has been formed around the porphyrin tem-plate. Analysis with nuclear magnetic reso-nance, mass spectrometry and chromato-graphy confirmed that virtually every alkenehad reacted with a neighbour. The wholesuperstructure is then sufficiently stable toallow the porphyrin template to be removedby cleaving the ester bonds. This leavesbehind a cavity, lined with eight carboxylic-acid groups, within the interior of the globu-lar molecule (Fig. 1b). The size and shapeof the cavity betray its origins from the porphyrin template.

The authors show that the dendrimerhost binds strongly to a test substrate — aporphyrin with four pyrimidine groups,with a total of eight basic nitrogen atoms inappropriate positions to form hydrogenbonds to the carboxylic-acid groups in thecavity. There is subtlety in the recognition,suggesting a well-defined binding pocket.The original porphyrin template with itseight hydroxyls does not itself bind becausethe combined sizes of a carboxylic acid and ahydroxyl group make too tight a fit.

However, the design is not perfect. Thereis flexibility in the dendrimer host structurethat leads to intramolecular hydrogen bondsbetween carboxylic acids in the bindingpocket. This results in an affinity that is simi-

lar for substrates irrespective of whether fouror eight hydrogen bonds are formed. Also,the lack of discrimination among differentisomers of a tetrapyridyl porphyrin suggeststhat the carboxylic-acid groups can movewithin the binding cavity and contact thenitrogen atoms in different positions.

Nonetheless, the overall concept outlinedby Zimmerman et al. is a promising one. Theidea of using a substrate (or at least a closeanalogue) as the template for the synthesis ofits own host will find many future applica-tions, for example in drug delivery and cata-lyst design, and in devising novel separationstrategies. Nature still has an edge in terms ofaffinity and selectivity, but Zimmerman andcolleagues’ approach will now permit theconstruction of artificial binding sites thatresemble those in globular proteins. ■

Andrew D. Hamilton is in the Department ofChemistry, Yale University, Box 208107,New Haven, Connecticut 06520-8107, USA. e-mail: andrew.hamilton@yale.edu1. Zimmerman, S. C., Wendland, M. S., Rakow, N. A., Zharov, I. &

Suslick, K. S. Nature 418, 399–403 (2002).2. Newkome, G. R., Moorefield, C. N. & Vögtle, F. Dendritic

Macromolecules: Concepts, Syntheses, Perspectives (VCH,Weinheim, 1996).

3. Wulff, G. Angew. Chem. Int. Edn Engl. 34, 1812–1832 (1995).4. Hawker, C. J. & Frechet, J. M. J. J. Am. Chem. Soc. 112,

7638–7647 (1990). 5. Trnka, T. M. & Grubbs, R. H. Acc. Chem. Res. 34, 18–29 (2001).

stage. So there have been hints that other proteins cannot substitute for dystroglycan,and that even reduced dystroglycan activitymight underlie the human disorders.

Michele et al.3 now provide evidence thatreduced (hypo) glycosylation of a-dystrogly-can is involved in the two diseases, as well asin a naturally occurring mouse mutant, themyodystrophy (myd) mouse. Muscle biop-sies from patients with MEB and FCMDrevealed normal patterns of b-dystroglycanbut no glycosylated a-dystroglycan. In elec-trophoresis, the core a-dystroglycan proteinshowed a shift in mobility, interpreted as achange in apparent molecular weight due toloss of sugar groups. Furthermore, the changein mobility of the a-dystroglycan componentwas identical for the MEB and FCMD sam-ples, implying that the different glycosyl-transferases mutated in these diseases affectthe same sugar residues on a-dystroglycan.

The authors further show that the hypo-glycosylated a-dystroglycan from patientswas impaired in binding proteins such aslaminin, agrin and neurexin — all of whichare components of basement membrane,the specialized sheet of extracellular matrixthat surrounds muscle and other cells. Simi-lar biochemical abnormalities were evidentin both the muscle and brain of a myd mousewith a mutation in a gene — the LARGE gene— which again is thought to encode a glyco-syltransferase5. Finally, Michele et al. findthat defects in neuronal migration in themyd mouse brain are like those seen in MEBand FCMD patients.

In sum, Michele et al.3 show that muta-

news and views

376 NATURE | VOL 418 | 25 JULY 2002 | www.nature.com/nature

Muscular dystrophies are genetic dis-eases that cause progressive muscleweakness. The best known is that

described by Duchenne, which affects boysand is evident from about five years of age,and which results from mutations in the geneencoding a protein called dystrophin. Anoth-er subclass is the congenital muscular dystro-phies, where muscle weakness is apparent atbirth or shortly afterwards. Two of thesefor which gene mutations have been foundare muscle–eye–brain disease (MEB) andFukuyama congenital muscular dystrophy(FCMD). Children carrying the faulty MEBor FCMD genes1,2 suffer from both muscleweakness and ‘cobblestone lissencephaly’, inwhich a flaw in neuronal migration results ina brain with a bumpy, cobblestone appear-ance and loss of the normal folding pattern.

How the two very different muscle andbrain defects arise in the same patient has notbeen known. Now, however, companionpapers by Michele et al.3 and Moore et al.4

(pages 417 and 422 of this issue) describe animpressive array of data that points to a

common mechanism. To function properlyin muscle, dystrophin has to form complexesthat include two components, a and b, ofanother protein, dystroglycan. Each of thesehas to be appropriately modified by glycosy-lation — the addition of sugar molecules byglycotransferase enzymes. The b-dystrogly-can in the membranous sheath of a musclecell, the sarcolemma, binds a-dystroglycan;in turn, a-dystroglycan binds to proteins suchas laminin in the extracellular matrix. Thetwo papers provide evidence that the defectunderlying muscle weakness and brain abnor-malities in both MEB and FCMD is disruptedglycosylation of a-dystroglycan (Figs1 and 2).

The MEB and FCMD genes both havesimilarity to genes known to encode glyco-syltransferases, although it was unclearwhich substrates of these enzymes are rele-vant to the congenital dystrophies. Likelycandidates, however, lie in the dystrophin–dystroglycan complex. Mutations in dystro-glycan have not hitherto been associatedwith human disease. But a ‘knockout’ of dys-troglycan in mice proves lethal at the embryo

Neurobiology

Full circle to cobbled brainM. Elizabeth Ross

A biochemical link between certain congenital muscular dystrophies andthe associated brain malformation known as cobblestone lissencephalyhas been elusive. But it looks as if that link has been found.

β

α

Laminin

Impaired binding

Sarcolemma

Dystroglycan

Muscle cell

Dystrophin

Extracellular matrix

SyntrophinsF-actin

Figure 1 Congenital muscular dystrophy anda-dystroglycan3. In muscle, dystrophin bindsboth F-actin in the cytoskeleton and the‘dystrophin glycoprotein complex’, whichincludes dystroglycan. a-dystroglycan is asecreted component that lies outside the musclecell. To function properly, it must be glycosylated— have sugar groups attached — and bind bothb-dystroglycan in the cell’s membrane, thesarcolemma, and proteins in the extracellularmatrix such as laminin. Failure of glycosylationimpairs binding to the extracellular matrix,destroying the muscle fibre over time.

© 2002 Nature Publishing Group

tions in three different glycosyltransferasesresult in similar biochemical abnormalitiesin a-dystroglycan, with associated changesin muscle tissue. Moreover, the glycosyl-transferase mutation in myd mice also resultsin a neuronal-migration defect that would beexpected of a congenital muscular dystrophyin which the brain is affected. One conse-quence in the mice is a disrupted basementmembrane — the glia limitans — of the glialcells that provide essential physical supportfor neurons in the brain.

Moore et al.4 make the story even moreconvincing. They took the next logical step:seeing whether knocking out dystroglycan,the proposed target of the glycosylationdefect, has the expected effects in the brain.Mice with complete loss of the dystroglycangene die as embryos, so Moore et al. made a‘conditional’ knockout that allowed them toinactivate the gene’s expression in glia andbrain neurons only.

The result is a viable mutant with normalmuscle but brain malformations similar tothose seen in MEB and FCMD, and also inWalker–Warburg syndrome, another con-genital muscular dystrophy. Most notably, theglia limitans is affected, presumably permit-ting the overextended migration of neurons inthe developing brain (Fig. 2). This is the mostimportant diagnostic feature of cobblestonelissencephaly. The mice also lack the usual fis-sure between the brain’s hemispheres, a char-acteristic of Walker–Warburg syndrome, andsuffer from an overabundance of glia. This lasteffect could arise in reaction to the loss ofbasement-membrane integrity, and the con-sequent inflammatory response, or could bedue to the altered development of neurons orglia, or both.

Dystroglycan may also work in the neu-

romuscular junction6, the structure at whichneurons and muscle cells connect and whichis similar to the synapses that connect neu-rons in the brain. So Moore et al.4 went on tolook at synaptic function in their knockoutmice. Their results from electrophysiologicalrecordings on brain preparations show that

the capacity for long-term potentiation, theprocess thought to underlie learning andmemory, is reduced under certain condi-tions. They suggest that this stems from theeffect of impaired dystroglycan in the post-synaptic mechanisms of neurotransmission.

Together, the two reports3,4 describe apowerful ‘full circle’ of investigation — fromknown human disease genes, to recognitionof similarities between mouse mutants andhuman disease, to discovery of common fea-tures in the disease processes and identifica-tion of a pathway that might be involved. Predicting the outcome of genetic-manipu-lation experiments in mice, and testing thosepredictions, completes the circle. The excit-ing part is that this is only the beginning ofunderstanding the neurobiology of the con-genital muscular dystrophies. Yet to come isthe unravelling of dystroglycan’s role insynapse formation in the brain and in theneuromuscular junction, as well as in neu-ronal migration and synaptic function. ■

M. Elizabeth Ross is in the Laboratory ofNeurogenetics and Development, Weill MedicalCollege of Cornell University, 525 East 68th Street,W605, New York, New York 10021, USA. e-mail: mer2005@med.cornell.edu1. Kobayashi, K. et al. Nature 394, 388–392 (1998).

3. Michele, D. E. et al. Nature 418, 417–422 (2002).

4. Moore, S. A. et al. Nature 418, 422–425 (2002).

5. Grewal, P. K. et al. Nature Genet. 28, 151–154 (2001).

2. Yoshida, A. K. K. et al. Dev. Cell 1, 717–724 (2001).

6. Grady, M. et al. Neuron 25, 279–293 (2001).

news and views

NATURE | VOL 418 | 25 JULY 2002 | www.nature.com/nature 377

Lamininαβ

Subarachnoid space

Marginal zone

Cortical plate

Glia limitans

Radialglial fibre

Neuron

Discontinuousglia limitans

Disorganized/displaced neurons

Normal Cobblestone lissencephaly

Dystroglycan

a b

Figure 2 Cobblestone lissencephaly and a-dystroglycan4. In brain development, thedystrophin–dystroglycan complex is present in both of the two principal types of brain cells, glia andneurons. a, During normal development, the top of the cerebral cortex is defined by a basementmembrane — the glia limitans — at the end of the radial glial fibres that guide neuron growth. Thisprevents neurons that form the marginal zone and cortical plate, two layers of the cortex, frommigrating out of the brain proper and into the subarachnoid space. b, In cobblestone lissencephaly,failure to glycosylate a-dystroglycan is associated with gaps in the glia limitans, and failure ofneurons to organize themselves within the cortical plate and their migration into the subarachnoidspace. It remains to be seen whether neuronal motility is also directly affected.

Like the gentle patter of raindrops, weexpect photons, the quanta of sunlight,to arrive at Earth at random intervals,

their arrival times distributed in just the same,natural way that customers arrive at a boxoffice to buy tickets for a play. A histogram ofthe number of people or photons arriving perunit time follows what is known as the Poissondistribution. But in 1909, in the first clear evidence for wave–particle duality, Einsteinpointed out1 that the width of this distribu-tion for sunlight contains both the Poissoncontribution of random arrival times, and asecond ‘wave-noise’ contribution, whichcauses the photons to arrive in bunches.

An experiment performed by Kiesel et al.2,reported on page 392 of this issue, shows thatthe same principle applies to a beam of coher-ent electrons — but with the opposite effect.In contrast to the bunching of photons3, thisexperiment confirms the theoretical predic-tion4 that a stream of coherent electrons

will ‘anti-bunch’, tending to become moreequally spaced than the classical Poisson pre-diction (Fig. 1, overleaf). In the quantumregime, electrons have an innate tendency toavoid each other, thereby demonstrating afundamental difference in the way light andelectrons interfere with themselves.

In the 1950s, the astronomers RobertHanbury Brown and Richard Twiss devisedthe famous experimental arrangementknown as intensity interferometry to studythese effects for photons in the form of lightfrom distant stars5. They split the light intotwo beams using a half-silvered mirror, thencompared the arrival times of photons attwo separate detectors. Due to bunching,they saw that coincident photon arrivalswere more likely than expected by chance.Hanbury Brown and Twiss realized theycould use this bunching to infer the angularsize of distant stars.

The similar experiment performed by

Quantum physics

Spaced-out electronsJohn C. H. Spence

In a stream of photons, the particles tend to bunch together, but electronsin a beam do the opposite. At last, this quantum effect for free electrons —the Hanbury Brown–Twiss anticorrelation — has been seen experimentally.

© 2002 Nature Publishing Group