Journal: bioRxiv
Article Title: Minute-scale oscillatory sequences in medial entorhinal cortex
doi: 10.1101/2022.05.02.490273
Figure Lengend Snippet: a. Schematic of calcium activity merging steps. We began by sorting the neurons according to the PCA method. Next, in successive iterations, or merging steps, we added up the calcium activity of pairs of consecutive neurons (merging step = 1) or consecutive ensembles (merging step > 1). b. Participation index (PI) as a function of merging step (mean ± S.D.). Black trace, example session in ; red trace, all 15 oscillatory sessions. The more neurons per ensemble, the higher the participation index of the ensemble. Note that the participation index plateaus after 5 merging steps, which corresponds to approximately 10 ensembles (Wilcoxon rank-sum test to compare the participation indexes in merging steps 5 and 6; Black trace: n =30 PIs in merging step 5, n =15 PIs in merging step 6, p = 0.23, Z = 1.20; Red trace: n =15 PIs in merging step 5 and 6, PIs of each merging step were averaged for each session separately, p = 0.14, Z = 1.49). c. Tuning of single cell calcium activity to ensemble activity calculated as the Pearson correlation between the calcium activity of each neuron and the activity of each ensemble for the example session in . Ensemble activity was calculated as the mean calcium activity across neurons in the ensemble. Each row is the tuning curve of one neuron, and neurons are sorted according to the PCA method. For each neuron, the calcium activity was positively correlated with a small subset of consecutive ensembles, and negatively correlated with the others. Pearson correlation is color-coded. d. The relationship between the calcium activity of each neuron and the activity of each ensemble was expressed by a Pearson correlation, as in (c). By repeating this calculation for all neurons across all ensembles, we could identify, for each neuron, the most representative ensemble (the one with maximal Pearson correlation). Left: 2D histogram of the most representative ensemble of each neuron and the ensemble it was assigned to based on the PCA sorting. Data are for the example session in . Each count is a neuron; counts are color-coded (484 cells). Right: The same 2D histogram calculated on one shuffled realization of the data for the example session in (484 cells). In the left diagram, note that the method for assigning cells into ensembles based on the PCA sorting correctly recovers the dependency between cells’ calcium activity and ensemble activity (higher number of counts along the diagonal). e. Same as (d), but for all neurons across all 15 oscillatory sessions (left, n = 6231), or one shuffled realization of the data (right, n = 6231). f. Probability distribution showing, for recorded data and shuffled data, the distance, in numbers of ensembles, between the assigned ensemble based on the PCA sorting and the most representative ensemble (as in d). The probability was calculated as the number of times that one given distance was observed in one session divided by the total number of recorded cells. Each count was one neuron. Note that the distance between the most representative ensemble and the assigned ensemble based on the PCA sorting reflects the periodic boundary conditions in ensemble activation and ranges from 0 to 5 ( x axis). 500 shuffled realizations per session were averaged and compared to the mean distance per session in the recorded data. The probability of finding small distances (lower than 2) was larger in the recorded data ( n = 15 sessions, for distances of 0 to 5 ensemble: p ≤ 3.4 × 10 −6 , range of Z : 4.64 to 4.67; Wilcoxon rank-sum test), suggesting that single cell calcium activity was maximally correlated with the activity of the ensemble it was assigned to. Blue, recorded data; orange, shuffled data. Error bars indicate S.E.M. g. Ensemble activity oscillated at the same frequency as the population oscillation. Ensemble activity was calculated as the mean calcium activity across neurons in the ensemble. Power spectral density was calculated on the activity of each of the ten ensembles from the example session in . Ensemble frequency was calculated as the peak frequency of the PSD, population oscillation frequency was computed as the total number of cycles (24 in this session) normalized by the amount of time in which the network engaged in the oscillation (~3600 s). The dashed line indicates the frequency of the population oscillation. Note that the dashed lines coincide with the peak of the PSD. h. Histogram showing the ratio between ensemble oscillatory frequency and population oscillation frequency in the session (calculated as in panel g; n = 150 data points given by 10 ensembles in each of the 15 oscillatory sessions). Each count is one ensemble. Note the two peaks at 1 and 2, indicating that ensembles tend to oscillate at the frequency of the population oscillation, or at an integer multiple of it. i. Anatomical distribution of recorded neurons for the example session in . The ensemble each neuron has been assigned to based on the PCA sorting is color-coded. Neurons indicated in red were not locked to the phase of the oscillation. Note that ensembles are anatomically intermingled. Dorsal MEC on top, medial on the right, as in . j. Box plot of pairwise anatomical distance between neurons within an ensemble and between those neurons and the rest of the imaged neurons, i.e. across ensembles. Data are shown for each ensemble of the session in (i) (Wilcoxon rank-sum test to compare the within and across group distances for each ensemble separately; n = 1125 pairwise distances in the within ensemble group, except for ensemble 10, in which n = 1326; n = 20928 pairwise distances in the across ensemble group, except for ensemble 10, in which n = 22464, 0.0005 ≤ p ≤ 0.9528, 0.06 ≤ Z ≤ 3.50). Symbols as in . Purple, distances between cells within one ensemble; green, distances between cells in different ensembles. k. Box plots of pairwise anatomical distance between neurons within one ensemble and across ensembles for the example session in (j) (left, n = 10 ensembles, p = 0.57, Z = 0.57, Wilcoxon rank-sum test) and across 15 oscillatory sessions including the example session in (j) (right). For each session the means for each of the “within” and “across” groups were computed across ensembles ( n = 15 oscillatory sessions, p = 0.93, Z = 0.08, Wilcoxon rank-sum test). Symbols as in (j). l. To quantify the temporal progression of the population activity at the time scale at which the population oscillation evolved, we calculated, for each session, an oscillation bin size. This bin size is proportional to the inverse of the peak frequency of the PSD calculated on the phase of the oscillation, and hence captures the time scale at which the oscillation progresses. The oscillation bin size is shown for each of the 15 oscillatory sessions. m. Schematic of the method for quantifying temporal dynamics of ensemble activity. For each session and each ensemble we calculated the mean ensemble activity at each time bin (oscillation bin size). Only the ensemble with the highest activity within each time bin (red rectangle) was considered. The number of transitions between ensembles in adjacent time bins divided by the total number of transitions was used to calculate the transition matrices in . n. The ensemble with the highest activity in each time bin, indicated in yellow and calculated as in (m), plotted as a function of time for the example session in . All other ensembles are indicated in purple. Notice that the transformation in (m) preserves the population oscillation. o. Box plot showing transition probabilities between consecutive ensembles for all 15 oscillatory sessions. The probabilities remain approximately constant across transitions between ensemble pairs ( n = 15 oscillatory sessions per transition, p = 0.56, χ 2 = 7.77, Friedman test), and there were no significant differences between pairs of transitions (Wilcoxon rank-sum test with Bonferroni correction, p > 0.05 for all transitions). Symbols as in . p. We further visualized the structure of the transitions in by using the transition matrix as an adjacency matrix to build a directed weighted graph. Nodes indicate ensembles (color-coded as in m). Edges (lines) between any two nodes represent the transition probabilities between any two ensembles. The thickness of the edge is proportional to the value of the transition probability, while the arrows on each edge indicate the directionality of the transition. Red edges indicate edges whose associated transition probability is significant. Edges with significant transition probability were only found between consecutive or nearby nodes as well as between the nodes corresponding to ensemble 1 and 10, once again mirroring the periodic boundary conditions in ensemble activation. In shuffled realizations of the data there were edges that corresponded to significant transition probabilities, but those were not between neighboring nodes. q. Scatter plot showing relation between oscillation score and sequence score. The oscillation score quantifies the extent to which the calcium activity of single cells is periodic and ranges from 0 (no oscillation) to 1 (oscillation). The sequence score quantifies the probability of observing sequential activation of 3 or more ensembles. Each dot corresponds to one session. The sequence score increases with the oscillation score, and is highest for oscillatory sessions. Note that non-oscillatory sessions display non-zero values of sequence score, indicating the presence of sequential ensemble activity also in sessions below criteria for oscillation. r. Percentage of sessions with significant sequence score in sessions classified as oscillatory vs non-oscillatory. In MEC sessions with oscillations, 100% (15 of 15) of the sessions showed significant sequence scores, while in MEC sessions without oscillations, 41% (5 of 12) of the sessions demonstrated significant sequence scores. For corresponding raster plots, see .
Article Snippet: Finally, to calculate the oscillation frequency of ensemble activity, the PSD was calculated (Welch’s methods, 8.8 min Hamming window with 50% overlap between consecutive windows, “pwelch” Matlab function).
Techniques: Activity Assay, Activation Assay, Transformation Assay, Sequencing
MathWorks Inc is a verified supplier 
Goldberger et al., 2000 ; 
