Figure 1. Firing activity of PV and SST interneurons during cortical up and down states in vivo.

(a) Fluorescence image showing a glass pipette (dotted white line) used for juxtasomal recordings from a PV positive interneuron (red cell) in a PVxTdTomato bigenic transgenic animal. (b) Representative traces of simultaneous LFP (top) and juxtasomal (bottom) recordings from an identified PV interneuron during up and down states. Pink and purple colours in the background indicate up and down states that were identified from the LFP signal, respectively. The white background colour indicates indeterminate states (see Materials and methods). (c) Action potentials fired by PV cells in the three identified periods (up, down and indeterminate, p = 7E-8, one-way ANOVA, N = 16 cells from five animals). In this as well in other figures: grey dots and lines indicate single experiment; black dots and lines indicate the average value represented as mean ± s.e.m; n.s., p>0.05; *p<0.05; **p<0.01; ***p<0.001. (d) Percentage of active up states and active down states (p = 3E-16, unpaired Student’s t-test, N = 16 cells from five animals). (e) Number of spikes fired by PV cells per single up or down state (p = 8E-6, Mann-Whitney test, N = 16 cells from five animals). (f–g) Same as in a, b but in SSTxTdTomato bigenic transgenic animals. (h–j) Same as in c-e for SST interneurons. In h, p = 2E-8, Friedman test, N = 19 cells from seven animals. In i, p = 1E-7, Mann-Whitney test, N = 19 cells from seven animals. In j, p = 9E-3, Mann-Whitney test, N = 19 cells from seven animals.

DOI: http://dx.doi.org/10.7554/eLife.26177.002

Figure 1—figure supplement 1. Up and down state detection from the LFP signal.

(a–c) Representative example of up/down state detection results. Up/down state periods (pink and purple shadows, respectively) as detected from membrane potential are shown in a, whereas states detected in the LFP signal (and based on the decision variable trace shown in c) are reported in b. In c, the up state threshold is marked by the dotted pink line, whereas the down state threshold by the purple dotted line. (d) Ratio between power spectral densities (PSD) of LFP in up/down states (representative cell in upper panel, mean ± s.d. of all cells in lower panel). The light blue square indicates the 10–40 Hz range used in our algorithm. (e–g) ROC curves for the detection of up (pink) and down (purple) states using either S_delta or S_comb are reported in e (representative cell). Up/down state ROC curves are shown either in the top left or in the bottom right half of the xy plane, respectively. The area under ROC curve (AUC) using different frequency bands for the computation of S_delta is shown in f. The AUC for the various metrics (i.e. S_delta or S_comb) or phase computation methods (i.e. Hilbert transform or interpolation) are shown in g. Red rectangles highlight the combinations of parameters giving the maximum performance. (h–i) Statistical distributions of instantaneous phase during up/down states (up in h, down in i, black curves), state onset (in green), and state offsets (in red). Upper panels refer to state detection based on membrane potential, lower panels to state detection based on LFP. (j) True positive (TPR) and false positive rates (FPR) obtained by detecting up/down states on S_comb traces ([0, 1]-[1, 3] Hz frequency bands, instantaneous phase computed by Hilbert transform, and thresholds set as described in Materials and methods).

DOI: http://dx.doi.org/10.7554/eLife.26177.004