Review
Hebb and homeostasis in neuronal plasticity

https://doi.org/10.1016/S0959-4388(00)00091-XGet rights and content

Abstract

The positive-feedback nature of Hebbian plasticity can destabilize the properties of neuronal networks. Recent work has demonstrated that this destabilizing influence is counteracted by a number of homeostatic plasticity mechanisms that stabilize neuronal activity. Such mechanisms include global changes in synaptic strengths, changes in neuronal excitability, and the regulation of synapse number. These recent studies suggest that Hebbian and homeostatic plasticity often target the same molecular substrates, and have opposing effects on synaptic or neuronal properties. These advances significantly broaden our framework for understanding the effects of activity on synaptic function and neuronal excitability.

Introduction

In the quest to explain how the nervous system encodes information, neuroscientists have uncovered a bewildering array of cellular mechanisms by which experience can modify the properties of neuronal networks. Information transfer across a synapse is a complex process that depends on presynaptic release of neurotransmitter, transduction by postsynaptic receptors, and integration of many synaptic responses into a sequence of action potentials via voltage-gated ion channels. Nearly every phase of this process can exhibit activity-dependent plasticity, and often different experimental protocols produce seemingly contradictory effects on any given parameter of synaptic function. A principle that may help illuminate this contradictory literature is to view plasticity as occurring in two forms that can have diametrically opposite effects: Hebbian, correlation-based mechanisms that progressively modify network properties; and homeostatic mechanisms that promote network stability.

These two forms of plasticity are opposite sides of the same coin. Correlation-based plasticity, such as long-term potentiation (LTP) and long-term depression (LTD), is thought to be crucial for information storage because it produces associative changes in the strength of individual synaptic connections. Such plasticity is prone to instability, however, so LTP and LTD are probably insufficient to explain activity-dependent development and learning. Correlation-based learning rules are unstable because once a synaptic input is potentiated it becomes easier for the presynaptic neuron to depolarize the postsynaptic neuron and make it fire, and this promotes further potentiation of that synapse. In addition, potentiation of some inputs will increase the net excitatory synaptic drive to the postsynaptic neuron, making it easier for other inputs to depolarize the neuron and promoting potentiation of previously ineffective synapses. In order to harness the ability of Hebbian mechanisms to selectively modify synaptic connectivity, there must be additional learning rules that stabilize the properties of neuronal networks.

In principle, a number of mechanisms are capable of stabilizing activity when synapse number and strength are changing dramatically. For example, the cycle of increasing correlation produced by synaptic potentiation would be short-circuited by any mechanism that stabilized postsynaptic firing rates 1, 2. An alternative mechanism would be to raise the threshold for LTP and lower the threshold for LTD as postsynaptic activity rises, so that LTD would be promoted and synaptic strengths would fall again [3]. A wealth of experimental evidence is now beginning to accumulate that suggests that these and other strategies are employed by central networks to maintain stability of network function; in addition, it is becoming clear that most targets of Hebbian plasticity are also regulated in a homeostatic manner. Importantly, both the mechanisms and substrates of these two forms of plasticity share important components, suggesting that they may be inextricably intertwined at the molecular level. In this review, we discuss recent advances in our understanding of homeostatic plasticity in central networks, and its mechanistic and functional relationship to Hebbian plasticity.

Section snippets

Conservation of activity levels in neuronal networks

It has now been established in a number of systems that networks of neurons can adapt to changing activity patterns by altering the level of synaptic transmission or the array of voltage-dependent conductances expressed by component neurons. For example, in both invertebrate central pattern generators and vertebrate spinal networks, pharmacological blockade of rhythmic activity engages compensatory mechanisms that cause activity to resume after a period of hours to days 4, 5, 6. Similarly,

Synaptic scaling

One mechanism that could help maintain relatively constant activity levels is if neurons increased the strength of all excitatory connections in response to a prolonged drop in firing rates, and vice versa. Such bi-directional plasticity of AMPA-mediated glutamatergic synaptic currents has recently been demonstrated in cultured cortical and spinal networks, and occurs through a scaling up or down of the strength of all of a neuron’s excitatory inputs 7, 8. This form of plasticity, termed

Activity and AMPAR trafficking: global and local receptor regulation

While the differences between LTP/LTD and synaptic scaling are profound, there are also interesting similarities. The preponderance of evidence to date suggests that synaptic scaling and at least some forms of LTP/LTD are expressed as changes in the number of postsynaptic AMPARs clustered at synapses. Tetanic stimulation in hippocampal and thalamocortical slices can convert developing synapses in which only NMDA currents are present into synapses in which both NMDA and AMPA currents can be

Activity-dependent regulation of neuronal excitability

Most studies of plasticity underlying learning and development have focused on changes in synaptic strength. But another potential substrate for activity-dependent plasticity is the rich array of voltage-dependent sodium, potassium, and calcium conductances that neurons express. The mixture and distribution of these conductances determines the integrative properties of the postsynaptic neuron, suggesting that if activity could selectively regulate the expression of these conductances, the

Activity-dependent regulation of synapse number

Might some forms of plasticity be expressed as changes in synapse number as well as changes in synapse strength? For both homeostatic and Hebbian forms of plasticity this issue has been controversial, and results from different investigators have varied widely. Prolonged changes in activity in hippocampal cultures, for example, have been reported to selectively modify the number of synaptic sites that express NMDARs but not AMPARs [28], or AMPARs but not NMDARs [29], whereas a recent study

Brain-derived neurotrophic factor and activity-dependent plasticity

The neurotrophin brain-derived neurotrophic factor (BDNF) may soon exceed calcium in the diversity of roles it has been postulated to play in the activity-dependent plasticity of central networks. Acutely, BDNF has been reported to modulate synaptic transmission and LTP 38, 39, 40, 41, 42, 43•• and to directly depolarize postsynaptic neurons 44, 45••, while longer exposures to BDNF regulate dendritic outgrowth [46], synaptic scaling of excitatory inputs [47], and intrinsic neuronal excitability

Conclusions

Evidence is mounting that many properties of central networks can be regulated in a homeostatic manner by long-lasting changes in activity. Recent work suggests that homeostatic plasticity can target both ionotropic glutamate receptors to regulate synaptic strength, and voltage-dependent ion channels to regulate neuronal excitability; it can also modulate the number of synaptic connections that neurons receive. Interestingly, each of these targets of homeostatic plasticity are also thought to

Acknowledgements

Supported by K02NS01893, R01 NS36853, and the Sloan Foundation.

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

References (54)

  • A.K. McAllister et al.

    Neurotrophins regulate dendritic growth in developing visual cortex

    Neuron

    (1995)
  • L.C. Rutherford et al.

    Opposite effects of BDNF on the quantal amplitude of pyramidal and interneuron excitatory synapses

    Neuron

    (1998)
  • G.G. Turrigiano

    Homeostatic plasticity in neuronal networks: the more things change, the more they stay the same

    Trends Neurosci

    (1999)
  • N. Chub et al.

    Blockade and recovery of spontaneous rhythmic activity after application of neurotransmitter antagonists to spinal networks of the chick embryo

    J Neurosci

    (1998)
  • M. Thoby-Brisson et al.

    Neuromodulatory inputs maintain expression of a lobster motor pattern-generating network in a modulation-dependent state: evidence from long-term decentralization in vitro

    J Neurosci

    (1998)
  • J. Golowasch et al.

    Network stability from activity-dependent regulation of neuronal conductances

    Neural Comput

    (1999)
  • G.G. Turrigiano et al.

    Activity-dependent scaling of quantal amplitude in neocortical neurons

    Nature

    (1998)
  • N.S. Desai et al.

    Plasticity in the intrinsic excitability of cortical pyramidal neurons

    Nat Neurosci

    (1999)
  • K.R. Leslie et al.

    Postsynaptic depolarization scales quantal amplitudes in cortical cultures

    Soc Neurosci Abstr

    (1999)
  • D. Liao et al.

    Activation of postsynaptically silent synapses during pairing-induced LTP in CA1 region of hippocampal slice

    Nature

    (1995)
  • S.H. Shi et al.

    Rapid spine delivery and redistribution of AMPA receptors after synaptic NMDA receptor activation

    Science

    (1999)
  • R.C. Carroll et al.

    Rapid redistribution of glutamate receptors contributes to long-term depression in hippocampal cultures

    Nat Neurosci

    (1999)
  • A. Nishimue et al.

    NSF binding to GluR2 regulates synaptic transmission

    Neuron

    (1998)
  • J. Noel et al.

    Surface expression of AMPA receptors in hippocampal neurons is regulated by an NSF-dependent mechanism

    Neuron

    (1999)
  • P.M. Lledo et al.

    Postsynaptic membrane fusion and long-term potentiation

    Science

    (1998)
  • A.J. Watt et al.

    Activity co-regulates quantal AMPA and NMDA currents at neocortical synapses

    Neuron in press.

    (2000)
  • G.G. Turrigiano et al.

    Activity changes the intrinsic properties of cultured neurons

    Science

    (1994)
  • Cited by (554)

    • Nature's learning rule: The Hebbian-LMS algorithm

      2023, Artificial Intelligence in the Age of Neural Networks and Brain Computing, Second Edition
    • Ketamine promotes adaption-induced orientation plasticity and vigorous network changes

      2022, Brain Research
      Citation Excerpt :

      Indeed, in mouse and cat, adaptation fails to change the average strength (p) of connections across the population despite the modification of the links between cell-pairs. During adaptation-dependent plasticity, neuronal circuits are subjected to two opposing mechanisms: the Hebbian mechanism that allows experience to change the properties of microcircuits and thus tends to destabilize the activity of neuronal networks by modifying both synapse number and strength, and another mechanism that stabilizes the total synaptic strength and number in the neuronal assemblies, which restores stability of their basic properties (Turrigiano and Nelson, 2000). It seems that a balance of the two forces is needed because the change following experience dependent-plasticity allows the microcircuits’ refinement with stability to maintain responsiveness to inputs (Shatz, 1990; Turrigiano, 1999).

    View all citing articles on Scopus
    View full text