In the CNIC, the DSIs of the recorded cells were also highly correlated with their CFs (Figure 2C). Based on the morphology of the cells successfully recovered after juxtacellular labeling or intracellular

labeling, we found that the neurons we recorded have flat-shaped dendrites and soma with diameters of ∼20 um (Figure 2E). It is reasonable to assume that our recording methods selected larger cells in the rat IC (Ito et al., 2009 and Poon et al., 1992). Upward FM sweeps evoked spikes strongly in the neurons with low CF, whereas downward sweeps evoked spikes robustly in the high CF neurons. For neurons showing stronger Baf-A1 mw direction selectivity (with an absolute DSI greater than 0.33), the temporal jitters were also strikingly smaller in the preferred direction than in the null direction (0.65 ±

0.45 ms versus 4.44 ± 3.45 ms [SD]), indicating that the precision of firing in DS neurons is sensitive to direction (Figure 2D). It has been suggested that spike waveforms of excitatory versus inhibitory neurons in the neocortex can be distinguished according to different peak versus trough amplitude click here ratios and peak-to-trough time intervals (Joshi and Hawken, 2006, Niell and Stryker, 2008 and Wu et al., 2008). However, the analysis of all the cells we encountered in the CNIC showed neither a bimodal distribution of peak-to-trough intervals nor a correlation of peak-to-trough intervals and DSI by this strategy (Figure S3B). To test whether the difference in spike precision for the responses to opposing directions is due to coincidental or scattered synaptic inputs or reflects a circuitry mechanism, we next dissected the major excitatory and inhibitory inputs to those DS neurons. To understand synaptic mechanisms underlying direction selectivity in the IC, we performed in vivo whole-cell recordings on identified DS neurons. Most of the previous studies on

the direction selectivity of FM sweeps were based either on analyzing membrane potential changes by current-clamp recordings or measuring synaptic inputs by voltage-clamp recordings (Gittelman et al., 2009, Ye et al., 2010 and Zhang Adenylyl cyclase et al., 2003). The former method cannot reveal neurons’ synaptic inputs directly, while the latter cannot demonstrate whether the output is also direction selective, so we applied both in vivo current-clamp and voltage-clamp whole-cell recordings to the same IC neurons. One of the major challenges of performing high-quality voltage-clamp recordings in the deep brain regions is the long traveling distance of recording electrodes through the brain tissue, which causes significant contamination of the electrode tips (Margrie et al., 2002). We designed a coaxial electrode system for deep brain-region recordings that is driven by separate micromanipulators (Figure S4A and Experimental Procedures). This system prevents electrode contamination by reducing the actual traveling distance of electrodes in the brain tissue.