NATURE REVIEWS | NEUROSCIENCE VOLUME 2 | JANUARY 2001 | 7

Traditional concepts of vertebrate neurogenesis, in which neuronsand glia arise from separate, irreversibly committed lineages, aregradually being eroded. It has been shown that a subset ofastrocytic cells in the adult brain can give rise to neurons,suggesting either that these cells are able to de-differentiate or thatthey are less committed to the glial lineage than was previouslysupposed. In a new paper in Development, Malatesta et al. showthat radial glial cells also demonstrate a rarely recognizedpotential for neuronal differentiation.

The radial glia appear in the developing cerebral cortex aroundthe time of onset of neurogenesis, which is embryonic day 11(E11) in the mouse brain. They send out long processes thatextend from the ventricular zone to the pial surface, forming ascaffold to guide neuronal migration. Their marker expressionprofile has placed them in the astrocytic lineage, and a role as glialprogenitor cells has already been established.

Using fluorescence-activated cell sorting, Malatesta et al.isolated individual radial glia from the mouse cortex at variousstages of embryonic development. Two different criteria wereused to select the cells: the expression of green fluorescent protein(GFP) under the human promoter of the astrocytic marker GFAPand the presence of long radial processes. These processes couldbe labelled from the pial surface of the intact brain using afluorescent retrograde tracer. It was confirmed that the sorted cells expressed markers characteristic of radialglia, including RC2, GLAST and BLBP. The differentiationpotential of the isolated cells was analysed in co-culture with ratcortical cells to replicate their normal developmentalenvironment. Of the cells isolated from the E14 cortex, halfgave rise only to neuronal clones under these conditions and therest generated mostly astroglia. By contrast, radial glia isolatedfrom the cortex later during development (E18) gave rise mostlyto astrocytes, suggesting that the differentiation potential of theradial glia becomes more restricted as development proceeds.Intriguingly, the radial glia demonstrating the highest level ofneurogenic potential were isolated at a stage during which thebrain is most active in neurogenesis, raising the possibility that these cells could be acting as neuronal precursors in vivo aswell as in vitro.

Heather Wood

References and linksORIGINAL RESEARCH PAPER Malatesta, P. et al. Isolation of radial glial cells by fluorescent-activated cell sorting reveals a neuronal lineage. Development 127, 5253–5263 (2000) FURTHER READING Doetsch, F. et al. Subventricular zone astrocytes are neural stem cells inthe adult mammalian brain. Cell 97, 703–716 (1999)ENCYCLOPEDIA OF LIFE SCIENCES Cells of the nervous system

Achtung baby!

P E R C E P T I O N

The study of the object processingabilities of infants is a growing areaof research that extends our under-standing of object cognition by com-plementing the numerous studies inadults and non-human primates.Not surprisingly, probing the mindof a 6-month old baby presents aseries of problems that have led tosome ingenious experimental proce-dures. One commonly used methodinvolves measuring the amount oftime that an infant will orientate to agiven stimulus. This ‘preferentiallooking’ paradigm provides a mea-sure of how interesting the infantfinds a given stimulus and therebyoffers a window into their previousexperience with that object and theirdeveloping perceptual and cognitiveabilities. Although this type ofbehavioural paradigm has been usedwith great success, the ability tomake more direct measurements ofbrain processing in infants wouldprovide a significant step forward forthis area of research and wouldstrengthen the link with the adultand non-human primate literatures.Csibra and colleagues provide anexample of how non-invasive imag-ing can be combined with more tra-ditional approaches to bring thedevelopment of object processingsystems in the infant brain into theneuroscience arena.

Their experiment focused on oneof the key questions in visual percep-

tion — if the visual features of anobject such as form and colour are, toa large degree, processed separately,how is this information ‘bound’together in the brain to allow the perception of a unified object? Onetheory suggests a prominent role forgamma-band (40 Hz) oscillatoryactivity in perceptual binding.

This form of binding can be stud-ied through the use of illusory per-cepts such as the Kanizsa square, inwhich an illusory square can beinduced between four appropriatelyorientated ‘pacman’ elements asshown in the figure. In adults the per-ception of illusory objects induces aburst of 40 Hz oscillations shortlyafter stimulus onset. Csibra et al. usedelectroencephalography to measurethe oscillatory brain activity of 6 and8-month old infants while theyviewed either a Kanizsa square or acontrol stimulus consisting of mis-aligned elements that did not inducethe illusory percept. There was anenhancement of gamma-band activi-ty in response to the Kanizsa squarein 8-month-old infants but not in 6-month-old infants. This activity was not observed in response to thecontrol stimulus.

This work is consistent with, andextends, the behavioural studies thatsuggest that perception of the Kanizsasquare begins at around 7 months ofage. But the wider importance of thiswork may reside in the demonstrationthat direct measurements of neuralprocessing can be undertaken in veryyoung infants during the performanceof behavioural tasks. By encouragingthe dialogue between cognitive devel-opment and cognitive neuroscience,this study may therefore bind thesetwo areas together and so open thedoors of infant perception.

Peter Collins

References and linksORIGINAL RESEARCH PAPER Csibra, G. et al.Gamma oscillations and object processing in theinfant brain. Science 290, 1582–1584 (2000)FURTHER READING Singer, W. & Gray, C. M.Visual feature integration and the temporalcorrelation hypothesis. Annu. Rev. Neurosci.18, 555–586 (1995)WEB SITE Gergely Csibra

Changing glia

N E U R O G E N E S I S

Two neurons generated from a mouse radial glial cell, double labelled with amouse-specific antibody (red) and the neuron-specific anti-β-tubulin-III(green). Image supplied by Paolo Malatesta, Max-Planck Institute ofNeurobiology, Germany.

© 2001 Macmillan Magazines Ltd