Antibodies against Synthetic Peptides Used to Determine the ...

THEJOURNAL OF BIOLOGICAL CHEMISTRY 0 1992 by The American Society for Biochemistry and Molecular Biologv, Inc.

Vol. 267, No. 1, Issue of January 5, pp: 644-648, 1592 nnted m U.S.A.

Antibodies against Synthetic Peptides Used to Determine the Topology and Site of Glycosylation of the cGMP-gated Channel from Bovine Rod Photoreceptors* (Received for publication, August 5, 1991)

Paulus Wohlfartt, Winfried Haaset, Robert S. MoldayO, and Neil J. Cooktn From the #Max-Planck-Institut fur Biophysik, Abteilungfur Molekulare Membranbiologie,Heinrich-Hoffmann-Strasse 7, 0-6000 Frankfurt am Main 71, Federal Republic of Germany and the §Department of Biochemistry, University of British Columbia, Vancouver, British Columbia V6T I W5,Canada

Peptides corresponding to amino acids 321-339 (peptide GS21) and 416-431 (peptide GS31) of the cGMP-gated channel from bovine rod photoreceptors were synthesized and used as antigens for the preparation of polyclonal antibodies. After affinity purification, both antipeptide antibodies were found to bind specifically to the channel protein after Western blotting, but only the antibody against GS21 gave satisfactory results on enzyme-linked immunosorbent assay and electron microscopy. Using immunocytochemistry, we were able to localize amino acids 321-339 to the extracellular sideofthe rodphotoreceptor plasma membrane. By synthesizing heptapeptides corresponding to amino acids 324-330 (peptide GS2s) and 420426 (peptide G S ~ S )we , were able to affinity purify antibodies specific for two N-glycosylation consensus sites in the channel protein. As assessed by Western blotting, antibodies against GS3s were found to bind to both the glycosylated and deglycosylated channel proteins, whereas antibodies against GS2s only bound to the channel protein after enzymatic deglycosylation. Together, these results allow the refinement of folding models for the cGMP-gated channel and implicate Asn327 as being the sole site of N-glycosylation.

Vertebrate rod photoreceptors hyperpolarize after illumination due to theclosure of cation-specific channels, directly gated by cGMP, present inthe plasma membrane of the outer segment (Fesenko et al., 1985; Yau and Nakatani, 1985; Yau and Baylor, 1989). A well-characterized enzymatic cascade links the photoactivation of the photoreceptor protein rhodopsin to the hydrolysis of cytosolic cGMP andchannel closure (Stryer, 1986). Since the activation of one rhodopsin molecule leads to the closure of thousands of channels, this cascade also serves to amplify the original light signal. In previous studies, we have succeeded in purifying and functionally reconstituting the cGMP-gated channel from

bovine ROS’ (Cook et al., 1986, 1987). The purified channel protein has a relative M,of 63,000 and is probably functional as a multimer (tetramer or pentamer). It is glycosylated at one site only (Wohlfart et al., 1989) and is post-translationally modified by specific cleavage and removal of a 92-amino acid peptide at theN terminus (Molday et aL, 1991). The channel exists exclusively in the ROS plasma membrane (Cook et al., 1989) and is directly associated with a cytoskeletal protein that shows immunocross-reactivity with spectrin (Molday et al., 1990). Cloning and sequencing of the rod photoreceptor cGMPgated channel (Kauppet al., 1989) revealed that this protein does not fit into either of the two classical channel families: (i) ligand-gated channels, which consist of apentamer of subunits with four putative transmembrane helices; or (ii) voltage-gated channels, which exist as homotetramers of subunits (orpseudo-subunits) containingsix putative transmembrane helices. Although the cGMP-gated channelhas a region withremarkable homology to the so-called “54 segment” voltage sensor of voltage-gated channels (Jan and Jan,1990), it displays virtually no voltage sensitivity (Yau and Baylor, 1989). In thisstudy, we have used antibodies directed against synthetic peptides corresponding to sequences within the cGMP-gated channel polypeptide to probe the topology and potential N-glycosylation sites of the cGMP-gated channel. The results permit the identification of Asn-327 as the sole site of glycosylation and allow the refinement of folding models for this novel channel protein. EXPERIMENTALPROCEDURES

Peptide Synthesis and Coupling-Peptides were synthesized using standard Fmoc (N-(9-fluorenyl)methoxycarbonyl)chemistry (Carpino and Han, 1972) and the solid-phase method (Merrifield, 1963). All peptides were synthesized with an additional cysteine residue at the C terminus to allow specific coupling. Peptides were purified by high performance liquid chromatography, and their correct molecular weights were confirmed by fast atom bombardment mass spectrometry. For antibody production, peptides GS21 and GS31 were coupled via the C-terminal cysteine to the carrier protein soybean trypsin inhibitor using the cross-linking reagent N-succinimidyl-3-(2-pyridy1dithio)propionicacid (Gordon et al., 1987). Antibody Production and Purification-Polyclonal sera against GS21 and GS31 were raised by subcutaneously injecting the trypsin inhibitor-coupled peptide a t multiple sites along the spines of New Zealand White rabbits (at least 0.2 wmolof peptide/immunization). The rabbits were boosted 1 month later and then at4-week intervals. Blood was collected 10 days after boosting. For the affinity purification of site-directed antibodies, the appropriate peptide was coupled

* This work was supported by Deutsche Forschungsgemeinschaft SFB 169 Projekte C3 and C4, the Fonds der Chemischen Industrie, the Leibniz Foundation, National Institutes of Health Grant EY 02422, and the Medical Research Council of Canada. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “adverThe abbreviations used are: ROS, rod outer segment(s); ELISA, tisement” in accordance with 18U.S.C. Section 1734 solelyto indicate enzyme-linked immunosorbent assay; PBS, phosphate-buffered sathis fact. 1To whom correspondence should be addressed. Tel.: 069-6704392; line; BSA, bovine serum albumin; Hepes, 4-(2-hydroxyethyl)-l-piperFax: 069-6704423. azineethanesulfonic acid.

644

Channel

cGMP-gated

to w-aminohexyl-Sepharose4B using the bifunctional reagent sulfosuccinimidyl-(4-iodoacetyl)aminobenzoic acid. The column was equilibrated at 4 "C with PBST (phosphate-buffered saline containing 0.05% (w/v) Tween 20), and antiserum (diluted10-foldwith the same buffer) was applied and allowed to circulate through the column overnight. After washing the column with PBST, antibodies were eluted with 0.1 M glycine HCl, pH 2.8, and then immediately neutralized with 1 M Hepes/KOH, pH 7.4. The antibodies were then dialyzed against PBS and concentrated by ultrafiltration. Aliquots were removed for the determination of protein concentration (Read and Northcote, 1981), and samples were then adjusted to contain 1% (w/v) BSA before storing at -70 "C. Enzyme-linked Zmmunosorbent Assay (ELISA)-Linbro E.I.A. I1 Plus microtitration plates (Flow Laboratories, Inc.) were coated with ROS membranes (0.1 mg of rhodopsin/well) in 50 mM sodium carbonate,pH 9.6, andthen blocked with 1% (w/v) BSA in PBS. Antipeptide antibody was applied at the appropriate dilution (in the presence or absence of peptide) in PBST containing 1%(w/v) BSA. After rinsing, anti-rabbit IgG-alkaline phosphatase conjugate was applied, and after further rinsing, antibody binding was detected in 0.1 M diethanolamine, pH 9.8, usingp-nitrophenyl phosphate as the substrate. The reaction was measured a t 405-450 nm using a Titertek Multiskan Plus MK I1 ELISA plate reader. Electrophoresis and Western Blotting-Electrophoresis and electrotransfer onto Immobilon membrane was carried out aspreviously described (Wohlfart et al., 1989). For Western blotting, membranes were blocked with 1%BSA in PBS, and antibody was then applied in PBST at theappropriate dilution. After washing, the membrane was incubated with anti-rabbit IgG-alkaline phosphatase in 1% BSA in PBS. After further washing, antibody binding was detected using 5-bromo-4-chloro-3-indolyl phosphate and nitro blue tetrazolium (Blake et al., 1984). Rod Outer Segment Membrane Preparation-ROS were routinely prepared using a discontinuous sucrose gradient procedure (Friedel et al., 1991).ROS plasma and disc membrane fractions were prepared by a ricin affinity procedure (Molday and Molday, 1987)without mild trypsinization. cGMP-gated Channel Purification and Deglycosylatwn-The cGMP-gated channel was purified from stripped ROS membranes as previously described (Cook et al., 1987). The purified fraction consisted of the channel protein (Mr 63,000) together with a spectrinlike protein (Mr240,000),which is directly associated with the cGMPgated channel (Molday et al., 1990). Partial deglycosylation of the channel protein was achieved by incubation with glycopeptidase F under nondenaturing conditions (Wohlfart et al., 1989). Electron Microscopy-Purified ROS were fixed for 1 h with 1% glutaraldehyde in 0.1 M sodium cacodylate and then absorbed onto Alcian blue-coated coverslips. The remaining glutaraldehyde was blocked with 2% glycine in PBS, and nonspecific binding was blocked with 1%BSA in PBS. After incubation with purified antibody in 1% BSA in PBS, samples were extensively washed with PBS andreacted with secondary antibody (either anti-rabbit IgG-colloidalgold or antirabbit IgG-horse radish peroxidase conjugate). The unbound second-

Topology

645

ary antibody was removed by extensively washing with PBS, and samples were then processed for electron microscopy as described elsewhere (Haase et al., 1990). In some experiments, ROS were lysed in 20 mM Tris buffer prior to antibody labeling as previously described (Molday et al., 1990). RESULTS

The strategy behind the synthetic peptide approach used in this study is outlinedin Fig. 1. It attempts to provide answers to the following two questions: (i) where is the sole site of N-glycosylation; and (ii) what is the orientation (ix. extracellular or cytosolic) of the segment linkingputative transmembrane helices H4 and H5 and thesegment linking putative transmembrane helices H5 and H6? The resolution of these two questions would provide important information on the topological distribution of the cGMP-gated channel polypeptide across the ROS plasma membrane. The primary sequence of the cGMP-gated channel contains five consensus sites( i e . Asn-X-Ser/Thr) for N-glycosylation. Two of these (Asn-90 and Asn-91) are eliminated by posttranslational processing of the channel protein, which leads to proteolytic cleavage between Ser-92 and Ser-93 (Molday et al., 1991). Of the three remaining sites, Asn-177 (GS1; see Fig. 1)seemed unlikely to be glycosylated since it is localized within the putative transmembrane helix H1. Furthermore, since the Ser-93 channel N terminus is known to be located on the cytosolic side of the ROS plasma membrane (Molday et al., 1991), should theH1 hydrophobic domain not be transmembrane, then Asn-177 would be located cytosolically and is therefore unlikely to be glycosylated. We therefore concentrated on Asn-327 and Asn-423, which correspond to potential glycosylation sites GS2 and GS3, respectively (Fig. 1). Polyclonal antisera were prepared against the two peptides GS21 and GS31 (see Fig. l),and sequence-specific antibodies (referred t o as anti-GS21 and anti-GS31) were affinity-purified before use. As shown by Western blotting (Fig. Z), both antiGS21 and anti-GS31 were found t o specifically bind to the 63kDa channel protein in purified extracts and ROS plasma membrane extracts. The ROS disc membrane was found to be devoid of immunoreactivity, in accordance with previous resultson the channelprotein's localization (Cook et al., 1989). The fact that both antibodies bindto the glycosylated channelprotein presumably means that the carbohydrate moiety, if present at Asn-327 or Asn-423, does not hinder antibody accessibility and that the peptide sequences flanking

-

e p peptide-Gm FIG. 1. Synthetic peptide strategy employed for production of antibodies specific for two candidate N-glycosylation sites and for topological refinement of channel protein folding models. Boldface hatched rectangles correspond to very hydrophobic putative transmembrane helices; lightfacehatched rectangles correspond to segments where the weak hydrophobicity makes their classification as putative transmembrane helices questionable. The S4 rectangle corresponds to theputative S4 segment voltage sensor of voltage-gated channels, which has a precursor or vestigial form in the cGMP-gated channel (Jan and Jan, 1990);the cGMP rectangle corresponds to the cGMP-binding site that must be on the cytosolic side of the membrane. The assignment of helices Hl-H6 is described by Kaupp et al. (1989);the threeconsensus N-glycosylation sites (GS1, GS2, and GS3) are designated by solid triangles. The peptides synthesized for this study are indicated by the boldface horizontal lines and correspond to thepotential N-glycosylation sites GS2 and GS3. Long peptides bear the suffix 1, and shortpeptides the suffix s.

cGMP-gated Channel Topology

646 a

'i

b

ac

b

ac

b

c

-

-

i45

CB

anti-GS21

anti-GS31

FIG.2. Western blotting using affinity-purified antibodies against GS21 and GS31. Lane a, purified channel extract; lane b, purified ROS plasma membrane;lane c, purified ROS disc membrane. Left, Coomassie Blue (CR)-stained gel; middle, Western blot using antibodiesagainst GS21; right, Western blot using antibodiesagainst GS31.

-.: 205

i

,,

a cb d

mFl

mnplr C

6&'i Ei

FIG.3. Specific inhibition of antibody binding to cGMPgated channel by synthetic peptide used for immunization as determined by Western blotting. Lane a, anti-GS31binding; lane b, anti-GS31 binding in the presence of 1 p~ peptide GS31; lane c, anti-GS2I binding; lane d, anti-GS21 binding in the presenceof 1 p M peptide GS21. FIG.5. Sidedness and specificity of anti-GS2l binding to ROS plasma membrane as determined by electron microscopy after pre-embedding labeling. a, immunogold labeling showsthat anti-GS21 binds to the extracellular side of the ROS plasma mem-

T-"J 0

,o'lo

10-9

IO-* IO-'

IO-^

Peptide (M) FIG.4. Specific inhibition of anti-GS21 binding by peptide GS21 as determined by ELISA. The binding of anti-GS21 to ROS membranes was assessed by ELISA at a constant concentration of antibody in the presence of different concentrationsof peptide GS21 and control peptideR-6 (Gordon et al., 1987).

the glycosylated asparagine residue are long enough to form epitopes. In Fig. 3, the specificity of the affinity-purified antibodies is demonstrated by the fact thatimmunoreactivity as determined by Western blotting is specifically blocked by adding the peptide against which the antibodies were raised. We attempted to further characterize the binding of anti-GS21 and anti-GS31 to thecGMP-gated channel protein by ELISA. In such experiments, ROS membranes were present as the solid phase, i.e. the channel protein was present in its essentially natural conformation. Of the two antibodies tested,only anti-GS2lbppeared to show satisfactory affinityand specificity for the channel protein under such nondenaturing conditions. In Fig. 4, the binding of anti-GS21 to ROS membranes is investigated. Antibody binding to the channelprotein was

brane; b, immunogold labelinga t higher magnification; c, immunogold labeling of ROS with anti-GS21 in thepresence of 1p~ peptide GS21; d, immunolabeling using peroxidase conjugate and diaminobenzidine substrate shows that anti-GS21 binds avidly to the ROS plasma membrane; e, peroxidase immunolabeling a t higher magnification; f, peroxidase immunolabeling with anti-GS21 in the presence of 1 pM peptide GS21. "

.

FIG.6. Determination of sidedness of anti-GS2l binding to plasma membrane of lysed ROS. Immunogold labelingis observed on the extracellular surface of the ROS plasma membrane and inverted plasma membrane vesicles. The cytoplasmic surfaces of the disc and plasma membranes are notlabeled. found to be specifically blocked by peptide GS21, with an IC50 of -10 nM. To determine the sidedness of the binding of anti-GS21 to the ROS plasma membrane, we performed electron microscopy using either horseradish peroxidase or colloidal gold

cGMP-gated Channel Topology a

b

CB

cd e f

anti- antiGS2sGS3s

FIG. 7. Bindin&of anti-GS2s and anti-GS3s to glycosylated and deglycosylated forms of the cGMP-gated channel as assessed by Western blotting. Purified channel extract was subjected to electrophoresis either before or after partial deglycosylation using glycopeptidase F (Wohlfart et al., 1989) and then subjected to electrotransfer. Western blotting was then carried outusingaffinitypurified antibodies anti-GS2s and anti-GS3s. Lanes a, c, and e, glycosylated channel; lanes b, d , and f , partially deglycosylated channel. Left, Coomassie Blue (CB)staining of glycosylated and partially deglycosylated channel extracts; middle, Western blot with anti-GS2s; right, Western blot with anti-GS3s.

n

647

plasmic surface of the plasma membrane. Localization of GS21 to the extracellular side of the ROS plasma membrane indicated that Asn-327 was a strong candidate for the site of N-glycosylation. To further investigate this possibility, we purified antibodies from the anti-GS21 and anti-GS31 polyclonal antisera using immobilized peptides GS2s and GS3s (Fig. 1) as affinity matrices. These short peptides contain the two potential glycosylation sites (Asn327 and Asn-423) flanked by only 3 aminoacids on each side. Antibodies, which were purified this way, are referred to as anti-GS2s and anti-GS3s, respectively. As shown in Fig. 7 (lanes c and e ) , only anti-GS3s was capable of binding to the glycosylated channel protein as assessed by Western blotting. Treatment of the cGMP-gated channelwith glycopeptidase F under nondenaturing conditions yields a partially deglycosylated channel extract exhibiting bands of 63 kDa (glycosylated) and 61kDa (deglycosylated) on sodium dodecyl sulfate gels (lanes a and b ) , the latter of which is devoid of lectinbinding properties (Wohlfart et al., 1989). Western blotting of such an extractshowed that theglycosylated channel only binds anti-GS3s (lanes c and d ) , whereas the deglycosylated channel is capable of binding both anti-GS2s and anti-GS3s (lanes d and f ). We therefore concluded that N-glycosylation of Asn-327 prevents anti-GS2sbinding, presumably by blocking the accessibility of the antibody to its binding site. The fact that the glycopeptidase F-treated form of the channel binds anti-GS2s suggests also that theside group of Asn-327 is not critical in antibody binding since deglycosylation with this enzyme leads to deamidation of the glycosylated asparagine, yielding aspartic acid (Plummer et al., 1984). DISCUSSION

The results presented hereprovide evidence that theamino acid sequence of the cGMP-gated channel that connects puB. tative transmembranehelices H4 and H5is extracellular and glycosylated at Asn-327. Together with the knowledge that (i) the C-terminal region contains the cGMP-binding siteand COOH must therefore be cytoplasmic and that(ii) the N terminus is cytoplasmic (Molday et al., 1991), we are now in a position to refine current folding models for this novel channel protein. Of the two models proposed by Kaupp et al. (1989), the sixhelix model is clearly inconsistent with our results since it requires that theloop joining H4 and H5 be cytoplasmic. The four-helix model proposed by these authors is shown in Fig. C 8A and isentirely compatible with our results. In Fig. 8 ( B and C ) , two further folding models consistent with ourdata for the cGMP-gated channelare shown and are N OOH based on the observation that the cGMP-gated channel contains a so-called S4 segmentthat, in the voltage-gated channel FIG. 8. Folding models for cGMP-gated channel polypep- family, is considered to be a voltage sensor and therefore tide within ROS plasma membrane. In each folding model, the Fig. 8B asconsensus sites for N-glycosylation, referred to as GS2 and GS3 in constitutes a putative transmembrane a-helix. the text, are labeled 2 and 3, respectively. The upper side of each sumes H5 tobe transmembrane and H6 tobe cytosolic. This model represents the extracellular side of the membrane, and the model is attractive in that itplaces the consensus glycosylalower side represents the cytosolic side. tion site Asn-423 (which we have concluded not to be glycosylated) on the cytoplasmic side of the plasma membrane, conjugates as second antibodies. Using the pre-embedding where glycosylation is known to rarely take place. Fig. 8C presents a folding model that reflects the current method, we found that anti-GS21 bound avidly to the extracellular side of the ROS plasma membrane(Fig. 5, a, b, d, and state of thinking on the topology of the voltage-gated channel e). The binding of anti-GS21 was found to be specific since family (Guy and Conti, 1990; Yo01 and Schwarz, 1991; Yellen antibody binding could be efficiently blocked by adding pep- et al., 1991; Hartmann et al., 1991). H5 is considered not to tide GS21 (Fig. 5, c and f ) . Lysed ROS, with the cytoplasmic be transmembrane; but since it has been implicated through surfaces of the disc and plasma membranes exposed, were site-directed mutagenesis studies as playing a role in pore come out of the membrane used to furtherdefine anti-GS21 binding specificity. As shown function, it is assumed to enter and in Fig. 6, labeling was only observed along the extracellular without actually crossing it. In thismodel, Asn-423 is located surface of the plasma membrane in such preparations. No extracellularly even though it is not glycosylated. significant labeling was observed on discs or on the cytoThe presence of a putative S4segment in the cGMP-gated

cGMP-gated Channel Topology

648

channel would appear to be morethan coincidental, especially given the fact that it is located exactly where one would predict (i.e. between H3 and H4) by comparison with the voltage-gated channel family. It should, however, be noted that the cGMP-gated channel shows virtually no voltage dependence. One reason for this couldbe the presence of negatively charged glutamate residues (Glu-267, Glu-281, and Glu-287), whichwould presumably negate the positively charged arginines, in the vicinity of the S4 segment. Indeed, it has been shown for voltage-gated channels that introduction or neutralization of the positively charged arginine residues in the S4 segment will decrease voltage sensitivity (Stuhmer et al., 1989; Papazian et al., 1991). It wouldbe interesting to neutralize the above-mentioned glutamates in the cGMP-gated channel through mutagenesis to see if it would be possible to introduce voltage dependence. In conclusion, although our results allow the refinement of current folding models for the cGMP-gated channel, it is still unclear in which family (ligand- or voltage-gated channels) this channel protein belongs. The presence of a putative S4 segment may attestto a vestigial (orperhaps precursor) voltage sensor, thereby making this channelan exotic member of the voltage-gated channel family. Further work, both topological and mutagenic, is needed to clarify this question. Acknowledgments-We thank Heidi Muller and Laurie Molday for technical assistance and Robert Gordon for helpful advice. REFERENCES Blake, M. S., Johnston, K. H., Russell-Jones, G. J., and Gotschlich, E. C. (1984) Anal. Biochem. 1 3 6 , 175-179 Carpino, L. A., and Han, G. (1972) J. Org. Chem. 3 7 , 3404-3409 Cook, N. J., Zeilinger, C., Koch, K.-W., and Kaupp, U. B. (1986) J. Biol. Chem. 261,17033-17039 Cook, N. J., Hanke, W., and Kaupp, U. B. (1987) Proc. Natl. Acad. Sci. U.S. A. 84,585-589

Cook, N. J., Molday, L. L., Reid, D., Kaupp, U. B., and Molday, R. S. (1989) J. Biol. Chem. 264,6996-6999 Fesenko, E. E., Kolesnikov, S. S., and Lyubarsky, A. L. (1985) Nature 313,310-313

Friedel, U., Wolbring, G., Wohlfart, P., and Cook, N. J. (1991) Biochim. Biophys. Acta 1061,247-252 Gordon, R. D., Fieles, W. E., Schottland, D. L., Hogue-Angeletti, R., and Barchi, R.L. (1987) Proc. Natl. Acud. Sci. U. S. A. 8 4 , 308312

Guy, H. R., and Conti, F. (1990) Trends Neurosci. 1 3 , 201-206 Haase, W., Friese, W., Gordon, R. D., Muller, H., and Cook, N. J. (1990) J. Neurosci. 10, 1486-1494 Hartmann, H. A., Kirsch, G. F., Drewe, J. A., Taglialatela, M., Joho, R. H., and Brown, A. M. (1991) Science 251,942-944 Jan, L. Y., and Jan,Y. N. (1990) Nature 345,672 Kaupp, U. B., Niidome, T., Tanabe, T., Terada, S., Bonigk, W., Stuhmer, W., Cook, N. J., Kanagawa, K., Matsuo, H., Hirose, T., Miyata, T., and Numa, S. (1989) Nature 342,762-766 Merrifield, R. B. (1963) J. Am. Chem. Soc. 85,2149-2154 Molday, R. S., and Molday, L. L. (1987) J. Cell Biol. 105,2589-2601 Molday, L. L., Cook, N. J., Kaupp, U. B., and Molday, R. S. (1990) J.Biol. Chem. 265,18690-18695 Molday, R. S., Molday, L. L., Dose, A., Clark-Lewis, I., Illing, M., Cook, N. J., Eismann, E., and Kaupp, U.B. (1991) J. Biol. Chem. 266,21917-21922

Papazian, D. M., Timpe, L. C., Jan, Y. N., and Jan, L.Y. (1991) Nature 3 4 9 , 305-310 Plummer, T. H., Jr., Elder, J. H., Alexander, S., Phelan, A. W., and Tarentino, A. L. (1984) J. Biol. Chem. 2 5 9 , 10700-10704 Read, S. M., and Northcote, D. H.(1981) Anal. Biochem. 116, 5364

Stryer, L. (1986) Annu. Reu. Neurosci. 9,87-119 Stuhmer, W., Conti, F., Suzuki, H., Wang, X., Noda, M., Yahagi, N., Kubo, H., and Numa, S. (1989) Nature 339,597-603 Wohlfart, P., Muller, H., and Cook, N. J. (1989) J. Biol. Chem. 2 6 4 , 20934-20939

Yau, K.-W., and Baylor, D. A. (1989) Annu. Rev. Neurosci. 1 2 , 289327

Yau, K.-W., and Nakatani, K. (1985) Nature 317,252-255 Yellen, G., Jurman, M. E., Abramson, T., and MacKinnon, R. (1991) Science 251,939-942 Yool, A. J., and Schwarz, T. L. (1991) Nature 3 4 9 , 700-704