Analysis of the correlation coefficients of ensemble responses at different time points with the initial activity patterns (time average of 0.75–1.25 s Obeticholic Acid after odor

onset) confirmed that ensemble activity patterns became more decorrelated over time in the awake versus anesthetized state (Figure 2G). We performed principal component analysis to explore how the different temporal dynamics of odor responses in the awake and anesthetized state contributed to the ability of mitral cell ensembles to distinguish different odors over time. Representations of odor-evoked activity as a function of time in principal component space revealed that ensemble activity patterns for different odors were more separated in the awake state, and the separation in the awake state significantly improved over time (Figure 2H). This temporal improvement of odor classification in the awake state was confirmed when we considered the fractions of correctly classified trials, using responses in increasing time windows. Considering the numbers of

responses that gave the same levels of correct classification in awake and anesthetized states during the first second of odor stimulus (awake: 15 responses; SNS-032 mouse anesthetized: 25), we find that odor classification improved over the course of odor stimulation more strongly in the awake state (Figure 2I). Thus, the improvement of classification efficiency in the awake state is partly due to the odor-specific temporal dynamics. Taken together, these results indicate that odor representations in the awake state are more sparse, temporally dynamic, and efficient, compared to anesthetized brain states. Granule cells the are a major class of GABAergic interneurons in the olfactory bulb that mediate mitral cell recurrent and lateral inhibition via dendrodendritic synapses (Isaacson and Strowbridge, 1998; Schoppa et al., 1998; Yokoi et al., 1995). Therefore, we next considered the possibility

that differences in mitral cell odor representations between awake and anesthetized states could reflect differences in granule cell activity. We expressed GCaMP3 by injecting a nonconditional viral vector in the olfactory bulb granule cell layer of wild-type mice. Several weeks after injection, dense sets of neurons in the granule cell layer were visible through a cranial window (Figure 3A). While the vast majority of neurons in this layer are granule cells, our sample probably contains a small fraction of short axon cells, a heterogeneous class of interneurons in the granule cell layer (Eyre et al., 2008; Pressler and Strowbridge, 2006). In awake mice, a large fraction of granule cells showed spontaneous increases in GCaMP fluorescence that are sometimes temporally correlated to one another (Figure 3B, top). Similar to mitral cells, odors activated ensembles of granule cells in awake mice (Figures 3C and 3D).