Surprisingly, restoration of binocular vision is again associated with inhibitory synapse loss, and increases the responsiveness to the previously deprived eye. These dynamics of inhibitory synapse turnover in adult V1 accompanying OD plasticity are very different from what has been described for excitatory synapses (Hofer et al., 2009). Upon MD of adult mice, dendritic spines

on layer 5 pyramidal neurons increase in density, while no changes in spine turnover are observed in layer 2/3 pyramidal cells (Hofer et al., 2009). Recovery of binocular vision several days later does not eliminate the newly formed spines on layer 5 pyramidal neurons, which is thought to leave a structural Selleckchem PLX4032 trace of the first OD shift that expedites a second shift induced later. It is thus possible Bortezomib that the changes in inhibition we observe facilitate the altered responsiveness of layer 2/3 pyramidal neurons without the need for extensive structural changes of their excitatory connections. The fact that deprivation and recovery, and thus a net decrease or increase of visual input, both increase inhibitory synapse loss makes the

interpretation that the changes represent a homeostatic response (Maffei and Turrigiano, 2008) aimed at counteracting the reduced input a less likely explanation for our findings. As has been described previously, inhibitory synapses were present on

shafts and on a minority of dendritic spines, which presumably also carry an excitatory synapse (Beaulieu and Somogyi, 1990 and Jones and Powell, 1969). Upon MD, inhibitory synapses on spines were lost at a much higher rate than shaft synapses. Reopening of the deprived eye caused a renewed increase in inhibitory synapse loss on spines, while it did not significantly affect shaft synapses. Previous studies employing EM have also noticed that inhibitory synapse densities on spines can rapidly change with sensory conditioning, deprivation, or whisker stimulation MRIP (Jasinska et al., 2010, Knott et al., 2002 and Micheva and Beaulieu, 1995), but could not distinguish whether this was due to inhibitory synapse loss or gain on stable spines or to the loss or gain of entire spines also carrying an inhibitory synapse. We show that in naive animals, only a fraction of spines with inhibitory synapses are formed de novo or lost together. The additional loss of inhibitory synapses after MD occurs almost entirely on persistent spines. Large dendritic spines have been found to carry higher efficacy excitatory synapses (Matsuzaki et al., 2001) and are more persistent than small spines (Trachtenberg et al., 2002). We found that in parallel, larger GFP-gephyrin puncta were also more persistent than small ones. This was true for puncta on spines and on shafts.