Cocaine triggered AMPA receptor redistribution is reversed in vivo by mGluR-dependent long-term depression Camilla Bellone 1 & Christian Lu ¨scher 1,2 Drugs of abuse induce long-lasting changes in neural circuits that may underlie core components of addiction. Here we focus on glutamatergic synapses onto dopamine (DA) neurons of the ventral tegmental area (VTA). Using an ‘ex vivo’ approach in mice, we show that a single injection of cocaine caused strong rectification and conferred sensitivity to the polyamine Joro spider toxin (JST) of AMPAR-mediated excitatory postsynaptic currents (AMPAR EPSCs), indicating the recruitment of receptors that lack GluR2. This qualitative change in transmission was paralleled by an increase in the AMPAR:NMDAR ratio and was prevented by interfering with the protein interacting with C kinase-1 (PICK1) in vivo. Activation of metabotropic glutamate receptors (mGluR1s) by intraperitoneal injection of a positive modulator depotentiated synapses and abolished rectification in slices of cocaine-treated mice, revealing a mechanism to reverse cocaine-induced synaptic plasticity in vivo. All addictive drugs increase dopamine concentrations in target struc- tures of the mesocorticolimbic projections 1,2 . Therefore, plasticity at glutamatergic afferents onto DA neuron of the VTA could participate in the incentive sensitization process thought to underlie addiction 3 . It is well established that the VTA is the locus of the induction of behavioral sensitization and contributes to conditioned place preference, both of which are models of core components of addiction. The tetanic stimulation of glutamatergic afferents onto DA neurons of the VTA can induce sensitization in drug-naı ¨ve rats 4 . On the basis of this finding, the synaptic plasticity of glutamatergic transmission onto DA neurons has been implicated as being the underlying cellular substrate 5 . In fact, glutamatergic synapses onto DA neurons undergo several forms of plasticity. For example, these synapses can express NMDAR- dependent long-term potentiation (LTP) 6 . However, NMDAR- dependent LTP at these synapses is of rather modest magnitude 7,8 . In contrast, in response to a single dose of cocaine 8 and several other addictive drugs 9 , the AMPAR:NMDAR ratio at excitatory synapses of the VTA strongly increases for up to 10 days 10 . Furthermore, repeated cocaine exposure over 5 days causes a loss of GABAergic inhibition that favors the induction of LTP (ref. 7). At first, these two observations seem contradictory, as an increase in the AMPAR:NMDAR ratio during cocaine administration may reflect AMPAR insertion associated with synaptic potentiation that would occlude further induction of LTP (ref. 11). In fact, using a pairing protocol to induce LTP, occlusion is observed after a single injection of cocaine 8 . This observation was recently challenged: in a spike timing–dependent plasticity protocol, and in the presence of GABA A receptor blockers, LTP was successfully induced in cocaine-treated rats 7 . Taken together, cocaine seems to drive a form of plasticity that shares some features with synaptic LTP (increase in the AMPAR:NMDAR ratio) but does not occlude further induction of LTP, which suggests that the two forms of plasticity have distinct mechanisms. A possible scenario that could account for both observations is the insertion of GluR2-lacking, calcium (Ca)-permeable AMPARs. The possibility of a switch to Ca-permeable AMPARs at this synapse after in vivo cocaine exposure is further supported by a number of indirect pieces of evidence from studies in rodents. The observation that cocaine exposure increases the expression of GluR1 in DA neurons of the VTA suggests the involvement of Ca-permeable AMPARs (refs. 12,13). Furthermore, whereas the overexpression of the AMPAR subunit GluR1 is sufficient to elicit sensitization in drug-naı ¨ve rats, this behavior is reduced when GluR2 is overexpressed 12 . Finally, in mice lacking GluR1, the context-dependent associative components of sensitization and conditioned place preference are abolished 14 . Here, in acute slices of the VTA, we directly tested the hypothesis that a single exposure to cocaine drives the synaptic insertion of AMPARs that lack GluR2. Such AMPARs, in addition to their calcium perme- ability, also show strongly rectifying synaptic responses (that is, currents at positive potentials are smaller than currents at symmetrical negative potentials) and are sensitive to polyamine toxins 15 ; therefore, their presence can be detected in acute brain slices using an ex vivo approach—namely, comparing AMPAR EPSCs of cocaine-treated mice to those of saline-treated or naı ¨ve mice. Our data indicated that the cocaine-evoked plasticity of excitatory synapses onto DA neurons of the VTA (refs. 8,9,14,16) involves a switch to synaptic AMPARs that lack GluR2. Received 1 February; accepted 13 March; published online 2 April 2006; doi:10.1038/nn1682 1 Department of Basic Neurosciences and 2 Clinic of Neurology, University of Geneva, 1 Michel Servet, CH-1211 Geneva, Switzerland. Correspondence should be addressed to C.L. (Christian.Luscher@medecine.unige.ch). 636 VOLUME 9 [ NUMBER 5 [ MAY 2006 NATURE NEUROSCIENCE ARTICLES © 2006 Nature Publishing Group http://www.nature.com/natureneuroscience