These data indicate that basic synaptic function matures normally but elimination of redundant CFs is impaired in GAD67+/GFP mice. We then investigated innervation pattern of CFs morphologically by anterograde labeling of CFs with dextran Texas red (DTR) (Figures 2B–2H, red) combined with immunofluorescence for a PC marker, calbindin (Figures 2B–2H, blue or ocher), and a CF terminal

marker, vesicular glutamate transporter type 2 (VGluT2) NSC 683864 clinical trial (Figures 2B–2H, green). In both mice, DTR-labeled CFs precisely followed the PC’s proximal dendrites and climbed up to the four-fifths of the molecular layer (Figures 2B and 2E). In control mice, terminals of DTR-labeled CFs were completely overlapped with VGluT2 immunoreactivity throughout dendritic arbors of each PC, indicating predominant mono-innervation patterns. At PC somata of control mice, VGluT2-positive terminals were rarely observed (Figures 2C and 2D), reflecting dendritic translocation of CFs during development. In contrast, PC somata of GAD67+/GFP mice were often associated with DTR-labeled/VGluT2-positive terminals

and DTR-unlabeled/VGluT2-positive terminals (red and green arrows in Figures 2F1 and 2G1, respectively). In some cases, proximal shaft dendrites were innervated by the two types of CF terminals (red and green SNS 032 arrows in Figure 2H1). These results indicate that reduction of GAD67 leads to incomplete pruning of surplus CFs, resulting in multiple innervation of PCs by CFs. To examine at which stage of postnatal development the impairment occurs in GAD67+/GFP mice,　we followed developmental course of CF innervation from P5 to P20. At P5–P6, just before the onset of CF synapse elimination, all PCs were innervated by four or more CFs in control and GAD67+/GFP mice (Figure 3A) with no significant difference in the frequency distribution of PCs as to the number of CF-EPSC steps (p = 0.635; Figure 3A). At P7–P9, the frequency distribution histograms of PCs were significantly shifted toward smaller numbers from those at

P5–P6, but no statistical significance was observed between the two mouse strains (p = 0.292; Figure 3B). At P10–P12, while nearly 70% of PCs were innervated by one or two CFs in control PAK6 mice, 60% of PCs remained innervated by more than three CFs in GAD67+/GFP mice. Control PCs were innervated by significantly fewer CFs than GAD67+/GFP PCs (p = 0.006; Figure 3C). At P13–P15 and P16–P20, the difference became even larger (p < 0.001). The proportion of PCs innervated by single CFs increased to about 70% in control mice, whereas nearly 70% of PCs remained innervated by multiple CFs in GAD67+/GFP mice (Figures 3D and 3E). To test whether functional differentiation into “strong” and “weak” CFs proceeds normally in GAD67+/GFP mice, we calculated the disparity index and disparity ratio (Hashimoto and Kano, 2003).