Consequently, projection neurons respond to excitatory odors differently;

TCs readily increase firing rate while MCs additionally show graded phase-advance (Figure 7). In this study, identifying the key morphological features (Mori et al., 1983) such as the soma and dendritic position was essential in elucidating functional differences clearly. Other means of distinguishing projection neurons in the olfactory TSA HDAC supplier bulb such as the depth of recording might be correlated with cell types and thus potentially show similar trends; so far, however, such attempts have failed to distinguish classes of neurons not overlapping in functional measures, such as the sniff locking. One reason for this could be that larger tufted cells such as deep tufted cells (e.g., cells

12 and 13 in Figure S2) are easily confused MS-275 price with MCs if recording depth were the sole measure of identification. More data will be needed to extend this analysis to potential subgroups of TCs, such as superficial, middle, or deep TCs. Our data so far showed no tendency for further distinction in phase locking (Figure 2K). Our method described here to identify MCs and TCs based on the sniff locking would help with further investigation of how the two populations may differ, even in cases where morphology is unavailable. Importantly, this is likely to extend to the awake state, where we observed similar strength and diversity of phase preference (Figure S1). We have investigated the mechanistic basis of the observed phase locking using a newly developed modeling approach that generates a large number of models with randomly chosen connectivity and selects for those that are consistent with the experimental data. The simplicity

of the network models has made it possible to sample a vast fraction of connectivity space. This allowed us to extract features of the network that correlate with phase properties consistent with experimental data. While we do not claim that we unequivocally found the actual connectivity implemented in the olfactory bulb, several robust features emerged from this selection procedure. The first observation is the strong feed-forward inhibition, specifically strong PGo-MC connectivity (Figure 6C), which may underlie the GABAergic component crucial in separating MC activation away from the TC activation. Second, the models suggest that Rolziracetam MCs are predominantly driven weakly or indirectly and shaped by inhibition. The robustness of the TC phase in turn points toward OSN inputs strongly and directly exciting TCs. A number of recent investigations suggest that the excitatory pathway to MCs from OSNs is rather indirect (Gire and Schoppa, 2009; Najac et al., 2011; Gire et al., 2012). It is exciting to note the consistency of our modeling results with this view. However we cannot exclude an important role of direct transmission from OSNs to MCs, for example by glutamate spillover onto MCs at higher input strengths (Najac et al., 2011).