DOWN state in the anterior cingulate and prelimbic areas in rats during immobility

Background UP-DOWN state is considering as a dominant electrographic pattern during immobility and slow wave sleep. This study is focused on the analysis of spatial distribution and neuronal correlates of the DOWN state in rats Methods Local field potentials and multiunit discharges were recorded bilaterally in the prefrontal cortex (PFC) and hippocampus of non-anesthetized, tethered rats (Sprague Dawley, males, weight 350-400g) with 256 channel, 4 shank silicon probes. We have focused our study on the analysis of the positive wave of slow oscillations (SOs), which is considered as the DOWN (silent) state of the UP-DOWN state in the anterior cingulate (AC), prelimbic (PL) areas of PFC and hippocampus during immobility. Results Our experiments showed that SOs occurred intermittently with a mean interval 1.4±0.8 (±SD) seconds. The SOs began with the DOWN state, and they were generated locally within AC or PL areas, or simultaneously in AC, PL and hippocampus bilaterally (generalized SOs). The DOWN state of local SOs in the AC was associated with a decreased rate of multiunit discharges. Similar waves in the PL area were associated with increased multiunit discharges. We observed high speed propagation of generalized SOs that occurred with 3-6ms delay within left and right PFC and less than 10ms delay between the PFC and CA1 area of hippocampus. All generalized SOs were associated with decreased multiunit discharges. Conclusion Our data support the hypothesis that neocortical networks are sufficient to generate focal SOs but the participation of external input is needed for occurrence of generalized SOs.


INTRODUCTION
Prefrontal cortex (PFC) is reported to participate in several different brain functions.
Subregions of PFC (anterior cingulate (AC) and prelimbic (PL)) have different anatomical connections, and play different roles in brain function. Compared to PL, AC receives a significantly larger percentage of intra cortical afferents and input from anterior, as well as "relay" nuclei of thalamus 59 , while PL receives more dense projections from the middle nuclei of thalamus 58,59 . Functionally, AC mostly plays a critical role in memory consolidation processes 19,20,35,41,46,47 while PL is necessary for retrieval of fear 4,11 .
Slow oscillations (SOs) are dominant patterns generated by PFC during immobility, slow wave sleep and light anesthesia. The two phases of SO reflect the existence of local network in a two states: the DOWN (silent) state is characterized by a suppression of the amplitude of gamma  activity associated with a decrease or silence of spike generation of recorded population of neurons 16,23,43,60 . The DOWN state is followed by the UP (active) state when the amplitude of gamma activity increases and it is associated by an increase of the rate of neuronal discharges 8,24,44,[51][52][53]55,60 .
Over the past 20 years the scientific community has devoted significant attention to the analysis of mechanisms of generation of the UP state, as well as its role in the process of consolidation of memory traces 1,10,23,24,33,38,42,55 . Much less attention has been paid to analysis of propagation of the DOWN state, specifically during sleep under un-anesthetized conditions. A goal of this study was to characterize SOs in AC and PL areas of PFC during immobility and their relation to SOs recorded in hippocampus. We carried out recordings with 4 shank silicon probes (total 256 recording sites) implanted bilaterally into the prefrontal cortex and hippocampus. We present data on the spatial distribution and neuronal correlates of the main local field potential (LFP) patterns in PFC, specifically focusing our attention on the DOWN state of the SOs. 4

MATERIALS AND METHODS
These experiments were carried out in accordance with relevant NIH guidelines and regulations.

All procedures were approved by the University of California Los Angeles Institutional Animal
Care and Use Committee (protocol 2000-153).

Silicon probe preparation.
Silicon microprobes each containing 64 recording contacts (type 64E, consisting of 64 100 µm 2 electrodes spaced linearly by 50 µm), were used for these experiments. The details of probes preparation were described elsewhere 45 . Each recording site was gold plated (Sifco 80535500) with constant-potential pulses (±2.5 V relative to a Pt wire reference, 1-5 s) to reduce impedance below 0.5 Megaohms to improve the signal-to-noise ratio 15

Surgery.
Experiments were performed on adult male Sprague-Dawley rats (Charles River, body weight 300-400g. Animals (n = 5) were anesthetized with 1.5-2.0 ml/min of isoflurane in 100% oxygen and fixed in a stereotactic surgery frame. Body temperature was kept between 36.6 and 38.0°C with a thermostatically controlled heating pad. The skull was opened and holes 1mm in diameter were drilled bilaterally above prefrontal cortex (AP=2.7; lateral 0.5mm from the sagittal suture), and above posterior hippocampus (AP=-2.5; lateral 4.0). After punching dura mater an array of silicon probes was attached to a stereotaxic manipulator, centered above each hole and slowly moved into the brain until the micromanipulator touched the surface of the skull. Ground and reference electrodes (stainless steel screws) were positioned in the cerebellum 2.0 mm posterior to lambda and 1.0 mm lateral from the sagittal suture. The system was fixed to the skull with dental cement. The incision was then sutured and treated with 0.25% bupivacaine as well as topical antibiotic ointment. Figure 1C illustrates a rat with implanted silicon probes connected to an amplifier

Data acquisition
On the fourth day after surgery electrical activity was recorded in each animal's home cage.
EEG/video data were recorded with wide bandwidth from 0.1 Hz to 3.0 kHz and sampled at 10 brains were then subjected to MRI imaging for localization of silicon probe tracks ( Figure 1D).

Data analysis.
Initially the identification of the location of silicon probes was performed on ex-vivo MRI sections (Fig. 1D). The fixed brain was scanned by the Bruker 7T animal scanner (Brain Research Institute, UCLA), using a 14mm diameter 1H radiofrequency coil. The 3D T2 RARE sequence with parameters of TR/TE = 1700/63 ms, 200×128×128 with voxel size 80 µm 3 and RARE factor = 6 was applied. Gamma event coupling analysis, described in our previous publications 3, 26 , was performed for more precise location of recording sites within PFC and hippocampus (Fig. 1E). It identifies the local networks and allows distinguishing the location of recording sites in the AC, and PL areas of PFC as well within different areas of hippocampus. In addition, current source density analysis (CSD) of the sharp wave-ripple complex identified layers within the CA1 area and dentate gyrus (DG) 7,61 . Specifically, the discrete second derivative across the depth electrode sites was computed on the averaged local field potentials (LFPs) with two-site spacing to reduce noise (Bragin et al., 1995(Bragin et al., , 1997Ylinen et al., 1995). The data were then further smoothed with spatial interpolation to produce the CSD map. Current sinks and sources associated with sharp waves and ripple oscillations in normal conditions provided precise landmarks for the identification of the recording sites located within pyramidal layer of the CA1 area and was extrapolated to identify layers in DG. In addition, unitary activity in the CA1 pyramidal layer provided further help for the depth calibration of the electrodes within hippocampus.
Data were then reviewed on the computer screen, and one hour files were selected from each animal, when it was immobile. The quality of selected files was verified by power spectrogram analysis of electrical activity recorded from the CA1 areas of hippocampus. If the spectrogram contained the peak at the frequency band 3-10Hz, which characteristic of exploration activity or rapid eye movement sleep 5,6 , they were rejected from the analysis. During 7 this immobile periods animals may be sleeping, and because we did not performed a sleep scoring, we will refer the selected files as recorded during the immobility.
Datapac 2K2 software (RUN Technologies, LLC) was used for detection of SOs.
Initially four recording sites located in the middle left and right AC and PL areas were selected.
After low pass 2Hz filtering (FIR, order 221) SOs exceeding 3 standard deviations (SD) were detected and then reviewed for artifacts. SOs that occurred in all 4 recording sites were considered as generalized SOs, and those that occurred in any single recording site were considered as local SOs. They formed different buffers for further analysis. For analysis of the delay of occurrence of SOs in different brain areas, generalized SOs were normalized and time difference between selected SOs was measured in the middle of the ascending slope of SOs using Datapac 2K2 software. We calculated a modified shift predictor for the event synchrony between left and right AC and PL, emulating procedures described in previous studies 36,49 . First, the distribution of delays from all measured events was calculated. Second, the random events were generated by assigning an equal number of events with bootstrap sampling of 1000 times.
Third, a normalized synchronization metric was computed by summing the measured values in a 1-3 ms window and then dividing by the corresponding area of the chance distribution. We then calculated whether the incidence of delays less than 3ms was greater or equal to the chance.
Multiunit discharges (MUD) of neuronal activity were selected after high pass 600Hz filtering and setting the threshold at 2 SD. For voltage versus depth analysis all detected events were averaged and values of the peak amplitudes were measured at each recording site. The values were normalized and plotted against each recording site.
Statistics. General descriptive statistics were performed to summarize the mean and standard deviations of two types of SOs. Histograms were created to describe the distributions of inter SOs intervals in the datasets. All the statistical analyses that performed in the current study were based on a significant criterion of p<0.05.

RESULTS.
Analysis of silicon probe tracks revealed that recording sites were located within layers 3-5 a distance of 2mm within AC and 1.5 mm within PL (see Figure 1D as example).
During the slow wave period, SOs occurred intermittently (Figure 2A (2) generalized, which occurred simultaneously and bilaterally in the AC and PL areas of the prefrontal cortex, and in the hippocampus.

Local slow oscillations.
Two types of local SOs were observed in the PFC. The first type, which constituted 24% of all SOs, had maximum amplitude in the AC area of prefrontal cortex. The second type, which constituted 11% of all SOs, had a maximum amplitude in PL (Figure 3,A,D). Figure 3C illustrates there was no significant change in the lower part (Figure 3, E).

Generalized SOs.
Generalized SOs were observed in 65% of all recorded SOs. They occurred simultaneously in left and right PFC as well as in the hippocampus (Fig. 4)   In both areas of prefrontal cortex these generalized SOs were associated with suppression of multiunit discharges (MUD) (histograms in the Figure 4,A & B, 5A). This suppression of MUD was more prominent close to the border between AC and PL areas. In the hippocampus the suppression of MUD was less prominent, was observed only in the CA1 (Fig. 4C, 5B) and was preceded by an increase of MUD associated with the negative wave in the hilus of the DG ( c1-c7, arrows). This almost simultaneous occurrence does not depend on the reference electrode, because it occurs only between some recording sites, but not between others, and it has a similar delay even in cases when it was recorded relative to other reference electrode (Figure 7 d, recordings 3&43 red and black were recorded relative to one reference electrode and purple and blue relative to another reference electrode). We calculated a modified shift predictor for the event synchrony between left and right AC and PL. The slope delays from a total of 1158 events were measured, and the random events were generated by assigning an equal number of events with bootstrap sampling of 1000 times. The histograms in Figure 7b show the normalized probability of the measured event delays (blue bars) and the bootstrap random sampled data (black bars). The occurrence of SOs within a 1-3 ms window is 4.6 times higher than chance.
To evaluate the synchrony of occurrence of SOs, we further estimate the SOs peak latencies between left and right AC areas in six different paired recording sites (named ch1 -ch6). The one-way ANOVA for the latency data indicated no statistically significant differences between LAC and RAC in all 6 selected recording sites (F (5, 876) = 1.38, p = 0.23, Figure 8).
Detailed results are summarized in the table 1. Our data indicated that the latency of SOs between LAC and RAC shows no changes in response to different recording areas, where we calculated delay of individual DOWN events. Examples of these events are presented in Figure   7a, were variability of slopes of ascending and descending phases is more than 30 ms. The histograms in Figure 7b show the normalized probability of the measured event delays (blue bars) and the bootstrap random sampled data (black bars). The occurrence of SOs within a 1-3 ms window is 4.6 times higher than chance.  The inset is the graphic representation of the measured delays, which is significantly shorter (p<0.001) in comparison to randomly selected events. c1-c8 examples of individual SOs recorded in these areas with recording sites separated by 200µm. Arrows indicate those events where time difference between left and right sites were less than 10ms. d. Ascending slope of SO b4. Recoding sites 3&43 colored as blue and purple are the same events as red and black, but recorded relatively to the reference electrode located in the thalamus.

DISCUSSION
The main finding of this study is a demonstration that during immobility generalized SOs Several publications showed that SOs is a wide spread electrographic pattern involved both neocortical and hippocampal areas of the brain 24,30,40,52,55 . Our findings extend these data by showing that this wide spread pattern begins with an initial DOWN state. It does not contradict the existing hypothesis of wide spread excitatory front of UP state 24 , and suggest that this initial UP state occurs as a rebound after an initial DOWN state According to numerous publications [see reviews 12,13,33 ], during the awake state and rapid eye movement sleep the membrane of neurons is slightly depolarized. The   16 electrophysiologically visible UP-DOWN state in the neocortex occurred when depolarizing influences from the sensory systems and brainstem areas decreased.
It is hypothesized that the DOWN state has occurred as a result of disfacilitation and temporal absence of synaptic activity 2,9,50,56 . Usually the occurrence of DOWN state is considered as a process of recovery after termination of the UP state 27,43,55 . The possible mechanism of the DOWN state as recovery process after the UP state is turning on of the hyperpolarization activated low threshold CA 2+ and Ih currents 28,32,48  that the DOWN state may occur initially form the baseline activity and is followed by the UP state. We cannot exclude the possibility that these initially occurring DOWN state has the same mechanisms it was proposed in previous studies, however it is not a recovery from the UP state.
The absence of the increase the synchrony of MUA before the DOWN state allow to suggest that DOWN state may occur as a result of initial inhibition of the neuronal activity. 16 showed that specific type of interneurons (GIN cells) fired persistently during the DOWN state. So the DOWN state could be an initial step in the occurrence of repetitive of UP-DOWN state activity.
Local SOs. Demonstration of local SOs in different areas of neocortex is in accordance with many previous data indicating the existence of local SOs and support the hypothesis that local neocortical networks are sufficient to generate SOs 23,30,40,43,54,60 . In the AC local DOWN state involved neuronal populations along 2mm of this area in the layers 3-5 and was associated with suppression of MUD during the ascending and descending phases of SOs. At present, mechanisms triggering the DOWN state remain unclear. 27 described an occurrence of long (100-300ms) IPSPs in the period about 250 ms before the onset of the DOWN state, which suggests the existence of interneurons, that trigger the silent period within a surrounding network.
Potentially, it may occur with involvement of GABAa and GABAb receptors 29 or D(1) -like dopamine receptors 31 . Recently two groups showed that inhibitory somatostatin containing interneurons (SOM-Ins) in the superficial layers of neocortex, which may be are Martinotti neurons 17 , play a key role in moving the network into the silent DOWN state. 21 found that chemogenetic activation of SOM-Ins increases slow wave activity. 39 , showed that during the transition into the SOs DOWN state SOM-Ins, but not parvalbumin-positive interneurons or pyramidal cells, increase frequency of discharges. Martinotti cells may form an electrically coupled network that exerts a coherent inhibitory influence on its targets and to play a role as "first responders" when cortical excitatory activity increases 17,25  However, as we described in our experiments, during immobility SOs occur intermittently and begin with the DOWN state. It cannot be a simultaneous release from the UP state and the initial inhibitory process should be involved in generation of intermittent SOs. These data indicate the existence of a common trigger, most likely in subcortical areas, which synchronize the 18 occurrence of generalized SOs. This hypothesis was suggested by 60 and was confirmed in 22 experiments . This common trigger, may interfere with local mechanisms triggering the DOWN phase and synchronizing their simultaneous occurrence of SOs in multiple brain areas. It was shown that D1-like receptors are essential for the occurrence of SOs 31 , so a potential candidate for this role could be substantia nigra, which has diffuse projections to multiple brain areas.
Another potential candidate could be the central lateral thalamic nucleus, which is as a key region in the subcortical arousal systems for maintaining the level of consciousness 18,37 . Both of these areas could synchronize local neocortical networks.
In conclusion, the results of our study have raised a number of questions for future experiments. Among them are unravel the mechanisms for simultaneous entry into the DOWN state in multiple brain areas, and what is the functional role of local SOs in the AC and PL cortices.