(4) chemical signaling brain

As long ago as 1904, the British sci-entist T. R. Elliot proposed cor-rectly that neurons (nerve cells)

often communicate with one anotherand with other cell types not electrical-ly but chemically. He suggested that anaction potential, or electrical impulse,propagating along an excited neurontriggers the release of chemicals (nowcalled neurotransmitters) from the ex-cited cell. In turn, the liberated chemi-cals may cause another cell to take in orextrude selected ions. By thus alteringthe ßow of charge across the membraneof this second cell, the neurotransmit-ters can give rise to a new impulse.

Since then, investigators have identi-fied perhaps 50 neurotransmitters andhave learned that a single neuron maysecrete several of them. Workers havealso struggled to explain just how neu-rotransmitters, particularly those in thebrain, manage to regulate ionic trans-port, and hence impulse production, inthe cells they inßuence.

The solution to this last problem hasemerged slowly, but research in mylaboratory at the Pasteur Institute inParis and in other laboratories duringthe past 25 years has made great head-way. We have ascertained that recep-tors for neurotransmittersÑwhich pro-trude from cell membranesÑplay a piv-otal role in mediating the conversion ofchemical signals into electrical activity.And we have begun to clarify how cer-tain major receptors carry out thischallenging task. Those molecules have

been found to constitute a remarkablesuperfamily of neurotransmitter recep-tors that are known as neurotransmit-ter-gated ion channels.

Much of this insight derives from in-tensive study in the 1970s and 1980sof a receptor initially isolated from theelectricity-generating organ of an elec-tric Þsh. Yet the story of how investiga-tors solved the mysteries of impulsetransmission by chemicals more prop-erly begins decades earlier, with the pi-oneering contributions of John New-port Langley of the University of Cam-bridge. In 1906 he proposed that bodilytissues bear receptors for drugs. In sodoing, he provided one of the Þrst sig-niÞcant clues to the means by whichneurotransmitters exert their eÝects.

Langley based his proposal on stud-ies he conducted into how the poisoncurare kills its victims. When he under-took these investigations, he was awarethat curare causes asphyxiation: itblocks motor nerves from inducing con-traction of respiratory muscles. Hewondered, though, whether it acted onthe nerves or on the muscles. To Þndout, he placed large doses of nicotine, asubstance that normally induces con-traction, directly onto strips of skeletalmuscle from chicken (at a site wheremotor nerves are normally connected).A contraction followed. Then he appliedcurare. The drug blocked nicotineÕs ac-tion. Langley concluded that curare in-teracts directly with muscle tissue,which displays on its surface an Òespe-cially excitable component,Ó or Òrecep-tive substance,Ó capable of combiningwith either nicotine or curare.

Neurobiologists now understand thatnicotine and curare couple with the partof the receptor molecule designed tobind to the neurotransmitter acetylcho-line. Bound nicotine serves as an ago-nist : it mimics the stimulatory eÝect of a naturally produced substance, ace-

tylcholine. Bound curare, in contrast, isa competitive antagonist: it issues nostimulatory signal and, at the sametime, prevents acetylcholine and nico-tine from doing so.

Despite the brilliance of the re-ceptor concept, its value eludedthe scientific community for de-

cades. Skepticism arose in part becausescientists lacked the tools for isolatingreceptors. Moreover, they had troubleimagining how binding of a chemical toa receptor molecule at the cell surfacecould inßuence the ßow of ions throughchannels in the cell membrane.

I helped to ease these objections inthe mid-1960s, when, as a graduate stu-dent working on my doctoral disser-tation, I suggested a theoretical solu-tion to this conceptual diÛculty. A fewyears earlier, structural studies of hemo-globin and various enzymes had indi-cated that these molecules includedseveral separate sites capable of associ-ating with other substances. IÑtogeth-er with my teachers Jacques Monod andFran�ois Jacob and their colleague JeÝ-ries WymanÑpostulated that certain en-zymes may be activated by allosteric,or indirect, means, by which binding atone site inßuences behavior of anothersite without any assistance from an ad-ditional source of energy. We assumed

58 SCIENTIFIC AMERICAN November 1993

Chemical Signaling in the Brain

Studies of acetylcholine receptors in the electric organs of fish have generated critical insights into how

neurons in the human brain communicate with one another

by Jean-Pierre Changeux

JEAN-PIERRE CHANGEUX has been di-rector of the molecular neurobiologyunit of the Pasteur Institute in Paris since1972. He is also a professor at the insti-tute and at the College of France. Chang-eux has won many awards for his contri-butions to neuroscience.

ACETYLCHOLINE RECEPTOR, which con-sists of five subunits (left), was the firstneurotransmitter receptor to be isolated.Later work showed it to include not onlyneurotransmitter binding sites but alsoan ion-transporting channel (right). (Thebeta and delta subunits and part of onealpha subunit have been cut away forclarity.) The channel is closed when thereceptor is at rest, but it opens rapidlywhen the two alpha subunits both com-bine with acetylcholine.

Copyright 1993 Scientific American, Inc.

that attachment of some substance toa docking site on an enzyme could prop-agate a conformational change through-out the enzyme, thereby rendering a dis-tant site able to act on a substrate (thesubstance transformed by an enzyme).

In my dissertation I noted brießy thatreceptors for neurotransmitters mightfunction similarly. They might containboth a neurotransmitter binding siteand a separate region that forms an ionchannel. Attachment of the neurotrans-mitter to the binding site could elicit aconformational change in the moleculethat would culminate in the opening ofits channel component. To evaluate themerit of this idea, my co-workers and Ihad to analyze the composition of somekind of receptor in detail. For this, weneeded a good supply. Unfortunately,no receptor had yet been isolated, andso that task became our mission.

Our choice of receptor was inspiredby discoveries made by David Nachman-sohn after he ßed Nazi Germany. In thelate 1930s, while at the University ofParisÐSorbonne, Nachmansohn and hiscolleagues showed that acetylcholinenot only induces muscle to contract, italso causes electricity-generating organsof electric Þsh to produce current. Fur-

thermore, the organs oÝer two particu-lar advantages for researchers. The con-stituent cells, called electrocytes, arehuge and thus relatively easy to handle.Additionally, they number in the billions,which means electric organs harbor anabundance of acetylcholine receptormolecules.

With these advantages in mind,we decided to isolate the ace-tylcholine receptor in the elec-

tric organ of the electric eel (Electropho-

rus electricus). First, we had to break upthe electrocytes in the organ to createpreparations that could be analyzedchemically. Michiki Kasai, now at OsakaUniversity, and I therefore ground upelectric tissue. Then we separated outmicron-sized fragments of membranesfrom the innervated regions of electro-cytes. LangleyÕs studies of muscle tis-sue suggested we would Þnd a highconcentration of the receptor in the in-nervated areas. These membrane frag-ments have a wonderfully useful prop-erty: they close up into microsacs, ortiny vesicles, that can be Þlled with ra-dioactively labeled sodium (Na+) andpotassium (K+) ions.

As would be expected if functional

copies of the receptor were present inthese microsacs, addition of acetylcho-line to a suspension of the vesicles dra-matically altered the ßow of ions intoand out of the vesiclesÑjust as occursin intact electrocytes when they respondto acetylcholine. Moreover, in agreementwith early suggestions of an allostericmechanism, no additional energy sup-ply was required for the reaction to takeplace. Hence, we felt reasonably surethat the receptor was present and func-tional in the microsac membranes.

Still, we needed some way to distin-guish the receptor from the rest of thematerial in the membranes. At that time,the only way to pinpoint a molecularspecies on a membrane was to radio-actively label a substance that homedto it and bound tightly. We were havingdiÛculty Þnding a suitable homing ma-terial when Chen-Yuan Lee of the Na-tional Taiwan University came to ourrescue.

Lee happened to visit my laborato-ry in the spring of 1970 to present hisresearch on the structure and action of snake venom. The bites of severalsnakes, such as the banded krait (Bun-

garus multicinctus) and the cobra, arefatal because their venom contains tox-

SUBUNITS

ALPHA ALPHA

ALPHA

CYTOPLASM

ALPHA

ACETYLCHOLINE

SODIUMION

BINDINGSITE

CELLMEMBRANE

POTASSIUMION

IONCHANNEL

BETA

GAMMA GAMMA

DELTA


ic molecules that, like curare, block sig-nal transmission by motor neurons.Among these molecules are alpha tox-ins. Lee reported that even at low con-centrations, alpha-bungarotoxin fromthe banded krait almost irreversiblyblocks the eÝects of acetylcholine onthe muscles of evolutionarily advancedvertebrates. I realized then that alpha-bungarotoxin might provide the speci-Þcity we needed for identifying the ace-tylcholine receptor on the microsacswe derived from the electric eel. To ourdelight, toxin supplied by Lee did ourbidding perfectly.

We could Þnally take up thetask of purifying the receptor,which we soon found was a

protein. By 1974 our group and severalothers had succeeded. My team reachedthis goal by applying a technique calledaÛnity chromatography. We created in-soluble beads to which arms that end-ed in a structural analogue of curarehad been attached. Then we dousedthe beads with microsac membranesthat had been dissolved in a detergentsolution to separate the constituentmolecules. The free receptor moleculesbound to the analogue, and the rest ofthe solution ßoated away. Next wepoured copies of the curare analogueover the beads. Now the receptor mole-cules bound preferentially to the addedanalogue and came oÝ the beads. Bypassing the resulting complexes of re-ceptor and analogue through a mem-brane permeable only to the curare sub-

stitute, we eliminated the analogue andacquired a pure supply of the receptor.

Eager to learn something about thestructure of our prize, we presentedJean Cartaud of the Jacques Monod In-stitute in Paris with samples to view inan electron microscope. He found thatwhen observed from above, the recep-tor resembled a rosette with a depres-sion in the center. Other analyses, by Ar-thur Karlin of the Columbia UniversityCollege of Physicians and Surgeons andby Michael Raftery of the California In-stitute of Technology, revealed that theoverall molecule is composed of Þveprotein chains, or subunits: two alphachains, which have an identical molecu-lar weight, and three chains called beta,gamma and delta, which vary in molec-ular weight. Moreover, Karlin demon-strated that the alpha subunits bear theprimary responsibility for recognizingacetylcholine. (Today it is clear that thebinding site on each of the two alphasubunits must be occupied in order forthe ion channel to open.)

These results intrigued us. They sug-gested that each subunit might formone ÒpetalÓ of the rosette and that thecentral depression might reßect the ex-tracellular entryway to a membrane-spanning ion channel. We needed moredata in order to test that idea, but inthe interim we had to cope with anoth-er nagging problem. Could we be surethat receptor molecules contained anion channel, not solely the site thatbound acetylcholine?

In 1974 my co-worker Gerald L. Ha-

zelbauer and I tackled this question byincorporating proteins from puriÞedextracts of microsacs into lipid mem-branes that enclosed radioactively la-beled sodium or potassium ions. Aswould be predicted if the channel werepresent, binding by acetylcholine trig-gered the ßow of ions, and binding byalpha-bungarotoxin and curare blockedchanges in ionic ßux. Later we con-Þrmed the results with puriÞed recep-tors themselves. Hence, by 1980 weand other groups had clearly shownthat the pure protein does indeed con-tain all the structural elements neededfor the chemical transmission of anelectrical signalÑnamely, an acetylcho-line binding site, an ion channel and amechanism for coupling their activity.

To attack the deeper problem of howthe acetylcholine receptor worked, wehad to decipher the sequence of its con-stituent amino acid. Such informationprovides clues to the shape adoptedwhen a protein, which is little more thana string of amino acids, folds in on it-self. And knowledge of the folded struc-ture oÝers clues to the functions of thevarious domains in that structure.

Introduction of automated sequenc-ing and genetic engineering techniquesin the late 1970s facilitated this eÝort.In 1979 my colleague Anne Devillers-Thi�ry, Donny Strosberg of the JacquesMonod Institute and I elucidated thesequence of the Þrst 20 amino acids atone end (called the amino terminal) ofthe alpha subunit in the acetylcholinereceptor of the European, or marbled,electric ray (Torpedo marmorata). Sub-sequently, Raftery and Leroy E. Hoodand their colleagues at Caltech identi-Þed essentially the same sequence inthe Californian electric ray (T. californi-

ca) and went further. When they char-acterized the 54 amino acids abuttingthe amino terminal of the alpha, beta,gamma and delta subunits, they unex-pectedly found striking similarity: 35 to50 percent of the sequence was identi-cal, or homologous, in all four subunits.

Molecular biologists interpret suchidentity to mean that the genes specify-ing the amino acid sequences of thesubunits are descendants of some an-cestral gene that duplicated twice (andunderwent subsequent alteration) inthe course of evolution. The homologyfurther implied that the complete sub-units were similar to one another andtherefore probably did arrange them-selves quasisymmetrically around acentral axis, forming the petals on therosette seen in the electron microscope.

By 1983 more complete informationhad emerged. Shosaku Numa and histeam at Kyoto University had solvedthe full sequences of the alpha and then


ELECTRIC RAY from the Torpedo genus harbors an electricity-generating organthat contains billions of copies of an acetylcholine receptor. Identification of the re-ceptorÕs amino acid sequence led to the discovery that acetylcholine receptors inthe muscle and brain of humans are structurally similar to the Torpedo receptor.


SCIENTIFIC AMERICAN November 1993 60A

Major structural features of the ace-tylcholine receptor have begun to

yield to scrutiny. Its five subunits (insetin top panel ), which can be depictedschematically as cylinders (top, center ),are each formed from a protein that hasfolded in on itself (detail at top right ).Every subunit includes a large hydro-philic (water-loving) region abutting theamino (NH2) terminal, as well as fourhydrophobic (water-hating) membrane-spanning segments: M1, M2, M3 andM4.The neurotransmitter binding sites,viewed from above in the middle panel(inset), consist primarily of amino acids(yellow spheres in detail ) residing in thelarge hydrophilic region of the alphasubunits. (Letters in the spheres repre-sent specific amino acids.) Neighboringsubunits contribute as well (pink sphere).The ion channel is composed of five M2segments (inset in bottom panel ) andcontains several rings of amino acids thataffect the functioning of the receptor.Among these, three negatively chargedrings (blue in detail showing two M2segments in a specific acetylcholine re-ceptor ) draw positively charged ions(not shown) through the channel. An un-charged, leucine ring (green ) at the cen-ter, where M2 segments probably bend,participates in closing the ion channelwhen the receptor becomes desensi-tized to acetylcholine. The electron den-sity map (bottom left ), showing a slicethrough the receptor, indicates the prob-able orientation of two M2 segments(dark bars ). Nigel Unwin of the MedicalResearch Council in Cambridge, En-gland, supplied that image.

Anatomy of the Acetylcholine Receptor

M2

RE

CE

PTO

R

ATTACHEDPROTEIN

MEMBRANE

OVERALL STRUCTURE

NEUROTRANSMITTER BINDING SITE

ION CHANNEL

BINDING SITE

ACETYLCHOLINE

AMINO ACIDS

LARGEHYDROPHILIC

DOMAIN NH2

NH2

M1

M1

W

W

Y

Y

A

VV

V V

VV

M M

FF

E

S S

SS

G GI

I

I

I

K K

TT

T T

AEE

D D

––

–

––

–

–

–

––

––

––

–

E

L

L

LL

L

L L

L

L

L

L

C C YY

δ

δ

β

β

α

α α

α

δ

γ

αγ

γ

M2

M2

M2M2M2

M2

M2

M4

M4

COOH

AMINO ACIDS

M3

M3

α δ

β

α

γ


the beta, gamma and delta subunits ofthe T. californica receptor. Three otherlaboratories, including mine, had alsoidentiÞed the sequence of the T. cali-

fornica gamma subunit and suppliedthat of the T. marmorata alpha subunit.Then Numa published the sequences ofall the subunits in the acetylcholine re-ceptor on human muscle. The musclereceptor turned out to differ little fromthose of electrocytes.

These studies put workers in anexcellent position to ascertainsomething about the architecture

of the folded subunits and how theymight Þt together. In trying to predictthe structure of a folded protein, scien-tists often scan the linear amino acidsequence for stretches that are rich ineither hydrophilic or hydrophobic ami-no acids. Hydrophilic substances areattracted to water, such as that in cyto-plasm or in the ßuids that bathe cells;hydrophobic, or water-hating, substanc-es prefer to associate with other hy-drophobic entities, such as the lipidsthat form cell membranes.

Every subunit that had been se-quenced began at the amino-terminalside with a large hydrophilic regionand housed four separate hydrophobicsegments of about 20 amino acids. Thehydrophobic areas are referred to, be-ginning with the one closest to thelengthy hydrophilic domain, as M1, M2,M3 and M4. This arrangement suggest-ed that every subunit chain snakedthrough the thickness of the cell mem-brane four times, so that all four of thehydrophobic regions spanned the mem-brane [see box on preceding page].

In one model, at least, the large hy-drophilic domain protruded into theextracellular space. So situated, thatdomain on the alpha subunits (the sub-units primarily responsible for grasp-ing acetylcholine) would be well posi-tioned to serve as a neurotransmitterbinding site. It further seemed reason-able to guess that one membrane-span-ning segment from each of the Þve sub-units associated with its counterpartsin the other subunits to form the ionchannel. The like segments could formsuch a channel if together they encir-cled the central axis of the rosette.

Later research conÞrmed this scenar-io and added important details. Moreimmediately, though, the sequencing ofthe subunit proteins enabled research-ers to dispel confusion over the opera-tion of acetylcholine receptors in thebrain. The trouble derived from the re-sponse of certain of these receptors tosnake venoms.

By the early 1980s investigators knewthat nicotine-sensitive, or nicotinic, ace-

tylcholine receptors were present in thebrain of higher vertebrates. Neurobiol-ogists were puzzled, however, when al-pha-bungarotoxin seemed to block thefunctioning of certain of the receptorsbut not others. What is more, a Bungar-

us toxin called neuronal bungarotoxinapparently attached to some cerebralreceptors but, again, not to others. (Thepicture is actually more complicated.The brain additionally includes a classof acetylcholine receptors named mus-carinic receptors that I shall not discusshere. Those receptors are very diÝerentfrom the nicotinic types. They areformed from a single protein chain anddo not include an ion channel. They ex-ert their eÝects through such intracel-lular mediators as G proteins.)

The smoke cleared when James W. Patrick and Stephen F. Heine-mann and their colleagues at the

Salk Institute for Biological Studies inSan Diego and Marc Ballivet of the Uni-versity of Geneva used their knowledgeof the structure of electrocyte and mus-cle subunits to decipher the amino acidsequences of cerebral subunits. The investigators guessed that the aminoacid sequences of the cerebral subunitsprobably resembled those of electro-cytes and muscle even though certaincerebral receptors behaved somewhatdiÝerently from their counterparts inelectrocytes and muscle. If the proteinswere similar, then the genes specifyingtheir amino acid sequences would besimilar as well. That being the case, aprocess called DNA hybridization couldbe expected to help isolate the cerebralsubunit genes and thereby uncover theamino acid sequences of the corre-sponding subunits.

DNA hybridization techniques cap-italize on a prominent characteristic of genes. Genes consist of two strandsof nucleotides (the building blocks ofDNA). One strand from a gene will read-ily combine, or hybridize, with the oth-er strand from the same or a closely re-lated gene. Aware of this propensity,the investigators hoped they could re-trieve the cerebral subunit genes froma larger pool of brain-derived DNA byÒÞshingÓ for them with ÒhooksÓ madeof nucleotide sequences that actuallydirect the synthesis of electrocyte ormuscle subunits. The procedure workedbeautifully. Seven alpha-subunit types(each of which is numbered) were foundto be produced in the brains of verte-brates, including humans. Three nonal-pha types, often classiÞed as beta sub-units, were discovered as well. This di-versity suggested that the variable re-sponses of cerebral receptors to snakevenoms derive from slight diÝerences

in the amino acid sequences of one ormore subunits.

Subsequent studies have demonstrat-ed that the protein products of most ofthe subunit genes identiÞed to date canyield functional receptors in living cellsif the cells make at least one alpha andone nonalpha variant. The experimentsyielding this conclusion often involvedinjecting the genes into the nucleus ofoocytes, or immature eggs, from thefrog Xenopus. In response, the protein-making machinery of the oocytes tran-scribed the genes into messenger RNAand, after transporting the RNA to thecytoplasm, translated it into the speci-Þed proteins. Then the proteins associ-ated with one another in groups of Þveto produce receptors.

Evaluations of many receptors pro-duced by Xenopus oocytes also demon-strated that substitution of one sub-unit variant for another in a receptorcan indeed change some properties ofthe receptor. As an example, neuronalbungarotoxin blocks the response toacetylcholine receptors composed ofbeta-2 and either alpha-3 or alpha-4subunits, but the toxin does not inter-fere with the activity of molecules com-posed of beta-2 and alpha-2 subunits.

At about the same time as the het-erogeneity of nicotinic acetylcho-line receptors was emerging, re-

searchers were busy attempting to de-cipher the structure and operation ofreceptors for other neurotransmitters.When that work began, few would haveguessed that the receptors for gamma-aminobutyric acid (GABA) and glycinewould have much in common with ace-tylcholine receptors. After all, nicotinicacetylcholine receptors excite cells byopening a channel permeable to cations(positively charged ions). Receptors forGABA and glycine, in contrast, facili-tate transport of chloride anions (ClÐ);the ßow of chloride anions into cellsinhibits generation of electrical impuls-es and can thereby counteract the ef-fects of excitatory receptors.

Nevertheless, studies conducted inthe 1980s revealed that glycine andGABA receptors consist of multiple sub-units. That in itself was not remark-able. More strikingly, however, HeinrichBetz of the University of Heidelbergand Eric A. Barnard of the Medical Re-search Council in Cambridge, England,respectively determined the completesequences of the glycine and GABA re-ceptors and found that the distributionof hydrophilic and hydrophobic do-mains strongly resembled that of nico-tinic acetylcholine receptors.

In other words, it began to seem like-ly that the subunits in the GABA and

60B SCIENTIFIC AMERICAN November 1993 Copyright 1993 Scientific American, Inc.

glycine receptors, in common withthose of nicotinic acetylcholine recep-tors, weave through the cell membranefour times. Evidence also indicated thatthe complete receptors for GABA andglycine carry both a neurotransmitterbinding site and an ion channel. Morerecent work suggests that some sero-tonin receptors have a similar architec-ture as well. These receptors, like ace-tylcholine receptors, control the cross-membrane transport of cations and arethus excitatory.

The architectural similarities amongthe receptors explain why neurobiolo-gists now consider acetylcholine, GABA,glycine and serotonin receptors to con-stitute the superfamily of geneticallyand structurally related neurotransmit-ter-gated ion channels. (Receptors forthe prevalent neurotransmitter gluta-mate may be distantly related. They in-clude a neurotransmitter binding siteand an ion channel but differ in struc-ture.) There is also evidence that neu-rotransmitter-gated ion channels areallosteric proteins. As would be expect-ed for allosteric molecules, the estimat-ed distance between the neurotransmit-ter binding site and the ion channel islargeÑabout 30 angstroms.

As was found for acetylcholine recep-tors, the subunits of other members ofthe superfamily come in multiple vari-eties. Hence, a GABA receptor in onepart of the brain might well have some-what diÝerent properties than does avariant elsewhere in that organ. For in-stance, benzodiazepines, which are soabundantly consumed as tranquilizers

by industrialized populations, potenti-ate the inhibitory action of only certainGABA receptor species. They do so bybinding to a site that is distinct fromthe GABA binding site.

As the precise inßuence on behaviorof every subspecies of every subunit inneurotransmitter-gated ion channels isdeciphered, pharmacologists should beable to design drugs that will selectivelyimpede or enhance those eÝects. Suchagents, in turn, might help ameliorateany number of debilitating conditions,including mood disorders, tissue dam-age associated with stroke and, perhaps,AlzheimerÕs disease.

Of course, to devise such drugs,researchers require a rather fullunderstanding of receptor struc-

ture. They need to know the speciÞcamino acids responsible for bindingneurotransmitters, for directing the ßowof ions in and out of cells and for other-wise modulating receptor function. Oneuseful way to gather such informationis known as aÛnity labeling. Sometraceable version of a molecule that in-teracts with a receptor is allowed to bindirreversibly to that target; the boundsubstance thus highlights the aminoacids that constitute the binding site.

Between 1988 and 1990, my col-leagues Michael Dennis, J�r�me Girau-dat, Jean-Luc Galzi and I uncoveredmuch of the acetylcholine binding siteby identifying amino acids in a Torpedo

receptor that were labeled by the com-pound pÐ(N,NÐdimethyl) aminoben-zenediazonium ßuoroborate, also called

DDF. We learned that several aromaticamino acids (those carrying ring-shapedside chains) are critical to DDF binding,and we conÞrmed binding by a pair ofcysteine amino acids identiÞed in a Tor-

pedo receptor by Karlin. The labeled ami-no acids are distributed within threedistinct regions of the large hydrophil-ic domain of the amino-terminal region.It became evident that they collectivelyform a kind of negatively charged cupin which the positively charged part ofacetylcholine could lodge.

What is even more exciting, we wenton to show that these amino acids ac-tually do play a critical role in receptorfunction. With Daniel Bertrand of theUniversity of Geneva Medical Center, weproved this point in a receptor that con-sists entirely of alpha-7 subunits fromthe chicken brain. (This receptor is oneexception to the rule requiring the pres-ence of both alpha and beta subunitsfor receptor formation.) SpeciÞc muta-tion, by what is called site-directed mu-tagenesis, of the amino acids that DDFlabeled in the Torpedo receptor strik-ingly impeded the alpha-7 receptorÕsresponse to acetylcholine.

Taken together, the aÛnity-labelingand mutagenesis studies conÞrm thatthe large hydrophilic region of the al-pha subunit is exposed to the extracel-lular environment. There it sits, readyto receive acetylcholine released fromnerve endings and to trigger the open-ing of the ion-transporting channel.

AÛnity labeling also delineated thestructure of the ion channel in a Torpe-

do receptor. DiÛcult analyses convincedus by the end of 1985 that the drugchlorpromazine attaches to amino acidson the membrane-crossing, M2 hydro-phobic segment of at least one sub-unitÑthe delta chain. This work, and asimilar report by Ferdinand Hucho andhis co-workers at Berlin University, sug-gested that the channelÕs inner wall isformed by Þve M2 segments, one con-tributed by every subunit.

Numa and Bert Sakmann, then at theMax Planck Institute for BiophysicalChemistry in G�ttingen, conÞrmed thispossibility. By site-directed mutagene-sis, they determined that at least threerings of negatively charged amino acids(especially glutamate) participate intransporting ions through the channel.Each ring lies in a plane parallel to thesurface of the cell and consists of Þveamino acids, one supplied by the M2segment of every subunit. A single ringresides at the extracellular surface ofthe membrane (at the top of the chan-nel). A second, termed the intermediatering, lies at the bottom of the channel,and the third ring lies directly belowthe second, in the cytoplasm proper.

60F SCIENTIFIC AMERICAN November 1993

UNFOLDED PROTEIN CHAINS (multicolored bars) that constitute subunits in re-ceptors for acetylcholine, GABA and glycine have much in common. All harbor alarge, extracellular hydrophilic domain, a smaller, cytoplasmic hydrophilic domainand four hydrophobic segments (M1, M2, M3 and M4) believed to span the cellmembrane. These similarities suggest that the molecules all belong to one super-family of structurally related neurotransmitter receptors.

SUBUNITVARIANT

ALPHA-1, FROM ACETYLCHOLINE RECEPTOR IN MUSCLE

EXTRACELLULARHYDROPHILIC

DOMAIN

CYTOPLASMICHYDROPHILIC

DOMAIN

ALPHA-4, FROM ACETYLCHOLINE RECEPTOR IN BRAIN

M1 M2 M3 M4

ALPHA-1, FROM GABA RECEPTOR

BETA-1, FROM GABA RECEPTOR

ALPHA-1, FROM GLYCINE RECEPTOR

BETA-1, FROM GLYCINE RECEPTOR


Given that the distribution of hy-drophobic subunits in the GABA, gly-cine and serotonin receptors matchesthat of acetylcholine receptors, we won-dered if their M2 segments formed thechannel in those receptors as well. Theydo, even though those receptors trans-port negatively, rather than positively,charged ions. The diÝerence in chargepreference apparently stems from vari-ance in just a few amino acids. WhenBertrandÕs team and mine transferredinto the alpha-7 receptor three M2 ami-no acids from a GABA receptor (includ-ing the one giving rise to the intermedi-ate ring), those few changes convertedthe alpha-7 receptor channel from acationic to an anionic transporter.

Whether a receptor carries ananion- or cation-transportingchannel, its main function is

to open that channel in response to sig-nals from a neurotransmitter. Yet neu-rotransmitter receptors have anotherfascinating skill as well. By alteringtheir conformation, they can apparent-ly increase or decrease their readinessto respond to neurotransmitters. In thatway, they can regulate the pool of re-ceptors available to respond to exter-nal signals and can thus inßuence theeÛciency of signal transmission.

My associates and I realized that re-ceptors could possess this regulatorypower when we began to consider aphenomenon noted by several investi-gators over the years. Receptors reactdiÝerently to discrete pulses of high-ly concentrated acetylcholine (such asthose usually delivered by neurons)than they do to the continuous avail-ability of lower concentrations (such as is provided in many experiments).

Excited neurons secrete large concen-trations of acetylcholine in discretebursts at synapses, the specialized junc-tions now known to connect neurons.The molecules freed during a singlepulse pour into the synaptic cleft (thespace separating communicating cells).Many of them make their way from theexcited, presynaptic cell to receptors onthe surface of a postsynaptic cell. Undernormal circumstances, the aÛnity ofmost receptor molecules for the neuro-transmitter is low. Consequently, imme-diately after acetylcholine binds to re-ceptors and causes the channel to open,the receptors release their hold on theneurotransmitter, which is promptlydegraded. Within milliseconds of beingbound, the receptors revert to theirclosed, unbound state and are ready toreact once again.

In contrast, when acetylcholine is sup-plied continuously to receptors, the receptor molecules begin to lose their

responsiveness. After initially openingthe ion channel, they slowly take on aÒdesensitizedÓ conformation over thecourse of seconds or minutes. That is,they bind avidly to acetylcholine butmaintain a closed channel and do nottransport ions. Even small concentra-tions of acetylcholine will be held bythese closed-channel receptors for rela-tively long periods, during which thereceptors cannot react to new signals.

So it seems that acetylcholine recep-tors can adopt at least three intercon-vertible states that can diÝer in theiraÛnity for the neurotransmitter and in the eÛciency of signal transmission.In addition to the high-aÛnity, desen-sitized state, in which the channel re-mains closed, there is a low-aÛnity,resting (but activable) state in whichthe channel is closed but easily openedif both alpha subunits are bound sud-denly by acetylcholine. The low-aÛni-ty, open-channel condition is the thirdstate. All three states switch back andforth spontaneously but at diÝerentrates than occur when acetylcholine ispresent.

Site-directed mutagenesis has helpedclarify the process by which desensi-tization occurs; it appears that leucineamino acids are involved. In aÛnity-la-beling studies carried out by my group,chlorpromazine labeled a ring of un-charged leucine amino acids near thecenter of the ion channel. When Ber-trandÕs team and mine replaced theleucines in this ring with a smaller un-charged amino acid, we created a recep-tor that resembled a normal desensi-tized receptor in that it bound tightlyto acetylcholine. Yet its channel wasfixed in an open state. This result im-plies that the leucine ring locks the ionchannel closed when the receptor is inthe desensitized conformation.

My associates and I have longpondered the beneÞts thatmight accrue to an organism

from bearing receptors able to adoptmultiple states. Of course, a desensiti-zation mechanism would protect recep-tor-bearing cells from becoming over-excited in response to dangerously highlevels of acetylcholine. But I believe that

SCIENTIFIC AMERICAN November 1993 61

DYNAMIC NATURE of the acetylcholine receptor is evident in its ability to adoptmultiple conformations. In the resting state (a), the receptor has low affinity foracetylcholine, and its ion channel is closed. If it is exposed briefly to a high con-centration of the neurotransmitter, it assumes the active, open-channel conforma-tion for milliseconds (b) before releasing the acetylcholine and reverting to theresting state. If acetylcholine is supplied continuously, resting and activated re-ceptors can slowly assume a desensitized state (c ). In this condition, the receptorholds acetylcholine with high affinity for seconds or minutes, maintains a closedchannel and will not respond to new pulses of acetylcholine.

ACETYLCHOLINE

CLOSED, RESTINGSTATE

a b

cOPEN, ACTIVE

STATE

FASTTRANSITION

SLOWTRANSITION

CLOSED, DESENSITIZEDSTATE


there is another explanation as well.In 1982 my colleague Thierry Heid-

mann and I further proposed that theability of acetylcholine receptors to alter their conformation slowly couldmore routinely serve to increase or de-crease the eÛciency of signal transmis-sion at a synapse. In so doing, such re-ceptors could participate in learning.Many theorists, following the lead ofDonald O. Hebb, postulate that learn-ing depends on changes in the eÛcien-cy of signal transmission across thesynapses linking two neurons that areactivated simultaneously.

Our hypothesis is far from proved,but it is plausible. If the ability to adoptmany states were important to regulat-ing signal transmission, this ßexibilityshould appear in other neurotransmit-

ter receptors as well. Research showsthat GABA, glycine and serotonin re-ceptors also are able to assume desen-sitized conformations.

The demonstration of state changesin other receptors is not the only sup-port for the possibility that receptorsregulate synaptic eÛciency. These mol-ecules sit in a particularly good posi-tion to control the degree of respon-siveness needed at any given moment.Crossing the cell membrane as they do,they are exposed to chemical and elec-trical signals issued both from outsideand from within the cell. If each ofthese signals pushed the receptor to-ward one conformation or another, theÞnal arrangement would reßect thesummed, or integrated, inßuence ofvarious, possibly contradictory, forces.

Among the signals that impinge onreceptors are the intracellular concen-tration of calcium ions and changes inthe electrical potential across the cellmembrane. Hyperpolarization of themembrane and elevation of the calci-um concentration in muscle acceleratedesensitization of the acetylcholine re-ceptor. Richard L. Huganir and PaulGreengard of the Rockefeller Universityfurther established in 1986 that phos-phorylation of the receptor promotesdesensitization.

If neurotransmitter receptors did infact control the eÛcacy of intercellularsignaling, we would anticipate that theywould be capable of increasing, notmerely decreasing, their sensitivity toneurotransmitters. Such potentiationhas been observed. Extracellular calci-um enhances the stimulatory effect ofnicotinic acetylcholine receptors in thebrain, and glycine enhances the effectof glutamate receptors.

In general, then, we suspect that if areceptor is tottering between two con-formations, one of which is sensitiveand the other of which is refractory tostimulation, the balance can be shiftedtoward one of the two states by chemi-cal or electrical signals. The resultingconformation, in turn, enhances or de-presses the ability of the receptor toconvey signals promptly.

Clearly, our understanding ofchemical signaling in the brainhas advanced dramatically in the

past quarter century. The decision toisolate the acetylcholine receptor froma Þsh organ was risky: if the eÝortfailed or if that receptor was unrelatedto any others, we would have wastedtime and energy. Fortunately, the gam-ble paid oÝ more than we could havehoped. The acetylcholine receptor inelectric Þsh was identiÞed, and its se-quence was deciphered. Then biotech-nology, especially recombinant DNAmethodology, enabled workers to char-acterize related receptors in humanmuscle and in the brain and to learnthat nicotinic acetylcholine receptorsare structurally related to, and form asuperfamily with, those responsive toGABA, glycine and serotonin.

Meanwhile it became clear that thesubunits composing each type of recep-tor are themselves variable: a receptorin one part of the brain may well pos-sess properties that diÝer from thoseof their immediate kin elsewhere in ce-rebral tissue. This likelihood raises thetantalizing possibility that drugs tar-geted to speciÞc receptors on deÞnedcategories of neurons can be developedfor the highly selective treatment ofsignaling disorders in the brain.


MANY FORCES can affect the conformation of a neurotransmitter receptor andhence the efficiency with which it responds to signals received from a neuron act-ing on it. Aside from the concentration and delivery rate of the neurotransmitter(a ), some influences include binding by additional neurotransmitters or other ex-tracellular chemicals (b ), changes in the electrical potential across the cell mem-brane (c ) and binding by intracellular signaling molecules (d ), such as ions.

NEURONAL MAN: THE BIOLOGY OF MIND.J. P. Changeux. Translated by LaurenceGarey. Pantheon Books, 1985.

FUNCTIONAL ARCHITECTURE AND DYNAM-ICS OF THE NICOTINIC ACETYLCHOLINERECEPTOR: AN ALLOSTERIC LIGAND-GAT-ED ION CHANNEL. J. P. Changeux in FIDIA Research Foundation NeuroscienceAward Lectures, Vol. 4. Raven Press,1990.

EXPLORATIONS OF THE NICOTINIC ACE-TYLCHOLINE RECEPTOR. A. Karlin in Harvey Lectures: 1989Ð1990, Vol. 85,

pages 71Ð107; 1991.THE FUNCTIONAL ARCHITECTURE OF THEACETYLCHOLINE NICOTINIC RECEPTOR EX-PLORED BY AFFINITY LABELLING AND SITE-DIRECTED MUTAGENESIS. J. P. Changeux,J. L. Galzi, A. Devillers-Thi�ry and D. Ber-trand in Quarterly Reviews of Biophysics,Vol. 25, No. 4, pages 395Ð432; November1992.

NICOTINIC ACETYLCHOLINE RECEPTOR AT9 A RESOLUTION. N. Unwin in Journal ofMolecular Biology, Vol. 229, No. 4, pages1101Ð1124; February 20, 1993.

FURTHER READING

˚

NEURONTERMINALS

ab

d

OTHERNEURO-TRANSMITTERNEUROTRANSMITTER

NEUROTRANSMITTERRECEPTOR

OTHERRECEPTOR

CHANGE IN MEMBRANEPOTENTIAL

c

CELLMEMBRANE

INTRACELLULARSIGNALING MOLECULE


(4) chemical signaling brain

Documents