Neurons extend one long axon, through which they transmit electrical impulses to other cells in the nervous system. Surprisingly, it seems that where the axon forms is determined entirely within the neuron.
Neurons act as electrical relays: they collect information from other neurons through multiple extensions called dendrites, and transmit this information through one long protrusion, the axon. But the mechanism that determines where the axon forms at the neuronal surface has been unclear — it might be determined by an extracellular cue or by some intrinsic polarity that exists in the neuron even before the axon begins to grow. On page 704 of this issue, Calderon de Anda et al.1 resolve this question by describing an unexpected correlation between axon outgrowth and the position of the centrosome — a structure that is involved in organizing the cell's internal scaffolding. The authors propose that where the axon forms is ultimately determined by the orientation of the neuron's final cell division, after which it becomes fully specialized.
Neuronal differentiation can be followed in cell culture. When neurons from a brain area called the hippocampus are plated onto coated coverslips, they follow a stereotypical sequence of differentiation events2. First, they form lamellipodia — highly dynamic protrusions that characteristically grow out of the leading edge of motile cells (stage 1). Shortly afterwards, they form four or five short extensions called neurites that have yet to gain the characteristics of axons or dendrites (stage 2). After 24 hours, one of the neurites extends rapidly and will become the axon (stage 3). The remaining neurites then acquire the characteristics of dendrites (stage 4), and finally, axons and dendrites form electrical contacts (stage 5).
What determines where the axon will form? To address this, Calderon de Anda et al.1 analysed cultured rat hippocampal neurons from day 16 of embryonic development, immediately after the neurons' terminal division and before they become fully differentiated. Using real-time analysis, the authors show a sequence of events that correlates axon formation with a pre-existing polarity in the undifferentiated neuron (Fig. 1a). After the terminal division, the centrosome comes to lie opposite the plane of cleavage. During differentiation, the first lamellipodium forms in the region of cell membrane overlying the centrosome, and this is where the first neurite will grow. Calderon de Anda et al. also show that the axon consistently forms from the first neurite that grows out after the terminal division. So it is the plane of the terminal division that determines the site of axon emergence.
The results are not artefacts of cell culture because the authors find a similar correlation for hippocampal neurons differentiating in situ and for neurons in the developing eye of the fruitfly Drosophila. They are also not merely a correlation but a causal link, because ablation of centrosomes by a process known as chromophore-assisted light inactivation blocks axon outgrowth in cultured Drosophila neurons (Fig. 1b). Furthermore, when the final stages of cell division are inhibited in hippocampal or cultured Drosophila neurons, the daughter cells contain two centrosomes. These cells form two long neurites that extend from positions directly overlying the two centrosomes (Fig. 1c). Although the experiments are carried out in different organisms, they indicate that centrosomes are required and sufficient for determining the position of axon outgrowth. Furthermore, they suggest the existence of a ‘stage 0’ in which cell polarity exists without any visible effect. This pre-existing polarity is used at later stages to direct neurite formation and axon specification.
These observations reveal exciting parallels between differentiating neurons and other cell types in which centrosomes initiate polarization. Shortly after fertilization, zygotes (single-celled embryos) of the nematode worm Caenorhabditis elegans become highly polarized along what will become the anterior–posterior axis3. This polarity is needed during the first cell division to segregate proteins differentially into what will become the two daughter cells, which will go on to have different fates. The axis of polarity in C. elegans zygotes is determined by the sperm entry-point, with polarization being initiated by an interaction between the centrosome (provided by the sperm) and the cell membrane3.
The similarity between C. elegans and neurons extends to the molecular level. Both polarity processes seem to involve an evolutionarily conserved set of proteins known as Par proteins3,4. In C. elegans, Par-3 and Par-6 and the atypical protein kinase C (aPKC) localize to the anterior cell membrane, whereas Par-1 and Par-2 are concentrated posteriorly. In hippocampal neurons, Par-3 and Par-6 are found only in the axon, and if they are overexpressed, other neurites are induced to assume an axon-like morphology (see ref. 4 for a review). Moreover, vertebrate relatives of Par-1 seem to enhance neurite outgrowth. Although the distribution of Par proteins before differentiation now needs to be examined, the findings of Calderon de Anda et al.1 suggest the existence of an evolutionarily conserved molecular machinery that polarizes cells and uses centrosome position as a reference point.
How does the centrosome influence the overlying cell membrane to induce neurite outgrowth at a particular position? The first morphological change during neuronal differentiation is the formation of lamellipodia. The process that creates these structures during cell migration is fairly well understood. It is driven by formation of a meshwork of actin proteins beneath the cell membrane. In fact, cell migration and neurite outgrowth might involve the same molecular machinery. It is remarkable that one of the first events in cell migration (at least in brain cells called astrocytes) is the reorientation of the centrosome to the future leading edge. This process is accompanied by a redistribution of the protein Cdc42 (a small GTPase, involved in cell signalling) and — like axon formation and C. elegans polarity — it requires Par-6 and aPKC (ref. 5). Although the function and distribution of these proteins during early neuronal differentiation (the hypothetical stage 0) are unknown, it is likely that they control the cross-talk between centrosomes and the cell membrane in neurons as well.
What happens downstream of the Par proteins? The protein Rac (another small GTPase) is primarily responsible for lamellipodium formation, and Par-3 interacts with the Rac activators Tiam1 and STEF (ref. 4). So Par-3 could be responsible for lamellipodium formation through localized Rac activation. Thus, by analogy to other cell-polarity events, we can already draw a molecular pathway for neurite outgrowth that can be tested in the hippocampal neuron culture model. It should be noted, however, that Drosophila axon outgrowth is independent of Par-6 and aPKC, and the proposed pathway must be verified experimentally before any further conclusions can be drawn.
The results of Calderon de Anda et al.1 imply that the orientation of the final neuronal division is essential for correct wiring of the developing brain. Although such orientation is undoubtedly vital in invertebrates, its relevance in vertebrates is unclear6. The mechanisms responsible for the orientation of cell division have only recently begun to emerge in invertebrates3. Identification of those mechanisms in vertebrates will allow us to manipulate the orientation of cell division and test the effects on axon outgrowth — an experiment ultimately required to confirm the mechanism proposed by Calderon de Anda and colleagues.
Calderon de Anda, F. et al. Nature 436, 704–708 (2005).
Dotti, C. G. et al. J. Neurosci. 8, 1454–1468 (1988).
Cowan, C. R. & Hyman, A. A. Annu. Rev. Cell Dev. Biol. 20, 427–453 (2004).
Wiggin, G. R., Fawcett, J. P. & Pawson, T. Dev. Cell 8, 803–816 (2005).
Etienne-Manneville, S. & Hall, A. Cell 106, 489–498 (2001).
Wodarz, A. & Huttner, W. B. Mech. Dev. 120, 1297–1309 (2003).