Most previous studies that identified clonally related neurons in vivo have used a retroviral labeling method (Walsh and Cepko, 1988 and Luskin et al., 1988), which labels only a handful of cells. To analyze clonally related sister cells, we used a Cre/loxP system (Magavi et al., 2012) in which all the progeny of a single cortical progenitor

(∼600 cells) were labeled. We believe that this complete labeling is important to study the relationship between orientation selectivity and lineage. First, neurons with significant responses and sharp selectivity are relatively rare—approximately 20% of mouse V1 cells (Ohki et al., 2005). By recording from ∼100 sister cells, BMS-354825 molecular weight we could measure functional properties from approximately 20 of these cells and estimate their preferred orientation distribution. If only a small number of sister cells were recorded, the probability of obtaining pairs of such neurons would be extremely low. Second, as previously reported (Ohki et al., 2005 and Kreile et al., 2011), there is often a bias in the distribution of preferred orientations in a local population of neurons in rodent visual cortex. With such

a bias, a small number selleck inhibitor of randomly chosen cells could have a similar orientation just by chance. Thus, analyzing a large number of lineage-related cells allowed us to focus on robustly tuned cells to prove that the distribution of preferred orientation of sister cells was significantly different from the other nearby neurons. Sclareol Contamination from the neuropil signal (Göbel and Helmchen, 2007) is another variable that could potentially confound the analysis of response selectivity. As described previously (Kerlin et al., 2010), the orientation tuning of the neuropil signal is similar to the average of the orientation tuning of local neurons and varies only slightly across 300 μm in the imaging field. When there is some bias in the preferred orientations of local neurons, the neuropil signal can be also tuned to the local

orientation bias and might contaminate signals from cell bodies. Since neuropil contamination becomes a larger part of the signal for weakly responsive cells, those cells may appear more similarly tuned if neuropil contamination remains. To avoid these effects of neuropil contamination, we subtracted the surrounding neuropil signal and selected only highly responsive neurons that were sharply tuned. Finally, we used pixel-based orientation maps (Figure 2C) to confirm that we used only cells with reliable responses, as sharply selective and highly responsive neurons clearly stand out in these orientation maps and have different responses than the surrounding neuropil. In some cases, we did not observe a difference in the distribution of preferred orientations between F+ and F− cells and the peaks of the distributions matched between F+ and F− cells.