Synaptic gain control and homeostasis
Introduction
Neural circuits and their elements adapt to changes in their environment. Over short time scales, adaptation serves to increase information transfer by neurons [1]. For example, if the response of a sensory neuron consistently falls outside of its dynamic range, the gain (the magnitude of the neuronal response for a given stimulus) is adjusted to bring the response back to the appropriate dynamic range [1]. For central neurons whose inputs arrive through synapses with other neurons, a major component of this adaptation occurs at the synapses themselves. Depending on the time scale of these changes, different terms such as plasticity, potentiation, augmentation or depression are used [2]. In effect, however, they all serve to control the gain of the synapse.
Short-term synaptic adaptation — over a time scale of milliseconds to seconds — has been related to neuronal function in a few cases 3., 4., 5., and is largely mediated by presynaptic changes [2]. Longer-term changes in synaptic strength also occur over a time scale of hours to days. Historically, much attention has been paid to Hebbian plasticity, in which presynaptic and postsynaptic activities locally interact at individual synapses to induce lasting changes in strength 6., 7.. Hebbian plasticity, under some conditions, risks runaway strengthening or weakening of synapses, which leads to a saturation of synaptic strength [8]. In effect, this erodes the dynamic range of a neuron, and is akin to setting the gain of a sensory neuron too low or too high. To avoid saturation, neurons are thought to have a way of re-establishing normal synaptic strengths while maintaining the relative strengths of all synapses. Such global renormalization of synaptic gain has been described in terms of homeostatic synaptic plasticity and has gained experimental support 9., 10., 11.. Below, we discuss the exact meaning of homeostatic plasticity, and review some of the recent experimental findings. For the sake of brevity and clarity, we confine ourselves mainly to mammalian systems. We also focus on synaptic homeostasis, leaving out other forms of neuronal compensation [9].
Section snippets
Homeostasis and synaptic adaptation
The concept of homeostasis arose in physiology some 70 years ago, the term first being coined by Cannon in 1932 [12]. Certain physiological parameters, core temperature and blood pressure for example, must be maintained within a narrow range — in fact, a range so narrow that it becomes a set point — for the survival of an organism. Environmental (or internal) changes that push the biological system away from its equilibrium also trigger compensatory changes that return the system to its set
Global or local rules?
A crucial question regarding homeostatic plasticity is whether or not such regulation of synaptic gain requires an explicit global or cell-wide renormalizing mechanism. Alternatively, local rules for synaptic change might be sufficient to account for homeostatic regulation. For example, the Bienenstock-Cooper-Munro sliding threshold model for synaptic plasticity can predict stability of neuronal activity in the face of Hebbian modification of synapses 15., 16.. In this model, as synapses get
What is altered when you push the envelope?
Mechanistic studies of homeostatic plasticity have typically relied on pushing activity levels to extremes — either a complete block or very high activity levels 10., 11.. These studies have found that there are different ways in which neurons adjust their synapses to adapt to these manipulations (Figure 2). In cortical cultures, chronic silencing of all neurons with tetrodotoxin (TTX) or with glutamate receptor antagonists results in a larger quantal amplitude, with little effect on the number
Mechanisms in homeostatic plasticity
What activity-related variable is sensed in synaptic homeostasis? Some obvious candidates are membrane depolarization, a measure of spike activity (mean rate or peak rate, for example), and average calcium influx. All of these variables are closely related and it may prove difficult to tease them apart.
At the Drosophila neuromuscular junction, homeostatic plasticity can be triggered by selectively hyperpolarizing individual muscle fibers, which suggests that altered membrane depolarization is
Single neurons versus populations
Investigation of homeostatic plasticity has typically involved drastic alteration of activity in large populations of neurons [10]. What happens when activity is manipulated more selectively in a small number of neurons? To address this, the inwardly rectifying potassium channel Kir2.1 was used to reduce activity selectively in individual neurons in an otherwise normal network [29••]. In line with previous expectations, homeostatic plasticity was observed when suppression began after synapses
Homeostatic synaptic plasticity in the brain
Is there evidence for homeostatic regulation of synapses in the brain? Activity appears to have little or no effect on the initial formation of synapses in most regions of the brain [49]. Although dispensable for the initial formation of synapses, activity is necessary for sculpting precise circuits by means of selective synapse elimination and consolidation.
In many regions of the brain, the number of release sites terminating on a given postsynaptic neuron increases during development, even if
Concluding remarks
There is now much evidence to suggest that when neuronal activity strays from a normal physiological range synaptic gain can be altered to regain normal activity. How exactly this is achieved is still not fully understood. In fact, there appears to be multiple ways to alter synaptic gain in the direction of homeostasis, including changes in pre- and postsynaptic properties, and in the number of synapses. It remains to be determined if these different ways of achieving synaptic homeostasis have
Update
Several additional studies of relevance have been published after the initial submission of this review. At the Drosophila neuromuscular junction, where homeostatic plasticity has been characterized extensively, Goodman and colleagues have found that CaMKII in the muscle is necessary to initiate retrograde signals that cause changes in the presynaptic boutons [65]. Roles for CaMKII isoforms, both α and β, in synaptic plasticity have also been suggested by recent studies in mammalian neurons
References and recommended reading
Papers of particular interest, published within the annual period of review, have been highlighted as:
- •
of special interest
- ••
of outstanding interest
Acknowledgements
We thank the members of our laboratory, particularly W Tyler, for insightful discussion. Research in our laboratory is supported by the National Institutes of Health, the National Science Foundation, the Pew Scholars Program, National Alliance for Research on Schizophrenia and Depression, the EJLB Foundation and the Klingenstein Fund in Neurosciences.
References (68)
Synaptic economics: competition and cooperation in synaptic plasticity
Neuron
(1996)- et al.
Hebb and homeostasis in neuronal plasticity
Curr Opin Neurobiol
(2000) - et al.
Cortical development and remapping through spike timing-dependent plasticity
Neuron
(2001) - et al.
Activity co-regulates quantal AMPA and NMDA currents at neocortical synapses
Neuron
(2000) - et al.
BDNF has opposite effects on the quantal amplitude of pyramidal neuron and interneuron excitatory synapses
Neuron
(1998) Reinsertion or degradation of AMPA receptors determined by activity-dependent endocytic sorting
Neuron
(2000)- et al.
Activity-dependent modulation of synaptic AMPA receptor accumulation
Neuron
(1998) - et al.
Homeostatic plasticity induced by chronic block of AMPA/kainate receptors modulates the generation of rhythmic bursting in rat spinal cord organotypic cultures
Eur J Neurosci
(2001) - et al.
Inactivity produces increases in neurotransmitter release and synapse size
Neuron
(2001) Forward and reverse genetic approaches to synaptogenesis
Curr Opin Neurobiol
(1998)