Decreased phase information transfer from the mPFC to the BLA: During exploratory behavior in CUMS rats

1. Introduction

Depression, a common mental disorder affecting an estimated 3.8% of the global population, is a significant contributor to the overall global burden of disease (World Health, 2021). It is characterized by persistent sadness and a lack of interest or pleasure in previously rewarding or enjoyable activities. It can also disturb sleep and appetite. In the worst cases, depression can even lead to suicide. Although depression is a highly heterogeneous syndrome, it is clear that negative stimulation, such as exposure to stress, increases the risk of depression (Ma et al., 2021).

The majority of the literature on depression confirms that one of the main characteristics of depression is a loss of excitatory prefrontal cortical control over the core limbic structures, such as the amygdala, resulting in aberrant processing of rewarding and aversive behaviors (Russo and Nestler, 2013). Accumulating evidence suggests that the medial prefrontal cortex (mPFC) is a key brain region in the regulation of behaviors and emotions (Pfarr et al., 2018; Chen et al., 2021). The basolateral amygdala (BLA), whose aberrations are a major contributor to psychiatric illnesses, such as depression and anxiety, is a pivotal hub that integrates sensory information from the cortical and subcortical areas and drives mood and emotional expression (Munshi and Rosenkranz, 2018; Zheng et al., 2021). A large body of anatomical evidence indicates that structural and functional synaptic changes, including reduced cortical volume, dendritic atrophy, and dendritic spine loss, occur primarily in the mPFC after stress (Gilabert-Juan et al., 2013) while the BLA is inversely altered (Mitra et al., 2005). It is widely accepted that dysregulation of the PFC amygdala drives stress-induced emotional pathology (Hultman et al., 2016). The study by Liu et al. (2020) reveals a dorsal mPFC (dmPFC) to BLA dysregulation based on the connectivity of specific cells with the mPFC in stress-induced anxiety. Another study demonstrated that activation of the mPFC Drd1 pyramidal cells or stimulation of their terminals in the BLA can produce an antidepressant response (Hare et al., 2019). In addition, the mPFC has been shown to suppress depression-induced amygdala-mediated hyperactive affective responses by recruiting BLA inhibitory interneurons (Rosenkranz and Grace, 2001; Bertholomey et al., 2022). Evidence has shown that the activation of Ventromedial prefrontal cortex (vmPFC)–pBLA inputs diminishes chronic unpredictable mild stress- (CUMS-) induced depression- and anxiety-like behaviors via activation of the pBLA Calb1 neuron (Yu et al., 2022). These results suggest that mPFC–BLA innervation regulates affective disorders in a BLA inhibition-dependent manner and that depression-like behavior may be associated with a circuit.

Exploration, a crucial component of human and animal behaviors, has been found to be declined in stressed rats (Jacinto et al., 2013; Botta et al., 2020; Dong et al., 2021). The mPFC and the BLA also appear in exploratory behavior. Indeed, it was previously found that stressed rats showed novel exploratory behavioral impairment and that stress affected theta oscillatory coherence within the amygdala-PFC network in the exploration of a novel environment (Jacinto et al., 2013). According to research, increased amplitude synchronization of the mPFC and the BLA in the theta range is related to learned fear and innate anxiety (Likhtik et al., 2014). A previous study reported that information transmission within the mPFC and the BLA predominates the expression of normal emotions (Bao et al., 2021). Subsequently, CUMS rats showed debased connectivity of the mPFC network and reduced information flow from the mPFC to BLA during free exploration (Qi et al., 2020). These outcomes reflect the consequences of amplitude information. In general, it has been suggested that neural oscillations have three basic properties: amplitude, frequency, and phase (Watrous et al., 2015). In contrast to numerous studies focusing on amplitude, phase as a mode of information transfer is largely unstudied for its role in cognition and mood. Philippe used mutual information to organize the different contributions of amplitude, phase, and frequency of oscillation in visual information coding and revealed that phase encoded more visual information than amplitude (Schyns et al., 2011). Phase decoding has the highest accuracy rate among auditory stimulus decoding, indicating that phase contains more abundant task information (Ng et al., 2013). The aforementioned studies reported the importance of phase information transfer in cognitive and emotional tasks, especially in cooperative processing between the brain regions. As previously described, it is central to determining how phase information transmission from the mPFC to the BLA changes during exploratory behavior in depressed rats.

To address this issue, we employed a CUMS-induced rodent depression model. The rats were allowed 10 min of free exploration in an open field, which was the most common test to investigate curiosity, anxiety, and behavior in models of psychiatric disorders (Fonio et al., 2009; Belovicova et al., 2017). Local field potentials (LFPs), which were considered a vital tool to analyse brain network, were obtained synchronously from the mPFC and the BLA in the open field test (OFT; Marceglia et al., 2007). We applied the weighted phase lag index (WPLI) to investigate the interaction between the mPFC and the BLA during exploratory behavior and determined the characteristic frequency of the phase information. Then, we extracted the phase component from LFPs and calculated the phase transfer entropy (PTE) from the mPFC to the BLA in the control and CUMS groups. This study aimed to provide insights into the phase transmission mechanism of the mPFC to BLA circuit in the context of exploratory behavioral disorders for CUMS-induced depression. This might provide a theoretical basis for the alleviation of depressive symptoms through phase-specific oscillatory neuromodulation in the mPFC.

2. Materials and methods 2.1. Animals

Adult male Sprague–Dawley rats (aged 10–12 weeks, weighing 300–350 g) were used in all experiments. The rats were purchased from the Experimental Animal Center of Tianjin Medical University in China and bred at the animal facility of Tianjin Medical University. The rats were housed on a 12-h light/dark cycle (lights on at 7:00 a.m.) with groups of 3–4 rats per cage at a stable temperature (23°C ± 2°C) and constant humidity (50% ± 5%), except for the CUMS model. All animals had free access to food and water. All procedures were conducted in accordance with the Guide for the Care and Use of Laboratory Animals and the Tianjin Medical University Animal Care and Use Committee (license number: TMUaMEC2021059).

2.2. Chronic unpredictable mild stress

According to a previous study (Willner, 1997; Higuchi et al., 2016), a CUMS procedure was performed. Briefly, rats were randomly exposed for 3 weeks to a variety of unpredictable mild stressors: 1-min tail pinch, 5-min cold swimming (4°C), 24-h food deprivation, 24-h water deprivation, inverted light/dark cycle, 12-h stroboscopic light flash (50 Hz flash frequency), 12-h cage tilting, and 12-h soiled bedding. Each stressor was administered once a week. Non-stressed rats were housed under normal conditions.

The sucrose preference test (SPT) and the forced swimming test (FST) were conducted to validate this model. CUMS-induced depressed rats were successfully prepared by decreasing the sucrose preference rate in the SPT and increasing the immobility time in the FST (Figure 1A).

www.frontiersin.org

Figure 1. Experimental scheme. (A) Scheduling for chronic unpredictable mild stress (CUMS) and validation of the CUMS model. (B) Experimental scheme. Both control and CUMS rats were implanted with multichannel microelectrode arrays in the medial prefrontal cortex (mPFC) and basolateral[[Inline Image]] amygdala (BLA). Neural signals were obtained while rats were roaming freely in the open field test (OFT). (C) Histological verification of recording sites in the mPFC, PrL region (top, adapted from Wang) and BLA (bottom). Local field potentials (LFPs) data during exploration (−2 to 2 s) in the mPFC (top) and BLA (bottom), respectively. The square refers to the trajectory of the electrode tips. Blue line, reference point (0 s).

2.3. Open field test

A gray zinc-aluminum alloy chamber (length × width × height, 100 cm × 100 cm × 50 cm) was used as an open field and divided into a central zone (60 cm × 60 cm) and a peripheral zone. During the test, rats were first placed in the center of the field with normal illumination (185 lux) and allowed to explore freely for 10 min. Activities of all rats were recorded using an overhead video monitoring system (Sony, Japan). Center entries, time spent in the central area, rearing times, and the total movement distance were tracked and analyzed using the behavioral analysis software (Ethovision XT8.5, Noldus, The Netherlands). After each test, the open field was scrubbed with 75% ethanol.

When the rats crossed from the periphery to the center, including 2 s in the peripheral zone and 2 s in the central zone, a trial of exploratory behavior was detected and the transition occurred in 0 s.

2.4. Surgery and electrophysiological recording

After the successful preparation of these two groups of rats, they were anesthetized with pentobarbital sodium (40 mg/kg, i.p.). They were kept at a temperature of 36°C using a heating pad. Rats were implanted with two multichannel microelectrode arrays (2 × 8 configuration, 50 μm nickel-chromium wires, < 1 MΩ) in the mPFC [+3.5 mm anteroposterior (AP), −0.6 mm mediolateral (ML), and −2.8 mm dorsoventral (DV) from the dura mater] and the BLA (−2.46 mm AP, −4.8 mm ML, and −8.5 mm DV from the dura mater. The aforementioned coordinates were obtained according to a rat brain atlas. The reference electrode was fixed to the skull with screws and dental cement. Postoperatively, in case of incision infection, erythromycin eye ointment and iodophor were applied to the wound suture. Animals were given sustained-release buprenorphine (1 mg/kg) as an analgesic and were allowed to recover for 1 week.

In vivo electrophysiological recordings were obtained synchronously from the signal acquisition system during a 10-min open-field exploration in which rats were engaged in activities recorded by overhead video (Figure 1B). Wireline-linked multichannel microelectrode arrays to the neural data acquisition system and neural signals from the mPFC and the BLA were obtained when the OFT was turned on. Wideband neural signals were recorded with a neurophysiological data acquisition system (Plexon, USA). LFPs were amplified (gain: 5,000), bandpass filtered (0.3–120 Hz), and sampled at 2 kHz.

2.5. Data analysis

Original LFPs were preprocessed with a 50-Hz notch filter and polynomial fitting to eliminate interference and correct to baseline (Figure 1C). In this study, LFPs were divided into five frequency bands: delta (0.5–4 Hz), theta (4–12 Hz), beta (13–30 Hz), low gamma (30–60 Hz), and high gamma (60–100 Hz). We analyzed the data from each trial. A trial of exploratory behavior was shown as a movement in which the rat crossed from the periphery to the center, including 2 s in the peripheral zone (-2–0 s) and 2 s in the central zone (0–2 s), with the occurrence of the transition in 0 s.

2.5.1. Weighted phase lag index

The weighted phase lag index, a novel measure of phase synchronization, is based on the imaginary cross-spectral component (Vinck et al., 2011). The phase lag index (PLI), proposed by the Stam group, estimated a non-equal probability of phase leads and lags between signals from each brain area, irrespective of the magnitude of phase leads and lags (Stam et al., 2007). However, the discontinuity in this index might hinder the sensitivity of PLI to noise and volume conduction, as small perturbed turn phases lead to phase lags and vice versa. The problem is more serious in the case of minor synchronization effects. Compared with PLI, WPLI extends PLI by weighting the contributions of phase leads and phase lags by the magnitude of the imaginary cross-spectral component and consequently alleviating the aforementioned discontinuity. To summarize, we can conclude two main advantages of WPLI: its lower sensitivity to additional noise and its high ability to detect true phase synchronization from real neural data.

To confirm the characteristic frequency band of the phase, we computed the WPLI between the mPFC and the BLA as follows. First, the cross-spectrum was computed using X and Y, which were transformed from x (a signal from the mPFC) and y (a signal from the BLA) by the short-time Fourier transform:

Xi(t,f)=∫−∞+∞[xi(τ)h(τ−t)]e−j2πfτdτ    (1) Yi(t,f)=∫−∞+∞[yi(τ)h(τ−t)]e−j2πfτdτ    (2)

where Xi(t, f) and Yi(t, f) are short-time Fourier forms of xi(t) and yi(t) from trial i and h(τ − t) is the Hamming window with a window length of 250 ms and a moving step of 50 ms, according to the data length from each trial. Moreover, the width of the frequency interval is set as 1 Hz. And, the cross-spectrum is defined as

Ci(t,f)=Xi(t,f)Yi*(t,f),    (3)

where Yi*(t,f) indicates the complex conjugate of Yi(t, f ).

Then, the WPLI between the mPFC and BLA is calculated as:

WPLI=|∑i=1nℑ(Ci(t,f))|∑i=1n|ℑ(Ci(t,f))|,    (4)

where ℑ(·) refers to extracting the imaginary component and Ci(t, f) is the cross-spectrum, as mentioned in Equation (3).

The weighted phase lag index ranges from 0 to 1. A WPLI of 1 signifies phase locking, where the instantaneous phase of one signal uniformly precedes or lags behind the other signals, while a WPLI of 0 indicates no coupling. If the phase lead or lag of the two signals is random, the WPLI will be close to 0. In conclusion, the higher the phase synchronization, the higher the WPLI.

2.5.2. Phase transfer entropy

Phase transfer entropy, a measure to detect the strength and direction of connectivity between neuronal oscillations, was first presented by Palus and applied to electroencephalogram (EEG) signal analysis by Lobier et al. (2014). In the work referenced earlier, the Lobier group verified that PTE was robust to nuisance factors inherent in the neural signals, such as noise and linear mixing. Its properties are high computational efficiency and a limited number of a priori parameters, which reduce computational costs and considerably enhance computational accuracy. A comparison of PTE and real-valued TE (broadband and narrowband TE) showed that PTE is more effective in discovering band-limited directed interactions. Thus, PTE is suitable for estimating directional phase-based connectivity in large-scale investigations. The flow of the phase information between time series extracted from neural signals is quantified by referring to the instantaneous phase sequence as transfer entropy input (Hillebrand et al., 2016).

First, a zero-phase-shift digital filter was used to obtain the specific frequency LFPs, and their instantaneous phase series was computed by the Hilbert transform as follows:

H(S(t))=1π∫−∞+∞S(τ)t−τdτ    (5) θ(t)=arctanH(S(t))S(t)    (6)

where S(t) represents the filtered LSPs for a given frequency, H(S(t)) is the Hilbert transform of S(t), and θ(t) is the instantaneous phase of S(t ).

Then, we define PTE from channels i to j with a given time lag δ as:

PTEij=SH(θj(t),θj(t−δ))+SH(θj(t−δ),θi(t−δ))                 −SH(θj(t),θj(t−δ),θi(t−δ))−SH(θj(t−δ))    (7)

where SH(·) indicates the Shannon entropy, θ(t) indicates the instantaneous phase, and t is the time. δ refers to the time lag coefficient (δ = 10 ms). The formula can be expanded as:

                PTEij=∑p(θj(t),θj(t−δ),θi(t−δ))      log2(p(θj(t),θj(t−δ),θi(t−δ))p(θj(t−δ))p(θj(t),θj(t−δ))p(θj(t−δ),θi(t−δ))),    (8)

where p(·) denotes probability.

A phase-space binning method is used to calculate joint marginal distributions of signals, as mentioned in Eq. (8). A single histogram-based probability function is built by appropriately binning the occurrence of a single, pair of, or three instantaneous phases in each trial. Bin width is defined based on Scott's choice (Scott, 1992):

w(a)=3.5σ(a)N13,    (9)

where w(a) denotes the bin width for phase time-series (a = θj, θj(t − δ) or θi(t − δ)), σ(a) the standard deviation (SD) of a, and N the number of a. Hence, the number of bins is: k(a)=2πw(a) and PTE is further represented as:

PTEij=∑k1=1k(θj(t))∑k2=1k(θj(t−δ))∑k3=1k(θi(t−δ))N(k1,k2,k3)Nlog2                                                N(k1,k2,k3)N(k2)N(k1,k2)N(k2,k3)    (10)

Finally, the mean PTE from the mPFC to BLA is defined as:

PTE=∑i=1Nm∑j=1NbPTEijNm×Nb,    (11)

where Nm and Nb are channel numbers from the mPFC and the BLA, respectively.

Here, we considered an approach to evaluate the statistical significance of each PTE value. We shuffled the time series from each trial and used the bootstrapped data to compute PTE to obtain a curve of the proxy PTE. This procedure was repeated 500 times for each trial. Then, we generated a distribution of peak PTE of the shuffled controls from one trial. The 0.95 percentile of the mean peak PTE of shuffled controls becomes the significance level. The significance of connections is assessed by comparing the actual against the mean and SD of the mean shuffled controls, which allows for further confidence in the measure of estimated causality.

2.6. Histology

After all experiments, the rats were deeply anesthetized and perfused with phosphate-buffered saline (PBS) and 4% paraformaldehyde. Brain sections were obtained with a vibratome (Vibratome, USA). Recording sites were observed under an optical microscope (Olympus, Japan) and verified on a rat brain atlas.

2.7. Statistical analysis

All data were expressed as mean ± standard error of the mean (SEM). Methods and the sample size for statistics are described in the figure legend. Data were analyzed via Student's t-test (the Wilcoxon matched-pair signed-rank test and the Mann–Whitney test for a comparison of two groups) and one- or two-way repeated measures analysis of variance (ANOVA) followed by post hoc multiple comparisons with Dunnett's and Bonferroni's test, respectively. Circular statistics, such as the Watson-Williams test, were also gathered to assess the significance of the phase distribution. Pearson's correlation was applied to determine the correlation between WPLI/PTE and the motion of rats. The statistical significance threshold was set at a p-value of < 0.05.

3. Results 3.1. CUMS induces depressive-like behaviors during OFT

Chronic unpredictable mild stress is a commonly used rodent model of depression; hence, we performed a modified CUMS paradigm to induce a model of depression in rats. After 3 weeks of CUMS treatment, the rats showed typical depressive-like phenotypes, as evaluated by the SPT and the FST (Supplementary Figure 1).

To identify the effects of prolonged CUMS on exploratory behavior, rats underwent a 10-min OFT. Although rodents showed thigmotaxis, CUMS rats still showed little motor activity, more grooming, and a low level of rearing (Figure 2A). The results showed that the behavioral parameters of CUMS rats in the OFT, including center entries (10 ± 1.125, n = 6 rats for control vs. 3.167 ± 0.703, n = 6 rats for CUMS, p < 0.01, Figure 2B), time in the center area (70.658 ± 20.610, n = 6 rats for control vs. 17.639 ± 4.886, n = 6 rats for CUMS, p < 0.05, Figure 2C), rearing times (20.5 ± 2.466, n = 6 rats for control vs. 7.5 ± 1.258, n = 6 rats for CUMS, p < 0.01, Figure 2D), and total movement distance (4,780.653 ± 687.37, n = 6 rats for control vs. 2,040.826 ± 123.613, n = 6 rats for CUMS, p < 0.01, Figure 2E), were significantly lower than those of control rats, indicating that CUMS could induce depressive-like behaviors and decrease willingness to explore in the open field.

www.frontiersin.org

Figure 2. Depressive-like behaviors induced by exposure to CUMS. (A) Representation of open-field exploration trajectories in the control (top) and CUMS groups (bottom). (B) Center entries during the OFT. (C) Time in center during the OFT. (D) Rearing times during the OFT. (E) Total movement distance during the OFT. Error bars indicate standard error of the mean (SEM). Two-tailed Mann–Whitney test, *p < 0.05, **p < 0.01.

3.2. Phase synchronization between the mPFC and the BLA in theta-frequency augments during exploratory behavior

Previous research has shown an increase in theta power in the mPFC and the BLA during a transition from the periphery to the center (Likhtik et al., 2014). The cooperative long-scale information processing could be well reflected in the phase of neural oscillations. Hence, our analysis focused on the phase component of LFPs in the two sites. In trials of exploratory behavior, WPLI with better noise robustness and less volume conduction interference was calculated at all frequencies. According to the aforementioned study, exploratory behavior was defined as the movement of rats going into the central zone from the peripheral zone, which contained 2 s in the center and 2 s in the periphery. The time-frequency diagram showed a prominent theta frequency component at approximately −1 s (where rats were in the peripheral area of the open field; Figures 3A, D; Supplementary Figure 2). The curves of WPLI over frequency in the two groups showed significantly higher phase synchronization within the mPFC–BLA circuit in the theta-frequency transition from the periphery to the center (Figures 3B, E). A one-way ANOVA revealed a main effect for frequencies in the two groups [F(4,25) = 8.223, p < 0.001, n = 6 rats 130 trials for control; F(4,25) = 3.981, p < 0.05, n = 6 rats; 130 trials for CUMS], and Dunnett's multiple comparisons test demonstrated that theta frequency had a statistic

留言 (0)

沒有登入
gif