Neural Development

Neural development is the mechanistic link between the shaping forces of EVOLUTION and the physical and computational architecture of the mature brain. A growing knowledge of how the genome is expressed in development in the progressive specification of cell fates and neural structures is being combined with a better understanding of the functional organization of the adult brain to define the questions of neural development in a way never before possible. Until recently, questions in neural development were problem-centered -- for example, how do axons locate a target?, how is a neuron's neurotransmitter specified?, how are topographic maps made? While most current research remains directed at such empirical problems, advances in our understanding of genetics and evolution have begun to give the questions of neural development a more principled structure.

The most surprising insight of molecular genetics and evolution regarding the brain is the extreme conservation of fundamental genetic and physical structures across vertebrate orders, and even across phyla. Specification of gene expression at the level of the individual neuron is too expensive of the genome; a better solution is to divide areas of the developing nervous system into domains specified by overlapping patterns of gene expression that, in combination, confer unique information to each zone. Such a solution was originally found to be operating in the control of head development in the fruit fly Drosophila. Through a mosaic pattern of expression, the HOM-C class of homeotic genes specify segmentation during Drosophila development (Lewis 1978), regulating expression of other genes that direct differentiation of structures along the anterior- posterior neuraxis. Since this discovery, these genes or their homologues have been found throughout the animal world; vertebrate homologues of the HOM-C genes, known as Hox genes, were found to delineate various aspects of segmentation of the vertebrate hindbrain and midbrain (Keynes and Krumlauf 1994). An immediate benefit from this descriptive work is a better understanding of the segmental architecture of the forebrain, which had been enigmatic and controversial. The overlapping pattern of Hox and other regulator gene expression allowed the first characterization of a continuous pattern of segmental architecture from spinal cord to olfactory bulb in the vertebrate brain, even in the elaborate forebrain (Puelles and Rubenstein 1993).

This type of conservation of developmental patterning has been apparent not only in fundamental segmental divisions but in many other features of development. In morphogenesis, for example, the same regulator gene (reviewed in Zuker 1994) is implicated in the proper development of eyes both in Drosophila and mammals, even though their common ancestor could not have had an image-forming eye. Formation of initial axonal scaffolding and the mechanisms of axon extension are strikingly similar in both vertebrate and complex invertebrate brains (Easter, Ross, and Frankfurter 1993; Goodman 1994; Reichert and Boyan 1997). Within mammals, and possibly most vertebrates, the sequence and relative timing of events in both early neurogenesis and process extension produce extremely predictable alterations in morphology as brains enlarge in various radiations (Finlay and Darlington 1995). That the structures that underlie cognitive processes in complex animals can be found, albeit in reduced form, in animals without such capacities is an important consideration for future theory building in COMPUTATIONAL NEUROSCIENCE.

The actual solutions found in neural development to clearly stated, logical problems seem bound to defy standard hypothesis testing. For example, a central and conspicuous feature of brain organization is the topographic representation of sensory surfaces and the preservation of topographic order as one brain structure maps to the next, though the information-bearing dimensions being mapped are sometimes unclear. How are such maps formed? A number of mechanisms could independently produce an acceptable solution: (1) the spatial relationship of elements in connecting maps could be passively apposed; (2) temporal gradients could map one element to another in an organized sequence; (3) neighboring elements in a map could actively recognize one another so that the map might travel in a coherent pattern of axons to its target; (4) different parts of the map might have different "road maps" to find the target; (5) the elements in the first map might recognize locations in the target map at varying degrees of specificity; (6) the map might develop from trial and error, based on experience; or (7) statistical regularities in the activity pattern of the input array could be used to confer order in the target array. In the highly studied, paradigmatic case of map formation in the retinotectal system begun with the work of Roger SPERRY over fifty years ago, every one of the logical possibilities described above has been shown to contribute to the formation of the adult map (Udin and Fawcett 1988), and unsurprisingly (given the multiplicity of mechanisms), multiple genes are required for its successful development (Karlstrom et al. 1996). The cause of such (at least conceptually) uneconomical solutions is unclear; because this is an evolved system, it could be an accretion of solutions and exaptations, spandrels on spandrels (Gould 1997). Different features of the solution come in at different developmental times, and as a whole this apparent redundancy of mechanism may be responsible for the robust nature of much of neural development.

Conversely, single mechanisms often appear in the solution to multiple developmental problems. Such a mechanism is the Hebbian, activity-dependent stabilization of the synapse. This single mechanism serves in diverse areas such as stabilization of the neuromuscular junction, refinement of topographic maps due to the correlation of firing of neighboring units in a topographic map, sorting of unlike from like inputs in such notable cases as the formation of ocular dominance columns in the VISUAL CORTEX, and basic associative learning both in development and in adulthood (Katz and Shatz 1996). A particularly interesting developmental use of this mechanism is the "retinal waves" described in the ferret where, prior to actual visual experience, the RETINA appears to generate its own highly self-correlated waves of activation that propagate through the nervous system and can initiate the various axonal sorting processes described above (Wong, Meister, and Shatz 1993). A challenge for future work is to describe how this mechanism might function in the creation of neural networks capable of detecting more complex aspects of information structure than temporal association.

The CEREBRAL CORTEX or isocortex has always commanded special attention as the structure of largest volume in the human brain, and the one most closely associated with complex cognitive skills. In this case, the principal developmental questions have been motivated by the adult functional architecture of the isocortex, a structure with a rather uniform architecture which nevertheless carries out a number of distinct and diverse functions. Is the isocortex a structure that performs some sort of standard transformation of its input, with differences in cortical areas (e.g., visual, motor, secondary sensory) arising epigenetically from interaction with the input, or are early regions in some way optimized for their future roles in adult isocortex?

Both positions capture aspects of the true state of affairs, as the following list of features of cortical development will illustrate. The neurons of the isocortex arise from a sheet of cells in the ventricular zone that is not fully uniform in its neurochemical identity or rate of neurogenesis (reviewed in Levitt, Barbe, and Eagleson 1997). Neuroblasts migrate out on radial glial cells to the cortical plate in a well-described "inside-out" settling pattern, thus potentially retaining positional information of the ventricular zone, though considerable dispersion also occurs (Rakic 1995). The cortical plate shows some neurochemical nonuniformity before it is innervated by any outside structure (Cohen-Tannoudji, Babinet, and Wassef 1994). The innervation of cortex by the THALAMUS is extremely specific and mosaic, while by contrast, the large majority of intracortical connectivity can be accounted for by the generic rule, "connect to your two nearest neighbors" (Scannell, Blakemore, and Young 1995). The specific outputs of discrete cortical areas emerge from a generic set of intracortical and subcortical connections in mid- to late development, dependent on activity. There are several notable examples of plasticity: the isocortex can represent and transform visual input artificially induced to project to the auditory thalamus (Roe et al. 1990), and also, early transplants of one cortical area to another can accept innervation and make connections characteristic of the new area (O'Leary, Schlaggar, and Stanfield 1992). However, neither the new innervation nor the new connectivity is identical to the unaltered state.

Thus, the isocortex does not fall clearly into either the equipotential or modular description, but rather shows evidence of some early channeling of the epigenetic landscape in the context of a great deal of equipotentiality. When the isocortex first encounters the information of the external world, the separation of modalities and the topographic mapping of surfaces available already in the thalamus is preserved and available. Overall, the evolutionarily conservative primary sensory cortices show the most evidence of early morphological and connectional specialization, while those areas of frontal and parietal cortex that proliferate in the largest brains show the least. The specification of intra cortical connectivity, both short- and long-range, occurs concurrent with early experience, and it seems likely that it is in this circuitry that the isocortex will represent the predictability and variability of the outside world.

References

Cohen-Tannoudji, M., C. Babinet, and M. Wassef. (1994). Early determination of a mouse somatosensory cortex marker. Nature 368:460-463.

Easter, S. S., L. S. Ross, and A. Frankfurter. (1993). Initial tract formation in the mouse brain. Journal of Neuroscience 13:285-299.

Finlay, B. L., and R. B. Darlington. (1995). Linked regularities in the development and evolution of mammalian brains. Science 268:1578-1584.

Goodman, C. (1994). The likeness of being: Phylogenetically conserved molecular mechanisms of growth cone guidance. Cell 78:353-373.

Gould, S. J. (1997). The exaptive excellence of spandrels as a term and prototype. Proc. Natl. Acad. Sci. U.S.A. 94:10750-10755.

Karlstrom, R. O., T. Trowe, S. Klostermann, H. Baier, M. Brand, A. D. Crawford, B. Grunewald, P. Haffter, H. Hoffmann, S. U. Meyer, B. K. Muller, S. Richter, F. J. M. van Eeden, C. Nusslei-Volhard, and F. Bonhoeffer. (1996). Zebrafish mutations affecting retinotectal axon pathfinding. Development 123:427-438.

Katz, L. C., and C. J. Shatz. (1996). Synaptic activity and the construction of cortical circuits. Science 274:1133-1138.

Keynes, R., and R. Krumlauf. (1994). Hox genes and the regionalization of the nervous system. Annual Review of Neuroscience 17:109-132.

Levitt, P., M. F. Barbe, and K. L. Eagleson. (1997). Patterning and specification of the cerebral cortex. Annual Review of Neuroscience 20:1-24.

Lewis, E. B. (1978). A gene complex controlling segmentation in Drosophila.Nature 276:565-570.

O'Leary, D. D. M., B. L. Schlaggar, and B. B. Stanfield. (1992). The specification of sensory cortex: Lessons from cortical transplantation. Experimental Neurology 115:121-126.

Puelles, L., and J. L. R. Rubenstein. (1993). Expression patterns of homeobox and other regulatory genes in the embryonic mouse forebrain suggest a neuromeric organization. Trends in Neurosciences 16:472-479.

Rakic, P. (1995). Radial versus tangential migration of neuronal clones in the developing cerebral cortex. Proc. Nat. Acad. Sci. U. S. A. 92:11323-11327.

Reichert, H., and G. Boyan. (1997). Building a brain: Developmental insights in insects. Trends in Neuroscience 20:258-264.

Roe, A. W., S. L. Pallas, J.-O. Hahm, and M. Sur. (1990). A map of visual space induced in primary auditory cortex. Science 250:818-820.

Scannell, J. W., C. Blakemore, and M. P. Young. (1995). Analysis of connectivity in the cat cerebral cortex. Journal of Neuroscience 15:1463-1483.

Udin, S. B., and J. W. Fawcett. (1988). The formation of topographic maps. Annual Review of Neuroscience 11:289-328.

Wong, R. O. L., M. Meister, and C. J. Shatz. (1993). Transient period of correlated bursting activity during development of the mammalian retina. Neuron 11:923-938.

Zuker, C. S. (1994). On the evolution of eyes: Would you like it simple or compound? Science 265:742-743.

Neural Development

See also

References

Further Readings