIRK1 channels. Partial cross-linking formed four channels were purified in an intact state and then completely adducts representing monomers, dimers, trimers, and tetram- cross-linked. Native I heteromultimers, recombinant ers. The molecular weight of the adducts increased with the KACh GIRK1 homomultimers, and recombinant GIRK4 homomul- total number of subunits cross-linked in a linear fashion, again timers formed complexes of ;234, 216, and 211 kDa, respec- supporting the conclusion that native atrial I heteromul- KACh tively, consistent with predicted molecular weights of ;226, timers, recombinant GIRK1 homomultimers, and recombinant 222, and 188 kDa, respectively. The use of cross-linking agents GIRK4 homomultimers all formed tetrameric complexes. and iodoacetamide potentially complicates the interpretation of Three species of dimers, GIRK1-GIRK1, GIRK1-GIRK4, and our results, because both of these reagents can covalently bind GIRK4-GIRK4, were detected when native I was partially KACh the protein and therefore might increase its apparent molecu- cross-linked, based upon interpretation of molecular weights lar weight. These agents may also alter the protein’s mobility and Western blotting. The formation of GIRK1-GIRK1 and characteristics by changing its charge, hydrodynamic proper- GIRK4-GIRK4 cross-linked subunits within the native I KACh ties, or SDS binding. Nonetheless, others have shown that the tetramer indicated that the tetramer is composed of two GIRK1 molecular weight of cross-linked complexes determined by subunits and two GIRK4 subunits in ;1:1 stoichiometry. The I Channels Are GIRK Tetramers 5277 KACh FIG.6. Partial cross-linking of pure, native I is most consistent with a tetramer composed of two GIRK1 subunits and two KACh GIRK4 subunits. Partial cross-linking produces adducts that represent monomers, dimers, trimers, and tetramers. A, pure, native I was KACh treated with 10 mM DMA for 15 min (lanes 1 and 4),2h(lanes 2 and 5), or DTSSP for1h(lanes 3 and 6). The products were analyzed by 3–10% SDS-PAGE and immunoblotted. Lanes 1–3 were immunoblotted with anti-GIRK4 antibodies; lanes 4 – 6 correspond to lanes 1–3 when stripped and immunoblotted with anti-GIRK1 antibodies. B, anti-GIRK1 (aGIRK1) and anti-GIRK4 (aGIRK4) antigenicity profiles were created by densitom- etry scanning of lanes 2 and 4 of A. Likewise, profiles created when native I was treated with either iodine (C), DMS (D), or DSS (D, inset) are KACh shown. gly, glycosylated. 1:1 stoichiometry was supported by two additional experimen- ometry is 1:1, and rule out a fixed 3:1 or 1:3 GIRK1:GIRK4 tal approaches. First, silver-stained gels of purified I stoichiometry. KACh yielded a 1:1 GIRK1:GIRK4 staining intensity ratio after cor- Experimental results similar to those shown here might rection for their respective molecular weights. Both the purifi- have resulted from a random assembly of GIRK1 and GIRK4 cation scheme and cross-linking experiments assured that only subunits. This is an intriguing possibility, considering that in complete tetrameric heteromultimers were examined by densi- our heterologous COS7 expression system, the stoichiometry of tometry. Second, immunoblotting of native I was consist- GIRK1:GIRK4 varied directly with the ratio of DNA trans- KACh ent with 1:1 stoichiometry provided that the blotting antibodies fected, and would provide yet another way to contribute to K were first standardized against a known ratio of recombinant channel diversity. However, this interpretation would require GIRK1 and GIRK4 proteins. Thus, three different experimen- that the 3:1 and 1:3 GIRK1:GIRK4 pools were of equal size to tal methods support the conclusion that GIRK1:GIRK4 stoichi- be compatible with the 1:1 stoichiometry determined by densi- 5278 I Channels Are GIRK Tetramers KACh TABLE II Partial cross-linking of GIRK channels 1, GIRK1; 4, GIRK4. Molecular weights are given as a mean in kDa 6 standard deviation (n, number of trials). Adduct formed Tetramer Trimer 1–1 1–4 4–4 1 4 kDa Native I 237 6 4 180 6 4 131 6 4 113 6295 6362 6342 6 2 KACh (n 5 5) (n 5 5) (n 5 5) (n 5 5) (n 5 5) (n 5 5) (n 5 5) GIRK1 homomultimers 231 179 109 6 10 52 6 3 (n 5 2) (n 5 2) (n 5 3) (n 5 3) GIRK4 homomultimers 223 6 8 159 6 12 92 6343 6 5 (n 5 4) (n 5 4) (n 5 4) (n 5 4) Natl. Acad. Sci. U. S. A. 85, 5723–5727 tometry. Furthermore, large pools of complexes with 1:3 and 4. Soejima, M. & Noma, A. (1984) Pflugers Arch. 400, 424 – 431 3:1 stoichiometries would not be consistent with the observed 5. Koumi, S. J. & Wasserstrom, J. A. (1994) Am. J. Physiol. 266, H1812–H1821 nearly identical anti-GIRK1 and anti-GIRK4 tetrameric adduct 6. Ito, H., Hosoya, Y., Inanobe, A., Tomoike, H. & Endoh, M. (1995) Naunyn- Schmiedebergs Arch. Pharmocol. 351, 610 – 617 profiles. Thus, we currently favor the simplest interpretation of 7. Land, B. R., Salpeter, E. E. & Salpeter, M. M. (1980) Proc. Natl. Acad. Sci. the data, which is a fixed 2:2 GIRK1:GIRK4 stoichiometry U. S. A. 77, 3736 –3740 8. Matthews-Bellinger, J. & Salpeter, M. (1978) J. Physiol. 279, 197–213 rather than a random association model. 9. Fertuck, H. C. & Salpeter, M. M. (1976) J. Cell Biol. 69, 144 –158 Previous work has demonstrated that GIRK4 in expression 10. MacKinnon, R. (1991) Nature 350, 232–235 systems may form homomultimeric ion channels (27, 37, 39), 11. Liman, E., Tytgat, J. & Hess, P. (1992) Neuron 9, 861– 871 12. Li, M., Unwin, N., Stauffer, K., Jan, Y. & Jan, L. (1994) Curr. Biol. 4, 110 –115 whereas putative GIRK1 homomultimers are not functional 13. Schulteis, C., Naomi, N. & Papazian, D. (1996) Biochemistry 35, 12133–12140 (37–39 ). Moreover, GIRK1, by itself, does not localize to the 14. Li, M., Jan, Y. N. & Jan, L. Y. (1992) Science 257, 1225–1230 membrane (28). One possible explanation for these findings is 15. Tinker, A., Jan, Y. & Jan, L. (1996) Cell 87, 857– 868 16. Yang, J., Jan, Y. & Jan, L. (1995) Neuron 15, 1441–1447 that GIRK1 is unable to assemble with itself to form a tet- 17. Clement, J., Kunjilwar, K., Gonzalez, G., Schwanstecher, M., Panten, U., ramer, and is instead shuttled into a degradative pathway as Aguilar-Bryan, L. & Bryan, J. (1997) Neuron 18, 827– 838 18. Pessia, M., Tucker, J., Lee, K., Bond, C. & Adelman, J. (1996) EMBO J. 15, monomers or aggregates upon translation. It thus appears that 2980 –2987 GIRK1 must form heteromultimers with another GIRK family 19. Krapivinsky, G., Krapivinsky, L., Wickman, K. & Clapham, D. E. (1995) member to function. Wischmeyer et al. (30) suggest that GIRK1 J. Biol. Chem. 270, 29059 –29062 20. Wickman, K., Iniguez-Lluhi, J., Davenport, P., Taussig, R. A., Krapivinsky, homomultimers may not exist in vivo due to the spatial conflict G. B., Linder, M. E., Gilman, A. & Clapham, D. E. (1994) Nature 368, of bulky phenylalanines in the pore structure. In this study, we 255–257 show that recombinant GIRK1 subunits can form homotet- 21. Inanobe, A., Ito, H., Ito, M., Hosoya, Y. & Kurachi, Y. (1995) Biochem. Biophys. Res. Commun. 217, 1238 –1244 rameric complexes, although all evidence to date suggests they 22. Silverman, S. K., Lester, H. A. & Dougherty, D. A. (1996) J. Biol. Chem. 271, are not functional. 30524 –30528 In summary, the experiments presented here demonstrate 23. Tucker, S. J., Pessia, M. & Adelman, J. P. (1996) Am. J. Physiol. 271, H379 –H385 that GIRK channels form tetramers and that I is most KACh 24. McCormack, K., Lin, L., Iverson, L., Tanouye, M. & Sigworth, F. (1992) likely made up of two GIRK1 subunits and two GIRK4 sub- Biophys. J. 63, 1406 –1411 units. Although we cannot rule out the possibility that there 25. Langosch, D., Thomas, L. & Betz, H. (1988) Proc. Natl. Acad. Sci. U. S. A. 85, 7394 –73988 exist multiple pools of channels with varying stoichiometries, 26. Slaughter, R. S., Sutko, J. L. & Reeves, J. P. (1983) J. Biol. Chem. 258, the consistency of channel conductances and kinetics from sin- 3183–3190 27. Krapivinsky, G., Krapivinsky, L., Velimirovic, K., Wickman, B., Navarro, B. & gle-channel recordings in numerous species make this unlikely. Clapham, D. E. (1995) J. Biol. Chem. 270, 28777–28779 This study lays the foundation for future biochemical studies 28. Kennedy, M., Nemec, J. & Clapham, D. E. (1996) Neuropharmacology 35, on G binding stoichiometry, and determination of K channel 831– 839 bg 29. Slesinger, P., Reuveny, E., Jan, Y. & Jan, L. (1995) Neuron 15, 1145–1156 subunit arrangement around the pore. 30. Wischmeyer, E., Doring, F., Wischmeyer, E., Spauschus, A., Thomzig, A., Veh, R. & Karschin, A. (1997) Mol. Cell. Neurosci. 9, 194 –206 Acknowledgments—We thank Matt Kennedy and Kevin Wickman for 31. Huang, D. & Maero, S. (1997) BioTechniques 22, 454 – 458 critically reading the manuscript, Matt Kennedy for expertise and for 32. Davies, G. & Stark, G. (1970) Proc. Natl. Acad. Sci. U. S. A. 66, 651– 656 providing epitope-tagged GIRK1 and GIRK4, and Yiping Chen for pro- 33. Gaffney, B. (1985) Biochim. Biophys. Acta 822, 289 –317 viding technical assistance. 34. Richards, F. & Peters, K. (1977) Annu. Rev. Biochem. 46, 523–551 35. Aris, J. P. & Simon, R. D. (1983) J. Biol. Chem. 258, 14599 –14609 REFERENCES 36. Mourrain, P., Lasa, I., Guatreau, A., Gouin, E., Pugsley, A. & Cossart, P. 1. Papazian, D., Schwarz, T., Tempel, B., Jan, Y. & Jan, L. (1987) Science 237, (1997) Proc. Natl. Acad. Sci. U. S. A. 94, 10034 –10039 37. Velimirovic, B., Gordon, E., Lim, N., Navarro, B. & Clapham, D. E. (1996) 749 –753 2. Pongs, O., Kecskemethy, N., Muller, R., Krah-Jentgens, I., Baumann, A., FEBS Lett. 379, 31–37 Koltz, H., Canal, I., Llamazares, S. & Ferrus, A. (1988) EMBO J. 7, 38. Hedin, K., Nancy, L. & Clapham, D. (1996) Neuron 16, 423– 429 1087–1096 39. Krapivinsky, G., Gordon, E. A., Wickman, K. A., Velimirovic, B., Krapivinsky, 3. Iverson, L., Tanouye, M., Lester, H., Davidson, N. & Rudy, B. (1988) Proc. L. & Clapham, D. E. (1995) Nature 374, 135–141

Journal

Journal of Biological Chemistry – Unpaywall

Published: Feb 1, 1998

There are no references for this article.