connecting portion between the extracellularbinding pocket and the receptor’s transmem-brane domains. This leads to conformationalchanges within the membrane, forcing thegate to open. But in consequence there is an equivalent strain on the glutamate-binding pocket, facilitating an ensuing conformational rearrangement to escape tension. This causes rearrangement at thedimer interface and movement of the receptorinto the desensitized state.

In receptors that do not desensitize, the strengthened dimer interface preventsrearrangement within the ligand-bindingpocket. The linker between binding pocketand gate remains under tension and leaves thepore open as long as glutamate is bound. It istherefore no surprise that the glutamate-binding domain of the wild-type receptor andthat of the non-desensitizing mutant both crystallize into the same ‘open-channel’ state,because the link, and therefore the tensionforce, from the membrane is missing.

Finally, Sun et al. surprise us with crystalstructures of a mutant receptor that wasdesigned to disrupt the dimer interface. Theaccelerated desensitization properties of theintact mutant receptor nicely confirm theimportance of dimer affinity in desensitiza-tion. As expected, the structure does notdimerize at the usual interface. But it doesreveal another, previously unknown, sub-unit interaction — hinting at a possible wayin which pairs of dimers can assemble into atetramer. In this interaction, domain 1 con-nects with a different region (lateral to thatinvolved in dimerization) of domain 2 of aneighbouring receptor. If this interactionindeed takes place in the assembled receptorcomplex, the tetrameric structure representsa pair of dimers that are shifted parallel toone another.

This would be an unprecedented form ofchannel subunit assembly — one in which the dimers facing each other follow a differentsymmetry from the assumed (though notproven) fourfold symmetry of the receptor in the transmembrane channel pore. Thisarrangement may, however, provide a posi-tional clue to how different isoforms ofAMPA-receptor subunits assemble into acomplex in vivo8, as chemically equivalentsubunits may prefer to assume positions ofequivalent symmetry, namely diagonallyacross from each other. It is yet not clear howthese arrangements would affect receptor gat-ing. But it is clear that glutamate receptors arelikely to remain a rich source of surprises. ■

Christian Rosenmund and Michael Mansour are in the Department of Membrane Biophysics, Max-Planck-Institute for Biophysical Chemistry,D-37077 Göttingen, Germany.e-mail: crosenm@gwdg.de1. Sun, Y. et al. Nature 417, 245–253 (2002).

2. Dingledine, R., Borges, K., Bowie, D. & Traynelis, S. F.

Pharmacol. Rev. 51, 7–61 (1999).

3. Jones, M. V. & Westbrook, G. L. Trends Neurosci. 19, 96–101

(1996).

4. Armstrong, N., Sun, Y., Chen, G. Q. & Gouaux, E. Nature

395, 913–917 (1998).

5. Armstrong, N. & Gouaux, E. Neuron 28, 165–181

(2000).

6. Rosenmund, C., Stern-Bach, Y. & Stevens, C. F. Science

280, 1596–1599 (1998).

7. Stern-Bach, Y., Russo, S., Neuman, M. & Rosenmund, C.

Neuron 21, 907–918 (1998).

8. Mansour, M., Nagarajan, N., Nehring, R. B., Clements, J. D. &

Rosenmund, C. Neuron 32, 841–853 (2001).

news and views

NATURE | VOL 417 | 16 MAY 2002 | www.nature.com 239

Living organisms tend to be highly economical, using the same proteins orcell-to-cell signalling processes in differ-

ent contexts and for different purposes. Take,for example, the family of proteins known asNFATs. These are transcription factors — theyregulate gene expression — and they have traditionally been implicated in controllinggenes involved in immunity. The events lead-ing up to such regulation include an increase inthe levels of calcium ions in a cell; the NFATtranscription factors then move into thenucleus and form a complex with relevantregions of DNA (and, in some cases, with othertranscription factors1). Several different sig-nalling pathways are known to activate NFATby changing calcium levels, and to this list can now be added one of the pathways that istriggered by Wnt proteins, as Saneyoshi and colleagues2 describe on page 295 of this issue. Rather than being involved in immunity, however, these events are key to early vertebrate development.

Members of the Wnt protein family aresecreted from cells and picked up by receptor

proteins on the surface of cells, triggeringintracellular signalling pathways. These path-ways regulate cell proliferation, death, fateand behaviour in contexts ranging from earlyembryonic development to colorectal cancer.The best-understood Wnt pathway is the‘canonical’ one: here, Wnt-induced signallingsuppresses degradation of the b-catenin pro-tein, enabling it to accumulate in the nucleus3.Nuclear b-catenin then binds to particulartranscription factors (not the NFATs) andthereby modulates gene expression.

Over the past few years it has becomeapparent that certain Wnt proteins and theirreceptors (proteins of the Frizzled family)can also activate ‘non-canonical’ pathways.Although the details of the canonicalWnt/Frizzled pathway are firmly supportedby genetic, biochemical and cell-biologicalexperiments in many species, the non-canonical pathways are less well understood.Nonetheless, it is known that a non-canoni-cal ‘planar cell polarity’ pathway is used infruitflies to orientate cells4.

There also seems to be more than one

Developmental biology

Signalling polarityRandall T. Moon and Kavita Shah

Cell signalling pathways triggered by Wnt proteins control gene expressionand cell behaviour, especially during development. Work on frogs revealsanother specific role for these proteins, and the signalling involved.

Figure 1 Opposing signalling pathways, triggered by different members of the Wnt protein family, mayhelp determine the polarity of Xenopus embryos. a, Before fertilization, there is no evident distinctionbetween the dorsal and ventral sides of the egg. b, Fertilization by a sperm anywhere in the upper halfof the egg initiates a wave of Ca2+ ions, which originate in what will become the animal’s ventral side.c, During the first hour after fertilization, the cortex (the plasma membrane and associated structuresand molecules) rotates by about 30o relative to the inner cytoplasm. The Dishevelled (Dsh) protein, adownstream target of ‘canonical’ Wnt pathways, moves along filamentous tracks (microtubules) to the future dorsal side, where it will ultimately result in stabilization of the b-catenin protein in cellnuclei, leading to transcription of relevant genes. d, Saneyoshi et al.2 predict that sometime betweenfertilization and the establishment of dorsal–ventral polarity, NFAT is activated in response to Wnt-triggered rises in Ca2+ levels in cells on the future ventral side. This promotes ventral cell fates, andantagonizes the signal that promotes dorsal cell fates.

No dorsal– ventralpolarity Dorsal

fateVentralfate

Ca2+

NFAT

WntVentral

sideDorsalside

Unfertilizedegg

FertilizationCalcium wave

Corticalrotation

a b c d

Microtubules

Dsh

Dorsal—ventralpolarity established

Egg Embryo

© 2002 Macmillan Magazines Ltd

Wnt/Frizzled pathway at work in vertebrateembryos. Early studies of the South Africanclawed frog, Xenopus laevis, established thatthe overexpression of some Wnt proteinsstabilizes b-catenin and results in twinnedembryos, complete with two heads. Yet over-expression of other Wnt proteins perturbscell movements during gastrulation3 — acrucial stage of development that leads toformation of the three main tissue layers:endoderm, mesoderm and ectoderm. Thesedifferent effects hinted that there is morethan one Wnt pathway in vertebrates.

The idea was substantiated by furtherwork in zebrafish and Xenopus embryos,which suggested that the Wnts that perturbcell movements activate a non-canonicalWnt/Frizzled pathway3 involving increasesin intracellular Ca2+ levels and the conse-quent activation of the enzymes proteinkinase C (PKC) and Ca2+/calmodulin-dependent protein kinase II (CaMKII)5. Butquestions remain, including the relationshipbetween the non-canonical pathways infruitflies compared with vertebrates,whether the Wnt/Ca2+ pathway works in anycontext other than early embryos, andwhether it modulates gene expression oronly gastrulation movements.

Saneyoshi et al.2 enter this arena of uncer-tainty by proposing that if signalling from cer-tain Wnt proteins leads to rises in intracellularCa2+ levels, it may also activate calcineurin, a Ca2+/calmodulin-dependent protein phos-phatase, leading to the removal of phosphategroups from NFAT and hence its accumula-tion in the nucleus. And indeed, they find that Wnt5A and Frizzled-2 — the sameWnt–receptor combination reported to leadto intracellular Ca2+ increases in zebrafish andto activate CaMKII and PKC in Xenopus —strongly induces the movement of NFAT intothe nucleus in cells of Xenopus embryos.

The authors also observe that overexpres-sion of a constitutively active NFAT perturbsthe cell movements that occur during gastru-lation in Xenopus, and encourages cells to takeon characteristics typical of the belly (ventral)rather than the dorsal part of the embryo.Finally, this hyperactive NFAT antagonizes the duplication of the dorsal axis that can be induced by the canonical Wnt pathway. Similar effects have been seen3 with overex-pression of Wnt5A. So Saneyoshi et al.’s find-ings support the existence of the Wnt/Ca2+

pathway in Xenopus, and identify NFAT as alikely downstream target of this pathway.

What, then, is the most plausible role of normal Wnt/Ca2+ signalling and NFAT in early frog embryos? Saneyoshi et al.’s data lend much weight to the previous sug-gestion5 that this non-canonical pathwayhelps to determine a ventral fate for cells and opposes the canonical pathway, whichpromotes dorsal cell fates in early embryos(Fig. 1, previous page). In support of this,Saneyoshi et al. show that expressing a

presumed inhibitor of NFAT in frog embryos results in a new dorsal axis, aswould be predicted if the NFAT had beenpromoting ventral cell fates.

So the new data are a tantalizing hint thatWnt/Ca2+ signalling promotes ventral cellfates by activating NFAT. But there are questions to be answered before we can besure. First, does NFAT actually activate gene transcription in response to a Wnt/Ca2+

signal? Saneyoshi et al. show that, in the presence of one of its cooperative partners —the transcription factor AP-1 — a constitu-tively active Xenopus NFAT activates tran-scription of a ‘reporter’ gene inserted intoXenopus embryos. But it is not yet certainwhether normal NFAT would do so inresponse to Wnt/Ca2+ signalling, or whetherWnt5A leads to the transcription of NFAT-responsive genes, as would be predicted.There is evidence that Wnt5A regulates AP-1-dependent transcription as well asCa2+-mediated signalling6. So the Wnt/Ca2+

gene targets may encompass those regulatedby AP-1 alone, or those regulated by complexes of AP-1 and NFAT.

Second, Saneyoshi et al.’s data predictthat more NFAT proteins will be found in cellnuclei on the future ventral side of Xenopusembryos than on the future dorsal side. Ifthat is the case, does this nuclear accumula-tion depend on Wnt signals? The activity ofCaMKII is increased on the future ventralside in a Wnt-dependent manner5, but itremains to be seen whether the nuclear accu-mulation of NFAT is likewise increased.Third, we need to test the proposed role ofNFAT by knocking out its function, usinggenetic or molecular means.

Finally, it will be interesting to seewhether the Wnt5A-driven movement ofNFAT to the nucleus is unique to early verte-brate development, where all characteriza-tion of the Wnt/Ca2+ pathway has been conducted to date. Unpublished work fromChris Hughes (Univ. California, Irvine) suggests the contrary: that this Wnt protein also induces NFAT to accumulate in thenuclei of the immune system’s T cells. What-ever the answers, it is clear that it takes manyWnt pathways to make a frog — and, byextension, any vertebrate. ■

Randall T. Moon and Kavita Shah are at theHoward Hughes Medical Institute and Department of Pharmacology, University ofWashington School of Medicine, Seattle,Washington 98195, USA.e-mails: rtmoon@u.washington.edukvshah@u.washington.edu1. Macian, F., Lopez-Rodriguez, C. & Rao, A. Oncogene 20,

2476–2489 (2001).

2. Saneyoshi, T., Kume, S., Amasaki, Y. & Mikoshiba, K. Nature

417, 295–299 (2002).

3. Kuehl, M., Sheldahl., L. C., Park, M., Miller, J. R. & Moon, R. T.

Trends Genet. 16, 279–283 (2000).

4. Axelrod, J. D. & McNeill, H. Sci. World J. (in the press).

5. Kuehl, M., Sheldahl, L. C., Malbon, C. C. & Moon, R. T.

J. Biol. Chem. 275, 12701–12711 (2000).

6. Yamanaka, H. et al. EMBO Rep. 3, 1–7 (2002).

news and views

240 NATURE | VOL 417 | 16 MAY 2002 | www.nature.com

Daedalus

Augmented eggsThe size of an egg is one of evolution’sbattlegrounds. The chick wants the biggestpossible egg, to aid its development; themother wants the smallest possible egg, tohelp her to lay it. Neither party gets itsideal. Daedalus now has a compromise. Hewants the egg to expand after being laid.

Sadly, all birds’ eggs have a rigidcarbonate shell. So DREADCO biologistsare breaking fertilized eggs with great care. They are opening them into largerglass or plastic cavities, filled with distilledwater or nutritive solution and sealed to the unbroken section of shell withsilicone resin. The chick within will find plenty of room to develop further,before having to break its way out. Mother birds may not want to sit on andhatch augmented eggs, although somefemales appreciate outsize ones. Designproblems will loom large. It will probablybe easier to hatch augmented eggs in anincubator.

In normal conditions, most birds hatch fairly simple-minded offspring —Daedalus recalls Konrad Lorenz’s goslings,which followed him because he pretendedto be their mother. But a chick thatdevelops in an expanded egg would takelonger in development, and could be farshrewder than average. Even morecunning, imagine two eggs sealed togetherby a tube the diameter of one of them, the extra space filled with nutrientsolution. Would one chick be aware of the other? Would they develop as a cleverpair, tackling problems neither could work out alone? Might they even havesome subtle avian empathy? A whole new type of bird might result from thisresearch. It could emerge from theaugmented egg with much enhancedmental or physical powers.

Daedalus is not sure whether to try the idea on domestic fowl. They areselected for stupid tractability; bright ones could be very troublesome. But ducks seem to have a lot of enterprisealready. They could be ideal. The newimproved chicks might swim or fly better,or be stronger and more decisive. Theycould come to dominate their fellows. Even so, an augmented duck that came to dominate a flock of them would beunable to hand on its abilities to the nextgeneration. So Daedalus’s egg augmentersare concentrating on birds that haverecently lost the power of flight, such aspenguins, kakapos and certain rheas. Is the ability to fly still inherent in a chick? If so, could it be recovered by eggaugmentation? David Jones

© 2002 Macmillan Magazines Ltd