(Figure 1D). However, mEPSC amplitudes of the remaining synapses were normal in these mice (Figure 1E). These data suggest that PSD-95 is important for new synapse formation after EO, but in the absence of PSD-95, remaining synapses are still able to add glutamate receptors and be potentiated. To identify

the dendritic location of the EO-dependent synaptic plasticity, we examined the pattern of retinal and cortical afferent arborization on the dendrites of DOV neurons by anterograde labeling of retinal and VC afferents in eGFP mice at P29-30 when the number of high expressing cells was greatest (Figure 2A and Figure S2). VC axons were most dense in the deeper portion of the sSC, whereas contralateral retinal axons preferentially occupied more superficial positions (Figures 2B–2D). All eGFP-labeled DOV neurons were reproducibly located in the transition PARP phosphorylation zone between the two projections, and electrical stimulation of retinal and cortical axons where they enter together in the brachium of the

sSC could evoke unitary postsynaptic responses in these neurons (data not shown). We identified the potential locus of contact of retinal and cortical axons onto these neurons, by analyzing the colocalization of each afferent population and the eGFP-labeled Doxorubicin in vivo dendritic arbor (Figures 2E–2G). The degree of chance overlap was estimated by rotation of the images containing the particular afferent label 90° in the plane of section with respect to the images containing the GFP label (Supplemental Experimental Procedures). The mean size and number of overlapped pixel clusters in rotated images was significantly smaller than control

images. This is demonstrated in Figure 2E (arrowhead), where a portion of a labeled retinal axon is observed to course alongside a length of eGFP dendrite. DOV dendrites are highly branched with a variable dendritic branching structure and did not conform to standard definitions of secondary or tertiary branches used in traditional classification schemes for pyramidal neurons. To examine the relationship of these afferent projections to dendritic structure we subdivided the arbor regions ADP ribosylation factor by caliber at each branch point with successively thinner segments ranked from 1 to 4. Three-dimensional neurolucida reconstructions of two neurons, including the labeled neuron in Figure 2, illustrate the distribution of ranked segments (Figure 2H) (see Supplemental Experimental Procedures for classification details). Dendrite caliber rank is a significant factor affecting the distribution of potential contact points for both retinal (F statistic = 11.58, n = 43, p < 0.05) and cortical (F = 3.57, n = 37, p < 0.0001) axons (Figure 2I, significance between calibers assessed with Games-Howell post hoc).