Article Figures & Data
Figures
- Figure 1.
LFP characteristics of the medial MBs from conjoint hippocampal (HPC) and medial MB recordings. a, Histological reconstructions of the MB electrode tracks (red lines). Black and white heatmaps show the theta band power of the LFP through the depth of the electrode track. Associated red dotted line shows the peak LFP power of the LFP versus electrode depth. b, Normalized power spectral density (PSD; top) of the MB LFP showing peaks in the delta, theta, and first harmonic of theta and magnitude squared coherence between HPC and MB (bottom). c, Mean phase lag in an example awake session between HPC (red) and medial MB (black) LFPs showing ∼30 ms lag (top) and a histogram of cycle lags within the session (bottom). Red trace shows the mean lag between the LFPs from all animals combined. e, Example LFP traces, from conjoint MB and HPC recordings. Top traces, Mean running speed of 74 cm/s; bottom traces, 9 cm/s; illustrating change in frequency and cycle asymmetry. e, Theta frequency of HPC and MB at different running speeds. Lines show the best fits for each region (both sigmoid functions). f, Normalized power spectral density for medial MBs and HPC at different running speeds. Lines show the best fits of a sigmoid function for HPC and linear function for medial MBs. g, Comparison of ascending (trough-to-peak) and descending (peak-to-trough) theta cycle durations with running speed. h, Granger causality for HPC→MB and MB→HPC during active wakefulness. All scale bars in a are 250 µm.
- Figure 2.
Physiological characteristics of MB units during awake exploration. a, Firing characteristics of three representative medial MB units, showing speed-related changes in firing rate (i), theta phase MRV (ii), spikes per theta cycle (iii), AHV-related firing rate (iv), polar plot showing preferred theta phase (v), and spike waveforms (vi). Fits of linear and sigmoid functions are shown in black and red, respectively. Alongside speed/firing rate correlations, medial MB units also exhibited significant changes in strength of phase entrainment with running speed. b, Example raster plots of synchronously recorded medial MB units showing population-dependent firing with respect to running speed. Red trace shows the combined multiunit activity (MUA) of the units depicted in the raster plots. c, As demonstrated in the examples in a, a unit's firing rate by running speed and its spikes per cycle by running speed relationship were highly correlated across the population of medial MB units recorded, reflecting an active increase in firing rate, that is, an increase in spikes/cycle. d, Waveform asymmetry and peak–trough width of medial MB units (black) and hippocampal PYR units (red). Mean spike waveforms of medial MB (black) and hippocampal (red) units (top right). e, Example epoch of synchronously recorded medial MB (black) and HPC (red) LFPs with corresponding raster plots of medial MB units exhibiting phase entrainment of firing during awake exploration. f, Population phase preference of medial MB units (sorted by preferred phase), with each row representing the spike count phase histogram of a single medial MB unit.
- Figure 3.
Bursting neurons in the medial MBs. Using the approach of Simonnet and Brecht (2019), PCAs 1–3 of the LogISI and the first PCA of the 0.02 s autocorrelation explained 80.1 and 87.0% of the variance, respectively. a, Ward's agglomerative clustering identified two clusters (dendrogram) equivalent to the DB (n = 134; 24.7%; black) and SB (n = 409; 75.3%; red) of Simonnet and Brecht (2019); neighboring heatmaps show LogISI histograms (middle) and autocorrelations (maximum 0.02 s lag), ordered according to the clustering in a. In each case a row corresponds to a single medial MB unit. b, Left, Medial MB, DB units (black), had a higher mean peak in the LogISI ∼4–15 ms and a secondary peak shared by SB units (red) at ∼100 ms, corresponding to firing at a frequency within the theta band; Right, the mean 0.02 s autocorrelogram of DB units (black) showing a higher initial peak than SB units (red). The overlaid yellow trace in the four panels of b show the mean averages for each case. c, A scatterplot showing the burst probability of medial MB units, sorted according to the hierarchical clustering procedure, is higher in DB classified units (black) in a. Circle size represents unit mean firing rate. d, Boxplot showing that as a population, firing rates in DB units are significantly higher than SB units. e, Two example bursting medial MB units (DB#1, blue; DB#2; red) exhibiting cycle-skipping bursting (DB#1 in top example and DB#2 in the bottom trace), and nonskipping bursting activity (DB#2 in top trace and DB#1 in bottom example), at different points within the same recording. f, Left, The corresponding log 1/interburst interval histograms and theta bandpass filtered 0.5 s autocorrelograms of example units DB#1 and 2. Peaks between 4 and 6 Hz correspond to cycle-skipping bursting activity, while the peaks at 6–12 Hz correspond to cycle-by-cycle bursting. Right, Autocorrelograms for both example units show theta rhythmicity, while those of DB#2 show a characteristic theta-skipping trace, likely to correspond to more dominant cycle-skipping burst activity. g, Line plot showing the mean average LogIBI histogram of all medial MB units. Prominent peaks within the theta (6–12 Hz; green) and low theta (gray; 4–6 Hz) are highlighted.
- Figure 4.
Characteristics of medial MB unit physiology across the sleep–wake cycle. a, In AWK and REM, both theta-dominant states, Granger causality scores (within 6–12 Hz) were significantly higher in the HPC→MB direction than those in the reverse (top panels). Both HPC and medial MB LFPs had primary peaks in the theta band with significant coherence peaks at the corresponding frequency. b, Units that were phase entrained during AWK maintained their preferred phase during REM. Solid points represent SB units and open circles, DB units. c, Linear regressions of bursting probability of medial MB units across sleep–wake states. Medial MB units that exhibited complex bursting in AWK, also did so during SWS and REM. d, Comparison of LogISIs of DB and SB units during REM showing that DB exhibited higher counts of spikes within <∼15 ms than SB units, reflecting DB units’ preserved propensity to burst during REM. e, ei, Phase histogram with overlaid scatterplot of MRV length and preferred phase for each medial MB unit (e) alongside the mean normalized spike count phase histogram of the proportion of medial MB units (56%) that showed a phase entrainment to the downstate of 1–4 Hz oscillations during SWS (slow waves; ei). f, Representative example of a wavelet-transformed epoch of REM sleep falling between two SWS epochs, and the physiological characteristics used for REM (high theta/delta ratio; green) and SWS (high delta/theta ratio; blue) detection, both with no movement speed; note the brief post-REM wakefulness (black line showing running speed). fi, Example 5 s epochs of wideband HPC LFP from REM (green) and neighboring SWS (blue) epochs with corresponding raster plots of synchronously recorded MB (black) and HPC PYR (red) single units. g, AWK normalized firing rate scores of HPC (red) and medial MB (black) units within REM and neighboring SWS epochs. h, Synchrony across sleep state transitions for HPC (red) and medial MB (black) units. Dots show individual points and error bars show mean ± SEM for each region at each sleep state.
- Figure 5.
Hippocampal SWR-dependent firing in the medial MBs. a, HPC ripple-triggered average of the HPC (red) and the MB (black) LFPs. Individual MB black traces (n = 6) represent the mean for each animal. Top panel shows the medial MB population-wide firing rate change with respect to HPC-detected ripples. b, Peri-event time histogram showing the mean count of MB detected events with respect to putative SWRs detected in the hippocampal LFP (150–250 Hz; top); bottom, event-triggered average of high-frequency (100–250 Hz) events detected in the MB LFP. The resulting waveforms share the characteristics of putative MB response events observed by proximity to hippocampal SWRs (Fig. 5a) or through burst-triggered averages of SWR responsive firing units (Fig. 5g). c, Mean firing rate of medial MB units in the time window surrounding detected hippocampal SWRs. Top, middle, and bottom panels include units with significant positive, negative, or uncorrelated firing, respectively. d, Heatmaps with each row representing the convolved mean firing rate of a medial MB unit with respect to detected HPC SWRs. Top and bottom panels show the units (n = 107 and n = 21) with significant increases and decreases in firing rates, respectively. e, Representative example of a medial MB unit that exhibited a significant increase in firing rate around the detected HPC ripple event times. The SWR responsiveness index (SWR RI) of the unit was higher than the 95% CI of a shuffle distribution of SWR RIs (see Materials and Methods for details). The example is also a DB unit. Burst-triggered averages (bottom) of the HPC (red) and MB (black) LFP showed typical HPC SWR and a medial MB response events. f, Mean firing rate of DB versus SB medial MB units, with respect to HPC ripple events. Firing rates of both DB- and SB-classified units were significantly correlated with HPC ripple events; however, the mean magnitude of firing rate change was greater in the DB units. g, Mean average burst-triggered average of the HPC (red) and MB (black) LFP during SWS revealing large amplitude events corresponding to putative HPC SWRs and medial MB response events.
- Figure 6.
Summary charts of medial MB cell firing properties in wake and sleep. a–d, Donut charts displaying the proportion of cells with firing rates related to the labeled property, further denoted by label color. Proportions of a, AHV-dependent units (blue), separated by whether they are asymmetric (blue) or not (black), and the direction of movement by which asymmetric AHV units are modulated (clockwise, blue; counterclockwise, black) b, Speed-dependent units, their best fitting polynomial (linear or sigmoid fit), and the direction of the relationship between speed and firing rate; c, theta modulated units, the presence of theta cycle skipping, and the proportions of these units that are DB or not (SB); and d, SWR-responsive units, the direction of their SWR-modulation, and the proportion of SWR units exhibiting bursting.
- Figure 7.
Schematic of the MBs’ principle efferents and afferents. The collateral projections from the hippocampal formation (subiculum) to the retrosplenial cortex and MBs are shown. The MB efferent projections arise from collateral fibers but these are not depicted here.
Tables
- Table 1.
Statistical table detailing the data structure, type of statistical test and the test statistic or power of each inferential statistic used
Ref Figure Data distribution Test Power/test statistic a Normal Linear mixed effects −1.361 ± 16.870 b 1e Normal Linear regression R2 = 0.57 c R2 = 0.52 d Nonlinear curve fit—sigmoid function/BIC weighting R2 = 0.90/Sigmoid BICw = 1 e R2 = 0.88/Sigmoid BICw = 1 f 1f Normal Linear regression R2 = 0.54 g R2 = 0.45 h Nonlinear curve fit—sigmoid function/BIC weighting R2 = 0.67/Sigmoid BICw = 1 i R2 = 0.90/Sigmoid BICw = 0.02 j 1h Normal General linear model Diff: 0.48 ± 0.06 k 2c Normal Linear regression Slope: 0.92 ± 0.01, R2 = 0.93 l 2d Non-normal Wilcoxon signed rank z = 4.37 m z = 4.35 n 2f Circular Rayleigh Z Z = 0.47 o 3c Normal Linear mixed effects Diff: −0.070 ± 0.003 p 3d Normal Linear mixed effects Diff: −20.941 ± 1.125 q 3e,f Normal General linear model 6.13 ± 0.54 r 3g Normal Linear mixed effects Diff: 0.14 ± 0.05 s 4a Normal General linear model 0.84 ± 0.17 t 4b Circular Circular correlation r = 0.88 u 4c Normal Linear regression Slope: 0.73 ± 0.04 v Slope: 0.91 ± 0.05 w Slope: 1.1 ± 0.06 x 4e Circular Rayleigh Z Z = 57.75 y 4g (top) Non-normal Friedman χ2 = 29.93 z Dunn post hoc pairwise Diff: −0.25 ± 0.18a aa Diff: −0.28 ± 0.21a ab Diff: 0.01 ± 0.08a ac Wilcoxon signed rank −0.40 ± 0.13a ad −0.13 ± 0.23a ae −0.39 ± 0.20a af Dunn post hoc pairwise Diff: 0.04 ± 0.29a ag Diff: 0.0. ± 0.29a ah Diff: 0.07 ± 0.10a ai Wilcoxon signed rank 0.03 ± 0.31a aj −0.17 ± 0.41a ak −0.05 ± 0.26a al 4g (bottom) Non-normal Friedman χ2 = 23.29 am Dunn post hoc pairwise Diff: 0.00 ± 0.05a an Diff: −0.01 ± 0.04a ao Diff: 0.01 ± 0.01a ap Diff: −0.04 ± 0.03a aq Diff: −0.02 ± 0.03a ar Diff: 0.00 ± 0.01a as 5f Normal General linear model Diff: 0.024 ± 0.01 ↵a Differences for nonparametric tests are expressed as median ± IQR.
Letters in the “Ref” column refer to those denoted in superscript after each statistic in the main text.
In this issue
