Trends in Neurosciences
Volume 31, Issue 8, August 2008, Pages 377-383
Journal home page for Trends in Neurosciences

Opinion
Two opposing plasticity mechanisms pulling a single synapse

https://doi.org/10.1016/j.tins.2008.05.005Get rights and content

Homeostatic synaptic plasticity (HSP) has been suggested to act as a negative feedback mechanism responsible for globally and uniformly scaling (up or down) the strength of all synapses in the neuron, in compensation for chronically aberrant (too low or too high) levels of activity. Such global scaling preserves the relative strengths of synapses and thus keeps ‘Hebbian-like’ memory traces (long-term potentiations, LTP, or depressions, LTD). However, new experimental findings demonstrate that HSP can operate locally, controlling each synapse individually. Seemingly, this finding implies that HSP can abolish any modification of synaptic strength (erase LTP/LTD). We propose that dendrites offer an inherent solution to this ‘paradox’ and that in fact local HSP might confer upon the neuron several surprising benefits, which are demonstrated using computer simulations.

Introduction

The function and dynamics of neuronal networks are strongly dependent on the strengths of the synapses connecting neurons in the network. These connections are subject to continuous change and modification by various mechanisms of synaptic plasticity, affording remarkable adaptability and versatility to the neuronal network. Currently known plasticity mechanisms are highly diverse in their dynamics in terms of timescale, direction of change and spatial resolution. Thus, Hebbian long-term potentiation or depression (LTP/LTD) [1] operates at a short timescale of several seconds to several minutes (the time it takes for LTP/LTD to be elicited), in a positive feedback manner (already strong excitatory synapses, which tend to activate the postsynaptic neuron, tend to be further strengthened and vice versa) and at the spatial resolution of almost a single synapse (each synapse is modified individually; but see Ref. [2]). By contrast, homeostatic synaptic plasticity (HSP) 3, 4, 5 is a negative feedback mechanism (persistent overactivation of the neuron results in compensatory weakening of synapses and vice versa) operating at a substantially slower timescale on the order of hours to days. As for the spatial resolution of HSP, it is widely regarded as a global mechanism 3, 4 that proportionally scales all synapses of a given neuron up or down (Figure 1a, top) and can thus maintain the relative strengths of all synaptic inputs into the neuron, conserving the neuron's optimal operating regime while presumably still preserving memory traces (the relative synaptic strengths) 3, 6. Therefore, under HSP, all synapses in a neuron are predicted to change in unison according to a cell-wide global signal.

New experimental results 7, 8, 9, 10, 11 demonstrate, however, that, in certain systems, HSP, like LTP/LTD, is endowed with synapse-specific resolution, individually controlling each synapse's strength according to the local level of activity at the site of the synapse (Figure 1a, bottom). Seemingly, this finding implies that HSP will detrimentally erase any modification of synaptic strength such as produced by LTP or LTD, and thus abolish any change that the neuron's synapses might undergo. We offer a solution to this ‘paradox’ and propose that a synapse-specific HSP mechanism has many advantages over global HSP in regulating neuronal activity, such as supporting dendritic compartmentalization and selecting which spatial patterns of synaptic potentiation and depression will endure and which will be eliminated.

Section snippets

New experimental evidence supporting local HSP

The prevailing experimental paradigm for eliciting HSP has been the exposure of the entire neuron to a chronic uniform global increase or decrease in activity (e.g. by blocking overall excitatory or inhibitory synaptic input with neurotransmitter receptor antagonists; Figure 1b). The typical result obtained in such experiments has been a uniform scaling (up or down) of all synaptic strengths, leading to the conclusion that HSP is a global scaling mechanism [3]. However, under such experimental

The paradox of oblivion and its solution

The possibility of HSP being a local mechanism has been repeatedly rejected on the basis that it would lead to a seeming paradox 3, 15, 17: on the one hand, it is widely assumed that the whole point of (e.g. Hebbian) synaptic plasticity is to introduce long-lived changes in neurons and in the way that they (and the network they are part of) function (Figure 2a). On the other hand, HSP acting locally on specific synapses will presumably counteract and eventually erase any such changes in the

New roles for synapse-specific HSP

What are the potential benefits that a synapse-specific HSP mechanism could offer? The importance of a locally operating HSP mechanism was previously proposed in the context of Hebbian plasticity and local dendritic spikes. It was argued that without local regulation such as could be provided by local HSP, various dendritic subregions would become overdominant owing to the repeating firing of local dendritic spikes at these regions (giving rise to many very strong synapses via a spike

Conclusions

A key principle of the operation of neuronal networks is their ability to continuously and adaptively respond to the environment. One basic means for achieving this is implemented by the synapse – a miraculous device that connects between neurons and that is endowed with a rich collection of activity-dependent plasticity mechanisms. These mechanisms seem to operate at different timescales and spatial resolutions, and might even oppose each other in the direction of their effect. Some operate in

Acknowledgements

This work was supported by a grant donation by the Edmond J. Safra Foundation (‘Learning and Memory’) and by the NIH (IR01-MH59976–01A2) and the Israeli Science Foundation (10/256). I.R. was supported by a Horowitz Fund fellowship.

References (36)

  • P. Poirazi

    Pyramidal neuron as two-layer neural network

    Neuron

    (2003)
  • M. Häusser et al.

    Dendrites: bug or feature?

    Curr. Opin. Neurobiol.

    (2003)
  • F. Engert et al.

    Synapse specificity of long-term potentiation breaks down at short distances

    Nature

    (1997)
  • G.G. Turrigiano et al.

    Homeostatic plasticity in the developing nervous system

    Nat. Rev. Neurosci.

    (2004)
  • G.W. Davis

    Homeostatic control of neural activity: from phenomenology to molecular design

    Annu. Rev. Neurosci.

    (2006)
  • Q. Hou

    Homeostatic regulation of AMPA receptor expression at single hippocampal synapses

    Proc. Natl. Acad. Sci. U. S. A.

    (2008)
  • W. Ju

    Activity-dependent regulation of dendritic synthesis and trafficking of AMPA receptors

    Nat. Neurosci.

    (2004)
  • Cited by (0)

    View full text