Two complementary learning systems acquire and process episodic memories in order to learn adaptive models of the world that generalize beyond specific experiences and continue to incorporate information through the lifetime of an animal (Marr, 1971; McClelland et al., 1995; Kumaran et al., 2016; Yu et al., 2018). The hippocampus can form memory traces rapidly; however, over time, interactions between the hippocampus and neocortex are thought to extract statistical regularities from the organism’s experiences during wakefulness, and update neocortical models or schemas to increase their adaptive value (Kumaran et al., 2016; Morris, 2006; Moscovitch et al., 2016; Stickgold and Walker, 2013). During non-REM sleep, the interactions between the hippocampal and neocortical cell ensembles that underlie these processes are reflected in nested oscillations in the local field potentials (LFPs). The up-states of the slow oscillation (1–4 Hz) provide a processing window in which faster dynamics lead to synaptic plasticity in relevant cell populations. In the neocortex, some cells are preferentially active at the start or at the end of up-states (Gardner et al., 2013; Neske, 2015), and in the hippocampal CA1 region, sharp-wave ripple oscillations (SWRs, 100–275 Hz) occur primarily during up-states (Ji and Wilson, 2007; Isomura et al., 2006). Hippocampal SWRs are brief (up to ~100 ms) oscillations that correlate with the reactivation of spatial representations by hippocampal cell ensembles (Wilson and McNaughton, 1994; Davidson et al., 2009), which are thought to contribute the content in the consolidation of episodic memory (Buzsáki, 1996; Girardeau et al., 2009; Ego-Stengel and Wilson, 2010). Spindle oscillations (6–14 Hz) are coordinated with the slow oscillation and with SWRs (Siapas and Wilson, 1998; Sirota et al., 2003; Peyrache et al., 2011). Spindles are generated in the thalamus when the inhibitory cells of the thalamic reticular nucleus (TRN) evoke inhibitory post-synaptic potentials and rebound bursting in cells of the dorsal thalamus, which then entrain their postsynaptic targets in neocortex (Steriade et al., 1987; Contreras et al., 1997; Steriade et al., 1985). However, the role of spindles in gating and coordinating communication between hippocampus and neocortex remains to be explored at the cellular and systems level.

The coupling between non-REM oscillations appears to be causally linked to memory consolidation, as suggested by the disruption or enhancement of memory following closed-loop manipulations (Maingret et al., 2016; Latchoumane et al., 2017; Ngo et al., 2013; Marshall et al., 2006), and by the association of spindle dysfunction with perturbation of memory function in some neuropsychiatric disorders (Purcell et al., 2017; Manoach and Stickgold, 2019; Winer et al., 2019). However, we are far from understanding the specific mechanisms and single-cell dynamics involved in the hippocampal-neocortical interaction. In particular, information regarding the thalamic involvement is limited. Evidence from anesthetized animals suggests a coordination of thalamic firing with both the neocortical slow oscillation (Slezia et al., 2011; Contreras and Steriade, 1995) and with hippocampal SWRs (Lara-Vásquez et al., 2016), while fMRI results from studies in monkeys indicate an overall decrease in thalamic activity during SWRs (Logothetis et al., 2012). A recent study in naturally sleeping rats (Yang et al., 2019), found a decrease in activity in cells of the thalamic mediodorsal (MD) nucleus at the time of CA1 SWRs, particularly during SWRs uncoupled from spindles recorded in the skull EEG. This emerging evidence emphasizes the need to study the thalamus, but we have been lacking a multi-region approach (thalamus, neocortex, hippocampus) that would compare in the same preparation the dynamics of cell activity in each region to the sleep oscillations, and that would investigate the dynamics of the thalamic contribution down to the finer timescales relevant for plasticity.

We sought to address this gap by characterizing the fine temporal correlations of thalamic cells with sleep oscillations through extracellular recordings of single units and LFPs in the medial prefrontal cortex (mPFC), hippocampal CA1 region, and nuclei of the midline thalamus in freely behaving rats. We targeted these brain regions because of their anatomical and functional connectivity and their role in memory processes. The limbic thalamus has widespread connections with neocortical regions, and projections from the midline are particularly dense to mPFC (Vertes et al., 2006; Hoover and Vertes, 2007; Rubio-Garrido et al., 2009), a critical region for the recall of consolidated memories (Maviel et al., 2004). All the recorded nuclei share circuit, physiological and molecular commonalities (Guillery and Sherman, 2002; Phillips et al., 2019; Varela, 2014). The reuniens nucleus (RE) in the midline sends collaterals to CA1 and mPFC, making it a compelling candidate to coordinate mPFC-CA1 activity (Hoover and Vertes, 2012; Varela et al., 2014). In addition, although the functional association between RE and hippocampus has received a lot of attention (Dolleman-van der Weel et al., 2019), the hippocampus projects to rostral parts of the thalamic reticular nucleus and to prefrontal cortex (Zikopoulos and Barbas, 2012; Cavdar et al., 2008; Swanson, 1981), which could indirectly influence nuclei other than RE. Recording from several midline nuclei, we find evidence for a dual role of the thalamus in hippocampal-neocortical sleep interactions; first, an increase in thalamic firing at the down to up-state transition, which could influence the dynamics of the slow oscillation, and second, the sparse but consistent firing of thalamic cells within individual spindles suggests an interleaved activation of thalamocortical synapses that may promote a flexible integration of recently acquired memory traces into neocortical schemas during systems consolidation.

We used tetrodes to record extracellularly (single units, LFPs) from the midline thalamus, and from areas involved in recent (CA1) and remote (medial prefrontal cortex, mPFC) memory recall (Maviel et al., 2004; Varela et al., 2016). During the 1–5 hr recording sessions (in freely behaving rats), the animals remained in a quiet, square-shaped enclosure where they cycled through bouts of sleep and wakefulness; in six sessions from three rats, the animals also explored a radial maze after the recording of sleep, and were brought back to the sleep area after behavior for comparison of sleep activity before and after spatial exploration. LFPs and 333 units were recorded from seven rats; sessions from an additional animal in which no units were recorded were also included for LFP analyses for a total of n = 8 rats (Figure 1a displays a subset of tetrode locations in brain sections from one rat; Figure 1—figure supplements 1–2 document the trajectories in the brain for the electrodes and sessions used for analyses). We selected recording sessions in which there was at least one unit recorded in the thalamus and where reliable detection of sleep events could be performed (slow oscillation and spindle activity in mPFC, sharp-wave ripples, SWRs, in CA1; examples of simultaneously recorded LFP traces and units shown in Figure 1b). The majority of recordings in the thalamus were from units in the reuniens (RE), and ventromedial (VM) nuclei, and some from the mediodorsal (MD), paratenial (PT), rhomboid (RH) and centromedial (CM; detailed population numbers in Table 1). All these nuclei are bi-directionally connected to mPFC, and are functionally related to mPFC and other higher order cognitive regions, including CA1, which also receives direct input from RE (Dolleman-van der Weel et al., 2019; Cavdar et al., 2008; Swanson, 1981). Here, we will use the term limbic thalamus to refer to the thalamic nuclei that we recorded. See the Materials and methods section for target anatomic coordinates, procedures to identify tetrode location, and additional inclusion/exclusion criteria for the analyses reported below.

Figure 1 with 4 supplements see all

Download asset Open asset

To study the modulation of unit spikes in the three regions (mPFC, limbic thalamus, CA1) with non-REM sleep oscillations, we first identified times with both high delta (1–4 Hz) power in mPFC and high SWR (100–275 Hz) power in CA1 (as described in Methods); an example of such non-REM sleep state detection for a representative session is shown in Figure 1c. Within this state, we detected reference features that reflect non-REM sleep events that are key organizers of memory consolidation: K-complexes (KCs, marked by red bars in Figure 1d, mPFC LFP), the troughs of spindle cycles in mPFC (red dots in 1d), and SWRs in CA1 (red bars in Figure 1d, CA1 LFP). These LFP events are markers of cell population dynamics, namely, KCs mark the down-states of the neocortical slow oscillation, while SWRs correspond to the high-frequency oscillations that reflect memory-related reactivation of ensembles of CA1 place cells (Davidson et al., 2009; Girardeau et al., 2009; Ego-Stengel and Wilson, 2010; Wilson and McNaughton, 1994). Spindles (6–14 Hz) are generated through oscillatory bursting in the thalamic reticular nucleus (TRN), which is thought to then engage cells in the dorsal thalamus and, through them, neocortical networks (Steriade et al., 1987; Steriade et al., 1985; Steriade et al., 1993; Huguenard and McCormick, 2007). We considered that the time window of each spindle cycle has a similar duration to that of SWRs (up to ~100 ms) and sought to test if individual spindle cycles represent a window for multi-region coordination; we thus detected individual spindle cycles from the mPFC 6–14 Hz filtered LFP. The detection of individual cycles was also motivated by the occasional observation of few spindle cycles in succession (e.g. only 3–4 cycles following the second detected KC in Figure 1d), and by the broader variability in the morphology of the LFP trace in the spindle band compared to the more clearly discrete and consistent signatures of KCs and SWRs (examples in Figure 1d). KCs, spindle cycles, and SWRs were detected from the mPFC and CA1 LFPs after applying an amplitude threshold to the squared, filtered trace (1–4 Hz for KCs; 6–14 Hz for spindles; 100–275 Hz for SWRs), at a level of three (spindles) and four (KCs, SWRs) standard deviations above the average LFP voltage across sleep periods in each recording session (see Materials and methods for details). Although the spindle band overlaps with the theta band, we limited the detection of spindle cycles (and KCs and SWRs) to resting states; peri-spindle spectrograms confirmed that detected spindle cycles are functionally different from theta (Figure 1—figure supplement 3). Peri-event histograms of the LFP and local spiking activity, triggered on the timestamps at the time of largest amplitude of the detected events (trough for KCs and spindles, peak for SWRs), confirmed that the LFP detection identified events that reflect the modulated activity of local cell populations (Figure 1—figure supplement 4). Quantifications are reported as mean ± standard deviation (sd), unless otherwise indicated.

Figure 2a illustrates the temporal and phase relations between non-REM sleep events, with examples from the raw and filtered mPFC and CA1 LFPs. The overlaid filtered LFP traces (which have been scaled for display purposes), show that increases in SWR amplitude occur preferentially in the positive half-cycle of the delta-filtered LFP, and in the negative halves of the spindle-filtered trace (inset). The average distribution of delta and spindle phases (estimated with the Hilbert transform, see Materials and methods) at the time of SWRs is shown in Figure 2b for all sessions. Figure 2c shows the peri-event histograms for non-REM detected population events (SWRs, KCs, spindle cycles) across sessions and rats.

Figure 2 with 1 supplement see all

Download asset Open asset

Confirming previous results (Ji and Wilson, 2007; Siapas and Wilson, 1998; Sirota et al., 2003; Peyrache et al., 2011), we found that KCs, spindles, and SWRs are temporally and phase-coupled. SWRs preferentially occurred following and preceding KCs, phase-locked to the peaks of the delta-filtered LFP, and to the troughs and ascending phase of spindle cycles (circular mean and standard deviation of delta phase at the time of SWRs = −0.04 ± 0.43 rad; spindle phase = −1.72 ± 0.93 rad; n = 23 sessions, six rats; Figure 2—figure supplement 1 shows the delta and spindle phase distributions at the time of SWRs for all sessions included in this analysis).

Thalamic cells decrease their overall firing and increase bursting with CA1 SWRs

We next investigated the changes in the firing patterns of single units in the limbic thalamus with CA1 SWRs.

Most thalamic units presented a reduced average firing rate during rest compared to wakefulness. In 28 of 32 sessions, the animals (n = 7) alternated between sleep and active wakefulness (in the other four sessions rats were only in wakefulness for less than 5 min and were not included in this comparison), and the firing rate of thalamic units changed from an average of 10.10 ± 4.26 spikes/s during wakefulness to 7.73 ± 3.09 spikes/s during sleep (n = 40 units; p=1.31 × 10⁻⁶, Wilcoxon signed rank; Figure 3a).

Figure 3 with 1 supplement see all

Download asset Open asset

A decrease in firing rate may be due to the hyperpolarization of thalamic cells. Thalamic cells fire in two modes, burst and tonic, depending on the activation state of a transient conductance that relies on T-type calcium channels (Perez-Reyes et al., 1998; Gutierrez et al., 2001). Bursts correlate with thalamic hyperpolarization, which is a pre-requisite for T-channel activation, and they can enhance transmission in thalamocortical synapses and induce synaptic plasticity (Swadlow and Gusev, 2001; Leresche and Lambert, 2017). Therefore, analyzing burst firing can provide clues on the mechanism behind the decrease in firing rate in sleep. Bursting can be detected from extracellular recordings because of its characteristic combination of a relatively long inter-spike interval followed in rapid succession by a few spikes with brief inter-spike intervals (Lu et al., 1992), but the relation between thalamic bursts and SWRs has not been explored.

We detected bursts in the thalamus as groups of action potentials with less than 5 ms of inter-spike interval preceded by at least 70 ms with no spikes, a criterion based on the dynamics of the T-channels underlying burst firing in the dorsal thalamus (Gutierrez et al., 2001; Lu et al., 1992; Jahnsen and Llinás, 1984a). With this criterion, we found that 25% of the thalamic cells had at least 5% of their spikes in bursts during wakefulness, while 77.5% did during sleep (average percentage of spikes in bursts during wakefulness 4.60 ± 7.97%, and during sleep 18.65 ± 15.3%; p=1.8 × 10⁻⁷; Wilcoxon signed rank; n = 40 units, seven rats; Figure 3a). The increase in bursting was more pronounced around the time of SWRs in CA1 (25.52 ± 17.29% of spikes in bursts in a 2 s window around SWRs; p=1.8 × 10⁻⁷ Kruskall-Wallis; Figure 3a). We also estimated thalamic bursting after shifting the timestamps of each SWRs randomly by at least 2 (or 5 s) and calculated the total number of spikes and percent of burst spikes in the same 2 s window around SWRs. In both cases (>2 and>5 s shifts), the average number of thalamic unit spikes was significantly lower in the SWRs windows compared to the shifted timestamps (p<10⁻³), and the percent of spikes in bursts significantly higher (p=0.002 and p<10⁻³ respectively after a 2 s or 5 s shift; Wilcoxon signed rank). Because the T-channels responsible for burst firing need a sustained hyperpolarization for about 50–100 ms to be primed for activation (Gutierrez et al., 2001; Jahnsen and Llinás, 1984b), the increase in bursting suggests a hyperpolarization of the majority of cells in the limbic thalamus during sleep, and particularly in association with hippocampal SWRs.

Indeed, peri-SWR triggered histograms showed that SWRs occur in a background of decreased thalamic spiking. An example of a peri-SWR histogram for a unit from the RE nucleus is shown in Figure 3b, and the peri-SWR for the population of all thalamic units in Figure 3c; 55.32% of units in the thalamus showed a reduction in firing of more than two standard deviations (sd) from the null (average −4.23 ± 2.00 sd) in a 200 ms window around the time of SWRs occurrence. Instead, 90.18% of CA1 units increased their firing by 19.85 ± 18.65 sd in a 200 ms window around SWRs (4.91% of CA1 units showed an average decrease of 4.11 ± 1.59 sd), while the mPFC population included 30.95% of units showing an increase (average 6.22 ± 4.32 sd) and 40.48% a reduction in firing (average −3.22 ± 1.40 sd). The peri-SWR histograms for the population of CA1 and mPFC units are shown in Figure 3—figure supplement 1a. A few thalamic units (14.89%) showed a pure increase in firing in the 200 ms window around CA1 SWRs; this means an increase in firing larger than two standard deviations from the null distribution and that was not associated with a post-inhibition rebound. Another four units (8.51%) had a rebound in firing (larger than 2sd) that followed soon after a strong decrease in firing (below −2sd) (as the example in Figure 3b).

Using a complementary approach, generalized linear models (GLMs) of CA1 activity (SWRs and multi-unit firing rate) with the firing rate in thalamic units as covariate, showed that the firing in CA1 varied inversely with the firing of most thalamic units (61.7%, p<0.001; only one unit in the MD nucleus had a significant positive correlation with SWR rate). For the units that negatively correlated with SWRs, the models predict an average increase of 20.74% in the rate of SWRs for every one spike/s drop in thalamic firing, suggesting anti-correlated activity in the limbic thalamus and CA1 (instead, as expected, the rate of SWRs was positively correlated with the spiking activity recorded in CA1; p value < 5 × 10⁻¹⁸⁹ in all electrodes used for SWR detection). The decrease of spiking in thalamic units with SWRs was observed in RE and in other nuclei. Some of the strongest negative correlations with SWRs were observed in RE (unit in Figure 3b), and the size of the effect was marginally significant in RE compared to non-RE units (p=0.05). However, there was no difference between the proportion of SWR-correlated units in RE compared to non-RE units (p=0.159), suggesting that the decrease in thalamic firing with SWRs relates to general excitability changes in the state of thalamocortical networks in the limbic thalamus.

Temporal correlation of thalamic units with mPFC delta and spindle oscillations

Although we found a decrease in average thalamic firing and an increase in bursting that suggests a hyperpolarized thalamus during SWRs, it is unclear how the remaining thalamic spikes relate to the mPFC delta (1–4 Hz) and spindle (6–14 Hz) oscillation bands, and if there are specific temporal windows with selective increase in thalamic firing. To investigate this, we used two approaches: we calculated firing rates with respect to local events detected in the mPFC LFP (KCs, spindle troughs); and we estimated the phase-preference of unit spikes to the delta and spindle-filtered mPFC LFP.

Thalamic rebound following the down-state of the mPFC slow oscillation

Figure 3d shows the peri-KC histogram for a unit recorded in the mediodorsal (MD) nucleus. Units in the thalamus often fired asymmetrically with respect to KCs, with stronger firing in the few hundreds of milliseconds after the mPFC KC compared to before KCs (average of 2.48 standard deviations at 274.76 ± 135.33 ms after KCs, compared to 1.28 standard deviations at 263.31 ± 145.31 ms before KCs; n = 31 thalamic units; Figure 3e; Figure 3—figure supplement 1b includes the peri-KC histograms for units in mPFC and CA1). The trough of thalamic firing occurred on average after the mPFC KCs and after the trough of mPFC unit spikes, and coinciding with the trough of SWRs in CA1, suggesting that the neocortical slow oscillation trough leads both the thalamus and CA1 (thalamic unit firing minimum at 10.56 ± 130.75 ms after KCs, n = 31; mPFC units −7.70 ± 72.3 ms, before LFP KCs, n = 46; SWRs 10.33 ± 5.88 ms after KCs).

Previous evidence (Neske, 2015) suggests that the thalamus may have a specific role in the initiation of cortical up-states. To assess the specificity of thalamic firing to the down to up-state transition, we compared the firing of thalamic units before and after mPFC KCs. For the population of thalamic cells, 38.71% increased their activity above two sd (from the null distribution) following KCs, that is, in the down to up-state transition, compared to 9.68% of the cells with a significant peak preceding KCs, that is, in the up to down-state transition (19.35 % units had a significant peak both before and after KCs). Instead, CA1 and mPFC activity was more similar preceding and following KCs. SWR probability increased within the 250 ms around mPFC KCs (4.65 sd at 208.15 ± 90 ms before KCs, and 3.02 sd at 266.85 ± 100 ms after KCs; n = 23 sessions). This indicates a delta modulated increase in SWR probability in CA1, with the minimum rate of ripples following briefly after the trough of neocortical activity (10.33 ± 5.88 ms). The modulation of SWRs was paralleled by a modulation of firing in CA1 units (n = 111), which increased their activity both before and after KCs (2.32 ± 1.49 sd at −233.7 ± 146 ms before KCs, and 2.35 ± 1.69 sd at 273 ± 140 ms after KCs); on average, CA1 units presented a trough of activity after KCs (19.53 ± 132 ms). Activity in mPFC (n = 46 units) was modulated in a similar way, with increases in unit firing preceding and following the down-state of the slow oscillation (2.33 standard deviations at −267.30 ± 110.40 ms, before KCs; 2.31 standard deviations at 295.20 ± 120 ms, after KCs).

In other words, while units from the three areas showed no difference in the peak of the histograms in the time window following KCs (p=0.89, Kruskal-Wallis between regions), thalamic units had significantly lower firing before KCs compared to mPFC and CA1 units (p=0.007, Kruskal-Wallis). This suggests that thalamic cells contribute spikes primarily when a new cycle of the slow oscillation starts, while mPFC and CA1 units also fire at up-to-down transitions. We used an additional metric to quantify the differential firing of units in the three brain areas with respect to the time of mPFC KCs. For all the units, we calculated the difference between the maximum firing rate in the 0.5 s windows before and after KCs, normalized to the maximum firing in the two windows. This asymmetry index ranges between −1 and 1, with negative values indicating higher firing before KCs. The distribution of this index (Figure 3f) was biased to positive values for thalamic units, but was uniformly distributed for mPFC and CA1 units (average 0.24 ± 0.22 in thalamus; −0.004 ± 0.47 in mPFC; −0.06 ± 0.5 in CA1), consistent with a selective rebound of thalamic firing in the time window following mPFC KCs, relative to thalamic activity preceding the down-state.

Sparse thalamic firing and consistent activation of cell pairs during spindles

We then analyzed the firing at the timescale of mPFC spindles, and found the strongest modulations in the CA1 multi-unit activity (MUA, all spikes recorded by a given electrode). An example of thalamic unit firing modulation with spindles is shown in Figure 3g and the population histograms for the thalamus in Figure 3h. Additional peri-spindle-trough histograms for mPFC and CA1 are shown in Figure 3—figure supplement 1c. Across the sessions, 84.62% of the CA1 tetrodes showed an increase in spike activity (at least three standard deviations from null distribution in the −100 to 100 ms window of the peri-spindle-trough histogram), with 53.3% and 36% tetrodes showing a similar result in mPFC and thalamus. The increase was less clear and consistent in individual units, even when the local MUA had a clear modulation, which suggests a functional coordination at the cell population level with sparse engagement of individual cells. Spindle oscillations can vary in frequency (Gardner et al., 2013) and individual units may not contribute spikes in every spindle cycle, flattening the average correlograms. To investigate the oscillatory activity of thalamic units directly, we looked for peaks in the autocorrelation of single unit spikes.

The autocorrelations of thalamic units did not present strong peaks at the time intervals of spindle frequency, that is, in the time window between 50 and 150 ms, suggesting that the cells did not fire rhythmically in that frequency. Figure 4a illustrates the sparsity of thalamic cell firing during mPFC spindles; the panel shows a trace of mPFC LFP with several cycles of spindle oscillation (6–14 Hz filtered trace overlaid in black); the MUA recorded by one tetrode in the thalamus is shown in orange and the spikes of a single unit in magenta. The auto-correlograms next to the raw data were calculated with the spikes fired during sleep, and show the lack of spindle rhythmicity in the single unit (also when calculated with only burst spikes), in spite of the MUA modulation (which can also be seen in the spike raster). Similar results were obtained for all thalamic units, regardless of the nucleus, and including the units that showed the strongest rebound following KCs. The observation of peaks in the autocorrelation of the MUA was not just due to a larger amount of spikes compared to individual units; calculating the MUA autocorrelation in windows with smaller numbers of spikes (equal to the median number of spikes in units) preserved the shape of the MUA autocorrelation (Figure 4—figure supplement 1a–e). We calculated the time at the autocorrelation maximum (after the minimum that follows lag 0) and plotted the distribution for unit spikes and for MUA (all the spikes in sleep or a subsample matching the number of spikes of the sparsest unit recorded in the same session; Figure 4—figure supplement 1f). We found that the peak of the MUA autocorrelation occurs on average at 119.30 ± 78 ms, whereas the interval is longer for the units, 223.60 ± 113.30 ms (p<0.001, Wilcoxon Mann-Whitney). The peak of the autocorrelation occurred within the spindle band in 63.3% of the curves calculated with MUA (with subsampled spikes matched to the units) and in 17.02% of the units (Figure 4—figure supplement 1g shows the autocorrelation of units in which the maximum autocorrelation value falls within the lags that correspond to the spindle band, 6–14 Hz). The autocorrelations with peak values within the spindle band had significantly higher average correlation values for MUA spikes compared to units (MUA, 0.24 ± 0.20; units, 0.05 ± 0.01; p=0.0038, Wilcoxon Mann-Whitney). One thalamic unit that had a strong indication of sustained oscillatory activity showed a peak of the autocorrelation at 140 ms, or ~7 Hz, on the lowest range of the spindle band (Figure 4—figure supplement 1h).

Figure 4 with 2 supplements see all

Download asset Open asset

Because cells could be firing rhythmically only when they are bursting, we also calculated autocorrelations with only burst spikes, as well as only for the spikes that occurred in the 0.5 s time window immediately following KCs (when spindles are largest). Thus, these analyses were more restricted to conditions in which rhythmicity in the spindle band would be expected to be strongest, and still failed to reveal evidence of sustained rhythmic firing in this sample of thalamic cells. Overall, the results suggest that most thalamic units are not strongly oscillatory and that it is the combined spikes from multiple units that underlie spindle frequency oscillations.

The reduced rhythmicity in the thalamic units compared to the modulation of the MUA, raise the possibility that different groups of thalamic cells contribute to activity in different spindle cycles. A prediction from this hypothesis is that pairs of units should be more likely to be correlated within brief timescales than individual units with themselves. We tested this by calculating cross-correlograms of sleep spikes in sessions in which we recorded more than one unit in the thalamus (n = 39 pairs, four rats); 33% of the pairs had a peak above three standard deviations from the average cross-correlogram (average peak z-score 2.40 ± 1.21 sd). An example is shown in Figure 4b, which displays the cross-correlogram of two thalamic units during sleep; the plots show that the spikes of ‘unit 1’ follow reliably in a window of ~50 ms after those of ‘unit 2’ (reference), while neither of the units show a strong auto-correlogram in that period (Figure 4c).

All the pairs that showed a peak in the sleep cross-correlograms were correlated in a time scale consistent with the duration of spindle cycles (Figure 4b; Figure 4—figure supplement 2a for all pair cross-correlograms). We then identified the start and end of spindle cycles by finding the peak of the voltage trace before and after detected spindle troughs. The cross-correlograms between pairs of units showed similar values when calculated only with the spikes that occurred within spindle cycles (Figure 4d; Figure 4—figure supplement 2b for all pairs); we calculated additional cross-correlograms selecting the unit spikes that occurred during periods of wakefulness as well as during sleep periods between detected spindles (where no spindles were detected for at least 5 s). The strongest cross-correlations were observed during sleep spindles, suggesting sleep specific co-activation of thalamocortical networks (Figure 4—figure supplement 2b–d) and providing further support to the hypothesis of a co-activation of sparsely synchronous thalamic cells within spindles (average peak cross-correlograms with spindle spikes 2.74 ± 0.85 sd compared to 2.31 ± 1.18 sd with sleep spikes that occurred between detected spindles; p=0.015, Wilcoxon signed rank test).

The coordinated activity of cell populations drives amplitude and phase dynamics in the local field potentials (Buzsáki et al., 2012), and phase coordination has been suggested to underlie routing mechanisms within and across brain regions (Colgin, 2011; Hasselmo and Stern, 2014; Wilson et al., 2015). Therefore, to confirm a functional relation between thalamic spikes and mPFC non-REM oscillations, we assessed the phase locking of thalamic units to the delta and spindle frequency bands of the mPFC LFP.

Phase-locking of thalamic units to delta and spindle oscillations

We studied the phase-locking of thalamic units to the delta (1–4 Hz) and spindle (6–14 Hz) frequency bands by estimating the LFP phase using two complementary methods, the analytical signal and linear interpolation between consecutive troughs of the spindle and delta filtered LFPs (see Materials and methods for details). Figure 5a shows the estimated phase obtained with both methods for the spindle trace in Figure 4a. The unit fires sparsely (raster above the trace), but the spikes are likely to occur on the descending phase of the spindle-filtered and on the ascending phase of delta-filtered LFP, shown by the phase histograms calculated for the whole sleep session (Figure 5b). We found similar preferential firing for the population of thalamic cells. The phase histograms were fitted for all units to a circular Gaussian (von Mises distribution), with mean and concentration parameters that indicate the direction and depth of the modulation or phase-locking. In order to prevent sampling bias due to non-uniform baseline phase distributions (e.g. due to non-sinusoidal LFP morphology), the unit phase distributions were also normalized to the baseline phase distributions during sleep (Siapas et al., 2005).

Figure 5 with 1 supplement see all

Download asset Open asset

We found that the majority of thalamic units (67.74% of n = 31 units) showed preferential firing at the descending phases of the spindle-filtered mPFC LFP, as suggested by the positive values of the mean phase distribution (circular mean 0.77 ± 1.15 rad; p<0.05, Rayleigh test; Figure 5c). Thalamic units were phase-locked to the mPFC delta band, closer to the positive peak of the oscillation (64.52%, p<0.05, Rayleigh test; circular mean 0.30 ± 1.04 rad). The mean preferred phases were robust to the number of spikes; the circular mean was not different when the histograms were calculated with 10% of the spikes (p=0.68 for spindle and 0.07 for delta; Wilcoxon signed rank). The amplitude of the thalamic rebound after KCs (reported above) had a tendency to correlate with a preference for earlier phases of delta; although the correlation was not significant (p=0.08, r = −0.32), we did not observe such a trend when we regressed mean phase preference to the cell firing rate in a time window of the same length but before KCs (p=0.65, r = 0.08).

Interestingly, we found that the phase lags between the spikes of unit pairs that had the strongest cross-correlograms were more consistent than the phase lags between uncorrelated pairs. In the sessions with thalamic pairs (n = 39 pairs), we calculated the smallest difference between the phase of the spikes during non-REM sleep, and between the phase of the spikes in spindle cycles. The phase difference would be expected to be distributed between – 2π and 2π if units fire within one cycle of each other (positive or negative depending on their activation order), and units that are functionally locked are expected to show a more consistent, less variable phase relation. Figure 5d shows the phase lag distribution for the pair described in Figure 4, calculated both from all non-REM spikes (left panel) and from the spikes that occurred within spindle cycles (right panel); the distributions suggest that most of the spikes occur within less than half a cycle of each other; the slight asymmetry with respect to zero suggests the consistent firing of one of the units before the other. We found similar results (Figure 5—figure supplement 1) for the thalamic pairs that showed significant peaks in their cross-correlogram. The standard deviation of the distribution of phase lags was significantly lower for units with high spike correlations during sleep (pair cross-correlograms >3 sd) compared to the units that were not temporally correlated. The standard deviation of phase lag distributions within spindles was significantly lower for correlated pairs compared to uncorrelated pairs (2.15 compared to 2.50; p<0.001, Mann-Whitney U-test). These results demonstrate the phase-locked firing of thalamic cells within the period of the spindle oscillation.

Overall, the phase-locking results are consistent with the peri-event histogram analyses in suggesting coordinated thalamic firing with the neocortical delta and spindle oscillations, and demonstrate the phase coordination of pairs of thalamic units within individual spindle cycles.

Spindle activation of thalamic units modulated with CA1 SWRs

Because SWRs were also locked to preferential phases of the spindle oscillation (Figure 2b and Figure 2—figure supplement 1), we investigated the association between thalamus and SWRs within the timescale of spindle cycles. We calculated the ripple power within each spindle cycle, and classified cycles in two groups based on the median SWR power (detected spindle cycles and feature distributions for all sessions in Figure 6—figure supplement 1). Cycles with ripple power above the median were slightly shorter in duration (102.17 ± 2.01 ms compared to 103.22 ± 2.35 ms when ripple power was low; p<0.05 in 65% of the sessions, although not significant for the mean duration across sessions, p=0.11; Mann-Whitney U-test) and larger in amplitude (229.21 ± 102.14 µV versus 219.47 ± 94.32 µV; p<0.05 in 85% of sessions, p=0.54 across sessions; Mann-Whitney U-test).

Figure 6 displays the modulation of firing in one of the thalamic units with the strongest spindle modulation in our sample; the spindle-modulated spiking in this unit was enhanced with high SWR power in CA1. The peri-spindle-trough histogram of spikes is shown for all the spindle cycles in the session on the left panel (Figure 6a), and for the two functionally defined types of spindle cycles, with low and high SWR power (below or above the median for all spindles), in the middle and right panels. The spindle-trough-triggered histogram of CA1 SWRs for the same session is shown in blue, and confirms that the spindle classification identifies cycles with low or high probability of SWRs in CA1. Because the peri-spindle cycle histograms showed a pattern of both increase and decrease in the spike rate within the timeframe of spindle cycles, we calculated a modulation index as the difference between peak to trough firing relative to the mean in a 200 ms window centered at 0 on the peri-spindle-trough histogram. This index increased in the majority of thalamic units in spindle cycles with ripple power above the median (64.52%, n = 31). In some of the units, like the example in Figure 6, we observed a strong phase offset between SWRs and unit spikes (Figure 6b); however, this was not the case in all the units, and some had spindle phase preferences that overlapped with those of SWRs recorded in the same session. Likewise, in spite of an increase of modulation index with ripple power, units lacked specific increases of firing at the timescale of SWRs (Figure 6c shows the peri-SWR histograms for the same unit as in 6a-b). In addition, a subset of units (22.58%, n = 31) showed stronger modulation with low ripple power spindles, with modulation values that were at least as high as the average for cells that increased their modulation with high ripple power.

Figure 6 with 3 supplements see all

Download asset Open asset

The changes in firing modulation with ripple power in spindles could result from associated changes in spindle amplitude, or from a synergistic effect between spindles and ripples. We compared the peri-spindle cycle modulation index for spindle cycles classified based on spindle amplitude or on ripple power (above or below the median). The absolute index values were similar in spindles classified based on spindle amplitude or ripple power (p>0.3, Wilcoxon signed rank), but increased significantly from low to high amplitude spindles (p=0.005) compared to low to high ripple power spindles (p=0.25). The modulation also increased, although less, when comparing spindles with low amplitude and low ripple power to spindles with high amplitude and high ripple power (p=0.014). This suggests that the strongest firing modulation is due to changes in spindle amplitude compared to ripple power or to the combined effect, as expected from the role of thalamic cells in spindle generation (modulation index distributions for the population in Figure 6d). Although the modulation by ripple power is not as strong, it could provide an additional boost to the spindle related modulation. To find out if there are synergistic effects between spindle amplitude and ripple power, we detected the spindle cycles with the highest amplitude (upper 25% quartile) and compared the modulation of thalamic cells when ripple power was high or low (above or below median). We run a complementary analysis in which we detected spindle cycles with the highest ripple power (upper 25%) and compared the modulation of thalamic cells in cycles with low or high amplitude. Figure 6e illustrates an increased modulation with ripple power in the same thalamic unit from panels 6a-c. The peri-spindle histogram for spindle cycles in the upper quartile of the distribution of peak-to-trough amplitudes is split between those that have low or high ripple power; the two groups of spindles are not significantly different in amplitude (p=0.45), yet the unit increases its firing modulation with high-ripple-power spindles (top plots); the same unit shows reduced modulation with large spindle amplitude for spindles with the highest ripple power (bottom). Figure 6—figure supplement 2 includes additional examples of units that increase and decrease their spindle modulation with ripple power and spindle amplitude. In the population of cells, we did not find evidence of a significant effect of ripple power on the modulation value of thalamic firing in high amplitude spindles (p=0.94; Wilcoxon signed rank), while higher spindle amplitude led to higher modulation values on high ripple power spindles (p=0.04). However, although the modulation values were not significantly different, a majority (61.29%) of units showed a further increase in modulation index with ripple power in high amplitude spindles (average increase 18.15 ± 21.94%); instead, 32.26% of the units increased their modulation with spindle amplitude (in the spindles with highest ripple power; average increase 30.00 ± 20.97%), suggesting that when spindle amplitude is high, ripple power can enhance the modulation of thalamic cells further. The percent change in the modulation index with ripple power and spindle amplitude were positively correlated (Spearman correlation 0.51; p=0.003).

If there is a functional relation with spindle amplitude or ripple power, spindles in the upper quartiles should lead to stronger modulation. We then calculated the fraction of thalamic units in which the increase in modulation in spindles within the upper quartile was higher than the increase observed with the median classification. The increase in modulation index was higher in the upper quartile of spindle amplitude in 29% of thalamic units and with ripple power in 22.6% units. The absolute modulation values were higher for spindle amplitude compared to ripple power and positively correlated (average of 1.16 ± 0.57 in the upper quartile of spindle amplitude compared to 0.60 ± 0.26 in the upper quartile of ripple power; p=0.04, Mann-Whitney); units with stronger modulation with spindle power had higher values of ripple modulation (Spearman coefficient >0.7; p<0.04). For spindle amplitude, the lower modulation values also became lower with the quartile classification, leading to a wider spread of the distribution (when spindles were classified based on quartiles compared to median; p=0.04; F-test) which was not the case with ripple power (p=0.36; F-test; Figure 6f). These results also suggest that the strongest firing modulations in thalamic cells are related to the spindle oscillation, and that ripples in CA1 are associated with smaller changes in firing modulation in about 20% of thalamic cells.

Although we observed clear modulations of firing with spindles, the absolute number of spikes in single units did not change significantly with spindles low or high in ripple power, nor with SWRs coupled or uncoupled to spindles (Figure 6g).

We also observed a stronger modulation with ripple power in the cross-correlograms between unit pairs. We calculated the modulation index for the cross-correlograms of thalamic pairs and found that it increased by an average of 26.64% in 48.72% of the pairs in spindle cycles with high ripple power and decreased by 20.90% in the rest of the pairs (n = 39). Altogether, the data is consistent with the hypothesis that the up and down modulation of firing of thalamic cells during spindles is dynamically regulated in conjunction with changes in ripple activity in the hippocampus. This could influence spindle-specific thalamocortical synaptic plasticity and facilitate the integration of hippocampal memory traces into neocortical networks.

At broader time scales (seconds), we had found a decrease in thalamic firing and increase in bursting with SWRs (Figure 3). However, it is possible that only spindle-nested SWRs are correlated with thalamic units. To test this possibility, we calculated the cross-correlograms of thalamic units with SWRs that occurred within the time window of detected spindle cycles, and compared them to a sample of equal number of SWRs that did not occur within spindle cycles. We compared the mean spikes/s in the 100 ms central window of these cross-correlograms between thalamic single units and SWRs within or outside of spindles. We found low firing rates but no significant difference in thalamic firing near SWRs that occurred within detected spindle cycles (mean firing 3.83 ± 3.49 spikes/s with spindle-nested SWRs, compared to 3.99 ± 3.39; p=0.18, Wilcoxon signed rank; n = 31); only when broadening the window of the cross-correlograms to at least 1 s around SWRs, did we find a slight decrease in thalamic firing with SWRs nested in spindles (mean firing 3.73 ± 3.43 spikes/s with spindle-nested SWRs, compared to 3.94 ± 3.35; p<10⁻², Wilcoxon signed rank; n = 31).

In mPFC, previous work had shown increased correlation between SWRs and mPFC units outside of spindles. We calculated a peri-SWR histogram of mPFC unit activity with the same parameters used in Peyrache et al., 2011 (2 ms bins in a window of 400 ms before and after SWRs), and separated the peri-SWR histogram based on ripples that occurred within or outside spindle cycles (using a sample with the same number of coupled and uncoupled SWRs). We z-scored the resulting histograms to the same shuffled distribution (after randomizing all SWR timestamps). The resulting peri-SWR histograms are shown in Figure 6—figure supplement 3a; we found that there was no significant difference for the population of mPFC units between SWRs coupled or uncoupled to spindles (z-score 0.17 ± 1.26 for coupled-SWRs and 0.24 ± 0.86; p=0.46, Wilcoxon signed rank). However, the units with the strongest peri-SWRs correlograms (calculated with all SWRs), showed differential firing for SWRs coupled or uncoupled to spindles, such that their firing decreased with SWRs coupled to spindles (units with the strongest negative correlation had more negative values with SWRs uncoupled to spindles). We repeated the peri-SWR histograms for multi-unit spikes (all spikes recorded by one tetrode). With this analysis we found stronger correlation of mPFC spikes with SWRs coupled to spindles (z-score 0.83 ± 1.34 compared to uncoupled, 0.09 ± 1.53; p=0.002). We also found stronger correlations with CA1 firing (55.81 ± 28.40, 50.40 ± 30.51; p<10–3; Wilcoxon signed rank), and no significant difference in the thalamic tetrodes (−64 ± 1.42, 0.34 ± 0.95; p=0.17). Overall we find that, at the cell population level, both CA1 and mPFC increase their firing significantly during SWRs coupled to spindles, and that some mPFC cells may be more strongly modulated (up or down) with SWRs that are not coupled to spindles. This may reflect two different dynamics in mPFC, one driven by the spindle oscillation, which may engage cell population activity in coordination with CA1, and one, outside spindles, in which certain mPFC cells present larger changes in firing rate with SWRs.

Multi-region correlations after spatial exploration

To explore the possibility that awake behavior influences the multi-region coordination and thalamic contribution, we analyzed six sessions in which the rats (n = 3; two sessions from each rat) explored a radial maze (a 4-arm maze in one rat, 8-arm maze in the other two). The animals explored the maze freely for 30 mins, collecting pellet rewards at the end of each arm. In these sessions, we recorded an average of 52 min of sleep before and 78 min of sleep after spatial exploration to compare the non-REM sleep events and their correlation with single units before and after spatial experience. All the sessions showed an increase in the amplitude (peak-to-trough) and rate (per second) of SWRs during sleep after the maze exploration, as well as an increase in the amplitude of KCs (p=0.03, Wilcoxon signed rank), while spindles did not change significantly in number or amplitude (p>0.5). As a control for the possibility that these features change over time, we used another set of four longer recording sessions from two of the rats, in which we had recorded continuously (no maze exploration) for a similar amount of time. We compared the SWRs amplitude and rate in the first half to the second half of these sessions, finding no difference in amplitude nor rate between the first and second half of the recording (p>0.2); the total amount of sleep time was also not different in the first and second half of the recording, and not different from the pre-behavior or post-behavior sleep epochs in the other sessions with ‘run’ epochs (p>0.1 in all cases).

Although the number and amplitude of detected spindle cycles did not change, there was a significant increase in CA1 SWR power within spindle cycles (p=0.03, Wilcoxon signed rank), consistent with the overall enhancement of SWR activity in CA1. To find out if there is selective activation of behavior-relevant CA1 cells during spindles, we classified CA1 units recorded in these sessions (n = 42) into those that were active (n = 27) or not (n = 15) when the animal was on the maze (based on visual inspection during manual spike sorting, i.e., unit clusters present both on the maze and during sleep, or only in subsequent sleep; examples from one tetrode in Figure 7—figure supplement 1a). These two groups of CA1 units did not differ in firing rate in the sleep post-behavior (calculated as the mean number of spikes over the total sleep time, or as the reciprocal of the mean inter-spike interval; p>0.1, Mann-Whitney U-test). However, we found evidence that units that were active during maze exploration are more involved in SWR-spindle oscillations, as suggested by their higher firing rate and spiking at preferential spindle phases when they participate in SWRs nested in spindles. Figure 7a–b shows examples of the spindle temporal and phase organization of SWR spikes for two units active and two not active in run. Units that were active during maze exploration were more likely to be active within spindle cycles with high ripple power than units not active on the maze (p=0.009, Mann-Whitney U-test for the mean number of spikes in spindle cycles and p=0.039 for the percent of spindle cycles with unit spikes; examples in Figure 7a and population distributions in 7 c). A two-way ANOVA to compare the effect of the ‘RUN or not in RUN’ factor and ripple power being ‘low or high’ on the % of cycles with spikes and the mean number of spikes/spindle cycle found both factors to have a significant effect on the spike metrics (p<0.007 for ‘RUN compared to not in RUN’; p<0.01 for ‘low versus high ripple power’), with no interaction between the two factors (p>0.15). Pairwise comparisons found that the % of spindle cycles with CA1 unit spikes and the mean number of spikes were significantly higher for the combination of units ‘in RUN’ and cycles with high ripple power, compared to all other combinations (for the % of spindle cycles with CA1 spikes, p values were = 0.0002 when compared to units not in RUN and cycles with low ripple power; p=0.001 compared to RUN with low ripple power, and p=0.028 compared to non in RUN with high ripple power; p values for mean number of spikes were p=0.0002, 0.0007, 0.017 for the respective comparisons; multiple comparison Tukey-Kramer test).

Figure 7 with 1 supplement see all

Download asset Open asset

Likewise, the peri-SWR histograms of CA1 cells that had been active during run showed significantly higher peaks (mean 5.09 ± 4.23 spikes/s), compared to those units that were only active in the sleep post-behavior (mean 2.34 ± 2.09 spikes/s; p<10⁻², Mann-Whitney U-test; population distributions in Figure 7d, left). Although units active in run had higher firing rates within SWRs, we did not find differences between the maximum nor mean firing in SWRs that were nested within spindle cycles compared to a sample of an equal number of SWRs not coupled to spindles (p>0.3 for both units in run and units not active in run; Wilcoxon signed rank). However, the spikes fired by units active in run during SWRs nested in spindles occurred at preferential phases of the spindle oscillation; phase distribution examples are shown in Figure 7b for the units in 7a (circular mean = −1.48 ± 1.33; concentration = 0.63 ± 0.32; mean p value = 0.027 ± 0.05, Rayleigh test; n = 27 units). The phase modulation in the same spindle-nested SWRs, was not significant for units not active in run (circular mean = −1.20 ± 1.55; concentration = 0.51 ± 0.29; mean p value = 0.218 ± 0.28; n = 14 units, one unit was removed from this analysis due to low firing in SWRs). We used a two-way ANOVA to compare the effect of factors ‘RUN-not in RUN’ and ‘SWRs coupled-not coupled to spindles’ on phase locking, and found that Rayleigh’s Z was significantly higher for RUN units and when SWRs were coupled to spindles (p values 0.007 for ‘RUN-not in RUN’ and 0.015 for ‘SWRs coupled-not coupled to spindles’) with no interaction between the two factors (p=0.08). A pairwise comparison showed that the combination of units in RUN with spikes occurring in SWRs coupled to spindles had the highest Z values compared to all other cases (p=0.01 when compared to units not in RUN in SWRs coupled to spindles; p=0.0026 compared to units in RUN when not coupled to spindles; p=0.002 for units not in RUN and SWRs not coupled to spindles; multiple comparison Tukey-Kramer test). The distribution of Rayleigh’s Z for all conditions is included in Figure 7d. These results suggest an enhanced contribution of recently active CA1 units to SWRs, including those SWRs associated with thalamocortical spindles, when the spikes of units active in run are more strongly locked to the spindle oscillation.

We then characterized post-exploration changes in thalamic activity, and did not find significant changes induced by maze exploration in this sample of thalamic units (n = 9). Figure 8a shows an example of a thalamic unit with a post-KC rebound, which had a similar profile before and after behavior. Thalamic units also had similar peri-event histograms with mPFC spindle cycles before and after behavior (Figure 8b), and no change in spindle phase distribution mean or concentration values (p>0.3, Mann-Whitney U-test); the spindle modulation index increased on average after behavior although not significantly (0.52 ± 0.30 before maze exploration and 0.84 ± 1.03 after; p=0.43 Wilcoxon signed rank test). This is in contrast to an increase in the spindle correlations with SWRs and with CA1 units that were active on the maze. Figure 8c–d, shows examples of peri-spindle histograms of CA1 SWRs from the same session in which the thalamic unit in Figure 8b was recorded. Although the peri-spindle-trough histograms for the thalamic unit are similar before and after exploration (or even show a slight decrease in modulation), the CA1 unit (which was active during run) and the SWRs show enhanced modulation with spindles post-exploration. On average, we observed a trend for an increase in CA1 unit involvement in spindles after maze exploration. 60% of the CA1 units fired more spikes in high-ripple power spindle cycles after exploration compared to before (p=0.12), and fired in a larger proportion of high-ripple power spindle cycles (p=0.08; Wilcoxon signed rank; distributions for all the CA1 units in Figure 8e). A two-way ANOVA (to compare ‘Before-After’ experience and ‘Low-High ripple power’ spindles) yielded consistent results, finding a significant difference in engagement in spindles with ‘Low-High ripple’ power (p=0) and no significant effect for the ‘Before-After’ factor (p>0.06). It is possible that the lack of significant differences in cell contribution to high-ripple power before and after exploration is due to the low memory demand on this task.

Figure 8

Download asset Open asset

We tested the possibility of selective modulation of thalamic units with spindle cycles that engage CA1 units that were active during maze exploration (compared to spindle cycles with spikes from CA1 units that were not active during behavior). We calculated the peri-spindle histogram of thalamic spikes with spindle cycles containing spikes from the CA1 units that were active on the maze, and compared it to that of spindle cycles with spikes from CA1 units not active during exploration. We found no significant differences between the peak or the average firing values of the cross-correlograms between the two types of spindles and thalamic units (p>0.3, Wilcoxon signed rank), nor any differences between the mean number of thalamic spikes per spindle cycle or the percent of spindle cycles with spikes from thalamic units (p>0.1).

Another possibility is that thalamic cells specifically change their modulation with spindles when CA1 units are reactivating representations of spatial trajectories that were experienced while the animal was on the maze. During exploratory behavior, CA1 cells with neighboring place fields become active as part of theta sequences, showing strong phase-locked activation in the theta band when the animal is running (Figure 7—figure supplement 1b); these units then tend to be co-active in resting periods that follow behavior, often during SWRs (Davidson et al., 2009; Wilson and McNaughton, 1994). We selected pairs of CA1 units that had a clear co-modulation during run (n = 19 CA1 pairs; examples in Figure 7—figure supplement 1b), and calculated the peri-spindle histogram of thalamic spikes with spindles in which these pairs were active, compared to cycles in which they were not, and found no difference in thalamic activity (p>0.9). Because the sample of thalamic units in these sessions was small (n = 9), we repeated these analyses with all the background spikes (MUA) recorded by thalamic tetrodes, finding no significant changes (p>0.1).

In summary, these analyses suggest that spindles help organize the activation of recently active CA1 units during SWRs but failed to demonstrate a selective modulation of thalamic units with behaviorally relevant CA1 unit activation. It is possible that studying a larger sample of thalamic units, or after behavioral tasks with stronger memory demands, will reveal task-specific modulations between these three brain regions. Nonetheless, these results indicate that there is a spindle related organization of the spiking of task-relevant CA1 cells in spindle-nested SWRs. This could promote CA1-driven synaptic plasticity in mPFC networks during SWR reactivation during spindles.

Here, we report on the temporal correlation between cells in the limbic thalamus and the hallmark oscillations of non-REM sleep. We recorded single units and LFPs from functionally connected regions in the mPFC, CA1, and midline nuclei of the thalamus, including RE, MD and VM nuclei. We found that thalamic cells in these nuclei often reduce their firing and are more bursty with CA1 SWRs during non-REM sleep. Although firing in the thalamus becomes sparse, spikes are correlated with those of other thalamic cells and with LFP population events in mPFC and CA1. Specifically, thalamic units were phase-locked to the delta and spindle frequency bands recorded in deep layers of mPFC, and some thalamic cells showed a strong rebound of activity following the down-state of the mPFC slow oscillation (1–4 Hz). In addition, individual thalamic cells rarely contributed spikes in consecutive cycles of the spindle oscillation, but showed consistent co-activation with other thalamic cells within the period of spindle cycles. We also found that CA1 SWRs were strongly correlated with individual cycles of mPFC spindles, a correlation that was enhanced following spatial exploration. Furthermore, CA1 units that were active during behavior were more strongly activated during SWRs compared to units not active during behavior, and including during SWRs nested in spindle cycles, when they phase-locked to the underlying spindle oscillation. The evidence raises the hypothesis that sparse synchronicity in thalamic cell ensembles facilitates CA1-mPFC consolidation processes during non-REM sleep. The enhancement of ripple-spindle correlations after exploratory behavior and preferential engagement of recently active CA1 units in SWRs and their phase-locking to spindles, further suggest that individual spindle cycles provide time windows in which the interleaved activation of thalamic cells could promote incremental synaptic plasticity and facilitate the integration of recently acquired memory traces into neocortical networks.

Sparse firing of thalamic cells during spindles and its role in systems memory consolidation

Sparse firing of thalamocortical cells during spindles has been reported before (Huguenard and McCormick, 2007; Huguenard and McCormick, 2007; Steriade and Deschenes, 1984; von Krosigk et al., 1993; Crunelli et al., 2018; Steriade, 1994), but it was unclear how multiple cells in the dorsal thalamus would correlate with each other, and its relevance for the systems consolidation of memory has not been discussed. We find that pairs of thalamic cells are consistently phase-locked to the mPFC LFP within the period of spindle oscillations, suggesting that individual spindle cycles group and organize the activation of different thalamocortical ensembles at consistent phase lags. The repeated activation of thalamic cell ensembles through non-REM sleep could promote spike-timing-dependent synaptic plasticity. An intriguing hypothesis is that the sparse firing at the single-cell level and phase-locked firing of groups of cells could lead to the interleaved excitation of different sets of thalamocortical synapses in subsequent spindle cycles, and in coordination with the reactivation of recently active CA1 place cells. The activation of varying combinations of thalamocortical synapses with CA1 SWRs that engage recently active cells (Figure 9) may promote a broad range of synaptic associations at the neocortical level, make the neocortical trace more robust to noise, and facilitate generalization. Critically, the sparse activation could prevent plasticity saturation (Watt and Desai, 2010) and contribute to gradual synaptic changes that would allow for extended computations (Penagos et al., 2017) and facilitate continuous learning. One of the challenges in the development of artificial neural networks, has been to overcome the catastrophic interference that can occur during sequential training. While real-world settings require continuous learning, artificial network training algorithms are not yet optimized for these conditions. Sparsity and controlled weight changes have been suggested to facilitate robust representations and learning (Brunel and Hakim, 1999; Brunel and Hakim, 2008; Kirkpatrick et al., 2017; Hawkins and Ahmad, 2016). Future studies of thalamocortical sleep dynamics can provide clues on the biological implementation of continuous learning algorithms and help improve artificial networks and their applications.

Figure 9

Download asset Open asset

Behaviorally, the systems consolidation of episodic memory leads to the stabilization of memories and to the formation of mental models or schemas that have higher adaptive value for the organism than the original details of the experience (McClelland et al., 1995; Kumaran et al., 2016). The oscillatory interactions between hippocampus and neocortex during sleep support the extraction of statistical consistencies and integration of hippocampal memory traces into gist-like neocortical schemas that are applicable beyond the specific conditions in which a memory was acquired (Moscovitch et al., 2016; Penagos et al., 2017). Sparse thalamocortical population synchronicity during spindles may therefore promote incremental synaptic changes that optimize the interleaved extraction and incorporation of implicit information from recently acquired hippocampal memory traces into the neocortical representations.

Among the nuclei in the limbic thalamus, most of the evidence regarding their role on memory comes from studies of the RE nucleus. Pharmacological and optogenetic inactivation impaired performance in delayed spatial alternation memory tasks (Hallock et al., 2016; Hallock et al., 2013; Maisson et al., 2018; Viena et al., 2018). RE contributes to working memory by coordinating activity in CA1-mPFC, and through encoding trajectory specific and head direction information during spatial navigation (Ito et al., 2015; Jankowski et al., 2014). In addition, the mPFC modulates the firing mode of RE cells (Zimmerman and Grace, 2018) and can use this thalamic link to help retrieve memory traces (Jayachandran et al., 2019; Kupferschmidt and Gordon, 2018). Likewise, pharmacological blockade points to the role of RE in long-term memory (Loureiro et al., 2012). Although it has no direct projections to CA1, the MD nucleus also contributes to working memory possibly through its direct projections to mPFC (Bolkan et al., 2017). These studies provide key evidence to understand the role of higher order thalamic regions in active cognitive behavior. Recordings in RE suggest that, similar to its coordinating role in wakefulness, it may contribute to coupling oscillatory activity between mPFC and CA1 (Ferraris et al., 2018; Hauer et al., 2019). Our results extend these findings and point to spindle-specific modulation of cells in RE and other limbic nuclei of the thalamus as a function of CA1 ripple activity. The observed modulations in firing in higher order thalamic nuclei could enhance synaptic plasticity and memory. In addition, a few studies point to the role of RE in extinction (Ramanathan and Maren, 2019; Xu and Südhof, 2013); more generally, sleep can both increase and decrease synaptic strength (and recent evidence points to a role of sleep oscillations in forgetting [Langille, 2019; Kim et al., 2019]). Dynamic engagement in spindle oscillations could control which cell populations undergo up- or downregulation of synaptic strength.

In certain states, such as anesthesia or epilepsy (McCormick and Huguenard, 1992), cells in the thalamus may fire at the spindle population rhythm, and sparse synchronicity in the thalamus might be a distinguishing feature of normal resting states (Akeju et al., 2018). The disruption of thalamocortical oscillations has been proposed as a pathophysiological model for several neuropsychiatric disorders (Llinás et al., 1999; Fogerson and Huguenard, 2016). Oscillatory stimulation (mimicking the abnormally increased delta rhythm in schizophrenia) of thalamic projections to CA1 disrupts working memory in rats (Duan et al., 2015), and spindle irregularities (reduced or abnormally enhanced spindles) have been observed in children with neurodevelopmental disorders (Gruber and Wise, 2016), in schizophrenia (Wamsley et al., 2012), and in absence epilepsy (Lüttjohann and Pape, 2019). A careful balance in the excitatory and inhibitory thalamic networks is required to generate the appropriate population level synchronicity and function. Studying spindle dynamics at the cellular and population level will thus be critical to understand the pathophysiology behind these disorders, and to design therapeutic interventions that preserve normal synchronicity in the thalamus as well as the multi-region coordination that underlies cognitive function (Manoach et al., 2020).

We used eight male Long-Evans rats (400–500 g) in the experiments reported here (13 additional animals were implanted with multi-electrode arrays as part of the same project, but did not fulfil the criteria for inclusion in the current report due to lack of clear histological identification of electrode location in thalamus, or lack of electrophysiological signals in all three brain regions under study). Animals were housed individually, provided with food and water ad libitum, and monitored in a temperature-controlled room with a 12 hr light/dark cycle (lights on/off at 7:00 am/7:00 pm). These experiments were approved by the Committee on Animal Care at the Massachusetts Institute of Technology and conform to US National Institutes of Health guidelines for the care and use of laboratory animals.

Data files are available through the CRCNS website ('HC-24'; https://doi.org/10.6080/K0K35RVG); this includes data sets with raw data (LFP and units) recorded from 3 sessions, derived data (such as sleep-related events, like K-complexes, ripples and spindle cycles) as well as several matlab code to illustrate the main findings in the current manuscript. The data set and the documentation that describes it in detail is available through the CRCNS website ('HC-19' data set; http://crcns.org/data-sets/hc/hc-19).

1. 1. Varela C
   2. Wilson MA

   (2019) Collaborative Research in Computational Neuroscience

   Simultaneous extracellular recordings from midline thalamic nuclei, medial prefrontal cortex and CA1 from rats cycling through bouts of sleep and wakefulness.

   https://doi.org/10.6080/K0K35RVG

1. 1. Akeju O
   2. Hobbs LE
   3. Gao L
   4. Burns SM
   5. Pavone KJ
   6. Plummer GS
   7. Walsh EC
   8. Houle TT
   9. Kim SE
   10. Bianchi MT
   11. Ellenbogen JM
   12. Brown EN

   (2018) Dexmedetomidine promotes biomimetic non-rapid eye movement stage 3 sleep in humans: a pilot study

   Clinical Neurophysiology 129:69–78.

   https://doi.org/10.1016/j.clinph.2017.10.005

   • PubMed
   • Google Scholar
2. 1. Alitto HJ
   2. Weyand TG
   3. Usrey WM

   (2005) Distinct properties of stimulus-evoked bursts in the lateral geniculate nucleus

   Journal of Neuroscience 25:514–523.

   https://doi.org/10.1523/JNEUROSCI.3369-04.2005

   • PubMed
   • Google Scholar
3. 1. Batterink LJ
   2. Creery JD
   3. Paller KA

   (2016) Phase of spontaneous slow oscillations during sleep influences Memory-Related processing of auditory cues

   Journal of Neuroscience 36:1401–1409.

   https://doi.org/10.1523/JNEUROSCI.3175-15.2016

   • PubMed
   • Google Scholar
4. 1. Bolkan SS
   2. Stujenske JM
   3. Parnaudeau S
   4. Spellman TJ
   5. Rauffenbart C
   6. Abbas AI
   7. Harris AZ
   8. Gordon JA
   9. Kellendonk C

   (2017) Thalamic projections sustain prefrontal activity during working memory maintenance

   Nature Neuroscience 20:987–996.

   https://doi.org/10.1038/nn.4568

   • PubMed
   • Google Scholar
5. 1. Brunel N
   2. Hakim V

   (1999) Fast global oscillations in networks of integrate-and-fire neurons with low firing rates

   Neural Computation 11:1621–1671.

   https://doi.org/10.1162/089976699300016179

   • PubMed
   • Google Scholar
6. 1. Brunel N
   2. Hakim V

   (2008) Sparsely synchronized neuronal oscillations

   Chaos: An Interdisciplinary Journal of Nonlinear Science 18:015113.

   https://doi.org/10.1063/1.2779858

   • Google Scholar
7. 1. Buchanan KA
   2. Mellor JR

   (2010) The activity requirements for spike timing-dependent plasticity in the Hippocampus

   Frontiers in Synaptic Neuroscience 2:11.

   https://doi.org/10.3389/fnsyn.2010.00011

   • PubMed
   • Google Scholar
8. 1. Buzsáki G

   (1996) The hippocampo-neocortical dialogue

   Cerebral Cortex 6:81–92.

   https://doi.org/10.1093/cercor/6.2.81

   • PubMed
   • Google Scholar
9. 1. Buzsáki G
   2. Anastassiou CA
   3. Koch C

   (2012) The origin of extracellular fields and currents--EEG, ECoG, LFP and spikes

   Nature Reviews Neuroscience 13:407–420.

   https://doi.org/10.1038/nrn3241

   • PubMed
   • Google Scholar
10. 1. Cavdar S
    2. Onat FY
    3. Cakmak YO
    4. Yananli HR
    5. Gülçebi M
    6. Aker R

    (2008) The pathways connecting the hippocampal formation, the thalamic reuniens nucleus and the thalamic reticular nucleus in the rat

    Journal of Anatomy 212:249–256.

    https://doi.org/10.1111/j.1469-7580.2008.00858.x

    • PubMed
    • Google Scholar
11. 1. Chen Z
    2. Vijayan S
    3. Ching S
    4. Hale G
    5. Flores FJ
    6. Wilson MA
    7. Brown EN

    (2011) Assessing neuronal interactions of cell assemblies during general anesthesia

    Conf Proc Annu Int Conf IEEE Eng Med Biol Soc 2011:4175–4178.

    https://doi.org/10.1109/IEMBS.2011.6091036

    • Google Scholar
12. 1. Colgin LL
    2. Denninger T
    3. Fyhn M
    4. Hafting T
    5. Bonnevie T
    6. Jensen O
    7. Moser MB
    8. Moser EI

    (2009) Frequency of gamma oscillations routes flow of information in the Hippocampus

    Nature 462:353–357.

    https://doi.org/10.1038/nature08573

    • PubMed
    • Google Scholar
13. 1. Colgin LL

    (2011) Oscillations and hippocampal-prefrontal synchrony

    Current Opinion in Neurobiology 21:467–474.

    https://doi.org/10.1016/j.conb.2011.04.006

    • PubMed
    • Google Scholar
14. 1. Contreras D
    2. Destexhe A
    3. Sejnowski TJ
    4. Steriade M

    (1997) Spatiotemporal patterns of spindle oscillations in cortex and thalamus

    The Journal of Neuroscience 17:1179–1196.

    https://doi.org/10.1523/JNEUROSCI.17-03-01179.1997

    • Google Scholar
15. 1. Contreras D
    2. Steriade M

    (1995) Cellular basis of EEG slow rhythms: a study of dynamic corticothalamic relationships

    The Journal of Neuroscience 15:604–622.

    https://doi.org/10.1523/JNEUROSCI.15-01-00604.1995

    • Google Scholar
16. 1. Crick F

    (1984) Function of the thalamic reticular complex: the searchlight hypothesis

    PNAS 81:4586–4590.

    https://doi.org/10.1073/pnas.81.14.4586

    • PubMed
    • Google Scholar
17. 1. Crunelli V
    2. Lőrincz ML
    3. Connelly WM
    4. David F
    5. Hughes SW
    6. Lambert RC
    7. Leresche N
    8. Errington AC

    (2018) Dual function of thalamic low-vigilance state oscillations: rhythm-regulation and plasticity

    Nature Reviews Neuroscience 19:107–118.

    https://doi.org/10.1038/nrn.2017.151

    • PubMed
    • Google Scholar
18. 1. Davidson TJ
    2. Kloosterman F
    3. Wilson MA

    (2009) Hippocampal replay of extended experience

    Neuron 63:497–507.

    https://doi.org/10.1016/j.neuron.2009.07.027

    • PubMed
    • Google Scholar
19. 1. Destexhe A
    2. Hughes SW
    3. Rudolph M
    4. Crunelli V

    (2007) Are corticothalamic 'up' states fragments of wakefulness?

    Trends in Neurosciences 30:334–342.

    https://doi.org/10.1016/j.tins.2007.04.006

    • PubMed
    • Google Scholar
20. 1. Dolleman-van der Weel MJ
    2. Griffin AL
    3. Ito HT
    4. Shapiro ML
    5. Witter MP
    6. Vertes RP
    7. Allen TA

    (2019) The nucleus reuniens of the thalamus sits at the nexus of a Hippocampus and medial prefrontal cortex circuit enabling memory and behavior

    Learning & Memory 26:191–205.

    https://doi.org/10.1101/lm.048389.118

    • Google Scholar
21. 1. Duan AR
    2. Varela C
    3. Zhang Y
    4. Shen Y
    5. Xiong L
    6. Wilson MA
    7. Lisman J

    (2015) Delta frequency optogenetic stimulation of the thalamic nucleus reuniens is sufficient to produce working memory deficits: relevance to schizophrenia

    Biological Psychiatry 77:1098–1107.

    https://doi.org/10.1016/j.biopsych.2015.01.020

    • PubMed
    • Google Scholar
22. 1. Ego-Stengel V
    2. Wilson MA

    (2010) Disruption of ripple-associated hippocampal activity during rest impairs spatial learning in the rat

    Hippocampus 20:1–10.

    https://doi.org/10.1002/hipo.20707

    • PubMed
    • Google Scholar
23. 1. Fanselow EE
    2. Connors BW

    (2010) The roles of somatostatin-expressing (GIN) and fast-spiking inhibitory interneurons in UP-DOWN states of mouse neocortex

    Journal of Neurophysiology 104:596–606.

    https://doi.org/10.1152/jn.00206.2010

    • PubMed
    • Google Scholar
24. 1. Ferraris M
    2. Ghestem A
    3. Vicente AF
    4. Nallet-Khosrofian L
    5. Bernard C
    6. Quilichini PP

    (2018) The nucleus reuniens controls Long-Range Hippocampo–Prefrontal Gamma Synchronization during Slow Oscillations

    The Journal of Neuroscience 38:3026–3038.

    https://doi.org/10.1523/JNEUROSCI.3058-17.2018

    • Google Scholar
25. 1. Fogerson PM
    2. Huguenard JR

    (2016) Tapping the brakes: cellular and synaptic mechanisms that regulate thalamic oscillations

    Neuron 92:687–704.

    https://doi.org/10.1016/j.neuron.2016.10.024

    • PubMed
    • Google Scholar
26. 1. Frey U
    2. Morris RG

    (1997) Synaptic tagging and long-term potentiation

    Nature 385:533–536.

    https://doi.org/10.1038/385533a0

    • PubMed
    • Google Scholar
27. 1. Gardner RJ
    2. Hughes SW
    3. Jones MW

    (2013) Differential spike timing and phase dynamics of reticular thalamic and prefrontal cortical neuronal populations during sleep spindles

    Journal of Neuroscience 33:18469–18480.

    https://doi.org/10.1523/JNEUROSCI.2197-13.2013

    • Google Scholar
28. 1. Girardeau G
    2. Benchenane K
    3. Wiener SI
    4. Buzsáki G
    5. Zugaro MB

    (2009) Selective suppression of hippocampal ripples impairs spatial memory

    Nature Neuroscience 12:1222–1223.

    https://doi.org/10.1038/nn.2384

    • PubMed
    • Google Scholar
29. 1. Gonzalez CE
    2. Mak-McCully RA
    3. Rosen BQ
    4. Cash SS
    5. Chauvel PY
    6. Bastuji H
    7. Rey M
    8. Halgren E

    (2018) Theta bursts precede, and spindles follow, cortical and thalamic downstates in human NREM sleep

    The Journal of Neuroscience 38:9989–10001.

    https://doi.org/10.1523/JNEUROSCI.0476-18.2018

    • Google Scholar
30. 1. Gruber R
    2. Wise MS

    (2016) Sleep spindle characteristics in children with neurodevelopmental disorders and their relation to cognition

    Neural Plasticity 2016:1–27.

    https://doi.org/10.1155/2016/4724792

    • Google Scholar
31. 1. Guillery RW
    2. Sherman SM

    (2002) Thalamic relay functions and their role in corticocortical communication: generalizations from the visual system

    Neuron 33:163–175.

    https://doi.org/10.1016/s0896-6273(01)00582-7

    • PubMed
    • Google Scholar
32. 1. Gutierrez C
    2. Cox CL
    3. Rinzel J
    4. Sherman SM

    (2001) Dynamics of Low-Threshold spike activation in relay neurons of the cat lateral geniculate nucleus

    The Journal of Neuroscience 21:1022–1032.

    https://doi.org/10.1523/JNEUROSCI.21-03-01022.2001

    • Google Scholar
33. 1. Hallock HL
    2. Wang A
    3. Shaw CL
    4. Griffin AL

    (2013) Transient inactivation of the thalamic nucleus reuniens and rhomboid nucleus produces deficits of a working-memory dependent tactile-visual conditional discrimination task

    Behavioral Neuroscience 127:860–866.

    https://doi.org/10.1037/a0034653

    • Google Scholar
34. 1. Hallock HL
    2. Wang A
    3. Griffin AL

    (2016) Ventral midline thalamus is critical for Hippocampal-Prefrontal synchrony and spatial working memory

    Journal of Neuroscience 36:8372–8389.

    https://doi.org/10.1523/JNEUROSCI.0991-16.2016

    • PubMed
    • Google Scholar
35. 1. Hasselmo ME

    (2005) What is the function of hippocampal theta rhythm?--linking behavioral data to phasic properties of field potential and unit recording data

    Hippocampus 15:936–949.

    https://doi.org/10.1002/hipo.20116

    • PubMed
    • Google Scholar
36. 1. Hasselmo ME
    2. Stern CE

    (2014) Theta rhythm and the encoding and retrieval of space and time

    NeuroImage 85:656–666.

    https://doi.org/10.1016/j.neuroimage.2013.06.022

    • Google Scholar
37. 1. Hauer BE
    2. Pagliardini S
    3. Dickson CT

    (2019) The reuniens nucleus of the thalamus has an essential role in coordinating Slow-Wave activity between neocortex and Hippocampus

    Eneuro 6:ENEURO.0365-19.2019.

    https://doi.org/10.1523/ENEURO.0365-19.2019

    • PubMed
    • Google Scholar
38. 1. Hawkins J
    2. Ahmad S

    (2016) Why Neurons Have Thousands of Synapses, a Theory of Sequence Memory in Neocortex

    Frontiers in Neural Circuits 10:23.

    https://doi.org/10.3389/fncir.2016.00023

    • Google Scholar
39. 1. Hoover WB
    2. Vertes RP

    (2007) Anatomical analysis of afferent projections to the medial prefrontal cortex in the rat

    Brain Structure and Function 212:149–179.

    https://doi.org/10.1007/s00429-007-0150-4

    • PubMed
    • Google Scholar
40. 1. Hoover WB
    2. Vertes RP

    (2012) Collateral projections from nucleus reuniens of thalamus to Hippocampus and medial prefrontal cortex in the rat: a single and double retrograde fluorescent labeling study

    Brain Structure and Function 217:191–209.

    https://doi.org/10.1007/s00429-011-0345-6

    • PubMed
    • Google Scholar
41. 1. Huguenard JR
    2. McCormick DA

    (2007) Thalamic synchrony and dynamic regulation of global forebrain oscillations

    Trends in Neurosciences 30:350–356.

    https://doi.org/10.1016/j.tins.2007.05.007

    • PubMed
    • Google Scholar
42. 1. Isomura Y
    2. Sirota A
    3. Ozen S
    4. Montgomery S
    5. Mizuseki K
    6. Henze DA
    7. Buzsáki G

    (2006) Integration and segregation of activity in entorhinal-hippocampal subregions by neocortical slow oscillations

    Neuron 52:871–882.

    https://doi.org/10.1016/j.neuron.2006.10.023

    • PubMed
    • Google Scholar
43. 1. Ito HT
    2. Zhang SJ
    3. Witter MP
    4. Moser EI
    5. Moser MB

    (2015) A prefrontal-thalamo-hippocampal circuit for goal-directed spatial navigation

    Nature 522:50–55.

    https://doi.org/10.1038/nature14396

    • PubMed
    • Google Scholar
44. 1. Jahnsen H
    2. Llinás R

    (1984a) Ionic basis for the electro-responsiveness and oscillatory properties of guinea-pig thalamic neurones in vitro

    The Journal of Physiology 349:227–247.

    https://doi.org/10.1113/jphysiol.1984.sp015154

    • PubMed
    • Google Scholar
45. 1. Jahnsen H
    2. Llinás R

    (1984b) Electrophysiological properties of guinea-pig thalamic neurones: an in vitro study

    The Journal of Physiology 349:205–226.

    https://doi.org/10.1113/jphysiol.1984.sp015153

    • PubMed
    • Google Scholar
46. 1. Jankowski MM
    2. Islam MN
    3. Wright NF
    4. Vann SD
    5. Erichsen JT
    6. Aggleton JP
    7. O'Mara SM

    (2014) Nucleus reuniens of the thalamus contains head direction cells

    eLife 3:e03075.

    https://doi.org/10.7554/eLife.03075

    • Google Scholar
47. 1. Jaramillo J
    2. Mejias JF
    3. Wang XJ

    (2019) Engagement of Pulvino-cortical feedforward and feedback pathways in cognitive computations

    Neuron 101:321–336.

    https://doi.org/10.1016/j.neuron.2018.11.023

    • PubMed
    • Google Scholar
48. 1. Jayachandran M
    2. Linley SB
    3. Schlecht M
    4. Mahler SV
    5. Vertes RP
    6. Allen TA

    (2019) Prefrontal pathways provide Top-Down control of memory for sequences of events

    Cell Reports 28:640–654.

    https://doi.org/10.1016/j.celrep.2019.06.053

    • Google Scholar
49. 1. Ji D
    2. Wilson MA

    (2007) Coordinated memory replay in the visual cortex and Hippocampus during sleep

    Nature Neuroscience 10:100–107.

    https://doi.org/10.1038/nn1825

    • PubMed
    • Google Scholar
50. 1. Kim J
    2. Gulati T
    3. Ganguly K

    (2019) Competing roles of slow oscillations and Delta waves in memory consolidation versus forgetting

    Cell 179:514–526.

    https://doi.org/10.1016/j.cell.2019.08.040

    • PubMed
    • Google Scholar
51. 1. Kirkpatrick J
    2. Pascanu R
    3. Rabinowitz N
    4. Veness J
    5. Desjardins G
    6. Rusu AA
    7. Milan K
    8. Quan J
    9. Ramalho T
    10. Grabska-Barwinska A
    11. Hassabis D
    12. Clopath C
    13. Kumaran D
    14. Hadsell R

    (2017) Overcoming catastrophic forgetting in neural networks

    PNAS 114:3521–3526.

    https://doi.org/10.1073/pnas.1611835114

    • PubMed
    • Google Scholar
52. 1. Kloosterman F
    2. Davidson TJ
    3. Gomperts SN
    4. Layton SP
    5. Hale G
    6. Nguyen DP
    7. Wilson MA

    (2009) Micro-drive array for chronic in vivo recording: drive fabrication

    Journal of Visualized Experiments 20:1094.

    https://doi.org/10.3791/1094

    • Google Scholar
53. 1. Kumaran D
    2. Hassabis D
    3. McClelland JL

    (2016) What learning systems do intelligent agents need? complementary learning systems theory updated

    Trends in Cognitive Sciences 20:512–534.

    https://doi.org/10.1016/j.tics.2016.05.004

    • PubMed
    • Google Scholar
54. 1. Kupferschmidt DA
    2. Gordon JA

    (2018) The dynamics of disordered dialogue: prefrontal, hippocampal and thalamic miscommunication underlying working memory deficits in schizophrenia

    Brain and Neuroscience Advances 2:239821281877182.

    https://doi.org/10.1177/2398212818771821

    • Google Scholar
55. 1. Langille JJ

    (2019) Remembering to forget: a dual role for sleep oscillations in memory consolidation and forgetting

    Frontiers in Cellular Neuroscience 13:71.

    https://doi.org/10.3389/fncel.2019.00071

    • PubMed
    • Google Scholar
56. 1. Lara-Vásquez A
    2. Espinosa N
    3. Durán E
    4. Stockle M
    5. Fuentealba P

    (2016) Midline thalamic neurons are differentially engaged during Hippocampus network oscillations

    Scientific Reports 6:29807.

    https://doi.org/10.1038/srep29807

    • PubMed
    • Google Scholar
57. 1. Latchoumane CV
    2. Ngo HV
    3. Born J
    4. Shin HS

    (2017) Thalamic spindles promote memory formation during sleep through triple Phase-Locking of cortical, thalamic, and hippocampal rhythms

    Neuron 95:424–435.

    https://doi.org/10.1016/j.neuron.2017.06.025

    • PubMed
    • Google Scholar
58. 1. Leresche N
    2. Lambert RC

    (2017) T-type calcium channels in synaptic plasticity

    Channels 11:121–139.

    https://doi.org/10.1080/19336950.2016.1238992

    • PubMed
    • Google Scholar
59. 1. Lesica NA

    (2004) Encoding of natural scene movies by tonic and burst spikes in the lateral geniculate nucleus

    Journal of Neuroscience 24:10731–10740.

    https://doi.org/10.1523/JNEUROSCI.3059-04.2004

    • Google Scholar
60. 1. Llinás RR
    2. Ribary U
    3. Jeanmonod D
    4. Kronberg E
    5. Mitra PP

    (1999) Thalamocortical dysrhythmia: a neurological and neuropsychiatric syndrome characterized by magnetoencephalography

    PNAS 96:15222–15227.

    https://doi.org/10.1073/pnas.96.26.15222

    • PubMed
    • Google Scholar
61. 1. Logothetis NK
    2. Eschenko O
    3. Murayama Y
    4. Augath M
    5. Steudel T
    6. Evrard HC
    7. Besserve M
    8. Oeltermann A

    (2012) Hippocampal-cortical interaction during periods of subcortical silence

    Nature 491:547–553.

    https://doi.org/10.1038/nature11618

    • PubMed
    • Google Scholar
62. 1. Loureiro M
    2. Cholvin T
    3. Lopez J
    4. Merienne N
    5. Latreche A
    6. Cosquer B
    7. Geiger K
    8. Kelche C
    9. Cassel JC
    10. Pereira de Vasconcelos A

    (2012) The ventral midline thalamus (reuniens and rhomboid nuclei) contributes to the persistence of spatial memory in rats

    Journal of Neuroscience 32:9947–9959.

    https://doi.org/10.1523/JNEUROSCI.0410-12.2012

    • PubMed
    • Google Scholar
63. 1. Lu SM
    2. Guido W
    3. Sherman SM

    (1992) Effects of membrane voltage on receptive field properties of lateral geniculate neurons in the cat: contributions of the low-threshold Ca2+ conductance

    Journal of Neurophysiology 68:2185–2198.

    https://doi.org/10.1152/jn.1992.68.6.2185

    • PubMed
    • Google Scholar
64. 1. Lüttjohann A
    2. Pape HC

    (2019) Regional specificity of cortico-thalamic coupling strength and directionality during waxing and waning of spike and wave discharges

    Scientific Reports 9:2100.

    https://doi.org/10.1038/s41598-018-37985-7

    • PubMed
    • Google Scholar
65. 1. Maingret N
    2. Girardeau G
    3. Todorova R
    4. Goutierre M
    5. Zugaro M

    (2016) Hippocampo-cortical coupling mediates memory consolidation during sleep

    Nature Neuroscience 19:959–964.

    https://doi.org/10.1038/nn.4304

    • PubMed
    • Google Scholar
66. 1. Maisson DJ
    2. Gemzik ZM
    3. Griffin AL

    (2018) Optogenetic suppression of the nucleus reuniens selectively impairs encoding during spatial working memory

    Neurobiology of Learning and Memory 155:78–85.

    https://doi.org/10.1016/j.nlm.2018.06.010

    • PubMed
    • Google Scholar
67. 1. Manoach DS
    2. Mylonas D
    3. Baxter B

    (2020) Targeting sleep oscillations to improve memory in schizophrenia

    Schizophrenia Research 31:10.

    https://doi.org/10.1016/j.schres.2020.01.010

    • Google Scholar
68. 1. Manoach DS
    2. Stickgold R

    (2019) Abnormal sleep spindles, memory consolidation, and schizophrenia

    Annual Review of Clinical Psychology 15:451–479.

    https://doi.org/10.1146/annurev-clinpsy-050718-095754

    • PubMed
    • Google Scholar
69. 1. Marr D

    (1971) Simple memory: a theory for archicortex

    Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences 262:23–81.

    https://doi.org/10.1098/rstb.1971.0078

    • PubMed
    • Google Scholar
70. 1. Marshall L
    2. Helgadóttir H
    3. Mölle M
    4. Born J

    (2006) Boosting slow oscillations during sleep potentiates memory

    Nature 444:610–613.

    https://doi.org/10.1038/nature05278

    • PubMed
    • Google Scholar
71. 1. Maviel T
    2. Durkin TP
    3. Menzaghi F
    4. Bontempi B

    (2004) Sites of neocortical reorganization critical for remote spatial memory

    Science 305:96–99.

    https://doi.org/10.1126/science.1098180

    • PubMed
    • Google Scholar
72. 1. McClelland JL
    2. McNaughton BL
    3. O'Reilly RC

    (1995) Why there are complementary learning systems in the Hippocampus and neocortex: insights from the successes and failures of connectionist models of learning and memory

    Psychological Review 102:419–457.

    https://doi.org/10.1037/0033-295X.102.3.419

    • PubMed
    • Google Scholar
73. 1. McCormick DA
    2. Huguenard JR

    (1992) A model of the electrophysiological properties of thalamocortical relay neurons

    Journal of Neurophysiology 68:1384–1400.

    https://doi.org/10.1152/jn.1992.68.4.1384

    • Google Scholar
74. 1. Mölle M
    2. Yeshenko O
    3. Marshall L
    4. Sara SJ
    5. Born J

    (2006) Hippocampal sharp wave-ripples linked to slow oscillations in rat slow-wave sleep

    Journal of Neurophysiology 96:62–70.

    https://doi.org/10.1152/jn.00014.2006

    • PubMed
    • Google Scholar
75. 1. Morris RG

    (2006) Elements of a neurobiological theory of hippocampal function: the role of synaptic plasticity, synaptic tagging and schemas

    European Journal of Neuroscience 23:2829–2846.

    https://doi.org/10.1111/j.1460-9568.2006.04888.x

    • PubMed
    • Google Scholar
76. 1. Moscovitch M
    2. Cabeza R
    3. Winocur G
    4. Nadel L

    (2016) Episodic memory and beyond: the Hippocampus and neocortex in transformation

    Annual Review of Psychology 67:105–134.

    https://doi.org/10.1146/annurev-psych-113011-143733

    • PubMed
    • Google Scholar
77. 1. Neske GT

    (2015) The slow oscillation in cortical and thalamic networks: mechanisms and functions

    Frontiers in Neural Circuits 9:88.

    https://doi.org/10.3389/fncir.2015.00088

    • PubMed
    • Google Scholar
78. 1. Ngo HV
    2. Martinetz T
    3. Born J
    4. Mölle M

    (2013) Auditory closed-loop stimulation of the sleep slow oscillation enhances memory

    Neuron 78:545–553.

    https://doi.org/10.1016/j.neuron.2013.03.006

    • PubMed
    • Google Scholar
79. 1. Nguyen DP
    2. Layton SP
    3. Hale G
    4. Gomperts SN
    5. Davidson TJ
    6. Kloosterman F
    7. Wilson MA

    (2009) Micro-drive array for chronic in vivo recording: tetrode assembly

    Journal of Visualized Experiments 22:1098.

    https://doi.org/10.3791/1098

    • Google Scholar
80. 1. Ortuño T
    2. Grieve KL
    3. Cao R
    4. Cudeiro J
    5. Rivadulla C

    (2014) Bursting thalamic responses in awake monkey contribute to visual detection and are modulated by corticofugal feedback

    Frontiers in Behavioral Neuroscience 8:198.

    https://doi.org/10.3389/fnbeh.2014.00198

    • PubMed
    • Google Scholar
81. Book

    1. Paxinos G
    2. Watson C

    (1986)

    The Rat Brain in Stereotaxic Coordinates (Seventh Edition)

    Academic Press .

    • Google Scholar
82. 1. Penagos H
    2. Varela C
    3. Wilson MA

    (2017) Oscillations, neural computations and learning during wake and sleep

    Current Opinion in Neurobiology 44:193–201.

    https://doi.org/10.1016/j.conb.2017.05.009

    • PubMed
    • Google Scholar
83. 1. Perez-Reyes E
    2. Cribbs LL
    3. Daud A
    4. Lacerda AE
    5. Barclay J
    6. Williamson MP
    7. Fox M
    8. Rees M
    9. Lee JH

    (1998) Molecular characterization of a neuronal low-voltage-activated T-type calcium channel

    Nature 391:896–900.

    https://doi.org/10.1038/36110

    • PubMed
    • Google Scholar
84. 1. Peyrache A
    2. Battaglia FP
    3. Destexhe A

    (2011) Inhibition recruitment in prefrontal cortex during sleep spindles and gating of hippocampal inputs

    PNAS 108:17207–17212.

    https://doi.org/10.1073/pnas.1103612108

    • PubMed
    • Google Scholar
85. 1. Phillips JW
    2. Schulmann A
    3. Hara E
    4. Winnubst J
    5. Liu C
    6. Valakh V
    7. Wang L
    8. Shields BC
    9. Korff W
    10. Chandrashekar J
    11. Lemire AL
    12. Mensh B
    13. Dudman JT
    14. Nelson SB
    15. Hantman AW

    (2019) A repeated molecular architecture across thalamic pathways

    Nature Neuroscience 22:1925–1935.

    https://doi.org/10.1038/s41593-019-0483-3

    • PubMed
    • Google Scholar
86. 1. Puentes-Mestril C
    2. Roach J
    3. Niethard N
    4. Zochowski M
    5. Aton SJ

    (2019) How rhythms of the sleeping brain tune memory and synaptic plasticity

    Sleep 42:zsz095.

    https://doi.org/10.1093/sleep/zsz095

    • PubMed
    • Google Scholar
87. 1. Puig MV
    2. Ushimaru M
    3. Kawaguchi Y

    (2008) Two distinct activity patterns of fast-spiking interneurons during neocortical UP states

    PNAS 105:8428–8433.

    https://doi.org/10.1073/pnas.0712219105

    • PubMed
    • Google Scholar
88. 1. Purcell SM
    2. Manoach DS
    3. Demanuele C
    4. Cade BE
    5. Mariani S
    6. Cox R
    7. Panagiotaropoulou G
    8. Saxena R
    9. Pan JQ
    10. Smoller JW
    11. Redline S
    12. Stickgold R

    (2017) Characterizing sleep spindles in 11,630 individuals from the national sleep research resource

    Nature Communications 8:15930.

    https://doi.org/10.1038/ncomms15930

    • PubMed
    • Google Scholar
89. 1. Ramanathan KR
    2. Maren S

    (2019) Nucleus reuniens mediates the extinction of contextual fear conditioning

    Behavioural Brain Research 374:112114.

    https://doi.org/10.1016/j.bbr.2019.112114

    • PubMed
    • Google Scholar
90. 1. Redondo RL
    2. Morris RG

    (2011) Making memories last: the synaptic tagging and capture hypothesis

    Nature Reviews Neuroscience 12:17–30.

    https://doi.org/10.1038/nrn2963

    • PubMed
    • Google Scholar
91. 1. Rosanova M
    2. Ulrich D

    (2005) Pattern-specific associative long-term potentiation induced by a sleep spindle-related spike train

    Journal of Neuroscience 25:9398–9405.

    https://doi.org/10.1523/JNEUROSCI.2149-05.2005

    • PubMed
    • Google Scholar
92. 1. Rothschild G
    2. Eban E
    3. Frank LM

    (2017) A cortical-hippocampal-cortical loop of information processing during memory consolidation

    Nature Neuroscience 20:251–259.

    https://doi.org/10.1038/nn.4457

    • PubMed
    • Google Scholar
93. 1. Rubio-Garrido P
    2. Pérez-de-Manzo F
    3. Porrero C
    4. Galazo MJ
    5. Clascá F

    (2009) Thalamic input to distal apical dendrites in neocortical layer 1 is massive and highly convergent

    Cerebral Cortex 19:2380–2395.

    https://doi.org/10.1093/cercor/bhn259

    • PubMed
    • Google Scholar
94. 1. Saalmann YB

    (2014) Intralaminar and medial thalamic influence on cortical synchrony, information transmission and cognition

    Frontiers in Systems Neuroscience 8:83.

    https://doi.org/10.3389/fnsys.2014.00083

    • PubMed
    • Google Scholar
95. 1. Saalmann YB
    2. Kastner S

    (2009) Gain control in the visual thalamus during perception and cognition

    Current Opinion in Neurobiology 19:408–414.

    https://doi.org/10.1016/j.conb.2009.05.007

    • PubMed
    • Google Scholar
96. 1. Sadowski JH
    2. Jones MW
    3. Mellor JR

    (2016) Sharp-Wave ripples orchestrate the induction of synaptic plasticity during reactivation of place cell firing patterns in the Hippocampus

    Cell Reports 14:1916–1929.

    https://doi.org/10.1016/j.celrep.2016.01.061

    • PubMed
    • Google Scholar
97. 1. Schuman B
    2. Machold RP
    3. Hashikawa Y
    4. Fuzik J
    5. Fishell GJ
    6. Rudy B

    (2019) Four unique interneuron populations reside in neocortical layer 1

    The Journal of Neuroscience 39:125–139.

    https://doi.org/10.1523/JNEUROSCI.1613-18.2018

    • Google Scholar
98. 1. Seibt J
    2. Frank MG

    (2019) Primed to sleep: the dynamics of synaptic plasticity across brain states

    Frontiers in Systems Neuroscience 13:2.

    https://doi.org/10.3389/fnsys.2019.00002

    • PubMed
    • Google Scholar
99. 1. Siapas AG
    2. Lubenov EV
    3. Wilson MA

    (2005) Prefrontal phase locking to hippocampal theta oscillations

    Neuron 46:141–151.

    https://doi.org/10.1016/j.neuron.2005.02.028

    • PubMed
    • Google Scholar
100. 1. Siapas AG
     2. Wilson MA

     (1998) Coordinated interactions between hippocampal ripples and cortical spindles during slow-wave sleep

     Neuron 21:1123–1128.

     https://doi.org/10.1016/S0896-6273(00)80629-7

     • PubMed
     • Google Scholar
101. 1. Sirota A
     2. Csicsvari J
     3. Buhl D
     4. Buzsáki G

     (2003) Communication between neocortex and hippocampus during sleep in rodents

     PNAS 100:2065–2069.

     https://doi.org/10.1073/pnas.0437938100

     • PubMed
     • Google Scholar
102. 1. Sirota A
     2. Montgomery S
     3. Fujisawa S
     4. Isomura Y
     5. Zugaro M
     6. Buzsáki G

     (2008) Entrainment of neocortical neurons and gamma oscillations by the hippocampal theta rhythm

     Neuron 60:683–697.

     https://doi.org/10.1016/j.neuron.2008.09.014

     • PubMed
     • Google Scholar
103. 1. Slezia A
     2. Hangya B
     3. Ulbert I
     4. Acsady L

     (2011) Phase advancement and Nucleus-Specific timing of thalamocortical activity during slow cortical oscillation

     Journal of Neuroscience 31:607–617.

     https://doi.org/10.1523/JNEUROSCI.3375-10.2011

     • Google Scholar
104. 1. Staresina BP
     2. Bergmann TO
     3. Bonnefond M
     4. van der Meij R
     5. Jensen O
     6. Deuker L
     7. Elger CE
     8. Axmacher N
     9. Fell J

     (2015) Hierarchical nesting of slow oscillations, spindles and ripples in the human Hippocampus during sleep

     Nature Neuroscience 18:1679–1686.

     https://doi.org/10.1038/nn.4119

     • PubMed
     • Google Scholar
105. 1. Steriade M
     2. Deschênes M
     3. Domich L
     4. Mulle C

     (1985) Abolition of spindle oscillations in thalamic neurons disconnected from nucleus reticularis thalami

     Journal of Neurophysiology 54:1473–1497.

     https://doi.org/10.1152/jn.1985.54.6.1473

     • PubMed
     • Google Scholar
106. 1. Steriade M
     2. Domich L
     3. Oakson G
     4. Deschênes M

     (1987) The deafferented reticular thalamic nucleus generates spindle rhythmicity

     Journal of Neurophysiology 57:260–273.

     https://doi.org/10.1152/jn.1987.57.1.260

     • PubMed
     • Google Scholar
107. 1. Steriade M
     2. Nuñez A
     3. Amzica F

     (1993) Intracellular analysis of relations between the slow (< 1 Hz) Neocortical Oscillation and Other Sleep Rhythms of the Electroencephalogram

     J Neurosci Off J Soc Neurosci 13:3266–3283.

     https://doi.org/10.1523/JNEUROSCI.13-08-03266.1993

     • Google Scholar
108. 1. Steriade M

     (1994)

     Sleep oscillations and their blockage by activating systems

     Journal of Psychiatry & Neuroscience 19:354–358.

     • PubMed
     • Google Scholar
109. 1. Steriade M
     2. Deschenes M

     (1984) The thalamus as a neuronal oscillator

     Brain Research Reviews 320:1–63.

     https://doi.org/10.1016/0165-0173(84)90017-1

     • PubMed
     • Google Scholar
110. 1. Stickgold R
     2. Walker MP

     (2013) Sleep-dependent memory triage: evolving generalization through selective processing

     Nature Neuroscience 16:139–145.

     https://doi.org/10.1038/nn.3303

     • Google Scholar
111. 1. Swadlow HA
     2. Gusev AG

     (2001) The impact of 'bursting' thalamic impulses at a neocortical synapse

     Nature Neuroscience 4:402–408.

     https://doi.org/10.1038/86054

     • PubMed
     • Google Scholar
112. 1. Swanson LW

     (1981) A direct projection from Ammon's horn to prefrontal cortex in the rat

     Brain Research 217:150–154.

     https://doi.org/10.1016/0006-8993(81)90192-X

     • PubMed
     • Google Scholar
113. 1. Tononi G
     2. Cirelli C

     (2020) Sleep and synaptic down-selection

     European Journal of Neuroscience 51:413–421.

     https://doi.org/10.1111/ejn.14335

     • PubMed
     • Google Scholar
114. 1. Truccolo W
     2. Eden UT
     3. Fellows MR
     4. Donoghue JP
     5. Brown EN

     (2005) A point process framework for relating neural spiking activity to spiking history, neural ensemble, and extrinsic covariate effects

     Journal of Neurophysiology 93:1074–1089.

     https://doi.org/10.1152/jn.00697.2004

     • PubMed
     • Google Scholar
115. 1. Varela C

     (2014) Thalamic neuromodulation and its implications for executive networks

     Frontiers in Neural Circuits 8:69.

     https://doi.org/10.3389/fncir.2014.00069

     • PubMed
     • Google Scholar
116. 1. Varela C
     2. Kumar S
     3. Yang JY
     4. Wilson MA

     (2014) Anatomical substrates for direct interactions between Hippocampus, medial prefrontal cortex, and the thalamic nucleus reuniens

     Brain Structure and Function 219:911–929.

     https://doi.org/10.1007/s00429-013-0543-5

     • PubMed
     • Google Scholar
117. 1. Varela C
     2. Weiss S
     3. Meyer R
     4. Halassa M
     5. Biedenkapp J
     6. Wilson MA
     7. Goosens KA
     8. Bendor D

     (2016) Tracking the Time-Dependent role of the Hippocampus in memory recall using DREADDs

     PLOS ONE 11:e0154374.

     https://doi.org/10.1371/journal.pone.0154374

     • Google Scholar
118. Report

     1. Varela C
     2. Ahmad S

     (2019a) A Dendritic Mechanism for Dynamic Routing and Control in the Thalamus [Internet]

     Cosyne Abstracts.

     http://cosyne.org/c/index.php?title=Cosyne_19

     • Google Scholar
119. Report

     1. Varela C
     2. Wilson M

     (2019b)

     Simultaneous Extracellular Recordings From Midline Thalamic Nuclei, Medial Prefrontal Cortex and CA1 From Rats Cycling Through Bouts of Sleep and Wakefulness

     CRCNS.

     • Google Scholar
120. 1. Vertes RP
     2. Hoover WB
     3. Do Valle AC
     4. Sherman A
     5. Rodriguez JJ

     (2006) Efferent projections of reuniens and rhomboid nuclei of the thalamus in the rat

     The Journal of Comparative Neurology 499:768–796.

     https://doi.org/10.1002/cne.21135

     • PubMed
     • Google Scholar
121. 1. Vertes RP

     (2015) Major diencephalic inputs to the Hippocampus: supramammillary nucleus and nucleus reuniens – circuitry and function

     Progress in Brain Research 219:121–144.

     https://doi.org/10.1016/bs.pbr.2015.03.008

     • PubMed
     • Google Scholar
122. 1. Viena TD
     2. Linley SB
     3. Vertes RP

     (2018) Inactivation of nucleus reuniens impairs spatial working memory and behavioral flexibility in the rat

     Hippocampus 28:297–311.

     https://doi.org/10.1002/hipo.22831

     • PubMed
     • Google Scholar
123. 1. von Krosigk M
     2. Bal T
     3. McCormick DA

     (1993) Cellular mechanisms of a synchronized oscillation in the thalamus

     Science 261:361–364.

     https://doi.org/10.1126/science.8392750

     • PubMed
     • Google Scholar
124. 1. Wamsley EJ
     2. Tucker MA
     3. Shinn AK
     4. Ono KE
     5. McKinley SK
     6. Ely AV
     7. Goff DC
     8. Stickgold R
     9. Manoach DS

     (2012) Reduced sleep spindles and spindle coherence in schizophrenia: mechanisms of impaired memory consolidation?

     Biological Psychiatry 71:154–161.

     https://doi.org/10.1016/j.biopsych.2011.08.008

     • PubMed
     • Google Scholar
125. 1. Watt AJ
     2. Desai NS

     (2010) Homeostatic plasticity and STDP: keeping a neuron's Cool in a Fluctuating World

     Frontiers in Synaptic Neuroscience 2:5.

     https://doi.org/10.3389/fnsyn.2010.00005

     • PubMed
     • Google Scholar
126. 1. Wilson MA
     2. Varela C
     3. Remondes M

     (2015) Phase organization of network computations

     Current Opinion in Neurobiology 31:250–253.

     https://doi.org/10.1016/j.conb.2014.12.011

     • PubMed
     • Google Scholar
127. 1. Wilson MA
     2. McNaughton BL

     (1994) Reactivation of hippocampal ensemble memories during sleep

     Science 265:676–679.

     https://doi.org/10.1126/science.8036517

     • PubMed
     • Google Scholar
128. 1. Winer JR
     2. Mander BA
     3. Helfrich RF
     4. Maass A
     5. Harrison TM
     6. Baker SL
     7. Knight RT
     8. Jagust WJ
     9. Walker MP

     (2019) Sleep as a potential biomarker of tau and β-Amyloid burden in the human brain

     The Journal of Neuroscience 39:6315–6324.

     https://doi.org/10.1523/JNEUROSCI.0503-19.2019

     • PubMed
     • Google Scholar
129. Software

     1. Xclust WMA

     (2013) Wilson lab spike sorting, version eeb499a

     Github.

     https://github.com/wilsonlab/mwsoft64/tree/master/src/xclust\
130. 1. Xu W
     2. Südhof TC

     (2013) A neural circuit for memory specificity and generalization

     Science 339:1290–1295.

     https://doi.org/10.1126/science.1229534

     • PubMed
     • Google Scholar
131. 1. Yang M
     2. Logothetis NK
     3. Eschenko O

     (2019) Occurrence of hippocampal ripples is associated with activity suppression in the mediodorsal thalamic nucleus

     The Journal of Neuroscience 39:434–444.

     https://doi.org/10.1523/JNEUROSCI.2107-18.2018

     • PubMed
     • Google Scholar
132. 1. Yu JY
     2. Liu DF
     3. Loback A
     4. Grossrubatscher I
     5. Frank LM

     (2018) Specific hippocampal representations are linked to generalized cortical representations in memory

     Nature Communications 9:2209.

     https://doi.org/10.1038/s41467-018-04498-w

     • PubMed
     • Google Scholar
133. 1. Zikopoulos B
     2. Barbas H

     (2012) Pathways for emotions and attention converge on the thalamic reticular nucleus in primates

     Journal of Neuroscience 32:5338–5350.

     https://doi.org/10.1523/JNEUROSCI.4793-11.2012

     • PubMed
     • Google Scholar
134. 1. Zimmerman EC
     2. Grace AA

     (2018) Prefrontal cortex modulates firing pattern in the nucleus reuniens of the midline thalamus via distinct corticothalamic pathways

     European Journal of Neuroscience 48:3255–3272.

     https://doi.org/10.1111/ejn.14111

     • PubMed
     • Google Scholar

Author details

1. Carmen Varela

   1. Massachusetts Institute of Technology, Cambridge, United States
   2. Florida Atlantic University, Boca Raton, United States

   Contribution

   Conceptualization, Data curation, Software, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration

   For correspondence

   varelac@fau.edu

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-0398-2567

2. Matthew A Wilson

   Massachusetts Institute of Technology, Cambridge, United States

   Contribution

   Conceptualization, Supervision, Funding acquisition, Investigation, Project administration

   For correspondence

   mwilson@mit.edu

   Competing interests

   No competing interests declared

• 2,744

  views

• 444

  downloads

• 40

  citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.