Adding neurons to the adult mammalian brain.

A widely held dogma in neurobiology tells us that well before our professional careers begin, we can no longer add neurons to our brain-we can only lose them (1). A few exceptions to this deeply entrenched idea have been noted in nonprimate mammals over the past decade. The findings described by Lois and AlvarezBuylla (2) not only suggest another exception, they may lead to a challenge of its validity in primates. The vast majority of neurons in the mammalian brain, and indeed throughout the nervous system, is generated during a well-defined period of fetal or early postnatal life (3). Initially, neurons arise from the ventricular zone, a region of intense cellular proliferation surrounding the brain ventricles. Subsequently, as the volume of the brain increases, additional sites of proliferation are set up by progenitor cells that migrated from the ventricular zone. For example, subventricular proliferative zones appear adjacent to the ventricular zone, and in certain brain regions, such as the olfactory bulb, the hippocampal dentate gyrus, and the cerebellum, sites of neuronal proliferation are established well away from the ventricular lining. As development progresses, cellular proliferation, in the ventricular zone as well as in the extraventricular sites, diminishes greatly or ceases altogether. At this stage in development, the brain contains more neurons than it will ever have again. This is due not only to a lack of subsequent neuronal production but also to a normal phase of development during which about half of the originally generated population of neurons degenerate (3). As in mammals, in nonmammalian vertebrates neuronal proliferation predominantly takes place over a relatively brief, well-defined period of development. However, in many nonmammalian vertebrates, it has been recognized for some time that neurons are normally generated in certain parts of the central nervous system over the entire life of the animal. For example, in amphibians and fish, neurons are continuously added to the retina (4) and its main target in the brain, the optic tectum (5). More recently, Nottebohm and colleagues (6-8) have shown that in adult birds neurons continue to be generated in the forebrain ventricular zone and migrate long distances to populate a number of forebrain structures. Ryals and Rubel (9) have also shown that in adult birds sensory hair cells of the inner ear can be replaced following trauma, a particularly remarkable phenomenon since the proliferation of these peripheral neurons occurs in a normally quiescent population of precursor cells. The most often cited example of neuronal proliferation in adult mammals is the continuous turnover of sensory neurons in the olfactory epithelium (10, 11). However, even in mammals instances of neuronal proliferation in the adult are not limited to the peripheral nervous system. It has been well established that a low level of neuronal proliferation occurs in the brains of adult mammals, but the documented instances are restricted to the olfactory bulb and dentate gyrus of rats and rabbits (12-16). For example, in rats, the principal neuron type in the dentate gyrus, the granule cell, is slowly produced at the base of the granule cell layer and added to the existing population without any measurable loss of older cells (17). This lack of turnover results in a gradual, but substantial, increase in the number of granule cells in the rat dentate gyrus over the animal's life (18, 19). In both mammals and other vertebrates, the newly added neurons develop synaptic connections and often extend axons for appreciable distances (20-22). Is neuronal production in the normal adult brain and subsequent establishment of circuitry important to the animal? In some instances, it clearly is. Songbirds annually exhibit a seasonal loss and then regeneration of the neurons that populate forebrain nuclei involved in song production, a cycle controlled by circulating levels of the hormone testosterone (6). The replacement of the lost neurons is essential for song production and hence mating behavior (22). In other cases, such as the addition of neurons to the dentate gyrus, the benefits are not so clear. The dentate gyrus and its neighbor, the hippocampus, the major synaptic target of granule cell axons, are implicated in processes underlying learning and memory (23). If adding granule cells were particularly advantageous for performing these tasks, one might expect that granule cell addition would be characteristic of all adult mammals, especially those with enhanced cognitive abilities. How-