Communication Vol. 260 No. 7 Issue of April 10 pp. 3895-3898 1986 GI 1985 by The American Society of Bioiogical Chemisd, Inc.

Printed in U. S A .

THE JOURNAL OF BIOLOGICAL CHEMISTRY

Photoaffinity Labeling of Components of the Apamin- sensitive K+ Channel in Neuronal Membranes"

(Received for publication, December 11,1984) Michael J. Seagar$, Catherine Labbe-Jullie, Claude Granier, Jurphaas Van Rietschoten, and Franpois Couraud From the Laboratoire de Biochimie, Znstitut National de la Sante et de la Recherche Medicale, Unite 172, Facult6 de Medecine, Secteur Nord, Boulevard Pierre Dramurd, 13326 Marseille Cedex 15, France

An azidonitrophenylaminoacetyl m~no['~~I]iodoapa- min derivative was prepared which showed specific binding to rat neuronal membranes. UV photolysis lead to the irreversible occupation of binding sites. Photo- labeling of intact primary cultured rat neurones fol- lowed by membrane solubilization, sodium dodecyl sul- fate-polyacrylamide gel electrophoresis, and autora- diography revealed the covalent incorporation of ra- dioactivity into 3 main components with M. = 86,000, 30,000, and 23,000. Labeling was completely pre- vented by a competing excess of native apamin. Similar studies on purified synaptic membranes from the rat brain showed another labeling pattern with major bands corresponding to M. = 86,000 and 69,000. Al- though the reasons for the partial discrepancy between cultured embryonic neurons and an adult brain mem- brane fraction are not yet clear, we conclude that these proteins are intimately associated with the apamin binding site and are probably components of a type of Cas+-activated IC+ channel.

Hyperpolarizing K+ conductances controlled by changes in the free cytoplasmic Ca2+ concentration have been described in a variety of excitable cells where they are thought to be involved in the regulation of repetitive firing (I), as well as in other cell types where their physiological role is less clear (2). Evidence that several classes of Ca2+-activated cation-selec- tive channels may exist has been obtained from both electro- physiological and pharmacological studies (3,4).

Apamin, a 2000-dalton peptide neurotoxin from bee venom, was first shown to block the increase in K+ permeability resulting from receptor activation by a-adrenergic agonists and ATP in hepatocytes, which is mediated by a mobilization of intracellular Ca2+ (5). In mouse neuroblastoma and rat muscle cells, it blocks the slow Ca2+-dependent K+ current responsible for the hyperpolarization which follows the action potential (6, 7) but is inactive on the high conductance Ca2+- activated K+ channel widely studied by the patch clamp technique (8).

* This research was supported by Grant U.A553 from the Centre National de la Recherche Scientifique. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

$ To whom correspondence should be addressed.

We have recently demonstrated high affinity binding of m~no['~I]iodoapamin to intact primary cultured rat neurones which can be correlated to an inhibition of 86Rb+ efflux (9). We now report the use of a photolabile apamin derivative to identify the polypeptide chains associated with this type of K+ channel in rat neuronal membranes.

EXPERIMENTAL PROCEDURES

Preparation of 'Z61-ANPAA-apamin'-Apamin was purified from bee venom (10) and radioiodinated (9) and a m~no['~I]iodoapamin derivative with a specific radioactivity of 2000 Ci/mmoI was separated by SP(sulfopropy1)-Sephadex C-25 chromatography (11). In routine experiments, 200 pl of 10-20 nM mon~['~I]iodoapamin were adjusted to pH 8.5 by addition of NaOH and to a final concentration of 1 mg/ ml of bovine Serum albumin, in a final volume of 220 pl . All the following operations were then carried out in a dark room using a red photographic lamp. 4-Azido-2-nitro-phenylaminoacetylsuccinimidyl ester (a gift of Dr. K. Angelides, University of Florida, Gainesville) was then added in 10 p1 of acetonitrile. Molar ratios of reagent to reactive amino groups (of the m~no['~~I]iodoapamin plus the serum albumin) ranging from 2:1 to 101 have been used without any difference in the apparent molecular weight of the subsequently labeled protein bands. The reaction was allowed to continue for 3 h then quenched by dilution into the binding buffer. Apamin contains two modifiable amino groups, the a-NHz of Cys 1 and the c-NH2 of Lys 4 with pK values of 7.3 and 10.6, respectively, neither of which are necessary for biological activity (12). Our conditions should "a priori" favor modification of the a-NH2, although, however, some di- ANPAA-mono['261]iodoapamin may be formed.

Bwlogical Materiul-The brains from 16-day-old fetal Wistar rats were mechanically dissociated and cultured as previously described (9) in poly-L-lysine-coated 60-mm dishes (Corning) for solubilization experiments, or 24-mm multiwell plates (Flow Laboratories, Inc.) for kinetic binding experiments. Synaptic membranes from rat brains were prepared as described (13) except that the 1-1.2 M sucrose interface synaptosomal fraction was lysed by dilution into 20 volumes of 5 mM KPi, pH 7.4, and incubated for 30 min at 1 "C. Membranes were then sedimented at 35,000 X g for 30 min, resuspended in 1 ml of serum albumin free binding buffer/g of original tissue, and stored in liquid nitrogen. The P2 crude synaptosomal fraction is the 11,OOO X g pellet of the postnuclear supernatant, which was used immediately without freezing. Human red blood cell membranes were prepared by a published method (14). In certain experiments the protease inhibi- tors phenanthroline (1 mM), pepstatin A (1 p ~ ) , phenylmethylsul- fonyl fluoride (100 gM), iodoacetamide (1 mM), and EDTA (1 mM) were included from tissue homogenization onwards. Protein was assayed by the Bradford method (15) or a modified Lowry method (16) using a bovine serum albumin standard.

Photolabeling Procedure-The binding buffer for neuronal cultures contained 25 mM Hepes, 10 mM glucose, 140 mM NaCl, 5.4 mM KCI, 1.8 mM CaClz, 0.8 mM MgSO,, and 0.1% bovine serum albumin adjusted to pH 7.2 with Tris base. lZ61-ANPAA-apamin was included at a concentration of 0.15 to 0.4 nM in the presence and absence of 0.2 p~ native apamin. Incubation was carried out in the dark at 1 "C for 60 min, and cultures were then irradiated on ice for 5 min at about 5 cm from a ventilated 125-watt Philips mercury vapor lamp (X, = 356 nm). The cell layer was then washed 3 times with ice-cold incubation buffer, the last wash in the absence of albumin. Cells were harvested and homogenized in 2 ml of 0.32 M sucrose, 10 mM KC1, 0.5 mM Tris/HCl at pH 7.2, containing the same protease inhibitors as above. Nuclei were eliminated by centrifugation for 10 min at 600 X g and membranes were then pelleted at 150,000 X g for 30 min for counting and electrophoresis.

In kinetic dissociation experiments, cells were collected for y counting in 1.5 ml of 0.1 M NaOH.

The abbreviations used are: '2SI-ANPAA-apamin, 4-azido-2-nitro- phenylaminoacetylmono['*sI]iodoapamin; Hepes, 4-(2-hydroxyeth- y1)-l-piperazineethanesulfonic acid; SDS, sodium dodecyl sulfate.

3895

3896 Protein Components of the Apamin-sensitive K" Channel The binding buffer for membrane preparations contained 25 mM

Hepes, 10 mM KCl, and 0.1% bovine serum albumin adjusted to the desired pH with Tris base. For routine photolabeling 400 pg of membrane protein were incubated in 1 ml of buffer in 35-mm culture dishes containing 80-200 PM "'I-ANAA-apamin in the presence or absence of 0.2 p~ native apamin and then irradiated for 5 min, after 60 min at 1 "C. Membranes were pelleted at 11,000 X g in a Janetski bench centrifuge and washed 3 times by resuspension. In kinetic binding experiments 150 pg of membrane protein were incubated in 0.5 ml containing 80 pM '251-ANPAA-apamin until equilibrium. Dis- sociation was initiated by addition of native apamin to give a final concentration of 0.2 p~ and bound ligand was separated a t the indicated times by rapid filtration, under reduced pressure on Milli- pore EH cellotate filters.

Electrophoresis and Autoradiography-Membrane samples were resuspended in 30 pl of HzO and then solubilized and denatured for 5 min at 100 "c in a final volume of 80 p1 containing 40 mM Tris, 6 mM disodium ethylenediaminetetraacetate, 6% glycerol, 2% SDS, and 10 mM mercaptoethanol, pH 9. Subsequent disulfide bonding was prevented by addition of iodoacetamide to a concentration of 40 mM. Samples were then stored at -20 "C until SDS slab gel electrophoresis according to Laemmli (17) on 5-15% acrylamide gradient gels. Gels were stained with Coomassie Blue and dried and autoradiography was carried out using Kodak X-Omat films in cassettes equipped with intensifying screens at -80 "C. Autoradiogram scanning was carried out on an LKB Ultrascan Laser densitometer.

RESULTS

M~no['~~I]iodoapamin was modified as described and then immediately used for binding. Cell cultures or synaptic mem- branes were incubated with 80 PM lZ5I-ANPAA-apamin in the presence or absence of 0.2 PM native apamin. At this concen- tration the total binding to neuronal cultures was about 4 fmol of lZ5I-ANPAA-apamin/mg of protein with a nonspecific component of 1 fmol/mg of protein. Purified synaptic mem- branes bound 19 fmol of lZ5I-ANPAA-apamin/mg of protein, nonspecific binding accounting for 8 fmol/mg. Samples were then photolyzed or kept in the dark; a large excess of native apamin was added to initiate dissociation and release kinetics were followed. After UV irradiation a fraction of the saturable binding became nondissociable (Fig. 1).

In the case of neuronal cultures the release from nonirra- diated cells at 37 "C is rapid and complete whereas after photolysis about 40% of the occupied binding sites are non- dissociable (Fig. 1A 1.

Dissociation curves with synaptic membranes were esti- mated at 1 "C. At this temperature, dissociation is slower and, in nonirradiated samples, not yet complete at 160 min (Fig. 1B). However, comparison with photolyzed samples indicates a similar degree of covalent binding.

Analysis of solubilized membranes by SDS-polyacrylamide gel electrophoresis and autoradiography revealed the incor- poration of radioactivity into several bands (Fig. 2). The labeling pattern obtained on intact cells was somewhat differ- ent from that on brain membrane fractions.

Firstly, on cultured neurons 3 major proteins of M, = 86,000 f 2,000,30,000 f 1,000, and 23,000 f 2,000 were labeled (lune I ) . Faint and diffuse labeling was occasionally detected at about 59,000 and 44,000 (see Fig. 2, lane I, open arrows). The given M , are the mean f range from 4 independent experi- ments after substraction of 2,000 the approximate M , of lZ5I- ANPAA-apamin. In the presence of a saturating concentra- tion of native apamin (lane 2 ) all these bands completely disappeared.

In a second series of experiments we tested synaptic mem- branes from the rat brain prepared from pinched off nerve terminals, purified by discontinuous density gradient centrif- ugation, hypotonic lysis, and resedimentation, and which are then stored in liquid Nz. This preparation has a binding capacity of about 15 fmol/mg of protein which is 3 times

z 1 3 0 n

0

0 10 20 30

mrJ '

Time, min.

4 A B

si InA 0 I 5 0 100 150

Time, min. FIG. 1. Effects of photolysis on the dissociation of Y - A N -

PAA apamin from binding sites. Neuronal cultures (A) and synaptic membranes (B) were incubated with 80 PM IZ5I-ANPAA apamin for 60 min at 4 "C, then either photolyzed (0) or kept in the dark (0). 0.2 p~ native apamin was added and the remaining asso- ciated radioactivity was measured at the indicated times of incubation at 37 "C (A) or 1 "C ( B ) . Nonspecific binding, measured by initially adding 0.2 p~ native apamin with "'I-ANPAA apamin, has been subtracted. B, bound; Beq, bound at equilibrium.

higher than that of cultured neurons' and is of interest as it is probably the best source of apamin receptor from which to start purification. In this system, 2 major proteins with M , = 86,000 f 2,000 and 59,000 f 2,000 were labeled (Fig. 2, lanes 3 and 4). Results are from six independent experiments after subtraction of '251-ANPAA-apamin. A less intense band often appearing as doublet with M , = 44,000 f 1,000 was also observed; the relative labeling of this component seemed to vary from one membrane preparation to another.

A very similar labeling pattern was obtained in a more rapidly prepared, unfrozen P2 membrane fraction in which all operations were carried out in the presence of 5 protease inhibitors (lane 5 ) . Specifically bound lZ5I-ANPAA-apamin, that was not cross-linked by photolysis, dissociates during denaturation and migrates with the front ( M , - 2,000); this is clearly seen in lunes 1 and 5. In control experiments where receptors were occupied by unlabeled apamin (e.g. lanes 1 and 5) or were absent (lunes 7 and 8) , this component is not present.

The omission of mercaptoethanol during denaturation did ~ ~~~~

M. J. Seagar, C. LabbL-JulliB, C. Granier, J . Van Rietschoten, and F. Couraud, unpublished observations.

Protein Components of the Apamin-sensitive K" Channel 3897

1 2 FIG. 2. Photoaffinity labeling of

neuronal membranes with '"1- ANPAA apamin. Neuronal cell cul- tures were incubated with '251-ANPAA MrX1e3 apamin in the absence (lane 1 ) and pres- ence (lane 2 ) of a large excess of native apamin and then photolyzed, homoge- nized in the presence of protease inhibi- tors, and processed for SDS slab gel elec- trophoresis and autoradiography as de- 86. scribed under "Experimental Proce- dures." Similar experiments were carried E

out with purified synaptic membranes from the rat brain in the absence of E

protease inhibitors (lanes 3 and 4 ) , and a crude P2 fraction was prepared rapidly in the presence of protease inhibitors 30* and photolabeled immediately (lanes 5 23, and 6). As a control, human red blood cell membranes, which do not carry apa- min binding sites, were treated identi- cally (lanes 7 and 8). The open arrow- heads in lane 1 indicate minor labeled components (see text).

not modify the apparent M,, nor did the substitution of bovine y-globulin for serum albumin as the carrier during modifica- tion and in the binding buffer (not shown). Membranes from human red blood cells, which do not display apamin-sensitive K' fluxes (4) nor carry 1251-apamin binding sites (18), did not show any specific incorporation of radioactivity into macro- molecular constituents (Fig. 2, lanes 7 and 8).

We have previously shown that the specific binding of m~no['~~I]iodoapamin to cultured neurones requires extracel- lular K+ (9) and similar observations have been reported with rat brain membranes (11, 19). This property provides an additional control for the specificity of labeling. Mon0['~~1] iodoapamin specific binding to synaptic membranes increased by a factor of 2-3 when the K+ concentration was increased from 0 to 10 mM (Fig. 3B). When K+ was absent from the photolabeling medium, the incorporation of radioactivity into the 86,000,59,000, and 44,000 bands was reduced (Fig. 3A) . A densitometric scanning profile is shown in Fig. 3C.

DISCUSSION

The evidence presented argues for the fact that 1251-AN- PAA-apamin binds specifically to the neuronal plasma mem- brane binding site thought to be associated with a Ca2+- activated K+ channel. After photolysis about 40% of the associated ligand became covalently linked and slab gel elec- trophoresis revealed the specific labeling of several polypep- tide chains. It is not uncommon when photolabeling oligo- meric proteins to detect the labeling of more than one subunit. For example, the relatively small ligand, [3H]chlorpromazine, which blocks a permeability response to acetylcholine, labels all four subunits of the acetylcholine receptor in Torpedo murmorutu membranes (20). -

We think it is reasonable to assume that one of the 1251- ANPAA-apamin labeled chains carries the binding site whereas the others are subunits present in the immediate environment of the apamin receptor. Although it is as yet unknown whether apamin binds to an associated regulatory component or to the ion channel itself, some evidence in favor of the latter hypothesis has been obtained (2).

A component migrating with an apparent molecular weight of 86,000 was the largest polypeptide labeled and was a major band in both neuronal cultures and synaptic membranes and

B6*

3 4 5 6 7 8

86

591

4 4

it is therefore tempting to think that this chain carries the apamin binding site. As far as the other labeled chains are concerned, an obvious difference exists between the two sys- tems studied. Although we are as yet unable to present any further experimental evidence, some speculation as to the origin of this discrepancy could be constructive.

In experiments using purified membrane fractions which necessitate a relatively long preparation, proteolysis may take place while still leaving functional ligand binding domains. Artifacts of this kind are notoriously difficult to demonstrate and control. However, in experiments carried out at 1 "C on intact cultured cells, with the addition of protease inhibitors before homogenization, proteolytic artifacts should be avoided; no difference in the labeling pattern was observed when the protease inhibitors were omitted. No internalization of this ligand seems to take place in these conditions as dissociation is complete after a short incubation at 37 "C without photolysis. We can therefore be fairly confident that the 86,000, 30,000, and 23,000 chains are intact polypeptide chains, near to the binding site. In synaptic membranes, lZ5I- ANPAA-apamin labeled two major bands at 86,000 and 59,000. The weaker band at 44,000 fluctuated in intensity from one membrane preparation to another and may be a proteolytic fragment. The 59,000 chain, however, was still a very prominent component in rapidly prepared P2 membranes in the presence of protease inhibitors.

Cross-linking of '251-apamin to rat synaptosomal mem- branes with disuccinimidyl suberate has been reported as labeling a single protein of M , = 28,000-33,000 (21,22). This could well correspond to one of the lower molecular weight components that we have labeled on cultured neurones al- though we did not detect any major labeling in this range on synaptic membranes. It is possible that the discrepancy be- tween our present data and previous work (21, 22) is due to inherent differences between photoaffinity labeling and cross- linking methodologies. In the former the ligand is derivatized before binding, in the latter when already bound to its recep- tor; it is possible that the accessibility to modification of apamin's two primary amines is not the same in each case. Furthermore, the chemical remtivity of a photogenerated nitrene is not the same as that of a succinimidyl ester and, as a consequence, the probability for insertion into a nearby

3898 Protein Components of the Apamin-sensitive K+ Channel

v K+= 0

4 MIGRATION FIG. 3. Effect of K+ on the photoincorporation of laSI-

ANPAA apamin. A, synaptic membranes were photolabeled in the presence of 10 mM KC1 (lanes 1 and 2 ) or in the absence of added KC1 (lanes 3 and 4) . Nonspecific labeling was determined in the presence of 0.2 p~ native apamin (lanes 2 and 4 ) . B, the specific binding of m~no['~I]iodoapamin to synaptic membranes was deter- mined after a 60-min incubation in the presence of increasing con- centrations of K+. B, bound; Bo, bound in the presence of 30 mM KCl. C, the densitometric profile of the autoradiogram shown in A (lane 1,lO mM KCI; lane 3 without added KCl).

polypeptide will not be equivalent. More work is necessary to clarify this inconsistency.

One possible explanation for the difference between neu- ronal cells and brain membranes is that the labeled proteins detected in each system are all present in both systems but their accessibility to the photoactivated nitrene group is mod- ified by a change in their spatial inter-relationships induced by cell disruption. This kind of phenomenon has been re- ported in studies on the photolabeling of the adipocyte insulin receptor in which one subunit was labeled on whole cells but not on membranes (23). This hypothetical conformational transition of the apamin receptor does not seem to be related to a change in membrane potential. K'-induced depolariza- tion or preincubation with veratridine plus scorpion toxin,

which opens voltage-sensitive sodium channels, did not mod- ify the photolabeling pattern (results not shown).

As the 59,000 component was detected in adult tissue and the two smaller chains in primary cultured neurons, which is an embryonic system, another possibility is that its appear- ance is the reflection of functional maturation. This could also conceivably be related to morphological differentiation; the brain membrane fractions studied presumably contain neuronal plasma membrane of synaptic origin whereas, in the embryonic neuronal cells (6-15 days in culture), mature syn- aptic structures are quite scarce (24):

Although any conclusions as to the relationships existing between these affinity labeled polypeptides will need further characterization, we suggest that these proteins, associated with the apamin binding site, are subunits of an oligomeric structure involved in a transmembrane K+ flux.

Acknowledgments-We thank Paule Frachon for innovative tech- nical assistance, Paule Deprez for the cell culture, Her& Darbon and Emmanuel Jover for helpful advice, and Kim Angelides for the generous gift of reagents.

REFERENCES 1. Meech, R. W. (1978) Annu. Reu. Biophys. Bioeng. 7,1-18 2. Jenkinson, D. H., Haylett, D. G., and Cook, N. S. (1983) Cell

3. Marty, A. (1983) Trends Neurosci. 6,262-265 4. Burgess, G. M., Claret, M., and Jenkinson, D. H. (1981) J.

5. Banks, B. E. C., Brown, C., Burgess, G. M., Burnstock, G., Claret, M., Cocks, T. M., and Jenkinson, D. H. (1979) Nature 282, 415-417

6. Hugues, M., Romey, G., Duval, D., Vincent, J. P., and Lazdunski, M. (1982) Proc. Natl. Acad. Sci. U. S. A. 79,1308-1312

7. Hugues, M., Schmid, H., Romey, G., Duval, D., Frelin, C., and Lazdunski, M. (1982) EMBO J. 1,1039-1042

8. Romey, G., and Lazdunski, M. (1984) Biochem. Bwphys. Res. Commun. 118,669-674

9. Seagar. M. J.. Granier. C.. and Couraud. F. (1984) J. Bwl. Chem.

Calcium 4,429-437

PhySiOl. (Lo&.) 317,67-90

259; 149111495 . . . . .

10. Banks. B. E. C.. DemDsev. C. E.. Pearce. F. L.. Vernon. C. A.. and Wholley, T. E. (l9&) Anal: Biocheh. ll6,48-52 '

11. Hugues, M., Duval, D., Kitabgi, P., Lazdunski, M., and Vincent, J. P. (1982) J. Bwl. Chem. 257,2762-2769

12. Vincent, J. P., Schweitz, H., and Lazdunski, M. (1975) Biochem-

13. Catterall, W. A., Morrow, C. S., and Hartshorne, R. P. (1979) J.

14. Hanahan, D. J., and Ekholm, J. E. (1974) Methods Enzymol. 31,

15. Bradford, M. M. (1976) Anal. Biochem. 72,248-254 16. Lowry, 0. M., Rosebrough, N. J., Farr, A. L., and Randall, R. J.

17. Laemmli, U. K. (1970) Nature (Lo&.) 227,680-685 18. Cook, N. S., Haylett, D. G., and Strong, P. N. (1983) FEBS Lett.

19. Habermann, E., and Fischer, K. (1979) Eur. J. Biochem. 94,

istry 14,2521-2525

Biol. Chem. 254,11379-11387

168-172

(1951) J. Bwl. Chem. 193,265-275

152,265-269

R55-Rfa 20. Heidmann, T., and Changeux, J. P. (1984) Proc. Natl. Acad. Sci.

"- " _ U. S. A. 81.1897-1901

21. Hugues, M., Schmid, H., and Lazdunski, M. (1982) Biochem. Bwphys. Res. Commun. 107,1577-1582

22. Schmid-Antomarchi, H., Hugues, M., Norman, R., Ellory, C., Borsotto, M., and Lazdunski, M. (1984) Eur. J. Bwchem. 142,

23. Yip, C. C., Moule, M. L., and Young, C. W. T. (1982) Biochemistry

24. Yavin, Z., and Yavin, E. (1977) Exp. Brain Res. 29,137-147

1-6

21,2940-2945

P. Cau, personal communication.