News & Views | Published:

Developmental neurobiology

Cortical liars

Naturevolume 417pages605606 (2002) | Download Citation


Inhibitory cells known as interneurons constitute a significant proportion of the neurons in the neocortex of mammalian brains. As far as interneuron origins are concerned, humans may be the odd man out.

For more than a decade, the interneurons found in the neocortex of the human brain have been disturbing the peace of developmental neurobiologists. Interneurons represent roughly 20% of cortical neurons and are responsible for modulating the firing of the principal neurons known as projection neurons. They do this by producing the inhibitory neurotransmitter, GABA. But where exactly interneurons originate in the embryonic brain remains controversial. On page 645 of this issue1, Rakic and colleagues provide evidence that one source of these cells for the neocortex that is not so evident in other mammals provides the majority of interneurons in humans. This is a striking conclusion, for it might help to explain the greater complexity of the human brain and what can go wrong in mental illness.

The controversy over the origins of cortical neurons hinges in part on the difference between radial and tangential movement of the cells from their sites of origin to their final home in the layers that make up the neocortex. Until about ten years ago, accepted wisdom was that cortical neurons migrate only radially: this is a comfortable notion, for it implies that the neocortex is built according to a carefully crafted blueprint with only local, outwards migration to create the cortical columns.

However, that notion was upset by the demonstration, in rodents, that a significant percentage of neurons migrate tangentially some distance from their point of origin2. At that time, the identities of the tangentially migrating neurons were not known. But the fuss over migration patterns relates to the debate as to whether the embryonic neocortex is already carved into regionalized sectors3 or whether the emergence of functionally distinct areas in the adult is controlled by other mechanisms. Widespread and indeterminate tangential dispersion of cells would, on the face of it, scramble any embryonic blueprint.

A bigger surprise emerged when second-generation cell-lineage studies revealed that tangential migration is associated with the development of interneurons, whereas projection neurons seemed to be the product of radial migration4,5. At that time it was assumed that both kinds of neuron are born in the cortical ventricular zone (VZ), the main site of cell birth in the neocortex from which new cells cross an adjacent layer, the subventricular zone (SVZ), into the emerging cortical layers. An outline of embryonic brain geography is shown in Fig. 1.

Figure 1: Interneuron origins in the embryonic brain.
Figure 1

The graphic shows half of a cross-section through human (a) and rodent (b) forebrains. Rakic and colleagues1 provide evidence that, during development, about 65% of interneurons in the human neocortex originate locally in the ventricular zone/subventricular zone (VZ/SVZ) of the neocortex. In rodents, by contrast, the main source of interneurons seems to be long-distance tangential migration of cells from the VZ of a subcortical area known as the ganglionic eminence. The other 35% of human interneurons seem to be produced by this route. In humans, interneurons originating from the cortical VZ/SVZ express both the transcription factors Mash1 and Dlx1/2 (cells with red nuclei); interneurons from the ganglionic eminence express only Mash1 (white nuclei).

But why would interneurons — which constitute a minority of cortical neurons, yet have the same developmental environment as projection neurons — migrate in a different direction? There were thought to be at least two possibilities. Either the migration pathway itself was somehow involved in turning cells into interneurons, or the progenitors residing in the VZ were already specified to make either interneurons (which for some reason migrated only tangentially) or projection neurons (which migrated only radially).

Unexpectedly, neither explanation was in itself correct. Instead, studies on mutants lacking the gene-transcription factors Dlx1 and Dlx2 showed that in these animals cortical interneurons were severely reduced in number. What is more, most interneurons expressing Dlx1/2 were found to originate from a subcortical area6, the ganglionic eminence (shown in Fig. 1). This gives rise to subcortical structures such as the striatum, but the studies with mutant animals indicated that it also furnishes at least 75% of cortical interneurons7. Work with other species, including ferrets, produced similar observations, suggesting that this is an ancient mechanism of cortical development.

But what about interneurons in the embryonic human neocortex? The first hint that they might behave differently came from work on a different structure known as the thalamus, an important relay station for the neocortex. Letinic and Rakic8 discovered a migratory pathway that ferries inter- neurons generated in an external source, the ganglionic eminence, to the thalamus, and this pathway turned out to be absent from macaques and mice. Rakic and colleagues1 now report that the ganglionic eminence also produces interneurons for the human neocortex, but the contribution is minor (35%). Instead, and most notably, they find that 65% of the interneurons arise locally from the cortical VZ/SVZ (Fig. 1a).

This source has two defining qualities. First, the cells migrate as chains, using each other as scaffolding to glide along; second, they express Mash1, a transcription-factor marker for interneuron progenitors. Surprisingly, Rakic and colleagues found that Mash1-positive progenitors reside mostly in the SVZ — a secondary germinal zone traditionally associated with the generation of another, non-neuronal cell type, glial cells. This implies that the cortical environment can nurture and support two types of developing neuron that are born at adjacent addresses (projection neurons in the VZ, and interneurons in the SVZ).

In the most telling experiments, Rakic and co-workers1 infected cell progenitors in the cortical VZ/SVZ with a retrovirus carrying a gene tracer so that they could follow cell division and cell movement. By direct imaging of the infected cells, they showed that cells producing GABA also expressed the genes encoding Mash1 and Dlx1/2. But the idea that the interneuron source in the cortical VZ/SVZ is unique to humans needs to be counterbalanced by the finding that, in mice without Dlx1/2, a quarter of the cortical interneuron population still exists. The implication is that in other mammals interneurons are also generated locally in the cortical VZ.

Critics would also argue that the larger cortical interneuron source in humans might simply reflect a boosting of pre-existing developmental mechanisms. Indeed, Mash1 is expressed in certain cortical progenitors in mice9. Alternatively, the novel source of Mash1 progenitors in humans might arise from the evolutionary duplication of comparable cells in the ganglionic eminence. From work with rodents it seems that a single genetic switch is all that it takes — misexpression of Mash1 in cortical neurons results in their transformation to the interneuron type10. In this respect, laboratory studies with rodents might have recapitulated an event in human evolution.

Clarification of certain points raised by the new work1 will be required. For instance, Rakic and colleagues show that interneurons actually migrate tangentially in the VZ/SVZ at first, followed by radial migration in the zone beyond that. In rodents, interneurons arising from ganglionic eminences behave in a similar manner, initially travelling close to the VZ as depicted in Fig. 1b, and subsequently changing tack to the radial direction11. As Rakic and colleagues admit, they cannot exclude the possibility that some of the Mash1-positive cells in the VZ/SVZ might have arrived from the ganglionic eminence at earlier embryonic stages and then continued to divide locally.

What about the evolutionary aspect? The primate neocortex is disproportionately large compared with that of other mammals, whereas the subcortical structures show only linear increases in size12. Evolution of the relatively large neocortex would disturb the scaling of shared developmental processes between cortical and subcortical structures; and it would require a numerical increase in interneurons, which would also have longer distances to travel. Might the VZ of the ganglionic eminence (committed, of course, also to interneuron production for the thalamus) have been hard pressed to meet demand, resulting in the adaptational production of interneurons in the cortical VZ? There is also a temporal consideration. In humans, neuron migration into the neocortex occurs over a period of 200 days, compared with just 10 days in mice and 30 in ferrets. If interneurons were produced largely subcortically, as in rodents, it might be that the physical and molecular receptivity of the human cortical environment would become exhausted.

The debate about interneuron origins echoes lines from W. H. Auden's Ode to the Diencephalon: “How can you be quite so uncouth? After sharing the same skull for all these millennia, surely you should have discovered the cortical I is a compulsive liar.” The cortical I — the interneuron cell type — has long been seemingly lying to observers about its behaviour. And then there is the medical connection. Malfunction of the GABA system is a hallmark of disorders such as schizophrenia13, to which defects in interneuron migration can be a predisposing factor. So could those defects be responsible for the 'cortical lies' perceived by the sufferer? Even when origins of interneurons have been sorted out, questions such as this will continue to disturb the peace of neurobiologists.


  1. 1

    Letinic, K., Zoncu, R. & Rakic, P. Nature 417, 645–649 (2002).

  2. 2

    Maricich, S. M., Gilmore, E. C. & Herrup, K. Neuron 31, 175–178 (2001).

  3. 3

    Rakic, P. Proc. Natl Acad. Sci. USA 92, 11323–11327 (1995).

  4. 4

    Mione, M. C., Cavanagh, J. F. R., Harris, B. & Parnavelas, J. G. J. Neurosci. 17, 2018–2029 (1997).

  5. 5

    Tan, S.-S et al. Neuron 21, 295–304 (1998).

  6. 6

    Anderson, S. A., Eisenstat, D. D., Shi, L. & Rubenstein, J. L. Science 278, 474–476 (1997).

  7. 7

    Marin, O. & Rubenstein, J. L. Nature Rev. Neurosci. 2, 780–790 (2001).

  8. 8

    Letinic, K. & Rakic, P. Nature Neurosci. 4, 931–936 (2001).

  9. 9

    Nieto, M., Schuurmans, C., Britz, O. & Guillemot, F. Neuron 29, 401–413 (2001).

  10. 10

    Fode, C. et al. Genes Dev. 14, 67–80 (2000).

  11. 11

    Nadarajah, B., Alifragis, P., Wong, R. O. & Parnavelas, J. G. Nature Neurosci. 5, 218–224 (2002).

  12. 12

    Finlay, B. L. & Darlington, R. B. Science 268, 1578–1584 (1995).

  13. 13

    Benes, F. M. & Berretta, S. Neuropsychopharmacology 25, 1–27 (2001).

Download references

Author information


  1. the Brain Development Laboratory, Howard Florey Institute, The University of Melbourne, Parkville, 3010, Victoria, Australia

    • Seong-Seng Tan


  1. Search for Seong-Seng Tan in:

Corresponding author

Correspondence to Seong-Seng Tan.

About this article

Publication history

Issue Date


Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Newsletter Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing