Homeostatic plasticity in a reward processing region: accumbens neurons scale too! (Commentary on Sun & Wolf)

Abstract

There is an expanding literature that supports the idea that several aspects of addiction to drugs of abuse are behavioral expressions of pathological forms of learning and memory coded in brain circuits involved in reward processing. At a cellular level, it has now been well demonstrated that several drugs of abuse interfere with synaptic plasticity mechanisms in reward regions and thereby influence their function (Wolf et al., 2004; Kauer & Malenka, 2007). In this context, it is relevant to emphasize that whereas LTP and LTD, the two most studied forms of synaptic plasticity, are widely viewed as cellular models of learning and memory, it is also well appreciated that their implementation in neuronal network models carries an inherent stability problem (Turrigiano, 2008). The phenomenon of homeostatic synaptic scaling (O’Brien et al., 1998; Turrigiano et al., 1998) satisfactorily solves many of the issues related to this problem and has thus triggered considerable interest. This plasticity refers to the bi-directional ability of the overall synaptic strength of a neuron to actively compensate for changes in overall excitability levels, presumably to maintain firing rates within an optimal range (Davis, 2006; Turrigiano, 2008). Although homeostatic scaling can intuitively be expected to be found in all circuits where LTP and LTD are operative, including reward processing circuits, we still have an incomplete understanding of the degree of universality of this mechanism in the brain. Also, if such a mechanism were operant in brain reward circuits, would drugs of abuse affect it? If so, by what mechanisms? The study of Sun & Wolf (2009) published in this issue of EJN provides interesting progress along these lines. To begin addressing these questions, the authors sought to determine whether neurons of the nucleus accumbens in an in vitro culture preparation exhibit the core features of homeostatic synaptic scaling. This nucleus, an intricate part of the brain reward pathway implicated in addiction, is overwhelmingly composed of GABAergic medium spiny projection neurons and receives a strong dopaminergic input from the ventral tegmentum area. Experimentally, this network architecture brings an inherent complication for in vitro culture preparation as this structure does not provide an intrinsic glutamatergic drive required for the establishment of functional networks. To circumvent this issue, the authors used a clever co-culture system where accumbens neurons were co-cultured with prefrontal cortex pyramidal neurons (which provide strong excitatory input to accumbens neurons in situ). Using this system, the authors blocked synaptic activity for prolonged periods of time using pharmacological tools and observed a robust compensatory upregulation of surface expression of AMPA receptors (AMPARs), the hallmark of synaptic scaling. Importantly, this synaptic scaling in accumbens neurons was bi-directional, as increases in network activity lead to a reduction of surface expression of AMPARs. As such, these findings show that accumbens neurons, like cortical and spinal neurons, exhibit robust homeostatic synaptic scaling. The authors further worked out some interesting molecular details of this phenomenon. In an attempt to broadly recapitulate in a culture system the conditions of chronic cocaine use, the authors administered repeated dopamine treatments over several days (which is known to enhance surface expression of AMPARs; Wolf et al., 2004) before the scaling challenge. Intriguingly, the dopamine-induced increase in AMPAR expression occluded that induced by scaling challenge. The presence of an occlusion between two phenomena is usually interpreted as indicating some level of commonality between their expression mechanisms. It is possible that there might be fixed upper and lower limits of synaptic strength that cannot be surpassed by the scaling process, and that would be maximized by the dopamine treatment. The outcome of this interplay on overall network function at this point is unclear, so is its pathological role. Extrapolating a little, would cocaine essentially hijack the scaling process? Would it, in effect, modulate its dynamic range? Its gain? Regardless of these mechanistic minutiae that await clarification, this study of Sun & Wolf (2009) firmly establishes the presence of synaptic scaling in a brain region involved in reward and hints that drugs of abuse, like cocaine, may adversely affect its function. Altogether, just as we are trying to understand the computational features of the interaction between homeostatic synaptic scaling and classic synaptic plasticity that are relevant for information storage, this study shows that this challenge likely equally applies to our quest to understand the neural basis of addiction.