Psychiatric Times 
October 2005
Vol. XXII
Issue 11
The ability to adapt to changing environmental conditions is the most fundamental feature of the nervous system. At the neurobiological level, this adaptation corresponds to neuronal plasticity, defined as the lifelong ability of the brain to reorganize neural pathways based on new experiences. Reorganization is robust during early postnatal development and declines thereafter, but experiences shape neuronal networks to a more limited extent in the adult brain as well (Berardi et al., 2000). Plasticity may involve neurogenesis and neuronal death, formation or pruning of neuronal connections, or changes in the strength of existing connections. Data from the last decade have suggested that abnormalities in the development and information processing in the neuronal networks involved in emotional processes may at least partially underlie mood disorders (Castrén, 2005; Charney and Manji, 2004; Duman et al., 1997; Nestler et al., 2002). Therefore, plasticity of neuronal networks may be a necessary component in successful antidepressant treatments, including pharmacological and psychological therapies. Introduced here will be the principles of neuronal development and activity-dependent neuronal plasticity. The recent data suggesting the role of these processes in mood disorders and antidepressant action will be reviewed. Evidence supporting and contradicting the role of neuronal plasticity in depression has recently been reviewed elsewhere (Castrén, 2005).
Information in the brain is believed to be encoded in vast and overlapping networks of interconnected neurons. Since there are more neurons in the brain than there are DNA base pairs in the human genome, and each neuron makes a connection with hundreds of other neurons, it is obvious that the genomic information cannot code for more than a basic layout of neuronal connections in the brain. The fine-tuning of neuronal connectivity is achieved through neuronal plasticity, a process that depends on neuronal activity and environmental experiences (Katz and Shatz, 1996).
The Figure depicts the steps in the formation and refinement of neuronal networks. (Due to copyright concerns, the Figure cannot be reproduced online. Please see p44 of the print edition--Ed.) Network formation begins with "trial contacts" between individual neurons (Figure, top panel a) (Hua and Smith, 2004). These initial contacts are mainly thought to be formed stochastically, though it is probable that they are to some extent guided by genetic information. The fate of these trial contacts depends on how effectively they mediate relevant information. Active neurons release a neurotransmitter that, in turn, stimulates the release of a neurotrophic factor from the postsynaptic side (Figure, middle panel a). The bidirectional "discussion" over the synapse mediated by the neurotransmitter and neurotrophic factor leads to the stabilization of an active synapse, whereas inactive trial synapses, which cannot stimulate the release of the neurotrophic factor, get pruned away (Figure, lower panel a). In this way, those contacts that best mediate information get stabilized as new synapses (Cohen-Cory, 2002; Hua and Smith, 2004). Therefore, the connectivity and shape of a dendritic tree of an individual neuron are not determined genetically, but form in interaction with the environment, exactly as the branching and outline of a tree is shaped by the availability of light.
If two neurons innervate the same target neuron and are active at the same time (Figure, middle panel b), they can cooperate and get stabilized with a level of activity that is lower than that required for a single neuron, whereas neurons active off-beat are weakened and eliminated (Figure, lower panel b) (Poo, 2001; Thoenen, 1995). This is the basis for the formation of networks of coherently active neurons: "Neurons that fire together, wire together" (Katz and Shatz, 1996). Since neuronal activity reflects environmental stimuli, the activity-dependent neuronal plasticity gradually tunes the networks to optimally code for environmental information.
The critical process is not the formation of neurons or synaptic connections, which are made in excess, but the selection of the useful neurons and connections and the elimination of inactive ones, a process that optimizes signal-to-noise ratio. This could be likened to tuning a radio: Increasing volume (production of more neurons or synapses) does not help if the radio is tuned between stations, but optimal tuning (stabilization of active synapses and elimination of inactive ones) makes the broadcast intelligible even at low volume.
A well-characterized example of an activity-dependent network formation is the development of cortical maps in the sensory cortical areas to reflect their environmental input (Katz and Shatz, 1996). For example, if one of the eyes is covered during a critical period of postnatal development, the projections originating from this eye are inactive and get pruned in the visual cortex, which gradually gets almost exclusively innervated by connections originating from the open, active eye. If the covered eye is not opened before the end of the critical period, the dominance of the open eye gets stabilized and the covered eye remains amblyopic, even if it is opened in adulthood.