The reelin expression studies summarized above clearly demonstrate a drastic amplification of the mRNA and protein levels in mammalian CRc as compared to other reelin-positive neurons in the reptilian and avian MZ. It is tempting to assume that high levels of reelin are required during cortical spreading, and indeed a sustained supply of reelin-positive cells is observed in the human MZ throughout fetal development (Meyer and Goffinet, 1998). On the other hand, there is ample evidence that the concentration of reelin in the mouse cortical MZ is in large excess. The reeler (reelin-deficient) mutant phenotype only appears in brains of chimeric mice when reeler cells greatly outnumber normal cells (Mikoshiba et al., 1986; Terashima et al., 1986; Mullen et al., 1997; Yoshiki and Kusakabe, 1998). In spite of a loss of CRc and a drastic decrease of reelin concentration, homozygous p73 mutant mice do not have a reeler-like cortex (Yang et al., 2000). A normal CP develops in vitro in serum-free culture medium without the addition of exogenous reelin, and addition of reelin to reeler slice in vitro (Jossin et al., 2004), or ectopic expression of low levels of reelin in the VZ in transgenic reeler mice malformation (Magdaleno et al., 2002), are able to partially correct the reeler trait. These observations suggest that reelin may diffuse from sources other than CRc and that the expression of receptors and Dab1 may be more important than the site of ligand production.
Why did mammalian CRc amplify a production that appears so redundant in mice? As a way of explaining this apparent contradiction, we would like to suggest the following scenario. Amplification of reelin synthesis in CRc was necessary for the development of a foliated cortex, and stem mammals initially developed a moderately folded, not a lissencephalic, cortex. During evolution, some cortices, such as that of rodents, evolved secondarily into a lissencephalic type. Not being detrimental, elevated reelin production in CRc was not necessarily adjusted in parallel with the reduction of cortical surface. Other lineages, most notably primates, acquired increasingly more foliated cortex and, in humans, additional numbers of reelin-expressing cells became necessary. This idea is compatible with observations that cortical foliation can vary widely within closely related lineages. For example, in monotremes, Echidna has an elaborate, highly foliated cortex, whereas Platypus is almost lissencephalic (Rowe, 1990). Similar examples can be found in other phyla, including primates. Production of a gyrated mouse cortex is artificially accomplished by elegant, yet relatively simple, manipulations (Rakic, 2004), such as germline inactivation of caspases 3 and 9 (Kuida et al., 1996, 1998), increased expression of beta-catenin in transgenic mice (Chenn and Walsh, 2002, 2003), or incubation of embryonic cortex in vitro in the presence of lysophosphatidic acid (Kingsbury et al., 2003; Price, 2004). When such an increase in foliated cortex is observed, the cortical ribbon is nearly normal and, unlike the reeler cortex, shows that the production of reelin is largely sufficient. These observations indicate that the production of a gyrated cortex does not require extensive genetic modifications and could have evolved in any phylum, for example, by acquisition of more precursors of radial units in VZs.
As no living species are closely related to stem mammals, the hypothesis proposed above will always remain somewhat speculative. However, it predicts that, in a given lineage, the density of reelin-positive cells (per cortical surface area) should be higher in the embryonic brain of representatives with a smooth cortex than in those with a gyrated cortex. The mean density of reelin-posi-tive cells, averaged over large areas, including gyral crowns and sulci, could be compared in embryonic cortices of rodents versus animals that are thought to have evolved little, but have some cortical gyration, such as hedgehogs or spiny anteaters. Potentially the best possible test would be to compare the density of CRc in the embryonic MZ of Echidna and Platypus.
All components of the reelin and other signaling cascades were probably present in stem amniotes, available as basic building blocks for cortical evolution. Why then did significant cortical foliation occur only in mammals? Foliation correlates nicely with cortical volume and may be required to increase it beyond some threshold, but how is it achieved? Surely the amplification of reelin production in MZ cells was not the sole limiting factor, as the examples above indicate, nor was the necessary increase in the number of radial cortical units. We would propose that increased reelin synthesis and the development of an enlarged number of precursors and radial cortical units were not hard to achieve, but that the resulting increase in cortical surface did not occur widely because it was difficult to master for some unknown reason. One difficulty could be the coordinate growth of mesodermal components, such as blood vessels and the cranial envelope that must accompany brain growth. Another problem that had to be solved in order to evolve a laminarly organized and tangentially widespread cortex is that of increased neuronal excitability and susceptibility to seizures. A consequence of the highly geometrical arrangement of radial cortical columns is that it facilitates modification of the membrane potential by field effects (ephaptic interactions), largely believed to be involved in the oscillations of electrocortical rhythms such as the alpha or theta rhythms. This quasicrystalline arrangement presumably has advantages in terms of computational power, but also comes at a price, as ephaptic excitation facilitates the tangential spreading of activity and decreases the threshold for aberrant epileptic discharges (McCormick and Contreras, 2001).
Was this article helpful?